Vol. 80 No. 2 Autumn 2007
GEOGRAPHIA POLONICA
EXTREME METEOROLOGICAL AND HYDROLOGICAL EVENTS IN POLAND
EDITORS ZBIGNIEW W. KUNDZEWICZ JACEK A. JANIA
POLISH ACADEMY OF SCIENCES INSTITUTE OF GEOGRAPHY AND SPATIAL ORGANIZATION WARSZAWA
KKsisiąążżkka1.indba1.indb 1 22008-06-26008-06-26 110:54:410:54:41 PUBLISHED by the Institute of Geography and Spatial Organization, Polish Academy of Sciences, Warszawa, Poland.
EDITORIAL OFFICE: Twarda 51/55, 00-818 Warszawa, Poland Tel. (48 22) 6978-858, Fax (48 22) 620-62-21. CONTRIBUTIONS and CORRESPONDENCE related to editorial matters should be sent to the Editor: [email protected] COPYRIGHT by the Institute of Geography and Spatial Organization, Polish Academy of Sciences, Warszawa, Poland.
SUBSCRIPTION INFORMATION: Geographia Polonica (ISSN 0016-7282) is published twice a year in Spring, and Autumn by the Institute of Geography and Spatial Organization, Polish Academy of Sciences.
SUBSCRIPTION ORDERS for Geographia Polonica can be placed with: ARS POLONA, Obrońców 25; 03-933 Warszawa, Poland www.arspolona.com.pl (including single past and current issues)
ISSN 0016-7282
INFORMATION ON THE CONTENT (from 1964) and forthcoming issues are available at: www.igipz.pan.pl/wydaw/home.htm
GEOGRAPHIA POLONICA IS CURRENTLY ON LINE! from vol. 75 The two latest issues (full text) are available at www.igipz.pan.pl/wydaw/GPonline.htm
PREPARED FOR PRINT by Aleksandra Deręgowska Department of Cartography and Geographic Information Systems, Institute of Geography and Spatial Organization, Polish Academy of Sciences, Twarda 51/55, 00-818 Warszawa, Poland
PRINT Poligrafia Inspektoratu Towarzystwa Salezjańskiego Konfederacka 6, 30-306 Kraków, Poland
KKsisiąążżkka1.indba1.indb 2 22008-06-26008-06-26 110:55:400:55:40 CONTENTS
ARTICLES
JACEK A. JANIA and ZBIGNIEW W. KUNDZEWICZ—Preface ...... 5
ZBIGNIEW W. KUNDZEWICZ and JACEK A. JANIA—Extreme Hydro-meteorological Events and their Impacts. From the Global down to the Regional Scale ...... 9
TON W. DONKER—Access to and Re-use of Public-sector Environmental Data and Information. Policy Developments with a Focus on the European Hydro-meteoro- logical Scene ...... 25
PIOTR MATCZAK, ROMAN MAŃCZAK, ZBIGNIEW W. KUNDZEWICZ—Estimation of Damage Caused by Extreme Weather Events, with an Emphasis on Floods ...... 35
RAJMUND PRZYBYLAK, ZSUZSANNA VÍZI, ANDRZEJ ARAŹNY, MAREK KEJNA, RAFAŁ MASZEWSKI and JOANNA USCKA-KOWALKOWSKA—Poland’s Climate Extremes Index, 1951–2005 ...... 47
JOANNA USCKA-KOWALKOWSKA, RAJMUND PRZYBYLAK, ZSUZSANNA VÍZI, ANDRZEJ ARAŹNY, MAREK KEJNA, RAFAŁ MASZEWSKI—Variability to Global Solar Radiation in Central Europe During the Period 1951–2005 (On the Basis of Data from NCEP/NCAR Reanalysis Project) ...... 59
ANDRZEJ ARAŹNY, RAJMUND PRZYBYLAK, ZSUZSANNA VÍZI, MAREK KEJNA, RAFAŁ MASZEWSKI and JOANNA USCKA-KOWALKOWSKA—Mean and Extreme Wind Velocities in Central Europe 1951–2005 (On The Basis of Data from NCEP/NCAR Reanalysis Project) ...... 69
KKsisiąążżkka1.indba1.indb 3 22008-06-26008-06-26 110:55:400:55:40 ADAM ŁAJCZAK—River Training Vs. Flood Risk in the Upper Vistula Basin, Poland ...... 79
DAMIAN ABSALON, STANISŁAW CZAJA and ANDRZEJ T. JANKOWSKI—Factors Influencing Floods in the Urbanized and Industrialized Areas of the Upper Silesia Industrial Region in the 19th and 20th Centuries (the Kłodnica Catchment Case Study) ...... 97
KATARZYNA MAROSZ—Studies on Historical Floods in Gdańsk (a Methodological Background) ...... 111
EDMUND TOMASZEWSKI—Hydrological Droughts in Central Poland—Temporal and Spatial Patterns...... 117
ANDRZEJ CIEPIELOWSKI, EWA KAZNOWSKA— A Description of Hydrological Droughts in the Białowieża Primeval Forest in the Years 2003—2005 ...... 125
ADAM KOTARBA—Geomorphic Activity of Debris Flows in the Tatra Mts and in other European Mountains ...... 137
EWA SMOLSKA—Extreme Rainfalls and their Impact on Slopes—Evaluation Based on Soil Erosion Measurements (As Exemplified by the Suwałki Lakeland, Poland) ...... 151
ZBIGNIEW W. KUNDZEWICZ, ROMAN MAŃCZAK, IWONA PIŃSKWAR and MACIEJ RADZIEJEWSKI—Models of Impacts of Hydrometeorological Extremes...... 165
HALINA KOWALEWSKA-KALKOWSKA, MAREK KOWALEWSKI and BERNARD WIŚNIEWSKI— Application of Hydrodynamic Model of the Baltic Sea to Storm Surge Representation along the Polish Baltic Coast ...... 181
MAŁGORZATA SZWED, DARIUSZ GRACZYK, IWONA PIŃSKWAR and ZBIGNIEW W. KUNDZEWICZ—Projections of Climate Extremes in Poland...... 191
KKsisiąążżkka1.indba1.indb 4 22008-06-26008-06-26 110:55:410:55:41 PREFACE
JACEK A. JANIA and ZBIGNIEW W. KUNDZEWICZ
The present issue of Geographia Polonica cause for hope that the problem of the com- reviews a sample of results obtained through ercialised delivery of observational data for implementation of the Integrated Project scientific and educational purposes can be entitled ”Extreme meteorological and hy- resolved. Despite such major problems with drological events in Poland (The evaluation access to complete (gap-free) series of data, of events and forecasting of their effects especially in digital form, research work for the human environment)”. The project, within the project framework is being con- launched by the Ministry of Science and tinued with. Higher Education of the Republic of Poland A milestone for the Integrated Project in 2004, has as its aim an analysis and spa- was the convening of a conference on “Ex- tio-temporal assessment of main extreme treme hydrometeorological events in Poland meteorological and hydrological events in and their impacts—A European context” in Poland, using all the available data within Warsaw on 7–9 December 2006. This provid- an interdisciplinary framework that re- ed an opportunity for selected interim results lates to climatology, hydrology, oceanol- from the project to be presented, and made ogy, geomorphology, human geography and subject to extensive discussion. The partici- the economy. In the context of global-scale pation at the conference of speakers from studies, it is very important that the relation- other European countries made the forming ships between extreme events in Poland and of a broader perspective possible, allowing trends to ongoing climate changes be deter- the Polish findings to be seen in the context mined. The studies presented are devoted to of results from elsewhere, e.g. via projects a range of weather-related extremes, such as funded by the European Commission, and intense precipitation, floods, and geomor- global-scale considerations. Information on phic hazards like landslides and debris flow, the subject matter of the conference may be storm surges, droughts and extreme winds, found in Jania and Kundzewicz (2006). The as well as the impacts of all of these on hu- event was organized by the Faculty of Earth man existence (health and death hazards, Sciences, University of Silesia and the Insti- economic damage). The major obstacle in tute of Geography and Spatial Organisation, conducting these scientific studies has been Polish Academy of Sciences. considerable difficulty with the accessing The present volume is a collection of indi- of basic observational data from the mete- vidual contributions rather than a complete orological, hydrological and other stations review of the project. The issue starts with run by governmental institutions. Recently, a stage-setting contribution by Kundzewicz a better understanding of the importance and Jania (2007), reviewing extreme hydro- of the problem of natural extremes by some meteorological events and their impacts in governmental bodies in Poland has given a nutshell, on a range of scales (from the
KKsisiąążżkka1.indba1.indb 5 22008-06-26008-06-26 110:55:410:55:41 6 Jacek Jania, Zbigniew W. Kundzewicz
global down to the regional). In the next pa- impacts of the anthropogenic modification per, Donker (2007) reviews the vital problem of drainage basins on flood patterns. They of access to, and the re-use of, environmental consider the urbanised and industrialised data in the public sector in European Union areas of the Upper Silesia Industrial Region countries. This has been a problem of para- in the 19th and 20th centuries, with particu- mount importance in Poland. Indeed, a lack lar reference to the Kłodnica River Basin. of affordable access to hydrometeorological Marosz (2007) sketches the methodological data has jeopardized the attainability of the background to studies of historical floods in original objectives of the present Integrated Gdansk (including events dating back to the Project. Next, Matczak et al. (2007) deal pre-observational period), as well as the us- with the economic dimension by reviewing age of GIS in analysing and reconstructing valuation of losses caused by extreme weath- historical floods. er events, with particular emphasis on flood The further two papers examine droughts damages. Approaches based on restoration in parts of Poland. Judging by annual pre- value and market value, and methods ad- cipitation and the variability thereto, mete- dressing indirect tangible losses and intan- orological and hydrological droughts and gible damage are reviewed. low flows (streamflow droughts) are to be The subsequent ten papers are related seen as frequent phenomena during which to the analysis of observational records. water availability may achieve extremely low The final three papers then deal with vari- values. Tomaszewski (2007) analyses tempo- ous aspects of mathematical modelling and ral and spatial patterns to the hydrological forecasting. droughts and low flows occurring in cen- A very important part of the study of ex- tral Poland (in the basins of the Warta, the treme hydrometeorological events relates to Pilica, and the Bzura). Ciepielowski and Ka- the detection of changes in long time-se- znowska (2007) deal with recent hydrologi- ries of observational data. Three papers, by cal droughts (2003–2005) in the Białowieża Przybylak et al. (2007), Uscka-Kowalkowska Primaeval Forest, the largest and most et al. (2007) and Araźny et al. (2007) thus unique National Park in Poland. deal with extremes in such long time-se- The next two contributions deal with ries of meteorological variables extending geomorphological hazards capable of being between 1951 and 2005. Przybylak et al. triggered by extreme rainfall events. Kotarba (2007) seek changes in the index of climate (2007) discusses geomorphological instabil- extremes. Since Polish meteorological data ity processes in the Tatra Mountains and were not available at affordable cost, the compares them with the activity of debris following two papers deal with the results flows in other high European mountains. of an NCEP/NCAR reanalysis carried out Smolska (2007) reviews the geomorphologi- in the USA. The studies cover the region cal impact of extreme rainfall events in the of Central Europe. Uscka-Kowalkowska et Suwałki Lakeland (NE Poland), on the basis al. (2007) examine the variability to global of soil-erosion measurements. solar radiation, while Araźny et al. (2007) Two papers are devoted to aspects study mean and extreme wind speeds. of mathematical modelling in the analysis Floods have continued to be a major of extremes and their impacts. Kundzewicz weather-related hazard in Poland, with sev- et al. (2007) review alternative approaches eral destructive events having occurred even to the modelling of the impacts of hydrom- since the truly catastrophic deluge of July eteorological extremes, discussing such as- 1997. Łajczak (2007) examines the impact pects as: model taxonomy, the trade-off be- of human activity, and river training in par- tween accuracy and complexity, uncertainty, ticular, on the flood risk in the Upper Vistula and barriers relating to data (i.e. existence, River Basin, over a longer time-scale of a few accuracy, credibility and availability). Kow- centuries. Absalon et al. (2007) study the alewska-Kalkowska et al. (2007) present the
KKsisiąążżkka1.indba1.indb 6 22008-06-26008-06-26 110:55:410:55:41 Preface 7
application of the M3D_UG numerical drodynamic Model of the Baltic Sea to Storm model developed at the Institute of Ocea- Surge Representation along the Polish Baltic nography, University of Gdańsk, to a post- Coast. Geographia Polonica, 80, 2: 181–190. hoc analysis of storm surges along the Polish Kundzewicz, Z.W. and Jania, J. (2007), Extreme (southern) Baltic coast. Hydro-meteorological Events and their Im- The final paper, by Szwed et al. (2007) pacts. From the Global down to the Regional examines model-based projections of cli- Scale. Geographia Polonica, 80, 2: 9–23. mate (weather) extremes in Poland, in re- Kundzewicz Z.W., Mańczak, R., Pińskwar, lation to both intense precipitation and hot I. and Radziejewski, M. (2007), Models of and dry spells. A comparison of model-based Impacts of Hydrometeorological Extremes. information for the control period (1961– Geographia Polonica, 80, 2: 165–179. 1990) and for the future projection horizon Łajczak, A. (2007), River Training Vs. Flood (2070–2099) is made. Risk in the Upper Vistula Basin, Poland. Ge- ographia Polonica, 80, 2: 79–96. Marosz, K. (2007), Studies on Historical Floods REFERENCES in Gdańsk (a Methodological Background). Geographia Polonica, 80, 2: 111–116. Absalon, D., Czaja, S. and Jankowski A.T. (2007), Matczak P., Mańczak R., Kundzewicz Z. W., Factors Influencing Floods in the Urbanized (2007), Estimation of Damage Caused by Ex- and Industrialized Areas of the Upper Silesia treme Weather Events, with an Emphasis on Industrial Region in the 19th and 20th Cen- Floods. Geographia Polonica, 80, 2: 35–45. turies (the Kłodnica Catchment Case Study). Przybylak, R., Vizi, Z., Araźny, A., Kejna, M., Geographia Polonica, 80, 2: 97–109. Maszewski, R.,Uscka-Kowalkowska, J. (2007) Araźny, A., Przybylak, R., Vizi, Z., Kejna, M., Poland’s Climate Extremes Index, 1951–2005. Maszewski, R. and Uscka-Kowalkowska, Geographia Polonica, 80, 2: 47–58. J. (2007), Mean and Extreme Wind Veloci- Smolska, E. (2007), Extreme Rainfalls and their ties in Central Europe 1951–2005 (On The Impact on Slopes—Evaluation Based on Soil Basis of Data from NCEP/NCAR Reanalysis Erosion Measurements (As Exemplified by Project) Geographia Polonica, 80, 2: 69–78. the Suwałki Lakeland, Poland). Geographia Ciepielowski, A. and Kaznowska, E. (2007), Polonica, 80, 2: 151–163. A Description of Hydrological Droughts in the Szwed M., Graczyk, D., Pińskwar, I., Kundze- Białowieża Primeval Forest in the Years 2003– wicz, Z.W. (2007), Projections of Climate Ex- 2005. Geographia Polonica, 80, 2: 125–136. tremes in Poland. Geographia Polonica, 80, 2: Donker, T.W. (2007), Access to and Re-use of 191–202. Public-sector Environmental Data and Infor- Tomaszewski, E. (2007), Hydrological Droughts mation. Policy Developments with a Focus on in Central Poland—Temporal and Spatial Pat- the European Hydro-meteorological Scene. terns. Geographia Polonica, 80, 2: 117–124. Geographia Polonica, 80, 2: 25–34. Uscka-Kowalkowska, J., Przybylak, R.,Vízi, Jania, J. and Kundzewicz, Z.W. [Eds.] (2006), Z., Araźny, A., Kejna, M. and Maszewski, Extreme hydrometorological events in Po- R. (2007), Variability to Global Solar Ra- land and their impacts—European context. diation in Central Europe During the Period International Conference, Warsaw, Poland, 1951–2005 (On the Basis of Data from NCEP/ 7–9 December 2006. Book of abstracts. Sos- NCAR Reanalysis Project). Geographia Po- nowiec-Warszawa, 146 pp. lonica, 80, 2: 59–68. Kotarba, A. (2007), Geomorphic Activity of De- bris Flows in the Tatra Mts and in other Eu- Paper first received: March 2007 ropean Mountains. Geographia Polonica, 80, In final form: March 2008 2: 137–150. Kowalewska-Kalkowska H., Kowalewski M. and Wiśniewski, B. (2007), Application of Hy-
KKsisiąążżkka1.indba1.indb 7 22008-06-26008-06-26 110:55:410:55:41 EXTREME HYDRO-METEOROLOGICAL EVENTS AND THEIR IMPACTS. FROM THE GLOBAL DOWN TO THE REGIONAL SCALE
ZBIGNIEW W. KUNDZEWICZ*, JACEK A. JANIA** *Research Centre for the Agricultural and Forest Environment, Polish Academy of Sciences Bukowska 19, 60-809 Poznań, Poland [email protected] **Faculty of Earth Sciences, University of Silesia ul. Będzińska 60, 41-200 Sosnowiec, Poland [email protected]
Abstract: Despite the progress in technology, the risk of weather-related disasters has not been eradicated and never will be. On the global scale, disasters are becoming both more frequent and more destructive, annually causing material losses worth tens of billions of Euros, as well as several thousand fatalities. Furthermore, catastrophic weather events have been the subject of a rapid upward trend, with the value of material damage increasing by an order of magnitude over the last four decades, in inflation-adjusted monetary units. There is now an increasing body of evidence of ongoing planetary climate change (global warming), which has brought about consi- derable changes where extreme hydro-meteorological events are concerned, and is likely to lead to even more marked changes in the future. Typically, changes in extremes are more pronounced and exert more impact than changes in mean values. Among the extremes on the rise are the number of hot days and tropical nights; the duration and intensity of heatwaves; precipitation intensity (and resulting floods, landslides and mudflows); the frequency, length and severity of droughts; glacier and snow melt; tropical cyclone intensity and sea level and storm surges. In turn, a ubiquitous decrease in cold extremes (number of cool days and nights, and frost days) is pro- jected. Increases in climate extremes associated with climate change are likely to cause physical damage and population displacement, as well as having adverse effects on food production and the availability and quality of fresh water. A discussion of hydro-meteorological extremes and their impacts is therefore provided here in relation to a range of scales, and with the context for adaptation and mitigation also being alluded to.
Key words: extreme events; hydrometeorology; climate variability; climate change; climate chan- ge impacts
1. INTRODUCTION soil moisture; river flow; wind velocity; etc. attain extreme values. Such a situation may It is normal that, at times, hydro-meteorologi- jeopardize people and their settlements. cal variables such as temperature of air, water Despite the fascinating progress in tech- or the ground; precipitation intensity or total; nology, humankind continues to live with
KKsisiąążżkka1.indba1.indb 9 22008-06-26008-06-26 110:55:420:55:42 10 Zbigniew W. Kundzewicz, Jacek Jania
the hazards of extreme hydrometeorological enhancement of the greenhouse effect, and events, which may cause severe human and causing land-use changes (e.g. deforestation material damage. The risk of weather-re- in tropical areas reducing carbon sequestra- lated disasters has not been eradicated and tion). The global climate system has been never will be. In fact, on the global scale, driven out of its stable natural variability disasters are becoming more frequent and mode. As stated in the IPCC Fourth As- more destructive, causing material losses sessment Report (IPCC, 2007), most of the of tens of billions of Euros, as well as several observed increase in globally averaged tem- thousand fatalities annually. Catastrophic peratures since the mid-20th century is very weather events have been exhibiting a rapid likely (with a probability of over 90%) due to upward trend, increasing by an order of mag- the observed increase in anthropogenic gre- nitude over the last four decades, when ex- enhouse gas concentrations. pressed in terms of inflation-adjusted mon- The 1990s are likely to have been the etary units. warmest decade and the 20th century in- There are several categories of factors crease in temperature is likely to have been that may explain changes in hydro-mete- the largest occurring in any century over the orological hazards, and their impacts. The second Millennium in the Northern Hemi- principal categories of this kind are: (1) sphere (Watson and Core Writing Team, changes in the climate and atmospheric 2001). system; (2) changes in interactions between Twelve of the thirteen warmest years the atmosphere, the cryosphere and the globally in the 158-year global instrumental oceans; (3) changes in terrestrial systems temperature observation period occurred in (e.g. land-cover change: urbanization and the recent twelve years (see Table 1). deforestation); (4) changes in socio-eco- Only one of the last 13 years (1996) did nomic systems (e.g. land-use change, in- not make it on to the list of the top twelve hot- creases exposure and damages potential, test years. It was the 19th warmest year, still changing risk perception). The relative warmer by 0.137°C than the 1961–1990 mean importance of the above factors is site- and global temperature (cf. Table 1). The year event-dependent. 2007 also belongs to the short list of globally The aim of this paper is to provide a short warmest years (rank 8). Future warming will introduction to different aspects of the inter- depend on scenarios of socio-economic de- relationships between global climate changes velopment and on the mitigation policy (the and extreme meteorological and hydrologi- curbing of greenhouse gas emissions). Pro- cal events and their impacts on the regional jected temperature changes differ regionally, and local scales in Europe. Adaptation and being scenario- and model-specific, with the mitigation attempts are also considered, with range of global mean temperature increase a special emphasis being put on the so-called for the 2090s horizon being likely from 1.0 “short memory syndrome” where severe ex- to 6.3°C above the control period 1980–1999 tremes are concerned. (IPCC, 2007). Beside the temperature change, there have been ongoing changes in other climate- 2. CLIMATE CHANGE AND ITS IMPACTS related variables, such as sea level, precipita- tion (growth in some areas, decrease in oth- There is an increasing body of evidence on er areas of the globe), river discharge, soil the ongoing planetary climate change (global moisture, glacier and snow cover extents. warming) being attributable to human acti- Even more marked changes are projected for vities and caused by rising emissions of gre- the future. Projected precipitation changes enhouse gases (carbon dioxide, methane, ni- differ regionally, but are loaded with a high trous oxide, etc) leading to a buildup of the level of uncertainty, being model- and sce- said gases in the atmosphere and consequent nario-specific.
KKsisiąążżkka1.indba1.indb 1100 22008-06-26008-06-26 110:55:420:55:42 Extreme Hydro-Meteorological Events and their Impacts… 11
Table 1. Ranking of years with highest global mean annual temperature since 1860.
Ranking of particular years with the Deviation from the long-term mean highest global mean temperature temperature since 1860 Year (in the reference period 1961–1990)
1 1998 0.546 2 2005 0.482 3 2003 0.473 4 2002 0.464 5 2004 0.447 6 2006 0.422 7 2001 0.409 8 2007 0.403 9 1997 0.351 10 1999 0.296 11 1995 0.275 12 2000 0.270 13 1990 0.254 14 1991 0.212 15 1988 0.180 16 1987 0.179 17 1983 0.177 18 1994 0.171 19 1996 0.137
Data source: Jones (2007).
It is important to note that the Arctic gas), these having been previously immobili- system is very sensitive to climate change. zed in the frozen ground. Even when account is taken of the fact that In addition, a dynamic response of tide- the system is not located close to the terri- water glaciers to climatic (and sea) warm- tory of Poland, climate-cryosphere interac- ing is caused by increased melting of their tions play an important role in driving cli- surface. Faster flow of glaciers induced by matic changes on the global and European greater meltwater supply to their beds re- scales. Moreover, traditionally, Polish scien- sults in massive calving and faster transfer tists have been active in studying the Arctic of ice resting previously on land into the sea. system, and their achievements are visible in Such processes are responsible for more dis- the international context. tinct sea-level rise. Substantial loss of mass There are several important (positive) due to melting and ice transfer into the sea feedback mechanisms in the Polar/Arctic from the Svalbard glaciers (Jania, 2002) and track in the climate system (ACIA, 2005): other Arctic areas has been noted during • positive albedo feedback; a decrease the last two decades. While global sea rise in snow-cover area, and sea ice extent and 2 mm/yr was reported over the 40 years at the a retreat of glaciers affect albedo on the end of the last century (Cabanes et al, 2001), global scale, hence reducing the reflected the contribution due to glaciers was of the or- part of solar radiation and driving war- der of 0.2–0.4 mm/yr through the whole 20th ming; century (ACIA, 2005). In this respect a spe- • positive methane feedback; a thawing cial role has been played by the Greenland of permafrost leads to the release of large Ice Sheet—the largest ice mass in the North- volumes of methane (a powerful greenhouse ern Hemisphere. Recently published studies
KKsisiąążżkka1.indba1.indb 1111 22008-06-26008-06-26 110:55:420:55:42 12 Zbigniew W. Kundzewicz, Jacek Jania
(Dowdeswell, 2006; Rignot and Kanagar- precipitation variability. Where changes in atnam, 2006; Stearns and Hamilton, 2006) temperature produce changes in the timing suggest that the ice sheet responds more of streamflow, climate change effects are ge- rapidly to climate warming than previously nerally more marked than in areas in which thought, particularly by way of accelerated hydrological regimes are more sensitive to flow of several large outlet glaciers draining changes in precipitation. There are generally the southern part of the ice sheet. Extreme consistent patterns to changes in runoff and acceleration of Greenland ice-mass loss has water availability, with an increase at higher been noted by the satellite gravity survey for latitudes and in some wet tropical areas, and the period April 2002–April 2006. A trans- a decrease at mid-latitudes and in the dry fer of 248±36 km3/yr of ice to the sea has tropics. Climate change may cause increa- been detected, and constitutes an equivalent sed summer drying in continental interiors. contribution of 0.5±0.1 mm/yr to global sea- Semi-arid and arid areas are particularly ex- level rise (Velicogna and Wahr, 2006). The posed to the impacts of climate change on level of the contribution is ten times greater freshwater. The area of the Earth’s surface than that estimated earlier for the whole with a “very dry” status has been increasing Arctic. More intense melting and acceler- and is projected to increase further. The glo- ated glacier flow doubled the mass balance bal water cycle has accelerated, with conse- deficit of the Greenland Ice Sheet during quences for extremes. In many areas there the last decade, contributing to global sea- has been an increase in intense precipitation level rise and likely to be subject to intensi- which can be translated into an increase in fication. In consequence, low-lying seashore the flood hazard (Kundzewicz et al., 2007). zones will be affected more frequently and Global warming has brought about con- widely by storm surges. siderable changes in extreme hydro-me- In the longer term, a slowdown of ther- teorological events and is likely to lead to mohaline circulation is likely to partly com- even more major changes in the future. As pensate for further warming (ACIA, 2005). a result, several extremes will become yet These processes operate in the North At- more extreme. For instance, many presently lantic area and influence the climate of Eu- dry areas are likely to become drier, while rope, indirectly stimulating extreme events those that are now wet may become wetter. in remote regions. Such linkages (tele-con- As a rule, changes in extremes are more pro- nections) are a good example of interactions nounced and exert more impact than chang- between global warming and regional con- es in mean levels. sequences and, in turn, show how the influ- It can be expected that, in the warmer ence of processes observed in the Arctic on climate of the future, there will be a ubiq- a regional scale risks increases the range uitous decrease in cold extremes (number of extreme events worldwide. of cool days / nights / frost days). It is project- ed that many areas will witness increases in extremes as regards the number of hot days 3. EXTREMES AND THEIR IMPACTS— and tropical nights; the duration of heat- —THE GLOBAL SCALE waves; precipitation intensity; the frequency, length and severity of droughts; glacier and Ongoing climatic and non-climatic changes snow melt and tropical cyclone intensity have already influenced environmental con- (IPCC, 2007). Non-tropical cyclones and ditions on a global scale (and more specifi- the most intense storms may also increase, cally water resources) in a discernible way. while storm tracks may shift more poleward. Even more marked changes are projected for Impacts on air quality are likely. Stagnation the future. The most certain impacts of cli- of air masses in the summer may exacerbate mate change on freshwater systems are due air pollution problems (ozone and particu- to increases in temperature, sea level and late matter—soot, with cardiac and respira-
KKsisiąążżkka1.indba1.indb 1122 22008-06-26008-06-26 110:55:420:55:42 Extreme Hydro-Meteorological Events and their Impacts… 13
tory hazards), which will accumulate until nual precipitation is likely to decrease over a cleansing cold front comes. much of Europe (in particular, over South- Any regional increases in climate ex- ern and Central Europe). In much of South- tremes (storms, floods, cyclones, droughts, ern Europe, a joint effect of temperature etc.) associated with climate change are rise and a decline in precipitation is foreseen likely to cause physical damage, popula- for the summer. Global warming contributes tion displacement, and adverse effects on to heat stress and more drying (evaporative the economy of food production, freshwa- demand), exacerbating water stress. The risk ter availability and quality. Growing ad- of drought increases substantially in sum- verse health effects would include the risks mer, along with the risk of wildfires. of infectious disease epidemics in developing Mean annual precipitation is likely to countries. It is estimated that diarrhoeal dis- increase in northern Europe. However, the eases attributable to unsafe water and a lack intensity of rainfall events is projected to of basic sanitation already cause numerous increase even in regions in which the mean (nearly 2 million) deaths a year worldwide. annual precipitation is likely to decrease (cf. The projected increase in the frequency and Kundzewicz et al., 2006). severity of droughts would exacerbate the Extremely heavy and/or long-lasting situation and exert an adverse impact on hu- rainfall events cause geomorphic hazards man health. such as landslides and mud-debris flows in The consequences of globalization should the mountains. The geological structure be appreciated. As a result of the global vil- and lithological composition of the Polish lage effect, the number of Swedish citizens Carpathians are favorable for development (mostly Christmas tourists) killed by the tsu- of landslides. More than 95% of all regis- nami disaster of December 2004 in the Indi- tered landslides in Poland are located in the an Ocean coasts greatly exceeded the number Flysch Carpathians and, statistically, they of fatalities caused by all the natural disasters are as dense as one form per 1 km2. While within Sweden over many decades. Polish the majority of landslides in the Carpathians citizens were likewise among the fatalities occurred during the Pleistocene, the Late caused by the forest fires during the droughts Glacial and the early and middle Holocene in France in 2003 and in Greece in 2007. (Alexandrowicz, 1977; Margielewski, 2001), a reactivation of many of them and a crea- tion of new forms during the last decade have 4. EUROPEAN, REGIONAL been observed. Deforestation of mountain AND SUB-REGIONAL SCALES slopes and their cultivation (in cereal- and potato-growing) created favourable condi- Europe has warmed up considerably, especial- tions for water infiltration into slope-cover, ly in the last few decades, and further, stron- mantle and bedrock. Due to an increase in ger, warming is projected for the future by the number of extreme rainfall events, a re- climate models. The European Commission’s juvenation of older forms and occurrence perspective is focused on the juxtaposition of new landslides has been observed since of two scenarios: a controlled (around 2°C) 1996 (Rączkowski and Mrozek, 2002, Star- warming by 2100 if global mitigation policy kel, 2006). Large proportions of the land- becomes effective, and a very much stronger slide events are associated with regional and (possibly 5°C) warming, if a business-as-usu- local flood events in the area, like those in al approach prevails and atmospheric concen- 1997, 2000, 2001, 2002 and 2005. trations of greenhouse gases keep growing One of the most catastrophic landslides without effective mitigation. affected an area of 15 ha in Lachowice vil- Long-term precipitation trends have also lage in the Beskid Makowski range. Rapid been observed, and are projected for the fu- displacement of slope-cover masses de- ture, in many regions of Europe. Mean an- stroyed 14 buildings and a road during just
KKsisiąążżkka1.indba1.indb 1133 22008-06-26008-06-26 110:55:430:55:43 14 Zbigniew W. Kundzewicz, Jacek Jania
15 minutes on 27 July 2001. The eastern part ter-stressed regions, the possibility of offset- of this landslide was reactivated again a year ting declining surface water availability due later and additional 4 buildings were affect- to increasing precipitation variability may ed then. Damage caused to buildings and not prove a practical one. Sea-level rise will roads by landslides in Małopolskie Voivod- extend areas of salinization of groundwater ship had a value in excess of 173 million PLN and estuaries, resulting in a decrease in the (43 million Euro) in the years 2000–2001. availability of fresh water for both people Extremely heavy rainfall in summer 2002 and ecosystems (Kundzewicz et al., 2007). In again activated many landslide forms, many places, winter precipitation is increas- though it concentrated its downpour in ingly likely to fall as rain, rather than snow. the town of Muszyna and its surroundings, This may jeopardize winter sports, especial- causing large mud and debris flows. These ly in lower skiing domains. damaged up to 100 buildings there (Bejgier- In much of Europe, occurrences of very Kowalska, 2005). wet winters and of intense rainfall events will It is worth stressing here that a large become more frequent, with likely conse- part of the material damage caused by slope quences as regards flood risk. Increasing tem- mass movements in the Flysch Carpathians perature and variability of runoff are likely is co-induced by the location of new, larger to lead to adverse changes in water quality and heavier brick and concrete buildings on (turbidity increases, algal blooms, mobilizing old landslide slopes. Excavations on slopes and washing away of pollutants, favouring for constructional purposes and their under- of pathogens, and thermal pollution). cutting near the valley floor as the construc- Very severe material flood damage tion or modernization of roads takes place (above 20 billion Euros in value) was noted destabilize the slopes. Particular landslides on the European continent in 2002, con- and mudflows are damaging phenomena siderably exceeding records for any single of relatively limited extent in comparison year before. The floods in Central Europe with floods. Nevertheless, their density in in August 2002 alone (on the rivers Dan- the Carpathians shows the importance of se- ube, Labe/Elbe and their tributaries) caused vere local events whose frequency grows with damage exceeding 15 billion Euros in value. global climate change and the more frequent Only a year later, the summer (June to mid- occurrence of extreme rainfall. August period) of 2003 brought a disastrous Warming leads to changes in the sea- heatwave and drought across large parts sonality of river discharge in catchments of Europe, with temperatures exceeding in which much winter precipitation falls as the averages even by 3–5°C and annual pre- snow (e.g. in the Alps). Winter flows in- cipitation deficits of up to 300 mm, with the crease, while snowmelt occurs faster and result being an estimated reduction of 30% earlier (with peak flows coming earlier). over Europe in gross primary production There is less snow pack in spring and less of terrestrial ecosystems (Ciais et al., 2005). soil moisture in summer, and summer and The hot and dry conditions led to many very autumn flows decrease. The ongoing re- large wildfires. Many major rivers (in partic- duction of European glaciers will lead to ular in southern Europe) were at record low gradual longer-term decreases in the con- levels, resulting in a disruption of irrigation tribution glaciers make to river discharge and a cooling of power plants. (with the possibility of a river flow increase The 2003 European heatwave killed tens in the short term, including flooding due to of thousands of people (Koppe et al., 2004), rapid melt). Similar phenomena have been showing that even developed countries may observed in the majority of glacierized high not be adequately prepared to cope with ex- mountains over the Northern Hemisphere. treme heat. Such extreme events are usually As decreasing groundwater recharge is pro- amplified in large urban areas due to the jected over many areas, also in already wa- heat-island effect (caused by heat absorp-
KKsisiąążżkka1.indba1.indb 1144 22008-06-26008-06-26 110:55:430:55:43 Extreme Hydro-Meteorological Events and their Impacts… 15
tion in asphalt, concrete and building roof would be unlikely in the absence of anthro- surfaces). Heatwaves in Poland are not yet pogenic climate change (IPCC, 2007). How- perceived as a major disaster in the public ever, an individual extreme event, such as health perspective, yet one can expect that an extreme flood or a heatwave, can never they will increasingly become a hazard in be directly attributed to climate change. conditions of a warming climate combined What it is fair to state is that the probability with an ageing society. Systematic studies of such an extreme event of a given inten- on the influence of heat waves on the health sity (magnitude) is likely to increase in the of the urban dwellers in Poland have been future. Hence, the excess deaths caused by initiated (e.g. Kuchcik, 2003). Summer 2006 a heatwave can be linked indirectly with cli- in Poland was warmer than that of 2003 mate change. An increase in the frequency (when temperature records were broken in or intensity of heatwaves in the future warm- much of Europe) and in much of Poland, ing climate will increase the risk of mortal- July 2006 was the warmest month in the his- ity and morbidity, particularly in older age tory of observations. groups (sick people, lonely people), and Severe heatwaves in the Mediterranean among the urban poor. region and SE Europe, often with tempera- Gales, the disasters causing the larg- tures exceeding 40°C, occurred during the est insured material damage, play havoc summer of 2007. Usually, at least one seri- with northern and western Europe, at times ous heatwave occurs in Greece each sum- combining with coastal flooding. The storm mer (typically in August). However, in 2007, of 8 January 2005 blew down 75 million m3 three such heatwaves struck the Balkan of trees in southern Sweden, breaking the all- area, causing at least 700 additional deaths. time record. In Sweden, nearly 350,000 ho- Hungarian medical officials reported up to mes lost power and the problems persisted 500 heat-related fatalities across Hungary in over a longer time, as about 10,000 homes the second half of July 2007. were still without power after three weeks Drought and heat frequently go together (Wikipedia, 2006). The death toll in Scan- with wildfires, and these occurred in many dinavia was at least 17. Only a few days lat- places in south-eastern Europe in the sum- er, there was another gale across the north mer of 2007 (Fig. 1). In Greece alone, over of the British Isles, with windspeeds of up 3,000 wildfires were registered, causing to nearly 200 km/h, attendant fatalities and damage to forests, pastures and farmland, major socio-economic disruption (disrupted and producing a loss of up to 80 lives. At power supply, paralysed transport). least 10,000 farms were destroyed or seri- Gales in Poland are less frequent. How- ously damaged. Thousands of villagers were ever, gradual sea-level rise with superim- left homeless. Fire and smoke endangered posed storm surges is projected to cause settlements and summer holiday resorts in more frequent inundations in the area of the Greece, Bulgaria, Albania and even France. Baltic mouths of Polish rivers. Following the Greek citizens and foreign visitors ques- beginning of the verified observation series tioned the state’s ability to cope with ex- in 1950/51, the probability of storm-surge tremes. Summer months—usually a time for flooding about doubled towards the end leisure for tourists—became a period of hor- of the 20th century (Sztobryn et al., 2005). ror and traumatic experience. Sea-level rise and storminess are very im- Severe summer droughts have occurred portant for the erosion of coasts and poten- a number of times in Poland in the last dec- tial damage to the near-shore infrastructure. ades (e.g. in 1992 and 2006), often accom- Displacements of the Polish shoreline meas- panied by violent wildfires claiming human ured in the period 1875–1979 show erosion lives. along almost the complete length, except in The occurrence of a heatwave as extreme the Gulf of Gdańsk segment. More intense as that of summer 2003 over much of Europe erosion is predicted for the entire Polish
KKsisiąążżkka1.indba1.indb 1155 22008-06-26008-06-26 110:55:440:55:44 16 Zbigniew W. Kundzewicz, Jacek Jania
Figure 1. Wildfires (red dots) in the Southern Balkan region recorded on the MODIS satellite image taken on 25 July 2007 (© NASA, Visible Earth).
KKsisiąążżkka1.indba1.indb 1166 22008-06-26008-06-26 110:55:450:55:45 Extreme Hydro-Meteorological Events and their Impacts… 17
Table 2. Erosion risk on the Baltic coast of Poland and its prediction to 2050
Erosion rate in the period Distance 1875–1979 Prediction to 2050 along the (sea level rise by 2 (expected sea level rise coast1 mm/yr) by 6 mm/yr)
Rate Rate of shoreli- Erosion of shoreli- Erosion Regions ne position vulnera- ne position vulnera- and sub-regions change2 bility change bility of the coast (km) m/yr classes m/yr classes
Gulf of Gdańsk Vistula Lagoon bar 0.0–29.0 +0.15 low -0.12 low Wisła Przekop–Władysławowo 48.5–124.0 +0.11 low -0.08 low Hel Peninsula Gulfward shore H36.0–71.5 -0.21 low -0.30 medium Open sea shore H0.0–36.0 -0.46 medium -0.64 medium Open sea Karwieńska bar 134–144 -0.42 medium -0.59 medium Piaśnica–Sarbsko Lake bar 149–181.5 -0.66 medium -0.93 medium Sarbsko Lake bar–Gardno Lake bar 181.5–216 -0.91 medium -1.27 high Rowy–Ustka 217–233 -1.33 high -1.86 high Ustka W–Wicko Lake bar–Kopań Lake bar–Darłowo 233–270 -0.37 medium -0.52 medium Bukowo Lake bar–Jamno Lake bar 278–300 -0.42 medium -0.59 medium Ustronie Morskie–Dźwirzyno 319–345 -0.45 medium -0.63 medium Mrzeżyno–Dziwnów 352–386 -0.38 medium -0.53 medium Wolin cliff–Pomorska Bay 401–424 -0.47 medium -0.66 medium
1 beginning at the Polish-Russian border (except for the Hel Peninsula) 2 max. values for shoreline length segments ≥ 2 km (after Dubrawski and Zawadzka-Kahlau, 2006)
coast in future (Table 2), as a consequence The 1997 flood was indeed an extreme of more rapid further sea-level rise. event. Four issues of the principal and most Shore erosion, as a consequence of more influential weekly magazine POLITYKA frequent winter storm surges along the had flood-related cover stories (Fig. 2). Such Polish coast, results in the shrinking of the media attention has been without precedent beautiful sandy beaches available for tour- in Polish history. This illustrates the sever- ists. Therefore, this process could slowly af- ity of impact of the 1997 flood (55 fatalities, fect the attractiveness of the Baltic beaches 162,500 affected people and over $ 4 billion in Poland, counteracting the positive effect in material damage). The floods in 1998, of increasing temperature. 2001, 2002 and 2005 also caused fatalities Despite the warming climate, the main and affected thousands of people. killer among hydro-meteorological events in Poland remains the cold snap in winter, dur- ing which many people (some of them home- 5. WHAT CAN BE DONE? ADAPTATION less and drunk) freeze to death. According AND MITIGATION to data assembled by CRED (2006), the 2005/6 winter killed 233, while the 2001/2 A working definition of adaptation based winter killed 270 people in Poland. on the one accepted in the IPCC process Fewer fatalities but far greater material is: adjustment in natural or human systems damage in Poland have resulted from floods. in response to actual or expected changes,
KKsisiąążżkka1.indba1.indb 1177 22008-06-26008-06-26 110:55:490:55:49 18 Zbigniew W. Kundzewicz, Jacek Jania
Figure 2. The July 1997 flood attracted massive media interest.
which moderates harm or exploits beneficial including an unwillingness on the part of peo- opportunities. The taxonomy of adaptation ple to relocate (Kundzewicz et al., 2007). distinguishes classification into adaptation Both mitigation of (the causes of) cli- types (dichotomies), such as: anticipatory mate change and adaptation to (the effects (proactive; adaptation to ongoing changes) of) climate change are needed to avert or or reactive (to projected changes); autono- reduce adverse impacts. Adaptation strate- mous (spontaneous) or planned; private / gies can reduce vulnerability to changes in public, etc. climate at the local and regional levels. Miti- The capacity to adapt varies greatly across gation acts at a global level over longer time regions, societies and gender and income scales due to the inertia of the climate sys- groups (differences reflecting a number tem, slowing the rate of climate change and of factors, such as wealth, housing quality thus delaying the occurrence of impact and and location, level of education, mobility its magnitude. Most of the benefits of mitiga- etc.). Enhanced adaptability is needed, i.e. tion will not be obtained until several dec- an increase in the system’s coping capacity ades later, thus adaptation is needed to ad- and coping range (cf. Kundzewicz, 2007). dress near-future impacts. However, without There can be limits to adaptation (physi- mitigation, the increasing magnitude of cli- cal, economic, socio-political, or institutional). mate change will significantly diminish the Barriers to adaptation to floods via relocation effectiveness of adaptation. can be external, reflecting e.g. a lack of land Mitigation of climate change and ad- for relocation, as in Bangladesh; or internal, aptation to climate change and its impacts
KKsisiąążżkka1.indba1.indb 1188 22008-06-26008-06-26 110:55:490:55:49 Extreme Hydro-Meteorological Events and their Impacts… 19
are sometimes in conflict. For instance, In turn, water demand may be addressed desalination serves adaptation, but re- (reduced) by: quires a high energy input, hence adversely • improving the efficiency of agricultu- affecting mitigation—it drives the atmos- ral water use (e.g. “more crop per drop”); in pheric greenhouse gas concentration and particular—irrigation; warming. Afforestation serves mitigation • soil moisture conservation, e.g. through (carbon sequestration) but may play an ad- mulching; verse role where adaptation in some regions • recycling water (e.g. the re-use of wa- is concerned (due to transpiration of large ste water after treatment); amounts of increasingly precious water). • water-demand management through Enhancing water storage in reservoirs brings metering; co-benefits, being advantageous for both • promoting water-saving technologies; mitigation (hydropower without fossil-fuel • leak reduction; burning) and adaptation (weakening hydro- • market-based instruments, e.g. water logical extremes—floods and droughts), cf. pricing; Kundzewicz et al. (2007). Yet it may also • the re-allocation of water to high-va- have considerable disadvantages (barriers to lue uses; fish migration, resettlement, etc). • awareness raising. In general, Europe has a high adaptation Adaptation to the latter situation (too potential in socio-economic terms, due to its much water; intense precipitation, flooding, strong economies, high GDPs, stable growth, landslides, erosion) addresses options aimed moderate changes in numbers of inhabit- at reducing the load: ants, well-trained population with a capacity • enhanced implementation of structu- to migrate within the supranational organ- ral/technical protection measures, such as ism of the European Union, and well-devel- dikes, relief channels, enhanced water sto- oped political, institutional and technologi- rage; cal support systems. However, adaptation is • watershed management (“keeping wa- generally limited in the cases of the natural ter where it falls” and reducing surface runo- systems. Equity issues also arise, since the ff and erosion). more marginal and less wealthy areas (and Resistance may in turn be increased by: groups of people within them) are less able • flood forecasting and warning; to adapt (Kundzewicz et al., 2007). • regulation through planning, legisla- Many adaptation options address water- tion, and zoning; related problems exacerbated by climate • flood insurance; change, in particular the increasing vari- • the relocation of populations living in ability of water resources, i.e. increased fre- flood-risk areas; quency of occurrence of situations in which • flood proofing on location; there is too little or too much water. Adap- • flood plain protection measures. tation options for the former situation (too There are several adaptation strategies little water—water stress or drought) ad- when it comes to coping with floods, these dress (enhance) water supply by way of such being labeled as: protect, accommodate, or measures as: retreat (relocate), cf. Kundzewicz (2007). • the conjunctive use of surface water Strategies for flood protection and manage- and groundwater; ment may modify either flood waters, or sus- • increased storage capacity for surface ceptibility to flood damage and the impact water, groundwater, and rain water; of flooding. The EU Floods Directive (Com- • water transfer; mission of European Communities, 2006) • the desalination of sea water; obliges EU Member Countries to prepare pre- • the removing of invasive non-native liminary flood risk assessment (“taking into vegetation, account long-term development including
KKsisiąążżkka1.indba1.indb 1199 22008-06-26008-06-26 110:55:520:55:52 20 Zbigniew W. Kundzewicz, Jacek Jania
climate change”), and to develop flood risk able development; that is, they can protect maps and flood management plans. In some against both climate variability now and fu- countries, such as the Netherlands and the ture climate change (this refers in particular UK, flood design values have been increased, to “no-regret” strategies—doing things that based on early climate-change impact sce- make sense anyway. It is always good to save narios. In The Netherlands, measures to energy and water). Improved adaptation cope with an increase in design discharge to current climate variability would render from the Rhine from 15,000 to 16,000 m3/ societies better prepared to future climate s must be implemented by 2015, and it is change. planned that design discharge be increased to 18,000 m3/s in the longer term, in order to maintain a high level of protection, includ- 6. CONCLUDING REMARKS ing under conditions of climate change. However, dedicated and consequent Extreme meteorological and hydrological long-term disaster preparedness efforts are events affect human life and the environment jeopardized by a prolonged absence of disas- on different spatial scales. As demonstrated ter. This effect can be called a short-memory in the paper, a majority of the extreme events syndrome. In hydrology, the vicious circle il- have a direct effect locally (e.g. gales, storm lustrated in Fig. 3 is sometimes called a hy- surges, rain-induced local flooding, lands- dro-illogical cycle. In many countries, the av- lides and mudflows). Some extremes like erage time to the next large disaster is much heatwaves, droughts and major floods have longer than the duration of terms of office impacts on a regional scale. Only ocean le- of elected authorities. Since a large hydro- vel rise and its consequences can be directly meteorological emergency is not very likely observed globally. Our knowledge of clima- to happen during the short terms of office, te and environmental change on the global the attention of decisionmakers is focused scale is based on millions of measurements on more immediate, more burning (and made at particular locations, and observa- more certain) needs. tions of extreme events and their conseque- Many potential current adaptations are nces on local and regional scales (Fig. 4). consistent with the principle of sustain- Extreme meteorological and hydrological
Figure 3. Illustration of the short memory syndrome related to natural disasters.
KKsisiąążżkka1.indba1.indb 2200 22008-06-26008-06-26 110:55:530:55:53 Extreme Hydro-Meteorological Events and their Impacts… 21
Figure 4. Mechanism of reduction of the short memory syndrome by direct and indirect impulses of impacts of the extreme events on communities with the importance of the mass media spreading information.
events with their impacts on human life con- debates prior to elections in Poland. ditions in public perceptions are considered Scientific evidence for global warming a direct local experience. and associated extreme events has been The short memory syndrome influences built up as the ensemble of ground obser- the perspectives of local authorities and com- vations on a local and regional scale, com- munities in particular areas. However, the bined with remote-sensing methods usually media coverage of global and regional climate applied to supraregional areas. Positive and change enhances the awareness of threats negative feedbacks of differing intensity and caused by related extreme events. Timely dis- scale have been detected. Improved social semination of information on environmental awareness and understanding of such proc- disasters at different locations worldwide am- esses governing global-scale environmental plifies the societal effect of scientific reports changes—with a special emphasis on local on climate warming and its environmental and regional impacts—can help overcome consequences (e.g. IPCC, 2007). short memory syndrome. In this context, the The existence of short memory syndrome permanent care and attention of politicians, can also be observed among politicians and authorities at different levels and especially policymakers at the parliamentary and gov- the mass media, devoted to information on ernmental levels. The theme of threats re- the risk of hydrometeorological extremes lated to climate change and extreme events is of special importance. This would create had not been not present in public political enhancing conditions for the adaptation
KKsisiąążżkka1.indba1.indb 2211 22008-06-26008-06-26 110:55:540:55:54 22 Zbigniew W. Kundzewicz, Jacek Jania
of societies to ongoing and future climatic Dubrawski, R. and Zawadzka-Kahlau, E. (eds.) variability and change, preparing them to (2006), Przyszłośc ochrony polskich brzegow deal with extremes and to reduce their con- morskich [Future of protection of the Polish sequences. sea coasts]. Instytut Morski, Gdańsk, 302 p. IPCC (Intergovernmental Panel on Climate Change) (2007), Climate Change 2007: The ACKNOWLEDGEMENTS Physical Science Basis. Summary for Policy- makers. Contribution of the Working Group The present contribution has been prepared I to the Fourth Assessment Report of the within the framework of the research project Intergovernmental Panel on Climate Change, entitled “Extreme meteorological and hy- Cambridge University Press, Cambridge, drological events in Poland”, financed by the U.K.
KKsisiąążżkka1.indba1.indb 2222 22008-06-26008-06-26 110:55:570:55:57 Extreme Hydro-Meteorological Events and their Impacts… 23
Group II to the Fourth Assessment Report Stearns, L. and Hamilton, G. (2006), Dynamics of the Intergovernmental Panel on Climate of large tidewater glaciers in East Greenland: Change, Cambridge University Press, Cam- recent results from satellite remote sensing bridge, UK. (Also available at
KKsisiąążżkka1.indba1.indb 2233 22008-06-26008-06-26 110:55:570:55:57 ACCESS TO AND RE-USE OF PUBLIC-SECTOR ENVIRONMENTAL DATA AND INFORMATION. POLICY DEVELOPMENTS WITH A FOCUS ON THE EUROPEAN HYDRO-METEOROLOGICAL SCENE
TON W. DONKER Stakeholder and Contract Management, Royal Netherlands Meteorological Institute (KNMI), P.O. Box 201, 3730 AE De Bilt, Netherlands E-mail: [email protected]
Abstract: Thus far, the process of obtaining basic data and information from governmental agen- cies with a view to its being used in the information industry has seemingly been a troublesome and sometimes discouraging operation in most European countries. Re-users in both the public and private sectors are faced with extensive barriers reflecting increasing commercialization in the operations of government agencies, data protection, high license fees and a short-sighted appeal to profitability principles. However, a trend towards the provision of “free data” at delive- ry cost only now appears to be gaining currency both in society and with governments. It can be argued that “free data provision”, unlike short-term cost recovery policies, will generate optimal socio-economic benefits. But “free data” in our digital era is one side of the picture. The other is that national governments will be forced to re-think their role in the information society, and last but not least in their relationship with the private information industry. Neither full public dominance nor a private monopoly seem optimal solutions from the societal viewpoint.
Key words: data re-use policy drivers, data protection, government commercialization, informa- tion society, digital era, creation of derived and new data sets, knowledge-based economies, so- cio-economic benefits, information industries, value-added information services, decline of cost recovery policies, European Commission Directives
INTRODUCTION International Meteorological Organization (IMO), established in 1873, member nations Weather, climate and water cycles know no enjoyed worldwide unrestricted exchange national boundaries. Inevitably, interna- of nationally collected data and information tional cooperation on a global scale is cru- for over a century. This international col- cial to the development of meteorology and laboration survived World War II, the Cold operational hydrology, and to any reaping War and many other conflicts, demonstrat- of benefits from their applications there may ing that the WMO, as a specialized agency be. Under the auspices of the World Mete- of the United Nations, is also an outstand- orological Organization (WMO), estab- ing example of sustainable global coopera- lished in 1950, as well as its predecessor, the tion. However, in line with various drivers,
KKsisiąążżkka1.indba1.indb 2255 22008-06-26008-06-26 110:55:570:55:57 26 Ton W. Donker
the tide has turned (mainly in the European ment-commercialization raises significant Community) since the mid 1980s. Following obstacles. Thus far this has been a source the US example, the first EU-based private of ongoing conflict between the public and weather services have entered the European private sectors, and in some countries even market. Almost all National (Hydro) Mete- amongst public agencies themselves. orological Services (N(H)MSs) responded Last but not least, protective data policies by introducing competitive activity via their regularly hamper the process of the creation commercial arms, for reasons of protection. of new knowledge unnecessarily. In this way, This was (and remains) in line with current the further development of knowledge-based political finding and views to the effect that economies is discouraged. Overall, it can be “Public sector bodies must be required to concluded that most taxpayer-funded data in utilize their own resources and operate in the EU have become underexploited, some- a way that ensures that they are not a burden thing that cannot be said to be beneficial for to the tax-payer”. The consequence is that our societies as a whole. most European N(H)MSs nowadays oper- Now, however—some 25 years later—we ate under the dual disciplines of the market await radical policy changes in the near fu- and the budget. Such a hybrid foundation is ture. There is growing awareness at national not a stable one, lacks transparency, and puts and European administrative levels that (inter)national relationships under pressure. publicly-funded data and information (e.g. By virtue of their national WMO Mem- statistics, land surveys, data on health, the berships, all European N(H)MSs exercise environment, weather and climate) repre- a monopoly on actual national and inter- sent valuable public goods and resources. national data and information which are The socio-economic benefits to society are continuously exchanged through WMO maximized when such data and informa- telecommunications channels. This means tion are made available freely, or at least in- that private entrepreneurs and other (semi) expensively and as widely as possible. This public entities depend fully on operational “macro-approach” appears to be gaining data provision through their N(H)MSs. For currency among both the public and the pol- this reason a regulatory framework was cre- icymakers, and one can observe that the cost ated through ECOMET, established in 1995, recovery policies present in some countries a joint Economic Interest Grouping of Euro- are now in decline. pean N(H)MSs. Moreover all European nations are members of the European Centre for Me- THE IMPORTANCE OF ACCESS TO dium Range Weather Forecasts (ECMWF, AND THE RE-USE OF DATA FOR RESEARCH Reading, UK) and of the European Or- AND DEVELOPMENT ganization for the Exploitation of Meteoro- logical Satellites (EUMETSAT, Darmstadt, The atmospheric and hydrological sciences Germany) which produce very valuable data have progressed in the twentieth centu- and information. The Member States are ry towards a world-wide enterprise provi- represented by their N(H)MSs and jointly ding considerable benefits to individuals, settle the terms for data provisions to third business and governments. Through rese- parties. All this demonstrates the dominant arch and applications, these sciences provi- position of the N(H)MSs. de information that contributes to the pro- National policies very much focused on tection of life and property, economic and cost recovery and the application of profit- industrial vitality, management of air and ability principles are now putting severe water quantity and quality, national policies pressure on re-users of data. As in the cases concerning energy and the environment, of many other kinds of public data assets and so on. In our developed industrialized (e.g. the spatial and geographical), govern- societies we count increasingly on reliable,
KKsisiąążżkka1.indba1.indb 2266 22008-06-26008-06-26 110:55:580:55:58 Access to and Re-Use of Public-Sector Environmental Data… 27
factual information about the environment rithms had been set to eliminate data with and many other issues. This is the lifeblood extreme low ozone levels. NASA had been of all progress. A continuum of processing disregarding valid evidence for years. The and interpretation is needed if goals are to reanalyzed TOMS data confirmed that the be achieved, and we might distinguish the ozone loss first observed by the BAS was following consecutive phases and definitions real and occurred over most of the Antarctic (Resolving..., 2001): continent” 1) data—numerical quantities or other factual attributes derived from measure- This lesson learned here is that discover- ment, observation, experiment or calcula- ies in environmental science may go undetec- tion that lay the foundation of all research ted, sometimes for many years, simply because activities; they are unexpected. The only safeguard is 2) information—collected data assets constant vigilance and scrutiny of data and and associated explanations, interpretations methods with a view to their being analyzed or other textual material (i.e. meta-data) con- by as many scientists as possible. This is cerning a particular object, event or process; a particularly important reason for environ- (Note from the author: Both data and infor- mental scientists to anticipate full and open mation are generally denoted in this paper access to all environmental data upon which as PSI: Public Sector Information) scientific interferences are based. 3) knowledge—information, which is A free PSI policy for scientific research- organized, synthesized or summarized to ers in both the public and private domains enhance comprehension, awareness or un- would have many benefits of kinds that derstanding; may not always be valued in purely mon- 4) understanding—possession of a clear etary terms, but which are surely helpful as and complete idea of the nature, significance we strive for socio-economic sustainability or explanation of something. It is the power (OECD, 2006): to render experience intelligible by ordering • interdisciplinary, inter-sector, inter- particulars under broad concepts institutional and international research for It is self-evident that obstacles and the creation of new knowledge is promoted; conflicts at the PSI foundation may have • expensive duplication of research is harmful impacts. This can be demonstrat- avoided, and new research and types thereof ed by many examples like the discovery promoted; of the Antarctic Ozone Hole. Carver (1998) • open scientific enquiry is reinforced wrote: and diversity of analysis and opinion encour- aged; “A dramatic loss of ozone in the lower • the verification of previous results is stratosphere over Antarctica was first no- provided for; ticed in the 1980s by a research group from • new or alternative hypotheses and the British Antarctic Survey (BAS) that was methods of analysis can be tested for; monitoring the atmosphere using a network • studies on methods of data collec- of ground-based instruments. The drop in tion, measurement and calibration are sup- ozone levels was so large that at first the me- ported; teorologists thought their instruments were • the education of new researchers is faulty, although careful checks subsequently facilitated; confirmed their measurements. Meanwhile, • the exploration of topics not envi- data from NASA’s Total Ozone Mapping sioned by the initial investigators is provid- Spectrometer (TOMS) satellite failed to ed for; show a similar decline. The BAS results • the creation of new and derived data- spurred NASA scientists to re-examine the sets is permitted when multiple sources are TOMS data, and they found that their algo- combined;
KKsisiąążżkka1.indba1.indb 2277 22008-06-26008-06-26 110:55:580:55:58 28 Ton W. Donker
• the building of research capacity in abuse of their dominant position and unequal developing countries is promoted; treatment of the private sector. A major fur- • a maximising of the research poten- ther deficit is that valuable data and products tial of new digital technologies and networks may be transferred from the free Essential is facilitated, thereby providing greater re- Data Set to the Additional Set in order that turns from the public investment in research revenues from PSI sales be boosted. and education. WMO RESOLUTION 25 CG XIII (1999) A similar resolution was adopted for hy- REGULATIONS AND RECENT POLICY drological data to monitor and allocate wa- INITIATIVES ter resources and assessments of the risks of floods and droughts. Under Resolution WMO RESOLUTION 40 CG XII (1995) 25, a core set of data are freely available wit- WMO Resolution 40 was adopted in respon- hout conditions to use, while the remainder se to the onset of commerce in meteorology can be sold or restricted by the Member Sta- in Europe. te that collected the data (cf. www.wmo.ch). The resolution enshrines an understand- Some of the core data, such as river flow ing among WMO Member States that they records, are available through the WMO will endorse the free and unrestricted ex- Global Runoff Data Centre (Koblenz, Ger- change of data and products (in situ obser- many). However the Centre imposes three vations, data from satellites and models) for restrictions: the official public duties of N(H)MSs, the le- • the amount of data that can be re- gal tasks of other national public bodies (e.g. quested is limited; in defence) and (non-commercial) research. • users are not permitted to share the However, it includes a provision to the effect data with third parties; that an individual Member State may place • users must inform the Centre how restraints on the commercial re-use of data data will be used. and products by thirds parties, unlike the It is generally understood by experts that application for public duties (cf. www.wmo. the limiting of access to and re-use of all ch). available hydrological data severely hampers A two-tier classification of data and scientists’ ability to construct or validate products was introduced: global or regional models of the hydrologi- • WMO Essential Data and Products: cal cycle, land-atmosphere interactions and no conditions as regards use and re-use, biochemical cycles. Restricted PSI cannot be • WMO Additional Data and Products: shared amongst colleagues, and this under- conditions for commercial re-use generally mines the scientific practices upon which the expressed (apart from a recouping of de- research enterprise depends. Restrictions livery costs) in retribution for the creation also create major inefficiencies in the PSI of data, in other words a licence charge. systems, because the same data and infor- Additional conditions and licence fees may mation must be collected by multiple organi- also apply to e.g. broadcasting of derived zations. value added services through TV and the internet. THE EC DATABASE DIRECTIVE Resolution 40 was initiated factually by 96/9/EC (1996) the hybrid and “commercial” N(H)MSs in It is common for N(H)MSs to claim a protec- Europe. Outside the EU, criticism came tive policy in their capacity as PSI holders. from the USA and Japan, academia and the This is achieved via a doubtful and conte- private sector all over the world. One fear is stable appeal against the Database Direc- that Resolution 40 will not prevent cross-sub- tive of the European Parliament and of the sidies within the commercial N(H)MSs, the Council of 11 March 1996 on the Legal Pro-
KKsisiąążżkka1.indba1.indb 2288 22008-06-26008-06-26 110:55:580:55:58 Access to and Re-Use of Public-Sector Environmental Data… 29
tection of Databases (cf.http://ec.europa. tion of EU Member States. Deadlines have eu.prelex). been set for the review of implementation Indeed, this Directive was barely needed and evaluation. for private work though the growing under- standing now is that it should not or even can- DIRECTIVE 2003/4/EC ON THE PUBLIC not apply to PSI, funded by the tax-payers. ACCESS TO ENVIRONMENTAL INFORMATION The key elements of the Directive are open EC Directive 2003/4/EC (cf. http://ec. to conflicting interpretations and controver- europa.eu.prelex) prescribes free public ac- sy across the board (Hugenholz, 2001). cess without discrimination to public-sector For example Article 10, Term of Protec- environmental data and information, in or- tion, Para. 1 states: “The right provided for der to increase public awareness as regards in Article 7 shall run from the date of the environmental matters. The Directive states completion of the making of the database. that any natural and legal person has a right It shall expire fifteen years from the first of access to updated information held by or of January of the year following the date for public authorities, without his having to of completion”. state an interest. Explicitly, reference is made Following Para. 3 of this Article 10, any to the state of the elements of the environ- compiler who makes a database available to ment such as air and atmosphere, water, soil the public may continually renew the right and land, including environmental impact for additional 15-year terms with every ad- studies and risk assessments. Dissemination ditional investment in or extension of the is recommended in particular through “elec- database. This renewal covers the content tronic means” like the internet. of the entire database, and not just the new The Member States were expected to matter. comply with the Directive by 15 February All meteorological and hydrological data 2005, and not later than 14 February 2009, and in situ observations are collected in near the Member States will report on the experi- real-time, and by consequence their archives ences gained in the application of the Direc- are of a very dynamic nature with high re- tive. freshment rates. Compliance with this Di- An important “spin off” of this Directive rective would mean that meteorological and for European citizens is the current proposal hydrological data never can or will be dis- (COM 2006/15 final) of the European Par- seminated widely for free. Moreover, such liament and of the Council on the assessment compliance is in contradiction with more and management of floods in Europe. recent EC Directives which, on the contrary, encourage public access to, and the re-use of DIRECTIVE 2003/98/EC ON THE RE-USE PSI. OF PUBLIC SECTOR INFORMATION The PSI Directive (cf. http://ec.europa. THE EC GREEN PAPER (1999) eu.prelex) provides a general framework for The Green Paper issued by the Commis- re-use, i.e. use for commercial or non-com- sion of the European Communities in Janu- mercial purposes other than the initial pur- ary 1999 comes to the conclusion that PSI is pose within the public task for which the data a key resource for Europe, and suggests that or information were produced. However, the EU nations should more closely follow the exchange of data and information between model of US Federal Government policies public sector bodies purely in pursuit of their with regard to promoting broader access to public task does not constitute re-use within government databases (EC, 1998). in the meaning of the Directive. A few years after the Green Paper’s “Re-use of” includes more than “access publication, additional developments took to”. Thus the Directive provides for a mini- place, i.e. the transposition of more compel- mum harmonization, imposes non-discrimi- ling EC Directives into the domestic legisla- nation and equal treatment (especially
KKsisiąążżkka1.indba1.indb 2299 22008-06-26008-06-26 110:55:580:55:58 30 Ton W. Donker
regarding exclusive arrangements), seeks frastructure that delivers integrated spatial to prevent abuse of a dominant position information services to the users, in an in- (pricing) and cross-subsidies, and lays down ter-operable way and for a variety of applica- transparency requirements. With regard to tions. Possible services are the visualization fair trading, it is emphasized in Article 10 of information layers, overlay of information that “if data and information are re-used by from different sources, spatial and temporal a public sector body as input for its commer- analyses, etc. for the benefit of policy-mak- cial activities which fall outside the scope ers, planners and managers at EU, national, of its public tasks, the same charges and oth- and local levels, citizens and applicants in er conditions shall apply to the supply of the the private information industry. data and information for those activities as While the international (hydro) mete- apply to other users”. orological community already meets many An important aim is to remove the ex- INSIPRE requirements by virtue of tech- isting barriers for private entrepreneurs: nical WMO regulations, new technologies “wider possibilities of re-using public sector can enhance exchange and representations information should inter alia allow Europe- of data and information in a different format an companies to exploit their potentials and for comprehensive operations and research. contribute to economic growth and job crea- The Directive is still in the stage of concili- tion.” (Excerpt from Article 5). ation. The Conciliation Committee accom- The PSI holder may charge for creation, plished its joint text (PE-CONS 3685/2006) reproduction and dissemination, but the on 17 January 2007. Currently implemen- Directive recommends that public-sector tation teams are commencing with their bodies be encouraged to make PSI available tasks. at charges that do not exceed the marginal or incremental costs for reproduction and dis- tribution (preamble 14). The Member States SOCIO-ECONOMIC ASPECTS OF PSI were expected to comply by 1 July 2005. The EC shall execute a review of the application The vast economic potential of PSI has only before 1 July 2008 and shall communicate recently begun to be recognized in the eco- the results of this review, together with any nomic and public policy literature. With proposals for modification to the European respect to the growing challenge from econ- Parliament and the Council. omists, the European Commission’s Direc- torate General for the Information Society THE FORTHCOMING INSPIRE DIRECTIVE commissioned two studies. The general situation on spatial information in Europe is one of fragmentation of data- THE PIRA STUDY (2000) sets and sources, gaps in availability, lack The PIRA Study (Lit. 12) attempted to of harmonization between datasets and quantify the economical potential of PSI in formats at different geographical scales, Europe, and the extent to which it is being and duplications of information collection. exploited commercially, as well as to suggest These problems make it difficult to identify policy initiatives and good practices. The and re-use the data that are available. For- study observed that the European PSI mar- tunately, awareness is growing at all levels ket would not even have to double in size for of the need for quality geo-referenced in- governments to more than recoup (in addi- formation to support understanding of the tional VAT receipts) what they would lose by complexity and interactions between human ceasing to charge for (or license) PSI. activities and environmental pressures and The amounts of money involved are sig- impacts. This ambitious Directive (cf. http:// nificant. PIRA distinguished between gov- inspire.jrc.int) intends to trigger the crea- ernment investment in PSI (“Investment tion of a European spatial information in- Value”) and the value added by users in the
KKsisiąążżkka1.indba1.indb 3300 22008-06-26008-06-26 110:55:590:55:59 Access to and Re-Use of Public-Sector Environmental Data… 31
economy as a whole (“Economic Value”). was undertaken by an international research Economic Value could not be obtained consortium from November 2004 through directly, so aggregated data were used. April 2006. The main objectives were: PIRA estimated the Investment Value for • to develop, document and test the entire European Union at 9.5 billion a repeatable methodology for measurement EURO/year. The Economic Value was es- of PSI re-use; timated at 68 billion EURO/year. By com- • to perform a baseline measurement parison, the Investment Value for the USA of PSI re-use in the EU Member States and is 19 and the Economic Value is 750 billion Norway, including a comparison with the EURO/year. United States. Other PIRA conclusions are that: MEPSIR conducted the measurement in • the pursuit and monitoring of protec- 25 Member States of the European Union tive PSI policies is expensive. Charging for and in Norway, and investigated the condi- PSI (in some countries even between public tions of availability, accessibility, transpar- sector bodies and without any revenue for ency, accountability, non-discrimination, the Treasury) may be counter-productive, actual demand and economic results in six even from the short-term perspective of the main information domains, i.e. business, raising of direct revenue from government geographical, legal, social, transport, and agencies; meteorological/hydrological. • Government should make PSI avail- The economic results will eventually able in digital form at no more than the costs translate into direct results (more turnover of dissemination or direct delivery to indi- and employment for PSI re-users) and indi- vidual users; rect results (increasing commercial activity • Governments experience two kinds based on the re-use of PSI). Demand and of financial gain when they drop license economic performance were measured by charges: directly asking both PSI-holders and re-us- (i) higher indirect tax revenue from high- ers for key economic data under the current er sales of value-added products for which regimes. PSI served as “raw materials” and (ii) higher This allows for the generating of esti- income tax revenue and lower social welfare mates in billions of EUROs of: payments form net gains in employment in • the overall European PSI markets information industries. based on estimates from respondents; the Due to the circumstances at that time, market size estimated between 26 (median) PIRA had to confine its estimates to a lim- and 48 billion (upper limit) EUROs, ited number of individual and promising • the overall European PSI markets in-depth studies (land survey, meteorology, based on estimates of turnover (minus the publishers, patent & trade marks, business data license charges) from Public Sector In- services). Consequently, the individual val- formation; the market seize was estimated ues of PIRA might be more robust, but the between 12 (median) and 45 billion (upper subsequent aggregated value is less so. limit) EUROs. Nevertheless, the study should be suffi- MEPSIR found an inverse correlation cient to persuade policymakers of the need between the charges for PSI and the number for a serious rethink of European PSI policy of re-users: decreases in licence charges and its high priority. were more than offset by increases in the number of users seeing additional business THE MEPSIR STUDY (2006) opportunities. In the context of the preparations for the A repeat of MEPSIR within a few years PSI Directive (2003/98/EC) an extensive will most likely show a substantially in- study on Measuring European Public Sec- creased overall PSI market potential, but tor Information Resources (MEPSIR, 2006) with large spreading over market segments.
KKsisiąążżkka1.indba1.indb 3311 22008-06-26008-06-26 110:55:590:55:59 32 Ton W. Donker
OBSERVATIONS of licencees in and outside the EU has been growing. There are different practices among the Eu- • EUMETSAT is in the process ropean countries and a still imperfect under- of reducing license tariffs for Meteosat im- standing of the costs and benefits of making age data. Regulations and conditions will PSI freely available. Indeed, the issues are be simplified, and broadcasting of Meteosat complex and are of a different nature in dif- image products through TV and the Internet ferent fields: economic/financial, social/cul- will be encouraged. tural, organisational/institutional, manage- Other observations indicate that PSI ment, legal/regulatory and technological, policy issues and private-public settings are although some are cross-cutting. This is still beginning to move advantageously. preventing the full exploitation of the poten- • PSI surely has value, but its pricing tials behind PSI. is precarious and seldom clearly explicable. Nevertheless, at the national and inter- Perhaps this clarifies the striking difference national levels, there are initiatives and new of N(H)Ms licence tariffs for the same data practices removing obstacles towards unhin- asset (e.g. the price of an WMO standard- dered application of (hydro)meteorological ized in-situ observation). data and information by users unlike the The pricing and “selling” of publicly- N(H)MSs. funded data and information are more and • In most of the EU countries the ini- more criticized from an economic point tial licence charges have not risen over the of view. Economic analyses show that not past 10 years, or have been reduced in some all goods can be transacted through mar- cases; kets readily. PSI, a public good, is created • A growing number of N(H)MSs are for collective consumption or production, transferring chargeable data assets to the rather than private consumption or produc- free WMO Essential Data Set on a step by tion. Moreover, a public good is character- step basis (e.g. the United Kingdom, The ized by two attributes: non-depletability and Netherlands, Austria and Spain); non-excludability. Non-depletability means • The Norwegian Met. Service recently that the product in question cannot be used designated all data (in-situ observations, ra- up and is available to additional persons. If dar, models) as WMO Essential; (hydro)meteorological data are provided, • Most newcomers to the EU doubt the same data remain just as available as be- whether they should pursue a protective data fore for other users. Non-depletability is the policy; main reason that free use of PSI can be justi- • After much deliberation, the first li- fied: there is no additional social cost when cences for Internet dissemination of a “com- another person uses it, and there is no justi- mercial sensitive” West-European Weather fication for the disincentive to its use that is Radar Composite could be concluded (cf constituted by a substantial fee for that use. www.meteox.com). Non-excludabilty refers to the supplier side • The ECMWF Council decided to of the problem. It means that the goods in simplify the licence conditions and reduced question produce benefits from which others licence tariffs for the commercial re-use cannot be excluded and which cannot be eas- of global and regional atmospheric and ocea- ily constrained only to those who pay (e.g. nographic model output. It appeared in 2003 police services). that more than 95% of the numerical weath- • Debate in the public-private arena is er prediction data used by the private sec- gradually shifting from data costs to com- tor in Europe originated from free sources petition issues. Several cases have been in the USA (NOAA) and Japan. Meanwhile brought forward to National Competition the visibility of ECMWF, a Centre of Excel- Authorities and deal with unfair competi- lence, improved remarkably and the number tion by commercial arms of some N(H)MSs
KKsisiąążżkka1.indba1.indb 3322 22008-06-26008-06-26 110:55:590:55:59 Access to and Re-Use of Public-Sector Environmental Data… 33
which are believed not to pay in real money the upright national centre and holder of au- to the core for the re-use of data. thentic data and information and primary • Some N(H)MSs are willing to reduce provider; their licence fees as far as possible, or even • Treasuries should not simply regard to zero. But they fear non-compensation by N(H)MSs as traditional budgetary “items their National Treasury for the loss of rev- of costs”. In the emerging information era, enue. N(H)MSs can seamlessly participate in the • Some N(H)MSs are in a process of for- information industry, and generate benefits mal separation of their public duties and in an indirect way (e.g. tax revenues from commercial functions. One NMS achieved value-adding information services); that in 1999 through full privatization. • The Public-private relationship or per- haps duality must be fostered, but an under- standable and sustainable balance will not be CONCLUDING REFLECTIONS easy to achieve. However, the private sector or other applicants should not “pay twice” There are indications that, in the near future, for the use of data from the N(H)MSs, while data policies and public private relationships the public should not “pay twice” for the pro- in the European (hydro)meteorological vision of all essential news and information arena will gain momentum, due to internal related to weather and climate; and external drivers. Ongoing policy consid- • Not the delusions of the day but the erations are likely to show that the benefits long term interest should prevail for all par- on the national and European scales due to ties. open access and re-use far outweigh revenue that might be generated through cost-recov- ery policies. REFERENCES Next developments in information and communications technologies will endure in Carver, G. (1998), The Ozone Hole Tour, Part 1: a spectacular manner. It is difficult to imag- The History Behind the Ozone Hole, Univer- ine that, in this revolution, either N(H)MSs sity of Cambridge
KKsisiąążżkka1.indba1.indb 3333 22008-06-26008-06-26 110:56:000:56:00 34 Ton W. Donker
Policy, Working Party on the Information tion of Environmental Data (2001), National Economy, 82pp. Research Council, National Academy Press, PIRA International (2000), Commercial Exploi- Washington. tation of Europe’s Public Sector Information. Final Report for the European Commission, Paper first received: November 2007 Directorate for the Information Society. In final form: February 2008 Resolving Conflicts Arising from the Privatiza-
KKsisiąążżkka1.indba1.indb 3344 22008-06-26008-06-26 110:56:000:56:00 ESTIMATION OF DAMAGE CAUSED BY EXTREME WEATHER EVENTS, WITH AN EMPHASIS ON FLOODS
PIOTR MATCZAK, ROMAN MAŃCZAK and ZBIGNIEW W. KUNDZEWICZ Research Centre for the Agricultural and Forest Environment, Polish Academy of Sciences, 60-809 Poznań ul. Bukowska 19, Poland Email: [email protected]; [email protected]; [email protected];
Abstract: Damage caused by extreme weather events is projected to increase on account of cli- mate change. However, the assessment of losses is a weak point of the systems concerning pre- paredness for and management of weather extremes. Methods of ex post loss assessment are discussed here, with particular emphasis on floods. Approaches based on restoration value and market value are presented. Methods addressing indirect tangible losses and intangible damage are also reviewed. Restrictions and ambiguities connected with the methods are presented, and difficulties with data collection discussed.
Key words: Loss estimation, extreme weather events, floods, intangible damage, tangible damage
INTRODUCTION level to USD 9.2 billion annually between the 1950s and 1990s, with a significantly higher Data provided by the insurance industry re- insured fraction in industrialized countries, veal that losses caused by weather-related while the ratio of premiums to catastrophe catastrophes represent several tens of bil- losses fell by two-thirds (Mills et al., 2001). lions of USD per year on average, and have Inflation-adjusted economic losses due to been increasing rapidly in recent years. The catastrophic events increased 8-fold between number of major flood disasters in the 1990s the 1960s and 1990s, while insured losses in- was greater than in the three and a half dec- creased 17-fold (Mills, 2005). Insured and ades 1950–1985 taken together (Berz, 2001). total property losses (USD 45 billion and Annual global economic losses brought 107 billion in 2004, respectively) are rising about by major events have increased in val- faster than population growth, or economic ue by an order of magnitude in four decades; growth. Between 1970 and 1999, weather- from USD 4 billion in the 1950s to USD 40 related losses (adjusted for inflation) grew billion per year in the 1990s (all in infla- at a rate nine times faster than the popula- tion-adjusted 1999 USD), cf. Vellinga et al. tion. Over the 15 years at the end of the 20th (2001). The insured portion of these losses century (Mills et al., 2001), natural disasters rose even more strongly, from a negligible caused damage worth about 1 trillion USD,
KKsisiąążżkka1.indba1.indb 3355 22008-06-26008-06-26 110:56:000:56:00 36 Piotr Matczak, Roman Mańczak and Zbigniew W. Kundzewicz
three quarters of which was weather-related 3) Estimation of losses requires special and a fifth—insured. staff. Changnon (2003) argues that people It is likely that such trends in the late who estimate losses do not have adequate 20th century were related to the frequency skills and experience. and / or intensity of several types of extreme 4) Some losses are delayed, and indirect, weather event in the warming climate (IPCC, which makes assessments even more difficult 2007). Among extremes on the rise have and less precise. been heatwaves and large-area droughts. 5) The loss itself (e.g. a destroyed bridge) The frequency of heavy precipitation events should be expressed in economic terms (or the proportion of total rainfall accounted (monetary values). This is typically a dif- for by heavy falls) is likely to have increased ficult task. Moreover, for some losses (e.g. over most areas, and is very likely to increase cultural and environmental values) the very likewise in the future. Tropical cyclone activ- possibility of assessment in monetary terms ity is likely to have increased in some regions is problematic. since the 1990s and is also likely to increase 6) Alongside the losses, extreme events further in the 21st century. An increase in also bring positive effects (opportunities), intense precipitation leads to a greater risk e.g. prosperity for the construction indus- of rain-induced flooding. The latter effect try dealing with massive orders. This aspect will be exacerbated by increasing damage should also be taken into account. potential. As a result, existing data are fragmented If the public policies by which disasters and there are substantial uncertainties in are to be coped with are to be planned, the the assessments. For example, in the US, measurement of losses is an important is- 20-year losses due to tornados could only be sue, since it helps in the evaluation of public assessed with a margin of an order of mag- expenditure within the cost-benefit frame- nitude, with estimates ranging from 5.8 to work. Implementation of any adaptation or 58 billion USD (Changnon, 2003). Attempts mitigation measures involves a policy proc- to assess losses are based on the data which ess, and requires economic analysis of costs are available, even if their validity is at times and benefits. doubtful. Detailed damage surveys are un- Notwithstanding their importance, as- common and data are often therefore based sessments of the impacts of climate change on relief payments, insurance pay-outs or are only poorly developed (Feyen et al., even newspaper news. This process can lead 2006). Insurers have developed methods to large inaccuracies and a biased view. For of loss assessment, but they are reluctant to instance, the insurance industry started to make the data available (e.g. for scientific cover weather extreme events in the 1980s, research), because they are a part of their and in 1990–94 the insured losses reached business know-how (Genovese, 2006). Yet, 40 billion USD. As a result, however, both even within the insurance industry system- insurance and reinsurance sectors have suf- atic data collection is done in certain sectors fered, and many institutions have withdrawn only (Changnon, 2003). A range of prob- coverage in some areas. Thus, a reliance on lems relate to data available in the public insured losses as the source of information domain: representing total damage has its limita- 1) Data on extreme events are very tions. scarce—uncertainties in short time series This paper discusses the basic approach- are enormous; es to the estimation of damage caused by ex- 2) Even the most-advanced countries treme events. Problems with data collection lack a commonly-agreed system of data col- are reviewed, and possibilities for the assess- lection. Information is gathered using a vari- ment of intangible damage presented. The ety of methods and sources, and this renders analysis is referred to a particular example their comparability very complex. of floods, but it is sufficiently generic to also
KKsisiąążżkka1.indba1.indb 3366 22008-06-26008-06-26 110:56:000:56:00 Estimation of Damage Caused by Extreme Weather Events, with an Emphasis on Floods 37
be applicable to damage caused by other by reference to estimates of real damage. weather-related extremes. This is related, in one way or another, to the field collection of data. The division between the two approaches is not sharp. According BASIC METHODS OF LOSS ESTIMATION to Smith (1994), the synthetic methodology combines data from actual flood events, but Losses caused by extreme weather events also relies on some hypothetical analysis. As can be assigned to many categories, includ- a result, it is possible to assess both actual ing different aspects of the economy and flood loss and potential damage (Gissing social life. The scheme presented below and Blong, 2004). Nevertheless, in this pa- (Fig. 1) offers a typology of damage. The per it is basically ex post assessments that are primary division can be placed between tan- considered. gible and intangible losses, the former being There are two typical methods of asset easier to express in monetary terms than the valuation derived from accounting. Both use latter. The evaluation of intangible losses real market prices, but only one is directly can be very difficult, if not impossible. connected with the utility of an asset. The
Financial Damages Social costs can be estimated in dollars Tangible Intangible
Indirect Direct Indirect Direct – Environmental – Cultural and Financial Opportunity Clean-up – Homelessness heritage – Health impacts – Loss of – Non provision – Immediate – Unemployment – Mortality production of public removal – Death and – Reduced wages services of debris injury – Extra – Loss of and discarded – Stress and expenditures business items discomfort – Business opportunities disruption
Internal External Agricultural Infrastructural Structural – Contents of – External – Inundated – Road, bridges – Clearing and Buildings items land area – Railways repair of e.g. vehicles – Crops – Dikes buildings – Equipment – Pastures – Phone and – Livestock electric lines – Industrial plants
Figure 1. Types of loss which can be caused by extreme weather events (after: Guidance on the Assessment of Tangible Flood Damage, Queensland Government, Natural Resources and Mines, 2002).
The losses classified in Fig. 1 can be esti- methods are based on: mated using two basic procedural approach- 1) restoration value (reconstruction/re- es: ex ante and ex post. In the former case, production); the model prepared allows for an assessment 2) market value. of potential losses, assuming certain condi- The merits and weaknesses of these two tions. In the latter case, losses are assessed methods are presented in what follows.
KKsisiąążżkka1.indba1.indb 3377 22008-06-26008-06-26 110:56:010:56:01 38 Piotr Matczak, Roman Mańczak and Zbigniew W. Kundzewicz
VALUATION BASED ON RESTORATION VALUE Reconstruction value then represents an The restoration method is usually used in as- inaccurate value of loss. However, accept- set valuation, when there is no functioning ance of restoration value in flood damage market for valuing a good, or at least a close assessment is supported by the fact that substitute thereof. The concept of reproduc- the present value of an asset is not well tion value is that an asset or good is worth reflected in book value. Moreover, there as much as a the sum of values of its parts is sometimes no book value for individual (components), and of the work necessary farms, households and some important to restore usefulness and its ability to fulfil components of technical infrastructure its function to the level existing before the (e.g. public roads). This problem of differ- damage. Restoration value is the price of ob- ence between the gross book value and the taining exactly the same asset (including the restoration value is well presented for as- case in which there is no possibility of it be- sessment in the case of the flood damage in ing bought in a normal market). Poland in 1997 (cf. Table 1). Every asset has its book value. This ex- presses nominal value on a balance sheet Table 1. Damage caused by the 1997 flood from the date of introduction into accountan- cy reporting. Usage of the restoration meth- in Poland. od in ex post flood loss estimation has one important advantage as compared to book Total damage value. The book value does not necessarily According to According to stand for the amount of money that is enough gross book restoration Voivodship value value to restore an asset to its former state, while (province) restoration value directly quantifies this. in million PLN (1997) In terms of flood loss assessment, res- Total 1469.0 1947.8 toration value is broadly used in practice in ex post damage estimation. It is obvious Bielskie 23.0 25.2 that lost value of a particular asset is the cost Częstochowskie 224.8 241.6 of restoring it to its prior condition. Such Gorzowskie 2.4 22.9 a sum of costs provides good proxy value. In order to estimate its value, one has to pro- Jeleniogórskie 14.4 36.8 ceed in line with the following steps: Katowickie 178.5 173.3 1) assessment of the restoration cost of an asset; Krakowskie 19.9 19.7 2) assessment of the rate of depreciation Nowosądeckie 22.5 22.7 of an asset due to normal usage, technical and economic ageing and other factors; Opolskie 356.4 409.9 3) assessment of the net restoration cost Tarnowskie 12.1 16.9 as the difference between gross restoration Wałbrzyskie 33.0 133.0 cost and depreciation of an asset. Damaged assets have usually been Wrocławskie 554.9 817.7 worn out, to some degree, in the course Zielonogórskie 11.1 9.1 of normal activity by the date of occur- rence of a flood. Restoration is provided in Other provinces 16.9 19.0 excess to a former value, because material after Centrum Informacyjne Rządu [Government Infor- used cannot be worn out in the same pro- mation Center], 1998. portion. Moreover, restoration is usually connected with modernization, even if this is not the intention, but there is no other As can be seen from the table above, for opportunity to bring them back into usage. the reasons mentioned above, the restora-
KKsisiąążżkka1.indba1.indb 3388 22008-06-26008-06-26 110:56:010:56:01 Estimation of Damage Caused by Extreme Weather Events, with an Emphasis on Floods 39
tion value is greater than the book value. sessment was done mainly using restora- Restoration of an asset to its former state tion value and the incorporated cost of lost is obviously impossible. Evaluation of some benefits, as defined in comparison with the assets is not up-to-date and sometimes their business activity in the year before the oc- replacement requires considerably greater currence of the damage caused by the ex- expenses then their book value (e.g. when treme event. there are no spare parts). Simultaneously, any restoration is usually conducted with DATA COLLECTION modernization, which is understandable Valuation based on restoration data requires from the economic point of view, especially input data for analysis. Within the synthetic for entrepreneurs. Therefore, ex post dam- method, Smith (1994) differentiates be- age assessment may face distortion, as may tween: ex ante, which takes into account potential 1) existing data; and losses in terms of book value. Nonetheless, 2) data based on surveys by valuers and such a gap between these two values (book loss adjusters. and restoration) may show how modern The former method is based on classifi- a structure is. cation of residential buildings depending on Restoration value is based on a simple the number of floors, building age, and con- concept of the amount of money necessary tent (furniture etc.). Similarly, types of land to reinstate an asset to its former features. use are classified as commercial, industrial It is a well-defined market measure, in con- etc. (Lugeri et al., 2006). The approach is trast to the book value that is devoid of real based on maps combining data on land cov- depreciation (book depreciation does not er type, and the exposure (dependent on the provide a market decrease of value). How- value at risk—Lavalle et al., 2005). If such ever, the restoration value defined above a map is available, the estimation of losses is a measure which does not represent the is an easy procedure, providing that data on utility of an entire asset. Further disadvan- a geographical scale are known for an event. tages are connected with the following is- The crucial issue is the level of map resolu- sues: tion. The smaller the territorial unit and the 1) What prices of work and spare parts more detailed the classification, the more are most appropriate in calculating restora- precise can be the assessment. Due to the tion value? substantial input needed, this approach is 2) Should average market prices, or more important for ex ante than for ex post minima and maxima on the market be taken assessment. into consideration? The latter method, based on surveys, has 3) Should restoration value include to be used if no earlier data on the affected transaction and other costs (e.g. taxes)? territory and assets are available. This ap- Minimum prices are usually difficult to proach is based on valuation of actual flood obtain, and taking different levels for differ- damage in a dwelling, as done by a qualified ent assets leaves assessment distorted. loss adjuster or valuer. This is a basic ex post Assessment of losses via restoration method used in the insurance industry, but value is a widely-used method. It is rela- also in the collection of public data. Two ba- tively simple and effective. For instance, sic classes of losses are dealt with: structural losses caused by the 1997 flood in Wrocław (building and non-movable components were estimated for categories of loss for thereof) and contents (the movable con- the public sector, taking into account par- tent of buildings). In both cases, more pre- ticular buildings. For private households cise classifications are often made. Table 2 and companies assessments were based presents an example of the classification and on questionnaires from Poland’s Central distribution of losses among the classes for Statistical Office. For each case, the as- a flood case.
KKsisiąążżkka1.indba1.indb 3399 22008-06-26008-06-26 110:56:020:56:02 40 Piotr Matczak, Roman Mańczak and Zbigniew W. Kundzewicz
Data collection based on surveys offers VALUATION BASED ON MARKET VALUE better results when valuers are experienced Although restoration value may seem good, and unbiased, which is not always the ac- and provide quite appropriate results, it has tual case. Sometimes valuations are based one major disadvantage. It does not include on valuations by the victims. This method is assessment of the utility of an asset to its very susceptible, and overvaluation is typi- owner. The utility concept introduces the cal. Despite difficulties, there are databases personal relationship of an owner to the which already serve as a source for analysis value of an asset. The issue of the subjec- and modeling. There are also databases in tive character of personal utility is resolved Britain and other countries, such as Ger- by the observation of market value. A mar- many, where the HAWAS database contains ket value conveys the most adequate price, data on around 4000 buildings destroyed by keeping the assumptions of independence floods between 1978 and 1994. Kreibisch et and rationality on the part of parties to an al. (2005) undertook a questionnaire survey agreement. A market value should reflect among the victims of the 2002 Elbe River the market situation on the date of valua- flood. As a result, information on the scale tion. The market provides a price of an as- of damage was gathered, allowing for the val- set that is neither under- nor overvalued, in uation of content per one m2 of living area. terms of its economic value to market par- However, as the authors admit, the accuracy ticipants. of data is not perfect due to a lack of data for Market value reflects the real price as an about half of the examined cases. effect of the demand and supply balance. It is based on the microeconomic assumption that no one will pay more than an asset is Table 2. Distribution of losses reported for 113 worth to him, and no one will sell for a lower residential damage claims for the Grand River price than something has a value to him/her. flood of May 1974 In a dynamic formulation, the rational ex- pectations theory constitutes a fundamental Categories of loss Percentage of total losses in considering market value as a best price Food 5.7 (Muth, 1961). One of the applications of the Fuel oil 0.1 concept of rational expectations is the effi- Furniture 13.9 cient markets theory of asset prices (Sargent, Appliances 2002). Large 11.4 In fact, this approach to valuation in- Small 1.9 cludes utility as well. Some may argue that Clothing 10.3 the price resulting from the market may be Television/radios 3.3 set inappropriately due to temporary dis- Recreational equipment 1.4 equilibrium. However, any market that com- Floor coverings 5.9 plies with the conditions listed before elimi- Structural repairs nates such disequilibrium almost as soon as Floors 5.5 it appears. Walls 6.6 Assessment of loss to any asset due to Furnace 7.9 an extreme event that uses market value is Other 17.1 based on two valuations: Cleaning 2.2 1) the market price of an asset before the Miscellaneous items 6.8 occurrence of the extreme; 2) the market value of a damaged asset. (based on McBean et al., 1986). The difference between these two valua- tions is the market-oriented loss assessment. One disadvantage of market valuation is the necessity that perfect market conditions
KKsisiąążżkka1.indba1.indb 4400 22008-06-26008-06-26 110:56:020:56:02 Estimation of Damage Caused by Extreme Weather Events, with an Emphasis on Floods 41
should exist. This is not observed for many rium. Such an assumption helps to provide goods, so such markets would be barely ob- D(C), P(C) ∈[0,1] and is consistent with ra- served for assets damaged by a flood. An- tional expectations of market participants. other disadvantage of this approach is con- The presented concept is obviously connect- nected with the possibility of a supply surplus ed to stage-damage functions (D(C) actually after a flood that lowers the equilibrium refers to a relative stage-damage function), price and results in undervaluation of assets. but includes utility into the valuation. The A simple explanation may release that disad- main disadvantage is the necessity, not only vantage. If some vendors put off their offer for a market for a particular asset to exist, for some time, they will obtain a higher price but also for it to be of liquid form and for later (than those vendors who do not post- there to be effective collection of informa- pone). Taking into account the discounting tion about the transaction. Moreover, more factor (inflation, liquidity cost etc.) at the complex assets e.g. buildings, do not form time after a flood (when a market is already a cohesive set of similar goods and consti- functioning), the higher price (plus possible tute rather one distinguishable tradeable extra costs, e.g. warehousing costs) would be object. In such a case, any price acquired much lowered—it is highly probable that the from a market only goes unbiased if there price could be even lower after discounting. are a large number of potential buyers (this This means that the equilibrium price seems generates an appropriate valuation that in- proper from the economic point of view. cludes utility). When considering the ratio of market value of damaged asset to value before dam- age, we should observe the same ratio as is ESTIMATION OF INDIRECT TANGIBLE presented by the stage-damage curve in par- DAMAGE ticular conditions in relative terms (actual to potential losses). It may be written in the fol- Estimation methods based on the restoration lowing relationship for asset i: value and on the market are generally easily applicable to tangible, direct damage. How-
Di (C) = Pi(C) + εi (C) ever, losses caused by extreme events not only include direct costs (physical damage, where: valued restoration costs, possessions, pro-
Pi (C) is the ratio of an asset’s price before duction means); but also indirect costs (eco- a flood to its price after having been dam- nomic interruption, environmental damage, aged in certain flood conditions (C—reflects cleaning, evacuation), and relief costs (food water depth, velocity etc., and indirectly the aid, heath care; sanitation) (Genovese, scale of damages to an asset) 2006). In such cases assumptions have to be
Di (C) is the ratio of actual damage to made. Clean-up costs can serve as an exam- potential damage ple. They are assessed in terms of hours or
εi (C) is a stochastic component days per household. In practice, there is a va- It is obvious that D(C) and P(C) ∈[0,1], riety of suggestions: 15–20 hours per house; by definition. The stochastic component 50–60 hours, or 5 person-days (McBean et should be heteroscedastic. This results al., 1986). The differences can be attributed from a high level of certainty in the evalu- to local conditions of a particular flood, but ation of fully and merely damaged by mar- also to imprecise assessment. ket participants. Moreover, for C which On the macro level, indirect losses (dis-
generates complete damage var(εi) tends to ruptions caused by extreme events) are often 0, and the same holds for an intact asset. If estimated as a fixed proportion of the direct
a market is efficient, the distribution of εi is ones. The study by Parker et al. (1987) shows symmetrical, which means that participants’ that indirect costs are overestimated due to in- misassessments offset each other to equilib- surance payments for business interruptions.
KKsisiąążżkka1.indba1.indb 4411 22008-06-26008-06-26 110:56:020:56:02 42 Piotr Matczak, Roman Mańczak and Zbigniew W. Kundzewicz
Smith (1994) recommends assessment of in- cept the compensation for the decrease in direct losses on the regional or national levels. quality of the environment; In this way a loss of e.g. a retailer is “compen- 2) the quasi-market, where the prices on sated” for by the increased sales of the busi- the real-estate market are indirect indica- ness outside the flooded area. This gives no tors of environmental quality; net loss on the regional and national level. 3) the hedonic price based on costs of travel to a national park for “consump- tion” of the recreational values of nature. ESTIMATION OF INTANGIBLE DAMAGE Assessment of the value of the envi- ronment is applied in various contexts: to Assessment of intangible damage poses measure the value of a particular species more problems than in the case of tangible (includingt those endangered by extinction); losses. Losses in terms of cultural heritage, particular examples of pristine nature (e.g. biodiversity, human health etc, caused by the Grand Canyon), etc. Despite several extreme weather events can be severe and drawbacks the methodology is well-devel- unquestionable, but expressing them in mon- oped and can be used for the assessment etary terms is difficult. The issue of whether of losses caused by extremes. intangible goods can be treated in economic Another area of intangible damages is terms has been discussed widely in the envi- human health. Hajat et al. (2003) point to ronmental protection context. Some authors several effects of floods on human health. argue that at least some features of nature These are: are of an intrinsic character, and are not 1) physical health effects (mortality; in- possible to measure. Others disagree, and juries; illnesses from flood-induced contam- propose methods by which to measure the ination of water supply; other flood-induced value of the environment, e.g. the value illnesses—respiratory problems and chronic of environmental services, cf. Costanza health effects); (b) mental effects—espe- et al. (1997). The same discussion also ap- cially post-traumatic stress disorder. plies to the assessment of intangible dam- Numerous suicides linked with floods age caused by extremes. First of all, the very (Czabański, 2005) constituted a special cat- definition of tangibility appears problem- egory of problems during and after the 1997 atical. For instance, Smith (1994) poses the flood in Poland. However, the health effects question of how to treat gardens. Garden- are not obvious and evidence from different ing equipment is possible to value, but the countries is contradictory. A study on the loss of plants and lawns is more difficult and 1997 flood in the Czech Republic reveals these can be treated as intangible. The issue fewer hospital admissions and no suicide at- here is whether a garden can be treated as tempts, something which is attributed to an a tangible or an intangible. increase in social cohesion in flood circum- Natural resources are more of an intan- stances. gible characteristic. Since extreme events Nevertheless, in many cases, human have effects on biological systems, they can health losses caused by extreme events are disturb the robustness of ecological systems: tremendous. They can be presented in quan- provoking the extinction of populations, titative terms, as the number of casualties a decrease in biodiversity, etc. The assess- (Jonkman and Kelman, 2005; Jonkman, ment of ecological (natural) values in mon- 2005), morbidity rate, etc. In this case also, etary terms is of great importance, but is the however, there are attempts to quantify the subject of discussion (e.g. Pearce and Turner, effects in monetary terms. Like with meas- 1990). Methods of assessment are based on: urement of the impact of environmental 1) questionnaires measuring Willing- pollution on human health, losses caused ness-To-Pay for improvement of the quality by extreme events can be measured via the of the environment or Willingness-To-Ac- human capital concept. The health costs are
KKsisiąążżkka1.indba1.indb 4422 22008-06-26008-06-26 110:56:030:56:03 Estimation of Damage Caused by Extreme Weather Events, with an Emphasis on Floods 43
Figure 2. Relationship between the ratio of material losses in million USD to number of deaths and GNP per capita, in USD Source: Kundzewicz, Takeuchi, 1999.
quantified as loss of quantity and quality losses per fatality can be as low as USD of labour force. The inability to work brings 21 000, while in developed countries they costs of treatment and costs of absence from can be up to USD 400 million, i.e. by four work. It can be measured in terms of costs orders of magnitude higher. If floods strike of person-days in a specific sector and loca- developed countries, they may cause major tion. The approach focuses on the impact material damage, but the death toll in de- of the deterioration in human health on the veloped countries is far lower than in the economy, emphasizing decreased productiv- developing ones, because the former typi- ity at the macroeconomic level. There are cally have in place an efficient flood pre- also attempts to assess the costs of human paredness system (including a forecast and life in monetary terms (insurance compen- warning system). sation), but obviously this provokes ethical discussion. There are other attempts at the scalar- CONCLUSIONS ization of flood-related indicators. Since the number of fatalities and level of materi- The estimation of losses caused by extreme al damage are the most meaningful indices, weather events, such as floods, is difficult. one may use the ratio of material losses (in In economic terms, extreme events disturb million USD) to the number of deaths. In the market prices system. Services offered simple words, the indicator measures mate- during the event can be priced above the rial losses per death. Fig. 2 illustrates how market process at the normal time. A dis- this indicator varies for extreme floods, as torted market is not the only problem. In a function of GNP per capita (in USD). As many cases the damaged goods are not pure expected, there is a general pattern to this market exchangeable goods. This poses ad- relationship (Kundzewicz, Takeuchi, 1997) ditional difficulties with assessment. Even in that the indicator attains higher values in the case of tangible and direct losses, as- for wealthy countries and lower ones for sessment is normally difficult. However, the less wealthy countries. For catastrophic scale of damage and the probable increase floods in developing countries, material in the number of extreme events call for
KKsisiąążżkka1.indba1.indb 4433 22008-06-26008-06-26 110:56:030:56:03 44 Piotr Matczak, Roman Mańczak and Zbigniew W. Kundzewicz
data allowing for the preparation and appli- Genovese, E. (2006), A Methodological Approach cation of mitigation and adaptation meas- to Land Use-Based Flood Damage Assessment ures. Notwithstanding their weaknesses, in Urban Areas: Prague Case Study. EC Joint the presented approaches to the costing Research Centre, Institute for Environment of losses on the basis of market value or and Sustainability, Ispra, Italy. restoration (replacement) value allow as- Gissing, A. and Blong, R. (2004), Accounting for sessments in economic terms to be made. Variability in Commercial Flood Damage Esti- An open question remains related to the mation, Australian Geographer, 35(2): 209–222. defining of the efficiency of data collection, Hakat, S., Ebi, K. L., Kovats, S., Menne, B., Ed- i.e. how to collect the most useful data in wards, S. and Haines, A. (2003), The Human given circumstances. Health Consequences of Flooding in Europe and the Implications for Public Health: a Re- view of Evidence, Applied Environmental Sci- REFERENCES ence and Public Health, 1 (1): 13–21. IPCC (2007), Summary for Policymakers, in Berz, G. (2001), Climatic change: Effects on and Solomon, S., Qin, D., Manning, M.,Chen, Z., Possible Responses by the Insurance Industry., Marquis, M., Averyt, K.B., Tignor, M. and in: Lozán, J. L., Graßl, H., Hupfer, P. (Eds.) Miller, H.L. (eds.), Climate Change 2007: The Climate of the 21st Century: Changes and Physical Science Basis, Contribution of Work- Risks. Wissenschaftliche Auswertungen (in ing Group I to the Fourth Assessment Report cooperation with GEO), Hamburg, Germany, of the Intergovernmental Panel on Climate 392–399. Change, Cambridge University Press, Cam- Centrum Informacyjne Rządu (1998), Powódź bridge, UK and New York, NY, USA. 1997—straty, szacunek skutków kataklizmu, Jonkman, S. N. (2005), Global Perspective on działania rządu na rzecz odbudowy terenów Loss of Human Life Caused by Floods, Natu- zniszczonych przez powódź [Flood 1997 ral Hazards 34: 151–75. —Losses, Estimation of the Effects of the Jonkman, S. N. and Kelman, I. (2005), An Analy- Disaster, the Government’s Actions for the sis of Causes and Circumstance of Flood Dis- Reconstruction of the Areas Affected by the aster Deaths, Disasters 29 (1): 75–97. Flood], Przegląd Rządowy, nr 4 (82). Kreibich, H., Thieken, A., Petrow, Th., Mueller, Changnon, S. D. (2003), Measures of Econom- M. and Merz, B. (2005), Flood Loss Reduc- ic Impacts of Weather Extremes, Bulletin tion of Private Households due to Building of the American Meteorological Society, 84: Precautionary Measures—Lessons Learned 1231–1235. from the Elbe Flood in August 2002, Natural Costanza, R., d’Arge, R., de Groot, R., Farber, S., Hazards and Earth System Sciences 5: 117–26. Graso, M., Hannon, B., Limburg, K., Naeem, Kundzewicz, Z. W., Takeuchi, K. (1999), Flood S., O’Neill, R. V., Paruelo, J., Raskin, R. G., Protection and Management: Quo Vadimus? Sutton, P. and van den Belt, M. (1997), The Hydrological Sciences Journal 44(3): 417–432. Value of the World’s Ecosystem Services and Lavalle, C., Barredo J.I., de Roo, A., Niemeyer, Natural Capital, Nature 387: 253–260. S., Miguel-Ayanz, J.S.M., Hiederer, R., Gen- Czabański, A. (2005), Rezygnacja z życia w obliczu ovese, E. and Camia A. (2005), Towards an klęski żywiołowej. Powódź 1997 roku w Polsce European Integrated Map of Risk from Weath- [Resignation from Life Facing Natural Disas- er Driven Events. EC Joint Research Centre, ter. The Flood 1997 in Poland]. Wydawnictwo Institute for Environment and Sustainability, Rys, Poznań, Poland. Ispra, Italy. Feyen, L., Dankers, R., Barredo, J. I., Kalas, M., Lugeri, N., Genovese, E., Lavalle, C. and de Roo, Bodis, K., de Roo, A. and Lavalle, C. (2006), A. (2006), Flood Risk in Europe: Analysis of Flood risk in Europe in a changing climate. EC Exposure in 13 Countries. EC Joint Research Joint Research Centre, Institute for Environ- Centre, Institute for Environment and Sus- ment and Sustainability, Ispra, Italy. tainability, Ispra, Italy.
KKsisiąążżkka1.indba1.indb 4444 22008-06-26008-06-26 110:56:040:56:04 Estimation of Damage Caused by Extreme Weather Events, with an Emphasis on Floods 45
McBean, E., Fortin, M. and Gorrie, J. (1986), Sargent, T. I. (2002), Rational Expectations, The A Critical Analysis of Residential Flood Concise Encyclopedia of Economics,
KKsisiąążżkka1.indba1.indb 4455 22008-06-26008-06-26 110:56:040:56:04 POLAND’S CLIMATE EXTREMES INDEX, 1951–2005
RAJMUND PRZYBYLAK, ZSUZSANNA VÍZI, ANDRZEJ ARAŹNY, MAREK KEJNA, RAFAŁ MASZEWSKI and JOANNA USCKA-KOWALKOWSKA Department of Climatology, Nicolaus Copernicus University, Gagarina 9, 87-100 Toruń, Poland E-mails: [email protected]; [email protected]; [email protected]; [email protected]; [email protected]; [email protected]
Abstract: The paper seeks to synthesise contemporary (1951–2005) trends regarding the occur- rence of extreme meteorological events in Poland using the complex Climate Extremes Index (CEI) proposed by Karl et al. (1996). Poland’s CEI Was the greatest in the 1990s. The trend noted for it in the period from 1951 to 2005 is an upward one, but does not achieve statistical significance. Similar tendencies for the index have been observed in the 20th century for the USA (1910–90), the Russian Federation (1950–96), and Central Europe (1951–2000).
Key words: climate extremes index (CEI), temperature, precipitation, moisture index, Poland.
INTRODUCTION 2004). On the other hand, our knowledge A common practice in describing the climate of extreme climatic events and relevant con- of a given area is to present characteristics temporary changes resembles our knowl- for the main meteorological variables (e.g. edge of such events and trends in other air temperature, precipitation, cloudiness, parts of the world in still being limited, not- humidity) on the basis of their average val- withstanding significant progress in recent ues. Consequently, climate change estimates years (e.g. Dubicki et al. 1999; Bogdanowicz are also mainly given in terms of such statis- et al. 2005). The reasons for this lack of re- tical characteristics. Climate extremes and search are complex, though the main reason the changes occurring therein in recent years seemingly delaying research in this direc- have been investigated more rarely (e.g. in tion (not only in Poland) is the significantly the reports of the Intergovernmental Panel more limited availability (or even absence) on Climate Change (IPCC): Houghton et al. of homogeneous data with daily resolu- 1990, 1995 and 2001 or Heino et al. 1999). tion in comparison with monthly or annual The history of mean climate charac- mean data. A second, also very important teristics for Poland and the changes that reason relates to the lack of a consensus have characterized them in recent decades as regards the definition of extreme events are quite well known (Dubicki et al. 1999; (Easterling et al. 2000). Moreover, the large Fortuniak et al. 2001; Żmudzka 2003; number of existing extreme phenomena, Degirmendžić et al. 2004; Kożuchowski which sometimes show different trends in
KKsisiąążżkka1.indba1.indb 4477 22008-06-26008-06-26 110:56:050:56:05 48 Rajmund Przybylak, Zsuzsanna Vizi, Andrzej Araźny et al.
recent time series, do not allow us to state 2006). It will thus be possible to compare unequivocally whether an overall change results. has characterized the extreme events in the given area in the observed record. As a re- DATA, AREA AND METHODS sult, reliable spatial averaging of the results The construction of the index requires mete- obtained over larger areas (e.g. continents orological data of a daily resolution. The var-
and especially the globe as a whole) is very iables taken account of are: maximum (Tmax)
difficult and may even be impossible. None- and minimum (Tmin) temperatures, amount theless, since model outputs simulate more of precipitation (P), and potential evapora- extreme events for future climates (Meehl tion (EP). In addition, the number of days et al. 2000) and suggest that they will lead with precipitation ≥ 0.1 mm has also been to greater environmental and societal vul- referred to. Due to the limited availability of nerability, we should concentrate our efforts such data for Polish meteorological stations, on learning more about climate extremes, NCEP/NCAR Reanalysis data from 20 grid frequencies and intensities. A greater points have been used (Kalnay et al. 1996; awareness of these research needs among Kistler et al. 2001; http://www.cdc.noaa.gov/ Polish scientists has led to projects involv- cdc/reanalysis/) (Fig. 1). Throughout the ing interdisciplinary cooperation within area of Poland the grid points occur at almost an integrated framework called “Extreme every two degrees of longitude (1°52’5’’) and Hydrological and Meteorological Events in two degrees of latitude (1°54’3’’). As a result, Poland (Estimation of Events and Impact each grid point represents an equal surface Forecasting for the Human Environment” area. These surface areas are reduced in supported financially by the State Com- line with the proportion lying inside the Pol- mittee for Scientific Research, Grant No. ish border. Similar reductions were made for PBZ-KBN-086/P04/2003. We hope that each indicator used to calculate the Climate implementation of this project will signifi- Extremes Index (CEI). Because EP values cantly improve our knowledge of changes are not available in the NCEP/NCAR Re- affecting different extreme meteorological, analysis, they were calculated using the for- hydrological and geomorphological events mula from Ivanov (1958): in Poland in the 20th century. The main task of the present paper is to EP= 0.0018 (25+T)2 (100–f) examine whether any increased incidence of climatic extremes is to be observed for Poland over recent decades. This issue is problematic because, as we mentioned ear- lier, there are a myriad different extreme phenomena (including a large number of different indicators and indices of climate extremes). Analysis of all of them is simply impossible, but the idea of a comprehensive and synthetic index, proposed by Karl et al. (1996) to describe climate extremity would seem to be a good solution, permitting us to estimate the existing tendencies more re- liably. To this end, we decided to calculate such an index for the area of Poland. Thus far this index (or modifications thereof) has been calculated for the USA (Karl et al. 1996), the Russian Federation (Gruza et al. Figure 1. Study area and location of grid points 1999), and Central Europe (Przybylak et al. from which data have been taken
KKsisiąążżkka1.indba1.indb 4488 22008-06-26008-06-26 110:56:050:56:05 Poland’s Climate Extremes Index, 1951–2005 49
To calculate relative humidity (f), the month separately, while for indicators 4 and variable which is needed to obtain the EP, 5, yearly sums were used. A particular year we used the daily mean values for specific at a given grid point was considered extreme humidity available in the NCEP/NCAR Re- if the number of extreme events was greater analysis data set. than 10%. As was mentioned earlier, the construc- tion of the Climate Extremes Index is based RESULTS AND DISCUSSION on the proposal by Karl et al. (1996), but we To begin by describing the behaviour of par- introduced certain minor modifications to ticular indicators in the study period, a marked
it. To calculate the third indicator, the vari- rise in Tmax is to be observed from the begin- able P-EP was used for the long-term esti- ning of the 1990s. This is very clear in terms of mation of moisture anomalies, instead of the both average yearly values and the number of Palmer Drought Severe Index (PDSI). In days exceeding commonly used threshold val- turn, in the fourth indicator, the threshold ues, i.e. 25°C (hot days), 30°C (very hot days), of extreme high precipitation was reduced and 35°C (extremely hot days). The last kind of from 50.8 mm (two inches) to 15 mm, which days occurred very rarely prior to 1985. In two is more appropriate for Polish conditions years (1992 and 1994) their areally averaged and for the kind of data used. Moreover, to number for Poland oscillated between 4 and calculate first and second indicators, daily 6. In the remaining years the number of these data have been used, instead of the monthly extremely hot days never exceed 2.
data used by Karl et al. (1996). Cold Tmax (areally averaged annual mean
The Climate Extremes Index is the an- number of days with Tmax < 1% and Tmax <10%) nual arithmetic average of the following five in Poland show downward (if not statistically indicators of the percentage of the study significant) trends (not shown). On the other area: hand, the trends for the areally averaged an-
1. The sum of the percentage area of Po- nual mean number of days with Tmax >90%
land with (a) Tmax considerably below normal and Tmax >99% (warm Tmax) are positive and
and (b) Tmax considerably above normal, statistically significant (Fig. 2). The last dec- 2. The sum of the percentage area of Po- ade of the 20th century was very exceptional in
land with (a) Tmin considerably below normal this regard. In many years in this time span the
and (b) Tmin considerably above normal, number of days exceeding these two thresh- 3. The sum of the percentage area of olds were greater than 50 and 6, respectively. Poland with (a) P-EP considerably below Similar results for data taken from stations normal and (b) P-EP considerably above in Pomerania (north-west Poland) have been normal, noted by Filipiak (2004). 4. Twice the value of the percentage area Generally, similar results can be seen in of Poland with a considerably greater than Fig. 3, which presents the percentage chang-
normal proportion of P derived from ex- es for the area of Poland with Tmax below treme (≥ 15 mm) 1-day P events, the 10th percentile and above the 90th per- 5. The sum of the percentage area of centile from 1951 to 2005. The latter shows Poland with (a) a considerably greater than a statistically significant increase for this normal number of days with precipitation characteristic (Fig. 3b) while the former and (b) a considerably greater than normal shows a non-significant downward trend number of days without precipitation. (Fig. 3a). It is evident that, since the begin-
The extreme conditions were defined as ning of the 1990s, the warm Tmax values, if the upper (considerably above normal) and they occur in a given year, are quite wide- lower (considerably below normal) tenth spread and mostly extend over the whole ter- percentile of the local 50-year (1951–2000) ritory of Poland (100%). On the other hand,
period of records. In the case of indicators 1, cold Tmax values in this time span relate only 2 and 3, thresholds were determined for each to isolated parts of Poland, even though they
KKsisiąążżkka1.indba1.indb 4499 22008-06-26008-06-26 110:56:060:56:06 50 Rajmund Przybylak, Zsuzsanna Vizi, Andrzej Araźny et al.
Figure 2. Areally averaged annual mean number of days for Poland with Tmax above the 90th percentile (a) and above the 99th percentile (b) from 1951–2005.
frequently covered the whole area of the ward trend, but are not statistically signifi- country in the 1950s and 1960s. cant (Fig. 5a). The opposite behaviour (i.e.
In the study period, Tmin shows sig- upward trend) is seen for warm Tmin (> 90th
nificantly more limited changes in Po- percentile). Like cold Tmin, warm Tmin also
land than Tmax. Non-significant changes in shows non-significant changes. In recent
both areally averaged Tmin (upward trend) years, however, these are widespread, if
and the number of frosty (Tmin<0°C), cold they occur in a particular year covering the
(Tmin<-10°C) and very cold (Tmin<-20°C) whole, or almost the whole, area of Poland days (not shown) have been noted. Statisti- (Fig. 5b). cally significant changes were observed only The observed trends in selected areally for the areally averaged annual number of averaged mean annual characteristics for
days with Tmin > 90% and Tmin > 99% in daily Tmax and Tmin for 1951–2005, as shown Poland. The marked rise in the number of and described above, are in line with the such days has occurred mainly in the last results presented by Niedźwiedź and Us- 10 years, in which counts frequently exceed- trnul (1994) and Brázdil et al. (1996), who ed 50 and 8, respectively (Fig. 4). The per- analysed monthly mean data from mete- centage changes in the area of Poland with orological stations for Poland (1951–1992)
cold Tmin (< 10th percentile) show a down- and for central and south-eastern Europe
KKsisiąążżkka1.indba1.indb 5500 22008-06-26008-06-26 110:56:060:56:06 Poland’s Climate Extremes Index, 1951–2005 51
Figure 3. Percentage changes in the area of Poland with Tmax below the 10th percentile (a) and above the 90th percentile (b) from 1951–2005.
(1951–1990), respectively. On the other of days with a moisture index (P-EP) < 1st hand, opposite tendencies (i.e. greater rises percentile and 10th percentile (extremely
for Tmin than Tmax) have been found by Wibig dry and very dry days, respectively). As and Głowicki (2002) for Poland for the pe- a result, a statistically significant downward riod 1951–1998. The results in these papers trend is observed. For example, prior to 1990 may have been influenced by the different the areally averaged number of extremely periods analysed and the different kinds of dry days oscillated mainly between 1 and 2, data used (reanalysis or station data); in the while from 1990 on their number was very case of papers using station data, different often greater than 5. On the other hand, the sets of data could also have influenced the areally averaged number of extremely wet results. Thus, more research is needed to days and very wet days (P-EP above 99% resolve this discrepancy. and 90% respectively) show a statistically The extreme moisture conditions in Po- significant downward trend during the study land changed significantly from the 1950s period. In the period 1951–1980 the number to the turn of the 20th/21st centuries. There of extremely wet days oscillated mainly be- was a marked increase in the last 15 years tween 3 and 6, while during the last 25 years in the areally averaged annual mean number the figure was generally from 1 to 3 days.
KKsisiąążżkka1.indba1.indb 5511 22008-06-26008-06-26 110:56:060:56:06 52 Rajmund Przybylak, Zsuzsanna Vizi, Andrzej Araźny et al.
Figure 4. Areally averaged annual mean number of days for Poland with Tmin above the 90th percentile (a) and above the 99th percentile (b) from 1951–2005.
The percentage changes in the area Fig. 6a) analysing the number of days with of Poland with P-EP >90th percentile P ≥ 10 mm from 1951 to 1996 based on sta- (Fig. 6b) and P-EP > 99th percentile show tion data, and also by Heino et al. (1999) for large statistically significant downward the period 1951–1995 (see their Fig. 6 –Pots- trends. The opposite behaviour is seen for dam and Łódź stations). Mean counts of extremely dry and very dry days (Fig. 6a), days with extreme 1-day P totals (≥ 15 mm) which reveal statistically significant upward in the period 1951–2005 oscillated from trends. In recent years, if those kind of days 8.74 days in 1980 to 1.23 days in 1993. How- occur in a particular year, they are wide- ever, in a particular grid box (20°37’30’’E, spread and cover the whole, or almost the 52°22’48’’N) the extreme 1-day P totals var- whole, area of Poland ied from 20 days (1952) to 1 day (in many The areally averaged annual mean years). On the other hand, the percentage number of days with extreme 1-day P to- changes in the area of Poland with a con- tals (≥ 15 mm) shows a significant down- siderably greater than normal proportion ward trend in Poland from 1951 to 2005 of P derived from extreme 1-day P events (Fig. 7a). Similar results (i.e. downward are not characterized by any significant trends, though not statistically significant) trends (see Fig. 7b). However, it is clear that were found by Frich et al. (2002, see their greater than normal percentage values oc-
KKsisiąążżkka1.indba1.indb 5522 22008-06-26008-06-26 110:56:070:56:07 Poland’s Climate Extremes Index, 1951–2005 53
Figure 5. Percentage changes in the area of Poland with Tmin below the 10th percentile (a) and above the 90th percentile (b) from 1951– 2005.
curred in the 1950s, the 1970s, and the last tion show a statistically significant increase 10 years. In four years (1952, 1956, 1975 and (Fig. 8a). Such situations did not occur at 1995) about half of Poland received a great- all in the 1950s, while they were very com- er than normal proportion of P in extreme mon at the turn of the 1960s and the 1970s, (≥ 15 mm) 1-day P events. the 1970s and 1980s and during the 1990s. The areally averaged annual mean num- This means, of course, that in these last bers of days with precipitation (≥ 0.1 mm) three time spans dry conditions were often and without precipitation in Poland show observed over more than half of Poland. In statistically significant changes from 1951 to two years (1972, 1997) such conditions even 2005 (not shown). The former show a down- covered more than 80% of the area. On the ward trend, the latter an upward one. In both other hand, the percentage changes in the cases, the changes in the number of days be- area of Poland with a considerably greater tween the 1950s and the period 1991–2005 than normal number of days with precipita- amounted to 50. tion (P≥ 0.1 mm) do not show a statistically The percentage changes in the area of significant downward trend (Fig. 8b). Such Poland with a considerably greater than days were observed over quite a large area normal number of days without precipita- of Poland (mainly from 10% to 40%) before
KKsisiąążżkka1.indba1.indb 5533 22008-06-26008-06-26 110:56:070:56:07 54 Rajmund Przybylak, Zsuzsanna Vizi, Andrzej Araźny et al.
Figure 6. Percentage changes in the area of Poland with P-EP below the 10th percentile (a) and above the 90th percentile (b) from 1951–2005.
1985, while such days only occurred subse- tions) in Poland occurred (in ascending or- quently in 2000 and 2003. Particularly ex- der of extremity) in 1984, 1974, 1973, 2001, ceptional was 1983, during which a consid- 1988, 1982, 1959, 1964, 1958 and 1967. erably greater than normal number of days Similar results for the USA have been with precipitation were noted over 89.37% obtained by Karl et al. (1996) for the period of the area of Poland. 1910–1990, for the Russian Federation by Poland’s CEI does not show any statis- Gruza et al. (1999) for the period from the tically significant rise from 1951 to 2005 1950s to 1996, and for central Europe by Przy- (Fig. 9). The greatest extremity of climate bylak et al. (2006) for the period 1951–2000. was observed during the 1990s and at the turn of the 1960s and 1970s. The 10 most extreme years in descending order were: CONCLUSIONS 1971, 1997, 1998, 1995, 2003, 1993, 1957, 1980, 1963 and 1956. On the other hand, the 1. In the 1990s, a frequency of occur-
10 years with the most limited climate ex- rence of daily Tmax considerably above nor- tremes (i.e. the most stable climatic condi- mal was evident. As a result, its frequency
KKsisiąążżkka1.indba1.indb 5544 22008-06-26008-06-26 110:56:080:56:08 Poland’s Climate Extremes Index, 1951–2005 55
Figure 7. (a) Areally averaged annual mean number of days with extreme 1-day P totals (≥ 15 mm) and (b) percentage changes in the area of Poland with a considerably greater than normal proportion of P derived from extreme 1-day P totals (≥ 15 mm) from 1951–2005.
during the period 1951–2005 shows a sta- changes in the area of Poland with P-EP con- tistically significant upward trend. Similar siderably below normal / considerably above
changes were also noted for daily Tmin. The normal show strong increases and decreas- percentage changes in the area of Poland es respectively, then achieving statistical
with Tmax below the 10th percentile and significance in both cases. above the 90th percentile show downward 3. The areally averaged frequency of oc- and upward trends, respectively. Statisti- currence of extreme 1-day P totals (≥ 15 mm) cally significant changes only occurred in in Poland shows a statistically significant
the case of Tmax exceeding the 90% thresh- downward trend. On the other hand, the
old. In turn, Tmin reveals non-significant decreasing percentage of Poland with a con- changes. siderably greater than normal proportion 2. The indicator of moisture (precipita- of P derived from extreme 1-day P events is tion minus evaporation, P-EP) shows mark- non-significant. edly lower values in the last two decades of 4. A considerably greater than normal the study period. As a result, the study period areally averaged mean number of days with is characterized by a downward statistically precipitation has shown a clear decrease significant trend. Trends for the percentage in Poland in the last two decades. In the
KKsisiąążżkka1.indba1.indb 5555 22008-06-26008-06-26 110:56:080:56:08 56 Rajmund Przybylak, Zsuzsanna Vizi, Andrzej Araźny et al.
Figure 8. Percentage changes in the area of Poland with a number of days without precipitation
(a) and with precipitation (P ≥ 0.1 mm) (b) having a frequency above the 90th percentile from 1951–2005.
Figure 9. Year-to-year course of the CEI for Poland from 1951–2005.
KKsisiąążżkka1.indba1.indb 5566 22008-06-26008-06-26 110:56:090:56:09 Poland’s Climate Extremes Index, 1951–2005 57
study period, a strong negative, statistically Zmiany i zmienność klimatu Polski: ich wpływ significant trend for such days was noted. On na gospodarkę, ekosystemy i człowieka [Chang- the other hand, the percentage changes in es and Variability of Poland’s Climate: their the area of Poland with this kind of data do Influence on Industry, Ecosystems and Man], not show a statistically significant downward Ogólnopolska Konferencja Naukowa, Łódź, tendency. 315pp. 5. Poland’s CEI was greatest in the Filipiak, J. (2004), Zmienność temperatury powie- 1990s. Its trend in the period from 1951 to trza na Wybrzeżu i Pojezierzu Pomorskim w dru- 2005 is upward, though not significantly so. giej połowie XX w. [Variability of Air Tempera- Similar results for the USA have been ob- ture on the Polish Coast of the Baltic Sea and tained by Karl et al. (1996) for the period Pomerania Lakeland in the Second Half of the 1910–1990, for the Russian Federation by 20th Century], Instytut Meteorologii i Gospo- Gruza et al. (1999) for the period from 1950s darki Wodnej (IMGW), Warszawa, 216pp. to 1996, and for central Europe by Przybylak Fortuniak, K., Kożuchowski, K. and Żmudzka, et al. (2006) for the period 1951–2000. E. (2001), Trendy i okresowość zmian tempe- ratury powietrza w Polsce w drugiej połowie XX wieku [Trends and Periodicity of Changes ACKNOWLEDGEMENTS in Air Temperature in Poland in the Second Half of the 20th century], Przegląd Geofizycz- The research in the present paper has been ny, 4: 283–303. carried out within the framework of the re- Frich, P., Alexander, L.V., Della-Marta, P., search project entitled “Extreme meteoro- Gleason, B., Haylock, M., Klein, A.M.G. logical and hydrological events in Poland”, and Peterson, T. (2002), Observed Coherent financed by the Ministry of Science and Changes in Climatic Extremes During the Higher Education of Poland (PBZ-KBN- Second Half of the Twentieth Century, Cli- 086/P04/2003). We would like to thank John mate Research, 19:193–212. Kearns for assistance with the English. Gruza, G., Rankova, E., Razuvaev, V. and Buly- gina, O. (1999), Indicators of Climate Change for the Russian Federation, Climate Change, REFERENCES 42: 219–242. Heino, R., Brázdil, R., Forland, E., Tuomenvirta, Bogdanowicz, E., Kossowska-Cezak, U. and H., Alexandersson, H., Beniston, M., Pfister, Szkutnicki, J. (eds.) (2005), Ekstremalne zja- C., Rebetez, M., Rosenhagen, G., Rösner, S. wiska hydrologiczne i meteorologiczne [Hy- and Wibig, J. (1999), Progress in the Study of drological and Meteorological Extreme Phe- Climatic Extremes in Northern and Central nomena], Polskie Towarzystwo Geofizyczne, Europe, Climatic Change, 42: 151–181. Instytut Meteorologii i Gospodarki Wodnej, Houghton, J. T., Jenkins, G. J. and Ephraums, J. J. Warszawa, 492pp. (eds.) (1990), Climate Change: The IPCC Brázdil, R. and 13 co-authors (1996), Trends of Scientific Assessment, Cambridge University Maximum and Minimum Daily Temperatures Press, Cambridge, 365pp. in Central and Southeastern Europe, Interna- Houghton, J. T., Meila Filho, L.G., Callander, B. A., tional Journal of Climatology, 16: 765–782. Harris, N., Kattenberg, A. and Maskell, Degirmendžić, J., Kożuchowski, K. and Żmudzka, K.(eds.) (1996), Climate Change 1995: The E. (2004), Changes of Air Temperature Science of Climate Change, Cambridge Uni- and Precipitation in Poland in the Period versity Press, Cambridge, 572pp. 1951–2000 and their Relationship to Atmos- Houghton, J.T., Ding, Y., Griggs, D.J., Noguer, pheric Circulation, International Journal of M., van der Linden, P.J., Dai, X., Maskell, Climatology, 24: 291–310. K. and Johnson, C.A. (eds.) (2001), Climate Dubicki, A.., Gutry-Korycka, M., Kożuchowski, Change 2001: The Scientific Basis, Cambridge K., Lorenc, H. and Starkel, L. (eds.) (1999), University Press, Cambridge, 881pp.
KKsisiąążżkka1.indba1.indb 5577 22008-06-26008-06-26 110:56:100:56:10 58 Rajmund Przybylak, Zsuzsanna Vizi, Andrzej Araźny et al.
Ivanov, N.N. (1958), Atmosfernoje uvlaznienie Meehl, G.A., Zwiers, F., Evans, J., Knutson, T., tropiceskich i sopredelnych stran zemnego Mearns, L. and Whetton, P. (2000), Trends szara [Atmospheric Moisture of Tropical in Extreme Weather and Climate Events: and Adjacent Areas of the Earth], ANSSSR, Issues related to Modeling Extremes in Pro- Moskwa, 311pp. jections of Future Climate Change, Bulletin Karl, T.R., Knight, R.W., Easterling, D.R. and of the American Meteorological Society, 81: Quayle, R.G. (1996), Indices of Climate 427–436. Change for the United States, Bulletin of Niedźwiedź, T. and Ustrnul, Z. (1994), Maximum the American Meteorological Society, 77: and Minimum Temperatures in Poland and 279–292. the Variability of Atmospheric Circulation, Kalnay, E., Kanamitsu, M., Kistler, R., Collins, W., in Brázdil, R and Kolář, M. (eds.), Contempo- Deaven, D., Gandin, L., Iredell, M., Saha, S., rary Climatology, Brno, pp. 420–425. White, G., Woollen, J., Zhu, Y., Chelli- Przybylak, R., Vizi, Z., Araźny, A., Kejna, M., ah, M, Ebisuzaki, W., Higgins, W., Jano- Maszewski, R. and Uscka-Kowalkowska, J. wiak, J., Mo, K.C., Ropelewski, C., Wang, J., (2006), Indeks ekstremalności klimatu Euro- Leetmaa, A., Reynolds, R., Jenne, R. and py Środkowej w okresie 1951–2000 [Central Joseph, D. (1996), The NCEP/NCAR 40-year Europe Climate Extremes Index ,1951–2000], Reanalysis Project, Bulletin of the American Dokumentacja Geograficzna 32, Instytut Geo- Meteorological Society, 77: 437–471. grafii i Przestrzennego Zagospodarowania Kistler, R. and 12 co-authors (2001), The NCEP- (IGiPZ), PAN: 240–248. NCAR 50-Year Reanalysis: Monthly Means Wibig, J. and Głowicki, B. (2002), Trends of Min- CD-ROM and documentation, Bulletin imum and Maximum Temperature in Poland, of the American Meteorological Society, 82: Climate Research, 20: 123–133. 247–268. Żmudzka, E. (2003), Wielkość zachmurzenia Kożuchowski, K. (ed.) (2004), Skala, uwarun- w Polsce w drugiej połowie XX wieku [Cloud- kowania i perspektywy współczesnych zmian iness in Poland in the Second Half of the klimatycznych w Polsce [Scale, Conditions 20th Century], Przegląd Geofizyczny, 3–4: and Perspectives of Contemporary Climatic 159–185. Changes in Poland], Wydawnictwo ‘Bibliote- ka’, Łódź, 169pp. Paper first received: February 2007 In final form: November 2007
KKsisiąążżkka1.indba1.indb 5588 22008-06-26008-06-26 110:56:100:56:10 VARIABILITY TO GLOBAL SOLAR RADIATION IN CENTRAL EUROPE DURING THE PERIOD 1951–2005 (ON THE BASIS OF DATA FROM NCEP/NCAR REANALYSIS PROJECT)
JOANNA USCKA-KOWALKOWSKA, RAJMUND PRZYBYLAK, ZSUZSANNA VÍZI, ANDRZEJ ARAŹNY, MAREK KEJNA, RAFAŁ MASZEWSKI Department of Climatology, Institute of Geography, Nicolaus Copernicus University, Gagarina 9, 87-100 Toruń E-mails: [email protected]; [email protected]; [email protected] [email protected]; [email protected]; [email protected]
Abstract: The paper presents the variability to Global Solar Radiation (GSR) in Central Europe in the period 1951–2005. The basic material comprises the data from the NCEP/NCAR reanaly- sis of 35 grid points. The research shows a statistically significant increase in GSR income in the entire study period in the research area. This increase, which started in the late 1980s, was also observed across Europe as a whole. One of the reasons for this change might be the decrease in pollutant emissions to the atmosphere. Moreover, an upward trend in the numbers of days with GSR over the 90th percentile is to be observed in the study period, while the number of days with GSR lower than the 10th percentile shows a negative trend. In both cases, the recorded trends are statistically significant at the 0.05 level.
Key Words: global solar radiation, Central Europe, NCEP/NCAR reanalysis data
INTRODUCTION ity, it is essential to find out why they occur and, most importantly, why their frequency Since solar radiation is the main source is on the increase, especially in recent years of energy for both weather and climatic proc- (Przybylak et al. 2006). esses, changes to it in terms of either size or spatial and temporal distribution, may re- sult in significant changes in weather and AIM OF THE RESEARCH, STUDY AREA climate patterns, which, in turn, influence AND SOURCE MATERIALS human life and activity. The most significant changes involve both the scale and frequency The aim of this paper is to present the chan- of extreme meteorological phenomena. As ges in the level of Global Solar Radiation extreme weather conditions exert a negative (GSR) in Central Europe in the period influence over ecosystems and human activ- 1951–2005. This has been done on the
KKsisiąążżkka1.indba1.indb 5599 22008-06-26008-06-26 110:56:100:56:10 60 Joanna Uscka-Kowalkowska, Rajmund Przybylak, Zsuzsanna Vizi, et al.
basis of data from the NCEP/NCAR re- lowest sums were recorded in 1960. The larg- analysis (Kalnay et al. 1996) provided by est input of solar energy was recorded at the the NOAA/OAR/ESRL PSD, Boulder, end of the research period, especially in the Colorado, USA, from their website at 1990s, with 1992 and 1994 being the most http://www.cdc.noaa.gov/. Mean daily val- significant. Thus, between 1951 and 2005 ues for the GSR, expressed in W·m-2, were a statistically significant increase in annual recorded for 35 grid points within the study global solar radiation within the entire study area. The grid points were located at almost area was observed (p = 0.05) (Fig. 1). every two degrees of longitude and latitude. According to studies undertaken in Eu- Monthly and annual totals for the GSR rope on the basis of data directly from ac- (MJ·m-2) were calculated for every grid point tinometric measurements, in the first part for both individual years and the entire ob- of the study period a decrease in GSR was servation period. The trends to the changes recorded. The examples include the results were studied in terms of annual sums as of the measurements at stations located in well as for the individual months. Moreover, the Baltic Sea area (Russak 1990). Between studied for every grid point was the number 1969 and 1986, the stations Toravere (Esto- of times mean daily values for the GSR nia), Helsinki, Stockholm, Kołobrzeg, Taas- (W·m-2) were lower than the threshold value trup and Saint Petersburg reported a lower- of the 10th percentile, and how many times ing of the value of GSR by 6.4% to 16.0%. they were higher than that of the 90th per- At Toravere, a negative trend of -6.8% was centile. also recorded over a longer research period (1955–1986). Russak (1990) argued that the decrease in the value of GSR at Toravere RESULTS throughout the 32-year study period may be connected with an increase in the amount Within the study period, the lowest annual of low level cloud by 11% (1964–1986) and totals for the GSR were recorded in the 1950s a decrease in transparency of the atmos- and 1960s; in as many as 22 grid points the phere of 3.7%. A decrease in GSR of 7.1%
Figure 1. Long-term course for sums of global solar radiation for selected grid points in Central Europe in the period 1951–2005
KKsisiąążżkka1.indba1.indb 6600 22008-06-26008-06-26 110:56:100:56:10 Variability to Global Solar Radiation in Central Europe During the Period 1951–2005 61
was also recorded for Moscow, between 1958 and 1993. This phenomenon is also connect- ed with greater cloudiness and lower trans- parency of the atmosphere (Abakumova et al. 1996). Abakumova et al. (1996) also show that the GSR in the European part of Rus- sia, the Baltic States, Belarus and Ukraine from 1960 to 1987 point to a downward trend equal to more than 2% per decade. Data gathered for Europe as a whole at- test to a negative trend for GSR to the surface between 1950 and 1985. Later this trend was reversed, an increase in GSR even being ob- served (Wild et al. 2005). The reversed trend for changes in GSR was in line with the trend towards changes in the transparency of the at- mosphere. The latter has been growing since the mid-1980s, something that might be con- nected with decreasing pollutant emissions due to the introduction of stricter rules on en- vironmental protection and changes of manu- facturing technologies. It was followed by lat- er economic transformation in the countries of Central and Eastern Europe, which often meant closedowns or changes in technologies for the factories most burdensome to the en- vironment (Wild et al. 2005). The reverse trend for the GSR data was also noted for Sweden in the period 1983–1997 (Persson 1999). Stations located in various parts of the country recorded an Figure 2. Annual course for daily sums of global upward trend (equal to 7.2 per decade) for solar radiation (mean—m, highest —max and the GSR. This increase was predominantly lowest—min) for the selected grid points in the the result of reduced cloudiness, mainly in period 1951–2005 the summer months (Persson 1999). Variability to GSR for the years 1961– 1995 was also studied for the area of Poland (Bogdańska and Podogrocki 2000). The An upward trend for total GSR was re- research was based on measurements from corded for both annual values and for indi- an actinometric network of 7 stations of the vidual months (Table 1). A negative slope to Institute of Meteorology and Water Man- the linear trend for monthly GSR totals was agement (IMGW) located across Poland. In- only recorded for one grid point in March, creased GSR input was recorded for 5 of the six grid points in June and one grid point in stations; though for only one of them (War- September. This means that the magnitude saw) was the trend statistically significant. of incoming GSR in these months decreased The two remaining stations (Suwałki and over the 55-year study period. In none of the Zakopane) recorded decreasing values for above cases, however, was the decrease sta- GSR, though this tendency did not achieve tistically significant. In all other cases an statistical significance (Bogdańska and Po- increase in incoming GSR was observed, dogrocki 2000). mostly statistically significant. The most
KKsisiąążżkka1.indba1.indb 6611 22008-06-26008-06-26 110:56:110:56:11 Table 1. Trend coefficients (MJ·m-2 per month or year) for monthly and annual totals of global solar radiation for individual grids in Central Europe in the period 1951–20051
Grid Months po- Year ints2 J F MA MJ J A S O N D
1 0.144 0.215 0.422 0.855 1.550 1.612 1.122 0.795 0.329 0.061 0.119 0.089 7.315
2 0.082 0.158 0.213 0.749 1.314 1.605 1.046 1.078 0.369 0.089 0.093 0.053 6.850
3 0.039 0.103 0.083 0.429 0.661 0.532 0.539 0.660 0.175 0.060 0.088 0.053 3.421
4 0.025 0.076 0.068 0.297 0.374 -0.006 0.378 0.427 0.113 0.072 0.076 0.043 1.945
5 0.040 0.077 0.156 0.326 0.405 -0.094 0.507 0.358 0.169 0.122 0.057 0.024 2.146
6 0.149 0.206 0.355 0.665 1.437 1.644 1.185 0.757 0.295 0.045 0.115 0.087 6.940
7 0.091 0.155 0.226 0.736 1.296 1.603 1.027 0.971 0.337 0.097 0.106 0.072 6.717
8 0.042 0.114 0.107 0.525 0.725 0.619 0.583 0.650 0.203 0.068 0.100 0.071 3.809
9 0.023 0.081 0.083 0.439 0.482 0.089 0.446 0.453 0.151 0.083 0.084 0.055 2.470
10 0.034 0.058 0.142 0.458 0.517 -0.054 0.562 0.362 0.174 0.140 0.060 0.026 2.477
11 0.142 0.182 0.282 0.482 1.305 1.695 1.173 0.647 0.235 0.039 0.128 0.083 6.391
12 0.100 0.158 0.233 0.714 1.275 1.474 1.005 0.853 0.293 0.111 0.139 0.097 6.454
13 0.051 0.123 0.126 0.631 0.823 0.753 0.679 0.673 0.281 0.097 0.124 0.092 4.453
14 0.028 0.094 0.146 0.617 0.608 0.252 0.588 0.509 0.233 0.121 0.097 0.068 3.361
15 0.027 0.071 0.279 0.661 0.590 -0.054 0.691 0.363 0.153 0.176 0.060 0.026 3.043
16 0.153 0.250 0.314 0.431 1.184 1.588 0.953 0.468 0.111 0.039 0.167 0.112 5.770
17 0.118 0.171 0.190 0.624 1.166 1.380 0.844 0.696 0.238 0.112 0.166 0.114 5.819
18 0.070 0.130 0.135 0.676 1.018 0.733 0.752 0.685 0.278 0.127 0.150 0.093 4.848
19 0.037 0.094 0.164 0.699 0.838 0.312 0.678 0.576 0.248 0.134 0.113 0.063 3.955
20 0.017 0.062 0.268 0.695 0.633 0.085 0.616 0.387 0.157 0.134 0.054 0.024 3.133
21 0.163 0.325 0.338 0.364 1.041 1.411 0.718 0.360 0.028 0.042 0.184 0.136 5.111
22 0.130 0.204 0.189 0.522 1.048 1.212 0.668 0.599 0.185 0.109 0.171 0.131 5.169
23 0.090 0.147 0.156 0.651 1.094 0.674 0.723 0.677 0.270 0.153 0.159 0.098 4.891
24 0.055 0.102 0.167 0.692 0.948 0.313 0.693 0.612 0.284 0.151 0.123 0.067 4.207
25 0.026 0.068 0.218 0.655 0.643 0.102 0.590 0.426 0.234 0.108 0.066 0.040 3.175
26 0.172 0.408 0.356 0.281 0.880 1.166 0.469 0.321 -0.014 0.049 0.179 0.155 4.422
27 0.137 0.256 0.230 0.407 0.919 0.973 0.474 0.562 0.134 0.101 0.154 0.147 4.494
28 0.109 0.175 0.190 0.556 1.053 0.579 0.589 0.650 0.257 0.175 0.149 0.106 4.589
29 0.082 0.119 0.156 0.595 0.939 0.255 0.633 0.617 0.339 0.170 0.129 0.081 4.115
30 0.054 0.087 0.131 0.540 0.620 -0.004 0.613 0.483 0.383 0.094 0.095 0.073 3.169
31 0.163 0.363 0.355 0.262 0.948 1.092 0.394 0.400 0.033 0.074 0.179 0.162 4.425
32 0.136 0.260 0.269 0.241 0.896 1.035 0.456 0.672 0.177 0.102 0.150 0.132 4.525
33 0.093 0.157 0.156 0.437 0.984 0.600 0.592 0.712 0.240 0.161 0.139 0.094 4.366
34 0.055 0.082 0.063 0.494 0.924 0.236 0.635 0.610 0.314 0.173 0.122 0.074 3.781
35 0.021 0.034 -0.007 0.431 0.739 -0.059 0.600 0.390 0.392 0.145 0.097 0.074 2.857
1 values statistically significant at the 0.05 level are given in bold 2 grid points (numbering was made by columns from N to S, begining at the westernmost column, see also Figs. 5–10)
KKsisiąążżkka1.indba1.indb 6622 22008-06-26008-06-26 110:56:110:56:11 Variability to Global Solar Radiation in Central Europe During the Period 1951–2005 63
distinguished were the months at the turn Greater diversification in the course for of autumn and winter (November and De- annual GSR was visible when daily totals cember) and in spring (April and May). At were taken into consideration. The annual that time, an increase in GSR input signifi- course for GSR by daily totals (mean, high- cant at the 0.05 level was recorded at all, or est and lowest) for three selected grids is pre- nearly all, grid points. The smallest number sented in Fig. 2. The greatest variability over of grid points in which the increase in GSR the entire 55-year period has been found for input to the Earth’s surface was significant the annual course in the case of lowest daily characterised October and March. sums. Much lower diversity was noted for the The mean course for annual GSR courses of both mean and highest totals. The at every grid point reveals the highest val- reason for greater variability in the annual ues in June, and the lowest in December. course of lowest daily GSR totals as com- Of course, this is mainly connected to the pared with highest is probably cloudiness. height of the Sun above the horizon and The highest daily sums were probably noted the values obtained for cloudiness in these under a cloudless sky, while the lowest ones months. For all the grid points mean month- characterised an overcast sky. Optical con- ly values for GSR ranged from 38.8 MJ·m-2 ditions of the atmosphere are more differen- in December to 534.1 MJ·m-2 in June (Ta- tiate under an overcast sky than a cloudless ble 2). On average, over the entire research sky. As a result, the lowest daily sums for period it was spring that showed the lowest GSR show greater variability in their annual variability for GSR, the highest being noted course than the highest ones. in winter. Considering individual months, Changes in the level of GSR in Decem- the lowest value for the variability coeffi- ber, i.e. the month of the lowest mean sums cient was recorded in August (3.9%), the for GSR, and in June, i.e. the month with the highest in the winter months—December highest sums on average, are presented in and February (5.8%). Fig. 3 and 4. The highest mean sums of solar
Table 2. Statistical characteristics of areally averaged sums for global solar radiation (MJ·m-2) in Central Europe in the period 1951–2005
Parameter J F M A M J J A S O N D Year
Upper 55.50 105.60 227.10 354.90 507.80 548.10 525.70 403.30 256.00 145.60 64.40 40.40 3202.40 quartile Lower 51.80 98.00 212.60 333.10 469.40 520.90 492.40 380.70 241.30 134.80 59.30 37.30 3071.80 quartile Highest 60.40 117.50 243.90 381.50 548.80 590.30 601.50 421.50 270.00 157.10 69.00 42.70 3354.50 mean value Lowest 47.50 88.10 200.70 312.20 445.10 483.10 461.30 355.00 226.40 122.80 54.70 33.30 2965.30 mean value Mean 53.80 102.20 220.60 346.40 491.40 534.10 510.00 391.50 248.70 140.20 62.00 38.80 3139.70
Skewness -0.30 0.00 0.30 0.20 0.10 0.20 0.60 -0.20 0.20 -0.10 -0.10 -0.60 0.40
Kurtosis -0.22 0.63 0.11 -0.32 -0.55 -0.08 0.95 -0.16 -0.65 -0.21 -0.63 -0.22 -0.36
Variance 7.73 34.96 83.24 240.46 563.50 585.75 774.05 229.45 119.82 57.28 11.49 5.01 8429.50
Standard 2.78 5.91 9.12 15.51 23.74 24.20 27.82 15.15 10.95 7.57 3.39 2.24 91.81 deviation Variability 5.20 5.80 4.10 4.50 4.80 4.50 5.50 3.90 4.40 5.40 5.50 5.80 2.90 coefficient (%)
KKsisiąążżkka1.indba1.indb 6633 22008-06-26008-06-26 110:56:120:56:12 64 Joanna Uscka-Kowalkowska, Rajmund Przybylak, Zsuzsanna Vizi, et al.
Figure 3. Long-term course for areally averaged sums of global solar radiation in December in Central Europe in the period 1951–2005
Figure 4. Long-term course for areally averaged sums of global solar radiation in June in Central Europe in the period 1951–2005
radiation in December was recorded in 1992 spatial distributions for the linear trend and equalled 42.7 MJ·m-2; similar values coefficient for the number of days of that (42.0 MJ·m-2) being recorded in 1968 and kind are presented in Fig. 5 and 6 respec- 1991. The lowest values were noted in the tively. 1950s, with especially low incoming GSR in In the first case, the values for the linear 1954 (33.3 MJ·m-2), as well as in 1959, 1958 trend coefficient range from -0.60 to -0.15, and 1955. In June the situation was quite which means the number of days with GSR similar; the highest mean sums for GSR not exceeding the threshold value of the were recorded in 1992 (590.3 MJ·m-2) and 10th percentile decreased over the entire 1994 (588.2 MJ·m-2), while the lowest came study area (Fig. 5). The greatest decrease at the beginning of the research period, es- is recorded in the north- western part of the pecially in 1956 (483.1 MJ·m-2). study area, i.e. over southern Scandinavia During the research period there was (-0.60), while the smallest is in the south a decrease in the number of days of mean of the study area (-0.15). In Poland the larg- GSR (W·m-2) not exceeding the value est rate of change (-0.53) is recorded in the of the 10th percentile, and an increase in Bay of Gdańsk area (the grid point coordi- the number of days which exceeded the nates of ϕ=54.29°N, λ=18.75°E). 90th percentile. In both cases these trends In the latter case, however, the values for are statistically significant at p < 0.05. The the linear trend coefficient range from 0.27
KKsisiąążżkka1.indba1.indb 6644 22008-06-26008-06-26 110:56:120:56:12 Variability to Global Solar Radiation in Central Europe During the Period 1951–2005 65
Figure 5. Trend coefficients (days·year-1) Figure 6. Trend coefficients (days·year-1) for the number of cases of mean daily solar for the number of cases of mean daily solar radiation (W·m-2) not exceeding the 10th radiation (W·m-2) exceeding the 90th percentile in Central Europe percentile in Central Europe in the period 1951–2005 in the period 1951–2005
to 0.85 days a year (Fig. 6). The most rapid sula, while in Poland it is its north-western changes are recorded for the north-western part. The above contention finds support parts of the study area, i.e. in the southern in the observed decrease in the number part of the Scandinavian Peninsula, while of days on which the 10th percentile was not the slowest characterise the south-western exceeded, and the increase in the number part. In Poland, the highest values for the of days on which the 90th percentile was. On linear trend coefficient for the studied vari- the other hand, in the southern and south- able are recorded in the central part of the western part of the research area, the trend country and in the eastern section of the coefficients show the most limited changes Baltic coast. in GSR inputs. Distinct trends to changes are also re- In December (Fig. 9 and 10), the area corded in individual months. The maps with the most marked decrease in the number (Fig. 7, 8, 9 and 10) present the trends to of days with solar radiation not exceeding the changes in the number of days on which 10th percentile threshold and exceeding the GSR did not exceed the threshold value 90th percentile was in the north-east. In Po- of the 10th percentile, and those on which land, for the 10th percentile this is the area GSR exceeded the 90th percentile in the located north of: ϕ=52.38°N, λ=18.75°E and months of both the highest (June) and the ϕ=52.38°N, λ=22.50°E, while in the case lowest (December) levels of incoming solar of the 90th percentile it is the area of the Bay radiation. of Gdańsk and the western part of the Ma- In June (Fig. 7 and 8) the area of the zurian Lakeland. In December, the smallest largest increase in incoming GSR is the changes in incoming GSR in the study area north-western part of the study area, i.e. the were recorded in its southern and south- southern edge of the Scandinavian Penin- western parts.
KKsisiąążżkka1.indba1.indb 6655 22008-06-26008-06-26 110:56:130:56:13 66 Joanna Uscka-Kowalkowska, Rajmund Przybylak, Zsuzsanna Vizi, et al.
Figure 7. Trend coefficients (days·month-1) Figure 8. Trend coefficients (days·month-1) for the number of cases of mean daily solar for the number of cases of mean daily solar radiation (W·m-2) not exceeding the 10th radiation (W·m-2) exceeding the 90th percentile in June in Central Europe percentile in June in Central Europe in the period 1951–2005 in the period 1951–2005
Figure 9. Trend coefficients (days·month-1) Figure 10. Trend coefficients (days·month-1) for for the number of cases of mean daily solar the number of cases of mean daily solar radiation (W·m-2) not exceeding the 10th radiation (W·m-2) exceeding the 90th percentile in December in Central Europe percentile in December in Central Europe in the period 1951–2005 in the period 1951–2005
KKsisiąążżkka1.indba1.indb 6666 22008-06-26008-06-26 110:56:170:56:17 Variability to Global Solar Radiation in Central Europe During the Period 1951–2005 67
DISCUSSION OF THE RESULTS significant at the same level in September. AND CONCLUDING REMARKS Mean annual values attest to an increase in GSR, but this is not statistically significant The research reveals an increase in incom- (Bogdańska and Podogrocki 2000). As far ing GSR in the study area between 1951 and as the grid point selected for comparison 2005. It must be stressed, however, that the for the period 1961–1995 is concerned, an study is based on NCEP/NCAR reanalysis increase statistically significant at the 0.05 data. Kalnay et al. (1996) used a numeri- level was also recorded in May, as well as in cal model for the calculation of these data. November and December. A decrease in ra- Thus the results presented here must be diation, similarly to Mikołajki, was recorded treated with caution. The comparison un- in September, though it was not statistically dertaken with data from selected meteoro- significant. The mean annual values show logical stations operating in Poland shows a statistically significant increase. that the NCEP/NCAR reanalysis data need correcting. Actinometric stations are often located in cities, while data from the ACKNOWLEDGEMENTS reanalysis refer to larger areas of various level of human transformation. Thus, the The research in the present paper has been station selected for comparison should be carried out within the framework of the re- located where the influence of human activ- search project entitled “Extreme meteoro- ity on the atmosphere is relatively low. Such logical and hydrological events in Poland”, a comparison was made for the IMGW ac- financed by the Ministry of Science and tinometric station in Mikołajki (ϕ=53.78°N, Higher Education of Poland (PBZ-KBN- λ=21.58°E) and the grid point coordinates 086/P04/2003). of ϕ=54.29°N, λ=22.50°E. The available data for 1961–1995 (Bogdańska and Podo- grocki 2000) for the individual months and REFERENCES for the year were used to compare them with the mean sums for GSR for the selected sta- Abakumova, G.M., Feigelson, E.M., Russak, V., tion and grid point. According to the results Stadnik, V.V. (1996), Evaluation of long-term of the comparison for the winter months changes in radiation, cloudiness, and surface (December and January) and for the end temperature on the territory of the former of autumn (November) the reanalysis data Soviet Union, Journal of Climate V 9: 1319– are overstated (by 9.6, 3.1 and 7.8% respec- 1327. tively). For the other months, however, the Bogdańska, B., Podogrocki, J. (2000), Zmienność reanalysis data are understated by a few całkowitego promieniowania słonecznego na to a dozen per cent (from 6.0% in April to obszarze Polski w okresie 1961–1995 [Variabil- 17.8% in September), excluding August for ity of global solar radiation in Poland in the which they are understated by 23.3%. In the period 1961–1995], Materiały Badawcze: Seria case of the reanalysis data, the mean annual Meteorologia 30, Instytut Meteorologii i Go- sum of GSR incoming to the Earth’s surface spodarki Wodnej, Warszawa. is understated by 13% if compared with the Kalnay, E., Kanamitsu, M., Kistler, R., Collins, measurement data from the IMGW network W., Deaven, D., Gandin, L., Iredell, M., Saha, for the research period (1961–1995). These S., White, G., Woollen, J., Zhu, Y., Chelliah, differences do not relate solely to mean val- M, Ebisuzaki, W., Higgins, W., Janowiak, J., ues; also referring to trends for changes in Mo, K.C., Ropelewski, C., Wang, J., Leetmaa, the amount of radiation. GSR measured A., Reynolds, R., Jenne, R., Joseph D. (1996), at the station in Mikołajki shows an increase The NCEP/NCAR 40-year Reanalysis statistically significant at the 0.05 level in project, Bulletin of the American Meteorologi- May, while yielding decrease statistically cal Society, 77: 437–471.
KKsisiąążżkka1.indba1.indb 6677 22008-06-26008-06-26 110:56:250:56:25 68 Joanna Uscka-Kowalkowska, Rajmund Przybylak, Zsuzsanna Vizi, et al.
Persson, T. (1999), Solar radiation climate in Swe- Russak, V. (1990), Trends of solar radiation, den, Physics and Chemistry of the Earth (B), cloudiness and atmospheric transparency 24: 275–279. during recent decades in Estonia, Tellus 42B: Przybylak, R., Vizi, Z., Araźny, A., Kejna, M., 206–210. Maszewski, R., Uscka-Kowalkowska, J. Wild, M., Gilgen, H., Roesch, A., Ohmura, A, (2006), Indeks ekstremalności klimatu Eu- Long, Ch. N., Dutton, E. G., Forgan, B., Kal- ropy Środkowej w okresie 1951–2000 [Index lis, A., Russak, V., Tsvetkov, A. (2005), From of climatic extremes in Central Europe in dimmining to brightening: decadal changes the period 1951–2000], in Gierszewski P., in solar radiation at Earth’s surface, Science, Karasiewicz M.T. (eds.), Idee i praktyczny Vol 308: 847–850. uniwersalizm geografii Geografia Fizycz- na, Dokumentacja Geograficzna 32, Instytut Paper first received: March 2007 Geografii i Przestrzennego Zagospodarowa- In final form: November 2007 nia (IGiPZ), PAN, Warszawa: 240–248.
KKsisiąążżkka1.indba1.indb 6688 22008-06-26008-06-26 110:56:260:56:26 MEAN AND EXTREME WIND VELOCITIES IN CENTRAL EUROPE 1951–2005 (ON THE BASIS OF DATA FROM NCEP/NCAR REANALYSIS PROJECT)
ANDRZEJ ARAŹNY, RAJMUND PRZYBYLAK, ZSUZSANNA VÍZI, MAREK KEJNA, RAFAŁ MASZEWSKI and JOANNA USCKA-KOWALKOWSKA Department of Climatology, Nicolaus Copernicus University, Gagarina 9, 87-100 Toruń, Poland E-mails: [email protected]; [email protected]; [email protected]; [email protected]; [email protected]; [email protected]
Abstract: The paper presents the results of research on variability of wind speed in Central Europe between 1951 and 2005. According to NCEP/NCAR reanalysis data from 35 grids, Cen- tral Europe has witnessed increases in mean wind speed as well as in the number of days with strong wind, both statistically significant. This attests to the fact that the number of extreme phenomena connected with high wind velocities has increased recently.
Key words: wind speed, NCEP/NCAR reanalysis, Central Europe.
INTRODUCTION events in other elements of the environment, be these hydrological or geomorphological, Extreme natural (inter alia climatic) phe- for example (Starkel 2003). nomena bringing both material and non- Recently, interest in high wind veloci- material damage have always been a part of ties in Europe, and the effects thereof, has human life. Initially, people were effectively grown significantly among geographers. powerless in the face of weather phenom- This is mainly in anticipation of an increase ena. However, as great material and human in strong winds (Houghton et al. (eds.) losses ensue, extreme meteorological phe- 2001), and has been reflected in more and nomena have been assuming importance more careful analysis showing a significant as issues for policy makers (Houghton et increases in wind speeds throughout Eu- al. (eds.) 2001). The last decade of the 20th rope. Thanks to its accessibility on the In- century brought a significant increase in ternet, most studies have been based around climatic extremes (Katz and Brown 1992; reanalysis of NCEP/NCAR data for selected Karl et al. 1999; Houghton et al. (eds.) 2001; areas of Europe at various isobar levels (e.g.: Przybylak et al. 2006). Extreme meteoro- Frank 2001, Larsén et al. 2006, Pryor et al. logical phenomena often result in extreme 2005, Marosz 2005, Pryor et al. 2006).
KKsisiąążżkka1.indba1.indb 6699 22008-06-26008-06-26 110:56:260:56:26 70 Andrzej Araźny, Rajmund Przybylak, Zsuzsanna Vízi et al.
Nowadays wind energy also arouses re- 90th and 99th percentiles were exceeded. newed interest, being one of the oldest re- Additionally, the spatial distribution of wind newable energy sources to be exploited (e.g. velocities in Poland based on meteorological Lorenc 1996, Soliński 1999, Pudlik 2003). stations (Lorenc 1996, 2005) was compared The authorities of the European Union are to that based on the NCEP/NCAR reanaly- assuming a steady increase in the role of re- sis data. newable energy sources play in total energy production. By 2020 about 20% of the en- ergy used in the European Union should RESULTS AND DISCUSSION be coming from renewable energy sources, which are theoretically eternal. As the obtained results show, there are marked spatial and temporal differences to mean annual wind speed in Central Europe. AIM, DATA, AREA AND METHODS Within the studied area this decreases from the south to the north-east (Figs. 1A-D). The aim of this study was to analyse mean Such spatial diversity to wind speed over and extreme wind velocities in Central Eu- Central Europe is connected with orogra- rope between 1951 and 2005. Due to a lack phy. Recorded wind velocities are higher of complete data from meteorological sta- where altitude is greater. They are about tions in the area, the NCEP/NCAR data twice as high in the Carpathians as in the available at www.cdc.noaa.gov were analysed lowlands and lakelands and on the Baltic (Kalnay at al. 1996). The collected data for coast. For most of the area the mean annual the years 1951–2005 include mean daily val- value for wind speed is of 2.6–3.6 m·s-1. The ues for both zonal (U) and meridional (V) highest mean annual values were recorded components of the wind velocity vector at on mountain peaks, i.e. in the Carpathians a height 10 m above ground level (a.g.l.), and Sudetan Mountains (over 5 m·s-1), cf. these being used to calculate mean values the lowest values in the lakelands of north- for wind speed. The data from 35 grids in east Poland and Lithuania (about 2.6 m·s-1) Central Europe were used in the research. (Fig. 1C). This area is covered with grids at intervals of Mean annual wind speed for individual almost two degrees longitude and latitude. years deviates markedly from the long-term The way every four grids are located leaves mean. In the entire analyzed period the the area between them of the same size (see lowest mean annual value for wind speed dots in Fig. 1, 3 or 6). (3.1 m·s-1) in Central Europe (area mean of The values calculated in this study in- all 35 grids) was recorded in 1952 and 1963, clude: monthly and annual mean values for cf. the highest (3.7 m·s-1) in 1983 (Fig. 2). wind speed, standard deviations and linear Wind velocities much higher than the long- trends. Maximum mean wind velocities for term mean values for the entire period were individual months and the entire year were predominantly recorded between 1983 and obtained. The number of days with strong 1998. Across Central Europe mean annual (> 8 m·s-1) wind was also estimated. Moreo- values of wind speed range from 2.2 m·s-1 ver, values for 1st, 10th, 90th and 99th per- (in 1960 in the western part of Belarus and centiles were calculated for the individual the eastern part of Poland’s Mazovian Low- grids. Also estimated was the number of land), to 6.7 m·s-1 (in 1983 over the western times the 1st and 10th percentiles were not part of the Carpathians). Between 1951 and reached, and the frequency with which the 2005, mean wind speed in Central Europe
KKsisiąążżkka1.indba1.indb 7700 22008-06-26008-06-26 110:56:260:56:26 Mean and Extreme Wind Velocities in Central Europe 1951–2005 71
Figure 1. Mean values for wind speed (m·s-1) in January (A), June (B) and annually (C), as well as maximum daily mean values for wind speed (D) in Central Europe, between 1951 and 2005
from all 35 grids grew by 0.03 m·s-1·10 years-1 rope was that recorded in the western (Fig. 2). This increase was statistically sig- part of Germany and the Czech Republic, nificant at α = 0.05. Trend coefficients for where the mean wind speed grew by over the mean annual values for wind speed in 0.06 m·s-1·10 years-1. the individual grids are positive, excluding The highest mean annual wind speed in two located in northern Lithuania (Fig. 3B). the entire studied area was recorded in Jan- The highest rate of change in Central Eu- uary 1993 on the top parts of the Western
KKsisiąążżkka1.indba1.indb 7711 22008-06-26008-06-26 110:56:260:56:26 72 Andrzej Araźny, Rajmund Przybylak, Zsuzsanna Vízi et al.
Figure 2. Course for mean annual wind speed (m·s-1) in Central Europe between 1951 and 2005
Figure 3. Standard deviation of mean daily wind speed (m·s-1) (A) as well as values of the linear trend coefficient for mean annual wind speed (m·s-1·10 years-1) (B) in Central Europe between 1951 and 2005
Carpathians (10.4 m·s-1), while the lowest They are the result of large horizontal gra- was the 1.4 m·s-1 in June 1999 and July 1988 dients in atmospheric pressure, frequent in the eastern part of the Polish Lowland, at that time of the year. During these as well as in August 1967 above the northern months mean wind speed over most of Cen- part of the Mazurian Lakeland. tral Europe varied between 3 and 4 m·s-1 The yearly course is characterised by (Fig. 1A). Pryor et al. (2005) obtained simi- highest mean wind velocities in the win- lar values for this area over a shorter period ter months; for most of the grids this was (1953–2001), on the basis of data from the January or, sporadically, December (Fig. 4). NCEP/NCAR reanalysis. In the Carpathi-
KKsisiąążżkka1.indba1.indb 7722 22008-06-26008-06-26 110:56:270:56:27 Mean and Extreme Wind Velocities in Central Europe 1951–2005 73
ans, wind speed during the winter months ues for standard deviation from the daily exceeds 7 m·s-1. In January (the month mean values for wind speed over the entire of highest mean wind speed), the highest study area range from 1.2 m·s-1 (over the mean values of wind speed anywhere in north-eastern part of the analysed area) to the area were recorded in 1983 and 1993 3.0 m·s-1 (over the southern part of Central (5.8 and 5.7 m·s-1 respectively), while the Europe) (Fig. 3A). The distribution of the lowest (3.2 m·s-1) characterized 1963 and highest mean daily values for wind speed 1973. Between 1951 and 2005, Central Eu- is similar to that for standard deviation as rope experienced a statistically significant regards wind speed (Fig. 1D). The highest (α = 0.05) increase in mean wind speed in value (19.6 m·s-1) was recorded on 22 Janu- January (0.08 m·s-1·10 years-1). ary 1993 in the western part of the Car-
Figure 4. Annual course for wind speed values (m·s-1) in Central Europe between 1951 and 2005
The lowest mean wind velocities were pathians in Slovakia. In most of the ana- recorded in summer, with the minimum in lysed area (24 out of 35 grids, excluding August (Fig. 4). The analysis of August (as the north-eastern part of the studied area) the month of lowest mean wind speed) over the highest values for mean daily wind ve- the entire area (35 grids) shows that wind locities exceeded 10 m·s-1. According to velocities were lowest in 1997 (at 2.2 m·s-1), the analysis, the values for the variability and the highest in 1959 and 1998 (3.2 m·s-1). coefficient of wind velocity are between During the entire studied period, Cen- 0.48 (in the western part of Ukraine) and tral Europe has nevertheless experienced 0.57 (the southern part of the Scandina- a small decrease in mean wind speed in vian Peninsula). August (0.01 m·s-1·10 years-1). Extreme phenomena such as strong winds Much greater diversity is to be noted (> 8 m·s-1) are important for the way peo- where daily mean wind speed is concerned, ple feel. The maximum frequency of strong especially during the cold season. The val- winds falls in winter, while the minimum
KKsisiąążżkka1.indba1.indb 7733 22008-06-26008-06-26 110:56:270:56:27 74 Andrzej Araźny, Rajmund Przybylak, Zsuzsanna Vízi et al.
is in summer. Strong winds in winter are over Europe over recent years. This upward connected with the fast movement of cy- trend to the activity of low-pressure systems clones over the Atlantic Ocean towards can be observed clearly over the northern Eastern Europe. The areas in which strong Atlantic (e.g. Gulev et al., 2001; Wang et al. winds occur most frequently are the moun- 2006). tain tops of the Carpathians (22–25% days Table 1 and Figs. 6 A-D present the of a year). In individual years, the number spatial distribution and mean, minimum of days with strong winds here may amount and maximum values for the 1st, 10th, to over 100 annually. The number of days 90th and 99th percentiles of daily mean with strong wind compared to the mean values for wind speed in Central Europe value for wind speed shows that areas of between 1951 and 2005. The isolines of high mean wind speed also record a large the individual percentiles correlate with number of days with strong wind. Year- the distribution of annual mean values to-year changes in the frequency of strong for wind velocities (Fig. 1C). Generally, winds between 1951 and 2005 in Central the study area is divided into two separate Europe are as presented in Fig. 5. Between parts along a NW-SE profile. Central Eu- 1951 and 2005, the smallest number of days rope is divided in the following way: the with strong winds over the entire studied value of 0.4 m·s-1 of the 1st percentile, the region was recorded in 1968, the largest value of 1.3 m·s-1 of the 10th percentile, in 1983. The number of days with strong the value of 5.0 m·s-1 of the 90th percen- winds in the analysed period is character- tile, and the value of 7.0 m·s-1 of the 99th ised by a statistically upward trend (0.7 days percentile (Fig. 6A-D). ·10 years-1) that achieves significance at The source data used for this analysis α = 0.05. Such results attest to an increased come from the calculations of a numerical number of extreme phenomena connected model, and should as such be treated with with high wind velocities over the analysed caution. Comparative analysis with data area. The increase in mean wind velocity from meteorological stations in Central and in numbers of days with strong wind Europe proved that data from the NCEP/ is connected with intense cyclonic activity NCAR reanalysis need correcting. The com-
Figure 5. Year-to-year course of number of days with strong wind (v > 8 m·s-1) in Central Europe between 1951 and 2005
KKsisiąążżkka1.indba1.indb 7744 22008-06-26008-06-26 110:56:280:56:28 Mean and Extreme Wind Velocities in Central Europe 1951–2005 75
Table 1. Highest (H), mean (M) and lowest (L) values for the 1st, 10th, 90th and 99th percentiles of daily wind speed (m·s-1) in Central Europe (35 grids), between 1951 and 2005
Percentiles Elements I II III IV V VI VII VIII IX X XI XII
1st H 1.1 0.8 0.9 0.5 0.6 0.5 0.6 0.6 0.8 1.1 1.2 1.1 M 0.5 0.5 0.5 0.4 0.4 0.3 0.3 0.3 0.4 0.5 0.5 0.5 L 0.3 0.3 0.4 0.3 0.2 0.2 0.2 0.2 0.2 0.3 0.3 0.3
10th H 3.3 2.7 2.9 2.0 1.8 1.6 1.7 1.9 2.5 3.1 3.4 3.2 M 1.7 1.5 1.5 1.2 1.2 1.1 1.1 1.1 1.2 1.5 1.6 1.7 L 1.2 1.1 1.2 0.9 0.9 0.8 0.8 0.8 0.9 1.1 1.1 1.2
90th H 12.0 11.0 10.4 8.9 7.9 7.4 8.0 8.4 9.7 11.0 11.4 11.7 M 6.8 6.3 6.1 5.3 4.7 4.5 4.6 4.5 5.1 5.8 6.4 6.6 L 4.7 4.6 4.4 4.0 3.4 3.4 3.3 3.2 3.7 4.2 4.6 4.6
99th H 15.0 14.4 13.2 12.9 11.2 10.8 11.3 11.8 13.5 14.3 14.6 15.1 M 9.1 8.8 8.2 7.4 6.5 6.3 6.3 6.3 7.0 7.8 8.5 8.7 L 6.3 6.3 6.0 5.5 4.7 4.7 4.6 4.6 5.2 5.7 6.0 6.0
parison was undertaken for Poland (Lorenc CONCLUSIONS 1996, 2005), for both annual and monthly means at the selected points. According to The detailed analysis of wind speed the analysis, the reanalysis data are under- values over Central Europe between 1951 estimated. For instance, the analysis of the and 2005 as based on data form the NCEP/ distribution of mean annual wind speed in NCAR reanalysis has allowed the authors to Poland (along the north-south profile), the draw the following conclusions: values from the meteorological stations • Between 1951 and 2005 a statistically from the years 1971–2000 (Lorenc 2005) significant increase in mean annual wind are higher: by the Baltic Sea: for example speed was recorded in Central Europe. Val- Elbląg by 0.7 m·s-1 and Hel by 1.6 m·s-1 , in ues for wind speed were exceptionally high central Poland: Łódź by 0.9 m·s-1 and War- in the 1980s and 1990s. saw by 1.5 m·s-1, while on Kasprowy Wierch • Considering the annual course, the in the Tatras by 1.7 m·s-1. The same kinds of highest mean wind velocities were recorded differences between the two data sets were in January, the lowest in August. Between also obtained for mean monthly values. The 1951 and 2005 a significant increase in Janu- presented values for wind velocities from ary wind velocities (and a slight decrease in the NCEP/NCAR reanalysis not only are August wind velocities) were recorded. too small, but their spatial distribution over • Significantly higher wind velocities Poland is also highly problematic. As far are to be noted in mountainous areas (the as spatial distribution is concerned, one of Carpathians and the Sudetan Mountains). the most important differences between the These velocities are about twice as high as in data from the synoptic stations and those the surrounding areas. from the NCEP/NCAR reanalysis is the • In the analysed period the annual aver- lack of high speed winds characteristic for age number of days with strong wind showed the Baltic Sea shore. a statistically significant upward trend.
KKsisiąążżkka1.indba1.indb 7755 22008-06-26008-06-26 110:56:280:56:28 76 Andrzej Araźny, Rajmund Przybylak, Zsuzsanna Vízi et al.
Figure 6. Values of the 1st (A), 10th (B), 90th (C) and 99th (D) percentile of wind speed (m·s-1) in Central Europe between 1951 and 2005
• Reference to the number of days with and the data obtained from the meteoro- strong wind compared to the mean value for logical stations in Central Europe show that wind speed makes it clear that areas with reanalysis data are underestimated by about a high mean wind speed also recorded a 1–1.5 m·s-1. great number of days with strong wind. • The comparative analysis of the wind velocities from the NCEP/NCAR reanalysis
KKsisiąążżkka1.indba1.indb 7766 22008-06-26008-06-26 110:56:290:56:29 Mean and Extreme Wind Velocities in Central Europe 1951–2005 77
ACKNOWLEDGEMENTS analysis Data, EWEC Conference, Athens,
KKsisiąążżkka1.indba1.indb 7777 22008-06-26008-06-26 110:56:290:56:29 78 Andrzej Araźny, Rajmund Przybylak, Zsuzsanna Vízi et al.
Energy], Wydawnictwo Instytutu Gospodarki Wang, X. L. L., Swail, V. R. and Zwiers, F. W. Surowcami Mineralnymi i Energią (GSMiE), (2006), Climatology and Changes of Extra- PAN, Kraków, pp. 145. tropical Cyclone Activity: Comparison of Starkel, L. (2003), Extreme Meteorological ERA-40 with NCEP-NCAR Reanalysis for Events and their Role in Environmental 1958–2001, Journal of Climate, 19: 3145– Changes, the Economy and History, Global 3166. Change, 10: 7–13. Paper first received: March 2007 In final form: December 2007
KKsisiąążżkka1.indba1.indb 7788 22008-06-26008-06-26 110:56:300:56:30 RIVER TRAINING VS. FLOOD RISK IN THE UPPER VISTULA BASIN, POLAND
ADAM ŁAJCZAK Department of Earth Sciences, University of Silesia, Będzińska 60; 41-200 Sosnowiec, Poland E-mail: [email protected]
Abstract: This paper assesses the effect of river training in the 20th century on the evolution of flood risk in the middle and lower courses of certain Polish mountain and upland rivers, and in the lowland Carpathian foreland. The overall anthropogenic impact on the flood risk is a combi- nation of two contradictory trends: (a) the shortening of the floodplain inundation time (between the levees) as a result of the deepening of the trained channel; and (b) the increasing height of the flood water and frequency of flood culminations, a result of the flood wave transformation. The author, in his flood risk analysis, regards the former trend as the more influential. The highest levels of all types of flood risks were found along the valley reaches with unembanked channels that displayed a tendency to reduce both their depth and gradient. This type of reach occurs im- mediately downstream of embanked reaches with a deepened channel. The author also addresses ways to mitigate flood risk levels, taking into account limitations stemming from local land devel- opment and legal conservation status.
Key words: flood risk, flood, upper Vistula basin, river training.
INTRODUCTION modifies, not just the duration and extent of flooding, but also the pace of concentration Flood risk should be understood as the like- and speed of flood wave movement. lihood of economic losses being suffered One of the purposes of river training is within a floodplain as a result of flood- to reduce the flood risk by accelerating the ing by overbank culminations of a moving drainage of a submerged floodplain along flood wave. It depends on the duration of a deliberately shortened channel. River the overbank discharge and the vertical and training projects trigger processes of sys- horizontal extents of flooding within the tematic deepening along the shortened and embankments. Contemporary flood risk is steepened reaches, and shallowing along modified by human activity, and its level is reaches with a less steep gradient (Brookes a combination of the hydrological effects of 1990; Łajczak 1995a). This accelerates the processes of long duration, such as drainage flow velocity and, in embanked rivers, also basin deforestation, agricultural expansion, increases the water level amplitude. Along urbanisation and river training. This activity the reaches with fast deepening channels
KKsisiąążżkka1.indba1.indb 7799 22008-06-26008-06-26 110:56:300:56:30 80 Adam Łajczak
the duration with an overbank water level Vistula and Nida Valleys, and on data sup- may tend to reduce. However, an increased plied by the Institute of Meteorology and concentration of flood waves and increased Water Management. water level amplitude in the upstream reach may increase the flood risk along those less steep reaches further downstream where the STUDY AREA embanked cross-section volumes are low (Punzet 1991; Wyżga 1993; Łajczak 1995a). In Poland there are two areas in which sum- It is possible to control floods on trained mertime or early springtime floods predomi- rivers effectively by means of river dams nate, and turn into catastrophic events every (Punzet 1959, 1973). This is however im- few years (Fig. 1). One covers the drainage possible on rivers utilised for navigation, basins of the mountain tributaries of the since this type of hydrotechnical structure is rivers Vistula and Odra and the foreland absent. courses of those rivers, while the other spans River training has been increasingly per- the Vistula’s and Odra’s upland and lowland ceived as a controversial activity and some- tributaries, including those located in fore- times also as a contradictory one that does land areas. Most of the upper Vistula river not produce the expected reduction in flood system features a predominance of summer risk (Kajak and Okruszko 1990; Andrews rain floods, both annually and in the long- and Burgess 1991; Finlayson 1991; Angel- term. On the River Raba, a typical Western stam and Arnold 1993; Łajczak 2006a, b). Carpathian watercourse, the flood risk is re- A heightened flood risk resulting from river stricted to the summer season, while on the training has only been noted over the last River Nida, the longest upland tributary of 20 years (e.g. Cooper et al. 1987; Howard the upper Vistula, it is restricted to the early 1992; Kajak 1993). This can be seen, in par- springtime, and less often the summertime. ticular, along those reaches that are becom- The foreland course of the upper Vistula has ing shallower. A heightened ground water a hydrological regimen driven primarily by table in the valley, one of the effects of the its mountain tributaries. Currently, the flood channel shallowing downstream of the fast- risk along this course of the river is limited deepening channel reaches (Żelazo 1993), to the summer months, but the early spring further extends the duration and extent of is also added, downstream of the River San excessive water content within the flood- confluence (Dynowska 1971; Ziemońska plain beyond the flood embankments. 1973; Punzet 1991). About a century ago, the beginning of hydrological records (on water levels and STUDY OBJECTIVE AND MATERIALS discharges) coincided with the first active and passive measures mitigating flood risk This paper looks at the change in the flood in the Upper Vistula valley. The records can risk in the upper Vistula drainage basin, as now be used to assess the evolution of flood- a result of river training during the 20th cen- risk patterns, whether caused by natural tury. The study focuses on three rivers run- or by human-induced processes. ning through mountains, foothills, uplands The attempts at river training olong the and lowland forelands (the Rivers Raba course of the River Vistula that form the and Nida, and the foreland course of the subject of this study started at the end of the Vistula). The paper is based on the litera- 19th c. and lasted throughout the 20th. They ture, the author’s own research in the upper changed the earlier trends in the channel
KKsisiąążżkka1.indba1.indb 8800 22008-06-26008-06-26 110:56:300:56:30 River Training vs. Flood Risk in the Upper Vistula Basin, Poland 81
Figure 1. The upper Vistula drainage basin on the map of Poland (A). River network against principal geomorphological units (B). a) limit of the upper River Vistula drainage basin, b) area dominated by summer rain floods, c) main rivers, d) Cracow (K) and Warsaw (W), e) river reaches (Vistula, Raba and Nida) analysed in the paper, f) water gauge at Zawichost (Z) and Puławy (P) measuring the outflow from the upper Vistula drainage basin, g) large dams. Principal geomorphological units in the basin: h) Carpathian Mts., i) Polish Uplands, j) Sub-Carpathian basins.
development of its tributaries, involving both (the Wisłoka, Wisłok and San), which often shallowing and broadening. The morphology feature bedrock channels along their moun- of the Carpathian river channels, which are tain courses in the Beskidy Mountains. The crucial for the supply of water and bedload erection of numerous rubble dams in the and suspended load to the Vistula, exhibit Carpathian Mts. reduced bedload transport regional differences (Klimek 1979). The along the upper courses of mountain riv- westernmost gravel-bedded tributaries down ers and streams, while river dams, the first to and including the River Dunajec are more of which was built in the 1930s, have been susceptible to change under the influence intercepting all of the bedload and most of of river training than the eastern tributaries the suspended material. As a result, the pace
KKsisiąążżkka1.indba1.indb 8811 22008-06-26008-06-26 110:56:310:56:31 82 Adam Łajczak
of shallowing downstream of the dams has Foothills and the lowland Carpathian fore- been reduced. This has been further assist- land (Fig. 2). The newly constructed river ed by a reduction in the supply of bedload training along the course of the river has led from the channel banks after their rein- to a shorter, narrower and deeper channel. In forcement with stone bands. From the flood- the aftermath of the river training measures risk point of view, the greatest hydrological the channel deepened by more than three me- consequences have been engendered by the tres in its middle and lower course, a typical deepening and narrowing of the channels, value recorded on the lower courses of other a process started in the early 20th c. On the Carpathian tributaries of the River Vistula. River Vistula and the lower courses of its The channel also increased in compactness mountain tributaries this process was mostly through a lowering of the width to average driven by channel shortening (Punzet 1981; depth ratio. This change resulted in an in- Klimek 1987; Wyżga 1993; Łajczak 1995a; creased water flow velocity, particularly dur- Wyżga and Lach 2002). Along the upper ing flood events, and consequently in greater courses of the Carpathian rivers, this devel- concentrations of flood waves and increased opment trend is assisted by a growing affor- speeds of flood wave travel (Wyżga 1993). As estation of their drainage basins, since the a further consequence, the differences be- mid 20th century, and by the trend for field tween the maximum flood water levels record- tracks in the depopulated areas of the Bes- ed during the same events on the lower course kid Niski and Western Bieszczady ranges to of the Raba and on the middle course of the revegetate (Izmaiłow et al. 2003). river have been increasing since the second River training has also included flood em- half of the 20th century. (Fig. 3). The faster bankments erected along the foreland course flow in the deeper channel results in a higher of the Vistula, along the lower courses of its bankfull discharge at the expense of overbank Carpathian tributaries and also locally along discharge. The narrow embankment-to-em- their intra-mountain-basin courses, along the bankment space in the lower course of the riv- lowland tributaries and some of the upland er additionally forces maximum water levels tributaries (Hennig 1991). The insufficient to become ever higher, while shortening the embankment-to-embankment space that re- duration of the flood wave event and increas- sulted has reduced the zone liable to flooding ing the flood wave velocity. Such a flood wave by a factor that is often larger than 10 (on the profile has a considerable impact on the flood River Vistula). Subsequent effects included situation along the River Vistula. higher maximum water levels on the River The Raba is one of those Carpathian riv- Vistula, (up to twice as high along the Cra- ers wherein the assessment of the impact of cow reach), and a faster travelling flood wave river training on the evolution of the flood (Soja and Mrozek 1990; Punzet 1991). risk is far from easy and straightforward. On the one hand, engineering has brought a reduction in flood risks along the trained CHANGES IN FLOOD RISK DURING reaches in terms of flood duration and area THE 20TH C. affected. Indeed, the flood wave travel times and the number of overbank discharge days CARPATHIAN RIVERS USING THE EXAMPLE have been reduced, and the floodplain in- OF THE MIDDLE AND LOWER RABA undation time is down dramatically. On the The River Raba provides an example of other hand, the greater flood wave speeds a gravel-bedded Western Carpathian river may, in certain situations, shorten the draining the Beskidy Mts., the Carpathian time local communities have to mount an
KKsisiąążżkka1.indba1.indb 8822 22008-06-26008-06-26 110:56:310:56:31 River Training vs. Flood Risk in the Upper Vistula Basin, Poland 83
Figure 2. The River Raba drainage basin. a) main rivers, b) limit of the drainage basin, c) selected water gauges in the middle and lower courses of the river, G- Gdów, P- Proszówki, d) partly or fully trained reach of the river, e) artificial lake Dobczyce (ZD), f) deepening [m] of the Raba channel at selected water gauges following river training in the 20th century, in parenthesis the value related to the second half of the 20th century (according to Wyżga 1993). Geomorphological units in the basin: g) Beskidy Mts., h) Carpathian Foothills, i) Sub-Carpathian basins.
adequate response, and therefore be con- UPPLAND RIVERS EXEMPLIFIED BY THE sidered to increase the flood risk. Addition- MIDDLE AND LOWER NIDA ally, river training has increased flood risks The River Nida is an example of an upland along the untrained reaches downstream, as river with a sandy channel and a low gradient reflected by the increasing maximum flood in its middle and lower courses (0.2–0.5‰). wave levels in the lower course during the Along this stretch, the river meanders with largest events. This trend is observed despite a meander coefficient that reaches 2.0 local- a parallel trend to reduce the volume of the ly. Its floodplain is regularly submerged dur- flood wave, as measured above the bank- ing springtime floods, and less often during full stage. A further flood risk improvement summertime, and the water can stagnate for could be achieved via effective use of the Do- over three months. During the period 1950– bczyce Dam, especially in view of the mostly 1995, the middle course (between Brzegi and positive experience with dams as flood man- Pińczów) was shortened and partly equipped agement instruments on other Carpathian with flood embankments, and extensive wet- tributaries of the Vistula (see: Punzet 1959, lands were also drained within the flood- 1973, 1991). plain, especially along a long anastomosing
KKsisiąążżkka1.indba1.indb 8833 22008-06-26008-06-26 110:56:310:56:31 84 Adam Łajczak
Figure 3. Hydrograph of selected flood waves in the middle and lower River Raba courses between Gdów and Proszówki.
reach near Umianowice (Łajczak 2006b). at Brzegi (Fig. 5). Because of the slow trans- The lower course of the river (downstream port of sand, the river has been observed from Pińczów) has remained untrained, to become shallower downstream around with a natural channel and floodplain Pińczów, initially in the old channel but more (Fig. 4). The river training has been aimed recently (since 1970) also in the new, initially at accelerating the draining of flooded areas deep channel. This process varies in speed, and the drying out of the floodplain. From and depends on the intensity of the upstream the point of view of flood control, the result river training work. The zone subject to the has been mixed. shallowing process is moving, but has not The river has begun to deepen its short- yet reached the next water gauge at Wiślica. ened channel, as recorded by a water gauge The slow downcutting trend along the mouth
KKsisiąążżkka1.indba1.indb 8844 22008-06-26008-06-26 110:56:310:56:31 River Training vs. Flood Risk in the Upper Vistula Basin, Poland 85
Figure 4. River Nida drainage basin (A) and middle and lower courses of the river (B). a) main rivers, b) limit of the drainage basin, c) water gauges, d) partly trained river reach, e) formerly anastomosing reaches of the river currently partly or fully trained, f) floodplain extent. Geomorphological units in the basin: g) Polish Uplands, h) Sub-Carpathian basins.
reach of the Nida below Wiślica is a result it shrank to less than ten days and increased of a similar trend on the Vistula since the again during the last decade (to 30 days per beginning of its training in the late 19th cen- year). The main cause of the increase in the tury. On the deepened reach of the Nida, the flood risk near Pińczów (the largest town bankfull discharge volume has been grow- on the river) observed since the mid-1990s ing, as recorded by the water gauge at Brzegi. seems to be the continuing shallowing of the During the period 1939–1990, this produced river channel, as a result of misguided train- a trend whereby the number of overbank dis- ing and drainage projects, and only partly charge days was reduced from 10 to less than due to the greater frequency of large flood 2 days. The extension of the floodplain inun- events. The engineering measures have re- dation time observed along this reach since sulted in effective drainage of large marshy 1990 is independent of a gradual deepening areas adjacent to the anastomosing reaches. of the channel, and is caused by the increas- The flood risks near Pińczów could be partly ing frequency of large floods. Along the shal- mitigated by a revitalisation of the drained lowing channel the duration of the flood- wetlands and by returning their functions plain flooding has been growing into long (Łajczak 2006b). In the mouth reach of the periods that are independent of long-term river, downstream from Wiślica, the Vistula river discharge patterns. The inundation backwater effect generates a very high flood time near Pińczów had been growing until risk level, manifested by the long duration 1969 (reaching 80 days per year), after which of floodplain flooding (up to 100 days) and
KKsisiąążżkka1.indba1.indb 8855 22008-06-26008-06-26 110:56:310:56:31 86 Adam Łajczak water levels (C) at water gauge stations in Brzegi, Pińczów and Wiślica. water levels (C) at gauge stations in Brzegi, Pińczów 1 as well the change of overbank IN Figure 5. Minimum annual water levels Hmin during 1939–2003 (A) and change in the bankfull discharge Qbf (B),
KKsisiąążżkka1.indba1.indb 8866 22008-06-26008-06-26 110:56:320:56:32 River Training vs. Flood Risk in the Upper Vistula Basin, Poland 87
the highest water levels recorded along the of 4 m) and large scale deposition in the entire river. The fluctuation in the Vistula bank zones. Local level differences in the water level in the area reaches up to nine channel cross-section increased three-fold metres, and the backwater effect can go up compared to the early 20th century. Con- to Wiślica when large flood events are taking versely, there were also cases of the channel place on the main river. With its long reten- becoming shallower, i.e. in the Oświęcimska tion of flood water, the Nida river has seen Basin upstream of the Przemsza and Soła a mixture of flood risk improvement and confluences and over a longer reach where deterioration, depending on the reach, as the river breaks through the Polish Uplands. a result of river training. The only significant The alternating pattern of deepened chan- changes in the risk levels include the flood- nel sections and sections becoming shal- plain inundation duration and territorial ex- lower is therefore a characteristic feature of tent. The river has no significant impact on the foreland River Vistula during the river- flood risks in the Vistula valley. training era (Fig. 6). Another characteristic Uniquely among major tributaries of the of the channel morphology’s impact on the upper Vistula, the Nida has a valley already discharge conditions is the continuing trend subject to several legal conservation statuses for the channel’s width to average depth (including as a Landscape Park, an ecologi- to decrease, as measured at water gauges cal corridor and a Natura 2000 area), which (Łajczak 1995a). As a result of an increased only allows environmentally sound methods speed of flow, caused by modifications to the of flood risk mitigation (Łajczak 2006a, b). cross- and longitudinal sections of the chan- These methods might include revitalisa- nel, the bankfull volumes continue to grow tion of the extensive wetlands, now mostly at various rates, as recorded by the gauges drained, and an expansion of the territory (Fig. 7). The bankfull volumes tend to dif- between the embankments to increase its fer ever more from the average medium-high volume. The earlier plans to erect a dam on discharge, to which they were similar prior the Nida upstream of Brzegi, and indeed to the first river training projects. During any training measures that would introduce the 20th c., this increase in the bankfull dis- large quantities of sand, should be seen as charge varied along the river course studied undesirable. and reached three-fold, depending on the scale of the cross-section change. THE FORELAND COURSE OF THE RIVER The combined average duration of the VISTULA flooding within the embanked floodplain The foreland course of the River Vistula is and the number of flood events vary greatly the most thoroughly investigated part of the along the stretch of river being investigated study area in terms of changes to the flood (Fig. 8). Both of these characteristics express risk caused by river training. The engineer- an increase in flood risk , and display a rela- ing projects, involving an approximately tionship wih the scale of the post-river train- 30% reduction in channel length, new stone ing cross-section evolution. They are highest spurs and bank protection, were designed in the reaches which have become shallower, to prepare the river for its role as the coun- and only slightly lower in the least deepen- try’s main waterway, and to protect adja- ing reaches. Conversely, the lowest values cent areas from flooding. During the 20th are observed along the most deepened chan- century, modifications to the cross-section nel reaches. As an example, the reach of the of the river channel included deepening by Vistula crossing the Polish Uplands gap has an average of two metres (and a maximum a frequency and combined duration of flood
KKsisiąążżkka1.indba1.indb 8877 22008-06-26008-06-26 110:56:320:56:32 88 Adam Łajczak
Figure 6. Change in channel depth following river training in the foreland River Vistula CH.L.CH. and average thickness of deposits in bank zones M.B.A. a) selected water gauges (1—Smolice, 2—Jagodniki, 3—Karsy, 4—Koło, 5—Zawichost). The confluences of the principal Carpathian tributaries are also indicated. Changes in the depth of channel until 1930, 1960 and 1990.
events up to 15 times greater than along the discharge fluctuations and was a result of Cracow reach of the river. channel deepening (Fig. 9). The trend is Since at least 1930, the foreland Vistula proportionally related to the deepening rate has displayed a trend whereby the overbank (Łajczak 1995a, 1999). However during the water levels have been shrinking. Until 1990, period 1931–1990, the duration of the over- this trend was unrelated to the long-term bank water levels along the most deepened
KKsisiąążżkka1.indba1.indb 8888 22008-06-26008-06-26 110:56:330:56:33 River Training vs. Flood Risk in the Upper Vistula Basin, Poland 89
Figure 7. Change to bankfull discharge Qbf compared to medium high discharge MHQ against minimum annual water levels Hmin at selected water gauges on the foreland River Vistula. Water gauges are numbered as in Fig. 6.
reaches of the channel displayed only a weak level duration after 1990, when a number decline, accompanied by generally low val- of large flood events occurred. During the
ues of the IN1 parameter, up to a maximum period 1931–1990, overbank discharge along of approximately ten days per year. This the most deepened Vistula channels oc- may mean that the decline in this parameter curred on average every second year, but along the most deepened reaches had already prolonged periods of up to ten years without started before 1930. Lack of data prevents such periods were also recorded. Further a similar analysis of the overbank water downstream, along less deepened reaches of
KKsisiąążżkka1.indba1.indb 8899 22008-06-26008-06-26 110:56:330:56:33 90 Adam Łajczak
Figure 8. Differentiation of duration of overbank IN1 water levels and number of flood events
with such water levels per year IN2, along the foreland course of the River Vistula. a) selected water gauges (1—Smolice,2—Jagodniki, 3—Szczucin, 4—Zawichost).
the river, the decline in the duration of flood- bank water levels along the foreland course ing on the floodplain was more marked. of the Vistula from 1931 to 1990 might sug- Downstream from the Dunajec confluence gest that such a trend may have prevailed
the IN1 parameter dropped from approxi- earlier on, i.e. in the late 19th century and mately 10–20 days per year during the years during the first three decades of the 20th 1931–1960 to less than ten days before 1990. century, when the pace of channel deep- Over the period 1931–1990, this section of ening was already advanced, such as near the river experienced overbank water levels Cracow. Downstream, where the deepen- nearly every year, which may mean that the ing process only started after the 1930s, the duration of the flooding of the embanked trend towards a rapid decline in the periods space is shortened more effectively during of floodplain flooding between the embank- the initial period of channel deepening fol- ments was only noted after 1930. Below the lowing river training than at an advanced confluence of the River San the trend to stage of change in the geometry of a trained a decline in the number of days with over- channel. Downstream of the River San bank water levels is not very advanced, and confluence, the Vistula channel is not sub- started only in the 1970–1980s because the ject to intensive deepening, and there are river training efforts along this stretch of the places where it may even be becoming shal- river starting later than elsewhere. lower. Overbank water levels were recorded Another result of the engineering every year of the study period—peaking at projects in the foreland course of the Vistula 50–70 days per annum during the 1931–mid is the restriction of flooding of the embanked 1960s period. Afterwards, the number of floodplain along the considerably deepened days with overbank water levels declined channel (> 2 m) to just the summer season. again to a maximum of 30 days annually in Where the deepening is only minor (< 1 m), the 1980s. or the channel is becoming shallower, this The lack of a clear-cut trend involving part of the floodplain may also be flooded a reduction in the number of days with over- during rapid thaws. Prior to river training,
KKsisiąążżkka1.indba1.indb 9900 22008-06-26008-06-26 110:56:330:56:33 River Training vs. Flood Risk in the Upper Vistula Basin, Poland 91
the entire course of the river floodplain ex- and Zawichost, and by a half between Smo- perienced flooding during both rapid thaw lice (the confluence of the River Skawa) and floods and summer floods. The foreland- Sierosławice (the confluence of the River course engineering projects effectively Raba), the latter being the most deepened eliminated the early-springtime flood risk channel section. The accelerated move- along the most modified reaches alone. Dur- ment of the flood waves along the deepened ing summertime, floods have only become reaches of the foreland Vistula is consider- shorter as a result of the training measures, ably influenced by the greater concentra- except along the upland gap reach. tion of similarly-induced flood waves on the
Figure 9. Change of duration of overbank water levels IN1 at selected water gauges on the foreland course of the River Vistula during 1931–1990. Water gauges numbered as in Fig. 8.
Another effect of the training of the Vis- lower courses of the Carpathian tributaries tula foreland is an increased concentration (Punzet 1991; Wyżga 1993). The post-river of flood waves, as a result of the accelerated training development of the foreland Vis- flow observed across the entire spectrum of tula channel has produced a trend for the water levels (Punzet 1991; Łajczak 1995a). volume of the flood waves to be reduced as According to Punzet, during the 20th c., the a result of their greater concentration and Cracow reach displayed a trend to a short- higher peaks (Punzet 1991). This could be ened duration and increased height of the partly explained by the effect that the bank- flood wave, as well as to an increase in the full discharge has been growing at the ex- frequency of extraordinary flood peaks. pense of the overbank discharge following A combined effect of these changes has been the advent of the river training projects, an to cause the flood wave travel time to short- effect previously overlooked. The trend for en along the foreland course of the river, in- a declining duration of the overbank water cluding by one-third between Goczałkowice level periods, initiated at various dates in
KKsisiąążżkka1.indba1.indb 9911 22008-06-26008-06-26 110:56:330:56:33 92 Adam Łajczak
the 20th c., has been observed along the • a trend towards an increasing frequen- considerably deepened reaches, and is inde- cy of exceptional flood peaks; pendent of the long-term fluctuations of the • an increasing speed of large flood Vistula discharge (Łajczak 1995a). waves. The erection of flood embankments has reduced the flooding zone up to 50-fold along the foreland course. Events whereby DISCUSSION AND CONCLUSIONS the water tops the new embankments and spills over are very rare, and limited to days As demonstrated by reference to the exam- with extreme water levels. Combined with ple of the trained foreland course of the the causes mentioned above, this dramatic Vistula, the alternating pattern of reaches reduction of the flooding zone has led to featuring varying degrees of cross-section increased water level amplitudes along the modification translates into hydrological course of the Vistula we are examining, effects, especially with regard to flood risk. especially after 1920 (Osuch 1991; Punzet This situation is typical for the entire course 1991). They are currently double the value of the river (Łajczak 1995a). The reach recorded before the embankments, and may downstream of the River San confluence, reach nine metres (Soja and Mrozek 1990). where the deposition is at its most intensive, Just as in the tributaries, assessment of stands out as having the greatest flood risk the foreland course of the Vistula in terms along the river (Jędrysik and Rusak 1982). of the impact of river training on flood risk This should be seen as a consequence of the is complex. Taking into account the duration shallowing and broadening processes, along of the overbank stages and the extent of the the entire channel length, that began at least flooding zone, the duration and area aspects as far back as in the 17th century, driven by of the flood risk were reduced during the 20th the deforestation of the drainage basin. Only century. This is a result of faster flood wave since the 20th century should this be seen as travel, and of the fact that minor summertime a consequence of the river-training triggered floods and all the thaw floods remain within development of the upstream reach (above the bankfull stage. The number of days with the River San confluence). Along the en- overbank discharge was reduced, with the tire foreland Vistula, as well as in the lower exception of reaches without any significant courses of the Carpathian tributaries and in deepening effect or with a shallowing trend. the upland and lowland tributaries, the cur- Just as in the Carpathian rivers, the increased rent flood risk is critically determined by the flood wave speed may, , be regarded as an ad- effects of river training measures. I would ditional risk factor in certain circumstances. therefore confirm Punzet’s conclusion (1991) The third feature of flood risk, i.e. the height that river training as broadly understood, of peak river water levels, demonstrates along the course of the Upper River Vistula a growing flood risk along the entire chan- has a large impact on the process of flood nel, within the embankments only, during the wave formation in the river. 20th century. Only extremely high water level The anthropogenic aspects to the cur- events threaten to spill over and undermine rent evolution of flood risks in the foreland the embankments, which then need mainte- course of the Vistula and in the middle and nance. This is a result of: lower courses of its tributaries involve an • a trend for peak river water levels dur- overlap effect of two contradictory trends: ing subsequent large flood events to increase, (a) the falling risk level in view of the short- and for their durations to shorten; ening of floodplain inundation time and
KKsisiąążżkka1.indba1.indb 9922 22008-06-26008-06-26 110:56:340:56:34 River Training vs. Flood Risk in the Upper Vistula Basin, Poland 93
territory, and (b) the growing risk from the On the Nida near Pińczów, the inundation increasing height of the flood water. These times were extended more than fourfold dur- two trends are typical over most of the river ing the period 1939–1969. In 1970, the con- course where its channel has been deepened. struction of an artificial channel reduced that The former of the two trends (a) seems to time to just a few days per year, but a back- be more significant for this analysis of flood lash reaction returned it to ten times more. risks. Indeed the benefit of the shortening of While at Pińczów the embankments reduce the inundation time over the unembanked the flooded area to a very narrow strip, just area or the undeveloped embanked flood- two kilometres downstream from the town plain is greater than the losses incurred, they disappear for the rest of the river course because of the rare cases of flooding of in- all the way to its confluence with the Vistula, habited and agriculturally used areas, as exposing a floodplain up to three kilometres a result of breaches in the adequately high wide. The historically unprecedented extreme levees. This can be illustrated by the fore- flood wave recorded during the 1997 summer land course of the Vistula between the con- flood not only topped the embankments along fluences of the Dunajec and Wisłoka rivers, many reaches, but remained high throughout where—after the erection of up to six-metre- the lower course of the river, causing a high high embankments—the floodplain inunda- degree of damage. Along its partly embanked tion time was reduced 10-fold and the area upland gap reach, the Vistula has a broad 20-fold during the period 1930–1990. The floodplain of up to one kilometre. Unlike its coinciding increase in the water level ampli- long upstream reach, this section of the river tude to nine metres was a result of the simul- displayed no reduction of flooding time, but taneous channel deepening and maximum a slow opposite trend after 1930. The num- floodwave level increase (Punzet 1981, 1991; bers of days with flood effects varied across Jędrysik and Rusak 1982; Soja and Mrozek a broad range of 10–70. The currently higher 1990; Hennig 1991; Osuch 1991; Łajczak flood risks observed along this reach are also 1995a, 2006a). The areas beyond the levees partly a result of the heightened culminating remain unprotected from exceptionally high floodwaves developing in the upstream reach floodwaves, despite the vertical extension of with its deepening channel (Łajczak 1995a,b, the embankments after the great floods, and 2006a,b). other maintenance measures. When looking at various flood risk miti- The greatest flood risk levels along the gation options in the foreland course of the river course analysed, in all three respects, Vistula and the middle and lower courses of were recorded in the valley sections coincid- its tributaries it must by assumed that the cur- ing with the channel reaches that displayed rent channel development trends, initiated a shallowing process, lower gradients and or accelerated by the training projects, will only partial or no embankments. This type of continue in the long term (Łajczak 1995a, valley section is typically found at the end of 1999). These options tend to be much more long stretches where the channel has been sig- limited in the valley sections with shallowing nificantly shortened, narrowed and deepened channels than those with deepened channels. (its slope gradient additionally increased) as In the former case the risk could be mitigat- a result of training measures, which normally ed by halting the channel deepening proc- included long uninterrupted embankments ess, or by even less realistic measures, such on both sides of a narrow floodplain. An illus- as a broadening of the distance between the tration of this pattern is found using certain embankments or an increase in their height. water gauges on the Rivers Nida and Vistula. In the latter case, featuring long reaches of
KKsisiąążżkka1.indba1.indb 9933 22008-06-26008-06-26 110:56:340:56:34 94 Adam Łajczak
deepened channels with fast-moving flood REFERENCES waves travelling along a narrow floodplain between the embankments, the risk could Andrews, J.H. and Burgess, N.D. (1991), Ration- only be mitigated by heightening the em- ale for the Creation of Artificial Wetlands, in bankments. Where the deepened channel Finlayson, C.M. and Larsson, T. (eds.), Pro- reach is relatively short there is an option of ceedings from Workshop: Wetland Manage- setting aside large polders beyond the em- ment and Restoration, Swedish Environment bankments, such as between the confluences Protection Agency, Report 3992, 24–32. of the rivers Soła and Skawinka along the Angelstam, P. and Arnold, G.W. (1993), Contrast- Vistula valley (Nachlik and Wit 1997; Nach- ing Roles of Remnants in Old and Newly Im- lik 1998; Wit 1999). The role of a polder can pacted Landscapes: Lessons for Ecosystem also be played by the often inundated flood- Reconstruction, in Saunders, D.A., Hobbs, plain along a shallowing river channel, such R.J. and Ehrlich, P.R. (eds.), Nature Con- as along the lower Nida. Halting a channel servation, 3, Surrey Beatty & Sons, Sydney, deepening process that provides sandy ma- 109–125. terial for a downstream shallowing reach is Brookes, A. (1990), Channelized Rivers: Pros- feasible on a small scale, such as along an pects for Environmental Management, Wiley, anastomosing reach of the middle-course Chichester, 1–326. River Nida. In this case it would require Cooper, J.R., Giliam, J.W., Daniels, R.B. and Ro- a parallel reinstatement of the nearby former barge, W.P. (1987), Riparian Areas as Filters wetlands, which could play the role of an ef- for Agricultural Sediment, Journal of Soil Sci- fective deposit accumulation zone (Łajczak ence Society of America, 51: 416–420. 2006b). Dynowska, I. (1971), Typy reżimów rzecznych Measures to mitigate the flood risk must w Polsce [Types of Hydrologic Regime of Riv- also take into account the valley’s conserva- ers in Poland], Zeszyty Naukowe UJ, Prace tion status. Where no legal conservation is Geograficzne, 28: 1–155. involved, such as in the Carpathian rivers Finlayson, C.M. (1991), Wetland Management or along the foreland course of the Vistula, and Restoration: Summary, in Finlayson, a more active floodwave management can be C.M. and Larsson, T. (eds.), Proceedings achieved by controlling discharge from arti- from Workshop: Wetland Management and ficial lakes. Ideally, this could produce a flat- Restoration, Swedish Environment Protection tened floodwave extended in time (Punzet Agency, Report 3992, 174–179. 1959, 1973, 1991). In the River Nida valley, Hennig, J. (1991), Zabudowa hydrotechniczna subject to several conservation statuses, only rzeki, in Dynowska, I. And Maciejewski, M. environmentally sound methods are feasible, (eds.) Dorzecze górnej Wisły [The Drainage such as reinstatement of the natural quali- Basin of the Upper Vistula River], Part II, ties of the anastomosing reaches and the ad- Państwowe Wydawnictwo Naukowe ( PWN), jacent wetlands, or expanding the territory Kraków-Warsaw, 119–184. between the embankments. Howard, A.D. (1992), Modelling Channel Mi- gration and Floodplain Sedimentation in Meandering Streams, in Carling, P.A. and ACKNOWLEDGEMENTS Petts, G.E. (eds.), Lowland Floodplain Rivers: Geomorphological Perspectives, Wiley, Chich- The author would like to thank the two ester, UK. anonymous reviewers for their critical com- Izmaiłow, B., Krzemień, K. and Sobiecki, K. ments on the manuscript. (2003), Rzeźba [Morphology], in Górecki A.,
KKsisiąążżkka1.indba1.indb 9944 22008-06-26008-06-26 110:56:350:56:35 River Training vs. Flood Risk in the Upper Vistula Basin, Poland 95
Krzemień K., Skiba S. and Zemanek B. (eds.) ki Wodnej, PAN, 15: 1–215. Magurski Park Narodowy [Magurski National Łajczak, A. (2006a), Regulacja rzeki a zagroże- Park], Proceedings of the Magurski National nie powodziowe, na przykładzie Nidy [River Park and Jagiellonian University, ? Kraków- training vs. flood exposure. The example of Krempna, 21–30. the River Nida, Poland], in Bartnik, W. (ed.) Jędrysik, M. and Rusak, M. (1982), Kompleksowe Infrastruktura i ekologia terenów wiejskich zagospodarowanie Wisły [A Complex Man- [Infrastructure and Ecology of Rural Areas], agement of the Vistula River], in Piskozub, Komisja Technicznej Infrastruktury Wsi, PAN, A. (ed.) Wisła. Monografia rzeki [The Mono- Kraków, 4 (1): 38–45. graph of the River Vistula], PWN, Warsaw. Łajczak, A. (2006b), Regulacja rzeki a zagroże- Kajak, A. (1993), The Vistula River and its Ripar- nie powodziowe, na przykładzie Wisły między ian Zones, in Hilbricht-Ilkowska, A. and Skoczowem i Puławami [River Training vs. Pieczyńska, E. (eds.) Nutrient Dynamics and Flood Exposure as Exemplified by the Seg- Retention in Land/water Ecotones of Lowland ment of the Vistula River between Skoczów Temperate Lakes and Rivers, Kluwer Academ- and Puławy], in Bartnik, W. (ed.) Infrastruk- ic Publishers, Dordrecht-Boston-London, tura i ekologia terenów wiejskich [Infrastruc- 149–157. ture and Ecology of Rural Areas], Komisja Kajak, A. and Okruszko, H. (1990), Grassland on Technicznej Infrastruktury Wsi, PAN, Kraków, Drained Peats, in Breymeyer A.(ed.), Man- 4 (1): 46–53. aged Grasslands, Elsevier Scientific Publish- Nachlik, E. and Wit, M. (1997), Hydrauliczne ers, Amsterdam, 213–253. podstawy projektowania pracy polderów dla Klimek, K. (1979), Geomorfologiczne zróżni- wspomagania ochrony Krakowa przed po- cowanie koryt karpackich dopływów Wisły wodzią [Hydraulic aspects of polder design [Geomorphological Differentiation of Chan- in flood protection of Kraków], Materia- nels of Carpathian Tributaries to the Vistula], ły XVII Ogólnopolskiej Szkoły Hydrauliki Folia Geographica, ser. Geographica-Physica, nt.’Współczesne problemy hydrauliki wód 12: 35–47. śródlądowych’, Gdańsk, 125–130. Klimek, K. (1987), Man`s impact on fluvial proc- Nachlik, E. and Wit, M. (1998), Możliwości i efekty esses in the Polish Western Carpathians, realizacji polderów zalewowych dla wspoma- Geografiska Annaler, 69 A (1): 21–26. gania ochrony Krakowa przed powodzią [The Łajczak, A. (1995a), The Impact of River Regula- potential for polder application and effects of tion, 1850–1990, on the Channel and Flood- existing polders as a Kraków flood protection plain of the Upper Vistula River, Southern measure], Materiały Konferencji Naukowej Poland, in Hickin, E.J. (ed.) River Geomor- nt.’Powódź w dorzeczu górnej Wisły w lipcu phology, Wiley, Chichester, 209–233. 1997 roku’, Kraków, Oddział PAN, 277–285. Łajczak, A. (1995b), Potential Rates of the Osuch, B. (1991), Stany wód [Water Stages], in Present-Day Overbank Sedimentation in the Dynowska, I. and Maciejewski, M. (eds.) Vistula Valley at the Carpathian Foreland, Dorzecze górnej Wisły [Drainage Basin of the Southern Poland, Quaestiones Geographicae, Upper River Vistula], part I, Państwowe Wy- 17/18: 41–53. dawnictwo Naukowe (PWN), Warszawa-Kra- Łajczak, A. (1999), Współczesny transport i se- ków, 159–166. dymentacja materiału unoszonego w Wiśle Punzet, J. (1959), Wpływ zbiornika w Goczał- i głównych dopływach [Contemporary Trans- kowicach na ostatnie wezbranie Małej Wisły port and Sedimentation of the Suspended [Impact of the Goczałkowice Reservoir on Material in the Vistula river and its main the Last Flood of the Small Vistula River], tributaries], Monografie Komitetu Gospodar- Gospodarka Wodna, 3: 23–28.
KKsisiąążżkka1.indba1.indb 9955 22008-06-26008-06-26 110:56:350:56:35 96 Adam Łajczak
Punzet, J. (1973), Wpływ zbiorników zaporowych latach [Evolution of the Fluvial System of the na prawdopodobieństwo występowania wiel- Middle and Lower Raba River (Carpathians, kich wód w dorzeczu górnej Wisły [Influence Poland) during the last 200 years], Dokumen- of Dam Reservoirs on Probability of Occur- tacja Geograficzna, 6, Instytut Geografii i rence of High Water Stages within the Upper Przestrzennego Zagospodarowania (IGiPZ), River Vistula Drainage Basin], Gospodarka PAN, Warszawa: 1–92. Wodna, 2: 46–49. Wyżga, B. and Lach, J. (2002), Współczesne wci- Punzet, J. (1981), Zmiany w przebiegu stanów nanie się karpackich dopływów Wisły— przy- wody w dorzeczu górnej Wisły na przestrzeni czyny, środowiskowe efekty oraz środki za- 100 lat (1871–1970) [Changes in the Course radcze [The Present-Day Channel Deepening of Water Stages of Rivers within the Upper of the Carpathian Tributaries to the River River Vistula Drainage Basin over the period Vistula—Causes, Environmental Effects and 1871–1970], Folia Geographica, ser. Geo- Prevention Methods], Problemy Zagospoda- graphica-Physica, 14: 5–28. rowania Ziem Górskich, 48: 23–29. Punzet, J. (1991), Przepływy charakterystyczne Ziemońska, Z. (1973), Stosunki wodne w polskich [Characteristic discharges], in Dynowska, I Karpatach Zachodnich [Hydrological Condi- and Maciejewski, M. (eds.), Dorzecze górnej tions of the Polish Western Carpathian Mts.], Wisły [Drainage Basin of the Upper Vistula Prace Geograficzne, 103, Instytutut Geografii River ], part I, Państwowe Wydawnictwo Na- i Przestrzennego Zagospodarowania (IGiPZ) ukowe (PWN), Warszawa-Kraków, 167–215. PAN, 1–127. Soja, R. and Mrozek, T. (1990), Hydrological Żelazo, J. (1993), The Recent Views on the Small Characteristics of the Vistula River, in Star- Lowland River Training, in Tomiałojć, L. kel L. (ed.) Evolution of the Vistula River Val- (ed.), Nature and Environment Conservation ley During Last 15 000 Years Part III, Geo- in the Lowland River Valleys in Poland, PAN, graphical Studies, Special Issue, 5: 45–62. Kraków, 145–154. Wit, M. (1999), Obniżenie kulminacji fali po- wodziowej przez polder Smolice do ochrony Paper first received: March 2007 Krakowa przed powodzią [The lowering of In final form: December 2007 the floodwave using the Smolice polder as a Kraków flood protection measure], Gospo- darka Wodna, 3: 98–100. Wyżga, B. (1993), Funkcjonowanie systemu rzecz- nego środkowej i dolnej Raby w ostatnich 200
KKsisiąążżkka1.indba1.indb 9966 22008-06-26008-06-26 110:56:350:56:35 FACTORS INFLUENCING FLOODS IN THE URBANIZED AND INDUSTRIALIZED AREAS OF THE UPPER SILESIA INDUSTRIAL REGION IN THE 19TH AND 20TH CENTURIES (THE KŁODNICA CATCHMENT CASE STUDY)
DAMIAN ABSALON, STANISŁAW CZAJA and ANDRZEJ T. JANKOWSKI Faculty of Earth Science, University of Silesia, Będzińska 60, 41-200 Sosnowiec, Poland E-mails: [email protected]; [email protected]; [email protected]
Abstract: The occurrence and pattern of floods in urban industrial areas depend on both the hydro-meteorological and physico-geographical properties of the catchment area and on the de- gree of anthropogenic transformation of land. The area selected for research is one of the larg- est urban mining-industrial districts in Europe, known as the Upper Silesian Industrial Region (USIR). Besides the ‘typical’ flood risk, which manifests itself in rivers overflowing their banks, this catchment is also threatened with floods that do not depend on meteorological factors but are caused by the formation of flood lands in areas transformed due to deep mining of hard coal. The pattern of floods in the catchment has also been influenced by changes in the forms of land use resulting from the growth of urbanized areas. Because of the increasing flood risk and the fact that it is impossible to build water storage reservoirs other possibilities of improving water retention capacity in the catchment have been indicated.
Key words: hydrology, flood, human impact, urban area, Poland.
INTRODUCTION to a greater extent on the effects of human economic activity than on natural factors. The occurrence and pattern of floods in The Kłodnica is a right-bank tributary urban industrial areas depend on both hy- of the Odra, draining a river basin area of drometeorological and physicogeographi- 1085 km2. The Bytomka, on the other hand, is cal properties of the catchment and on the a right-bank tributary of the Kłodnica, drain- degree of anthropogenic transformation of ing a river basin area of 144.5 km2. The present land. Urban mining-industrial areas have study concerns the upper part of the Kłodnica a specific pattern of floods and inundations river basin (catchment area A = 505 km2). outside river valleys because the occurrence Mean annual discharge of the Kłodnica and pattern of this phenomenon depends at the Gliwice gauging station (catchment
KKsisiąążżkka1.indba1.indb 9977 22008-06-26008-06-26 110:56:360:56:36 98 Damian Absalon, Stanisław Czaja and Andrzej T. Jankowski
area A = 444 km2), calculated on the ba- lies within the Kłodnica catchment (Fig. 1), sis of data from the years 1961–1999 reach- where detailed hydro-meteorological obser- es 6.41 m3 s-1, and the annual Bytomka vations have been carried out since the mid discharge at the Gliwice gauging station 19th century. The collected material made (A = 136.5 km2), calculated on the ba- possible the accurate evaluation of human sis of data from the same period amounts impact on the pattern of catastrophic floods to 2.61 m3 s-1. Individual discharge values and inundations in areas located outside amount, respectively to: 14.4 dm3 s-1 km-2 for river valleys. the Kłodnica catchment and 19.1 dm3 s-1 km-2 for the Bytomka catchment. Both catchments are currently urbanized FLOODING RISK to a great extent and intensive mining activ- ity is carried out there. Due to this fact both The whole area under research is located rivers also carry foreign water, from outside within the Silesian Upland. The geographi- the catchment (for the water supply system) cal location and elevation of this area above and deep mining waters originating from the sea level were decisive factors in determin- draining of workings in hard coal mines. ing the precipitation total, which is 700 mm The area selected for research is one of on average. The greatest water runoff occurs the largest urban mining-industrial districts in the thaw period (March, April), while the in Europe known as the Upper Silesian maximum discharge occurs after long-term Industrial Region (USIR). Its western part or torrential rains, usually in June and July
Figure 1. The Kłodnica catchment against the background of a digital terrain model
KKsisiąążżkka1.indba1.indb 9988 22008-06-26008-06-26 110:56:360:56:36 Factors Influencing Floods in the Urbanized and Industrialized Areas of the Upper Silesia… 99
(Czaja and Jankowski 1993; Absalon et al. ering of first-level water table—this usually 2001). High-water stages after long-term happens in places where Quaternary sur- and torrential rains pose the greatest flood- face deposits lie directly above the exploited ing threat in the USIR area. Carboniferous level. On the other hand, wa- While evaluating the flood patterns of ter-logging due to the formation of extensive this area in accordance with the criterion subsidence basins and hollows may occur in linking the level of the high-water stage with areas of the deep mining of hard coal. These the probability of peak discharge, the USIR basins are often filled with water from sur- district falls into the category of areas in face runoff or groundwater recharge. The which normal high water prevails, where the formation of basins and hollows in river val- flood discharge: leys also leads to local disruptions in falls of the ground.
Q50% ≥ Qmax ≥ (Q + Q50%) / 2, These processes result in changes in sur- and medium high water, where the flood face and underground water retention and discharge: in the conditions for infiltration and surface runoff. A major role in modifying the con- ditions of water circulation and therefore in Q10% ≥ Qmax ≥ Q50%, generating floods in the area in question has Great high-water, where the flood dis- charge: been played by changes in spatial develop- ment and the surface hydrographical net- work. In the mid 1730s forests and coppices Q50% ≥ Qmax ≥ Q10%, comprised about 45% of the Kłodnica catch- occurs very rarely, and catastrophic ment area, arable land, meadows and pas- floods, where the flood discharge: tures nearly 46% and ponds and reservoirs Q > Q , 6% (Wieland 1736). However, as early as in max 5% the 1830s, major changes in the development occurred only twice within the last 200 of this area were observed. Forests were cut years—in 1940 and 1997 (Fischer 1915; down on a large scale mainly for the pur- Powódź… 1967; Powódź… 1975; Ocena poses of the mining industry, which caused sytuacji… 1997). the diminution of their area by nearly 40% where: (Czaja 1999). There were also changes in Qmax—maximum instantaneous discharge (peak discharge), forest structure. Deciduous forests were sub- stituted by quick-growing pines and spruces. Qp%—maximum instantaneous discharge (probability p%). In the 19th century the dominant forms of land use were arable lands, meadows and Human economic activity radically pastures, which constituted over 70% of the transformed natural conditions of the oc- described area (Tables 1, 2). currence of floods in the USIR area. Ex- From the end of the 19th century on- tremely intensive development of mining wards, further changes in land development and industry and the urbanization of the of the catchment took place. The arable area have taken place since the mid 19th land, which had prevailed in the landscape, century. This activity has resulted in dis- was taken over for the purposes of construc- tinct changes in the geographical environ- tion and industrial development. In the 20th ment of the area, on the one hand resulting century, there was a steady decrease in the in orogenic drainage in areas where drain- area of arable land in favour of residen- age of coal mining floors also results in low- tial and industrial buildings and roads and
KKsisiąążżkka1.indba1.indb 9999 22008-06-26008-06-26 110:56:410:56:41 100 Damian Absalon, Stanisław Czaja and Andrzej T. Jankowski
Table 1. Land use in the Bytomka catchment
1827 1887 1960 2000
Land use km2 % km2 % km2 % km2 %
Forests 35.5 24.1 27.2 18.5 21.0 14.3 26.2 19.2
Arable land 93.4 63.4 89.5 60.8 66.6 45.2 44.8 32.9
Meadows and pastures 13.5 9.2 13.8 9.3 10.2 6.9 9.2 6.8
Urban area 4.7 3.2 7.5 5.1 18.7 12.6 25.2 18.5
Industrial area 0.0 0.0 2.4 1.6 7.3 5.0 11.2 8.2
Roads and railway lines 0.0 0.0 0.3 0.2 3.0 2.0 1.8 1.3
Green areas 0.0 0.0 0.6 0.4 3.8 2.6 8.6 6.3
Wasteland and post-industrial lands 0.0 0.0 5.7 3.9 14.7 10.0 8.0 5.9
Water reservoirs 0.2 0.1 0.3 0.2 2.0 1.4 1.2 0.9
Total 147.3 100.0 147.3 100.0 147.3 100.0 136.2 100.0
Table 2. Land use in the Kłodnica catchment
1827 1887 1960 2000
Land use km2 % km2 % km2 % km2 %
Forests 104.1 29.0 97.6 27.3 76.7 21.4 90.4 25.2
Arable land 208.5 58.2 207.7 57.9 182.4 50.9 135.0 37.7
Meadows and pastures 35.2 9.8 35.7 9.9 32.0 8.9 28.6 8.0
Urban area 8.2 2.3 10.6 3.0 33.3 9.3 55.6 15.5
Industrial area 0.0 0.0 2.4 0.7 12.2 3.4 20.3 5.7
Roads and railway lines 0.0 0.0 0.4 0.1 2.1 0.6 1.0 0.3
Green areas 0.0 0.0 0.5 0.1 7.5 2.1 11.2 3.1
Wasteland and post-industrial lands 0.0 0.0 3.3 0.9 10.7 3.0 13.0 3.6
Water reservoirs 2.4 0.7 0.2 0.1 1.5 0.4 3.3 0.9
Total 358.4 100.0 358.4 100.0 358.4 100.0 358.4 100.0
railway lines. These processes were par- hamlets with farm buildings. These took up ticularly intensive within the catchment about 2% of catchment area. By the end of of the Bytomka, the main tributary of the 19th century, the number and area of settle- Kłodnica, draining the most urbanized and ments had increased slightly. Rapid urban industrialized part of the Upper Silesian development occurred towards the end of Industrial Region (Tables 1, 2). Until the the 19th century, when numerous residen- end of the 18th century dispersed settle- tial districts for factory workers and miners ment prevailed here, with small villages and were built and former villages were turned
KKsisiąążżkka1.indba1.indb 110000 22008-06-26008-06-26 110:56:420:56:42 Factors Influencing Floods in the Urbanized and Industrialized Areas of the Upper Silesia… 101
into urban settlements. By this time, such some of the post-industrial lands for their areas were comprising about 10% of the By- afforestation and the creation of parks and tomka catchment, though only 3.5% of the green areas. Urban green areas have only Kłodnica catchment. The figures at the end been playing a role since the mid 20th cen- of the 20th century were 27% and 21%, re- tury. These areas currently account for over spectively (Figs 2, 3, 4 and 5). 6% of the Bytomka catchment.
Figure 2. Land use in the upper Kłodnica catchment in 1827
With the development of industry and The Kłodnica and Bytomka catchments urbanization the research area witnessed are not unusual in their transformation of rapid growth of degraded and post-indus- the land surface—as early as at the begin- trial land. The main reason was storage of ning of the 1970s, it was estimated that about mining waste (barren rock) and metallur- 9% of the area of the Upper Silesian con- gical waste (slag and post-flotation waste) urbation had been subject to 100% change mainly on agricultural land and grassland. (Żmuda 1973). In the 1980s the conurba- At the end of the 19th century, degraded ar- tion went through intensive transforma- eas constituted nearly 4%, and at the end of tions of water relations over an area of about the 20th century, 6% of the Bytomka catch- 1,600 km2 (Jankowski 1987). ment area. Changes in surface hydrographical net- A positive trend in the changes in spatial work are another important factor in the development aiming at amelioration of wa- formation of floods in the Kłodnica val- ter circulation conditions is reclamation of ley and the valleys of its tributaries, which
KKsisiąążżkka1.indba1.indb 110101 22008-06-26008-06-26 110:56:420:56:42 102 Damian Absalon, Stanisław Czaja and Andrzej T. Jankowski
Figure 3. Land use in the upper Kłodnica catchment in 1887
Figure 4. Land use in the upper Kłodnica catchment in 1960
KKsisiąążżkka1.indba1.indb 110202 22008-06-26008-06-26 110:56:440:56:44 Factors Influencing Floods in the Urbanized and Industrialized Areas of the Upper Silesia… 103
Figure 5. Land use in the upper Kłodnica catchment in 2000
are heavily transformed anthropogenically. of calculated hydrological parameters. De- Most of the fishponds and dam reservoirs pending on the extent of subsidence, its lo- on rivers and streams so numerous in the cation off the centre of the basin or hollow 18th and 19th centuries have disappeared, and the location of major sewerage collector something that significantly diminished the outlets areas came to be included into or ex- water retention properties of the catchment cluded from one catchment or another. This and hurried surface water runoff. At the phenomenon was observed in the Bytomka same time hydrotechnical works and regula- catchment, which lost 11 km2 (7.5% of total tion of rivers and streams have caused their catchment area) in the years 1960–2000. shortening, making them steeper and as The present occurrence of floods and in- a consequence quickened water runoff (Cza- undations in the USIR part of the Kłodnica ja 1999). It follows from the analysis of his- catchment area is connected mainly with torical topographic maps from the 19th and land deformations due to open pit and deep 20th centuries that the river network in the mining. Numerous subsidence basins and Bytomka catchment (excluding the Mikul- hollows are formed, both within active min- czycki stream catchment) was 89.9 km long ing fields and on their edges. Subsidence in 1827 and 36.4 km long in 2000. basins and hollows are formed within min- Mining activity and the development of ing fields as the result of the exploitation the sewerage network in the urbanized ar- of resources carried out using the roof-fall eas located in zones crossed by watersheds method. Land surface deformations in min- result in major changes in catchment area, ing subsidence zones often cause changes in something that influences the magnitude the direction of surface and underground
KKsisiąążżkka1.indba1.indb 110303 22008-06-26008-06-26 110:56:480:56:48 104 Damian Absalon, Stanisław Czaja and Andrzej T. Jankowski
water runoff, leading in consequence to and streams draining areas within the zones of flooding of land located beyond river val- hollows and extensive subsidence basins. The leys (Fig. 6). riverbeds have been sealed by building stone
Figure 6. Changes in surface features in the Kłodnica valley (transverse section): 1—flat waste heaps of barren rock, 2—local subsidence basins, 3—potential level of water in land depressions. Source: Szczypek, Wach (1987)
The specific character of the described and concrete troughs and flood embank- area results from the fact that, in natu- ments. Intensive ground subsidence made it ral conditions, normal and medium-sized necessary to raise embankments, which has floods prevailed there. Changes in geologi- resultedraised the surface of water in rivers cal structure (as the result of mining activ- much above valley-floors. In many cases the ity) caused great ground subsidence, which level of water in an embanked riverbed is in turn brought permanent inundations of a few metres higher than the valley-floor be- river valleys. This phenomenon is particu- yond the embankments. This often causes the larly dangerous in the built-up areas of the formation of wide flood-lands collecting pre- region. In times of flooding, inundations of cipitation waters in river valleys. Flood-lands considerable areas can be observed in min- forming along the embankments of rivers and ing subsidence zones, mainly in connection streams are particularly dangerous when it with obstructed flow of surface and ground rains. The scale of flooding risk in river valleys waters (Szczypek and Wach 1987; Wach and in the USIR area can be proven by the extent Szczypek 1996; Czaja and Wach 1999). of flooded land during the July 1997 flood. In Besides inundations due to the subsid- some sections of the Kłodnica and its tributar- ence and sinking of ground, the USIR area ies and other rivers of the region the extent of is threatened with flooding caused by the ob- flooding was wider than the theoretical range structed flow of surface waters into the river of inundation of these valleys with the prob-
network. These obstructions result from rivers ability of peak discharge of Q0.1%.
KKsisiąążżkka1.indba1.indb 110404 22008-06-26008-06-26 110:56:500:56:50 Factors Influencing Floods in the Urbanized and Industrialized Areas of the Upper Silesia… 105
THE HISTORY OF FLOODING
In the 19th and 20th centuries floods on the upper Kłodnica were quite frequent. The 1803 flood caused considerable loss to prop- erty, destroying quite a long section of the newly-built Kłodnicki Canal. Water in the river in the Gliwice area reached the level of 2.0—3.3 m, as was the case in 1803, 1903, 1913, 1915, 1925 and 1930. Unfortunately no discharge data are available for the period before 1911 (the gauging station was installed in 1908) and we only know the descriptions Figure 8. Gliwice during the 1940 flood of the magnitude, extent and effects of the – Krakowska Street area. floods from historical sources (Mann 1905, Fischer 1915, Knothe 1939). Since the moment when systematic obser- Such a rapid increase in discharge attests vations of water stages started to be carried to a considerable intensity of rainfall over out at the Gliwice gauging-station the peak a comparably small area, because the time flood discharges recorded in different years of discharge concentration should be taken were: 1913—61.0 m3s-1; 1915—83.0 m3s-1; into account. The flood of 19–20 August 1925—66.6 m3s-1; 1930—64.7 m3s-1. The 1854 was quite similar. It follows from inter- lar-gest flood was recorded on the Kłodnica polation of flood marks that the discharge in late May and early June 1940. In Gliwice, of the Kłodnica in Gliwice might then have the recorded water stage was H = 505 cm equalled 80 m3s-1. Since those floods there and discharge Q = 121.5 m3s-1. Within have been other large events, the next major 22 hours there was an increase in discharge one being the flood of July 1997. The esti- from 11.0 m3s-1 to 121.5 m3s-1, the result be- mated discharge of the Kłodnica on the Gli- ing flooding in large areas of Gliwice, mostly wice gauging-station was 88.1 m3s-1. Slightly in the vicinity of the present Silesian Techni- smaller discharges were recorded during cal University (Figs. 7, 8). floods in: 1968—52.1 m3s-1, 1972—50.3 m3s-1 and 1985—48.2 m3s-1 (Kompleksowy pro- gram… 1998). The origin of most floods in the Kłodnica and Bytomka catchments may be sought in the spring thaw period. Only the largest floods resulted from storm rainfall, as in May/June 1940. An analysis of the peak high water observed on the largest tributary of the Kłodnica—the Bytomka—was also carried out. The Bytomka catchment, as has been shown before, is one of the most urbanized and transformed areas in the Kłodnica Figure 7. The pattern to peak high water along catchment. Unfortunately, as regards this the Kłodnica in 1940 catchment, hydrological data are confined
KKsisiąążżkka1.indba1.indb 110505 22008-06-26008-06-26 110:56:510:56:51 106 Damian Absalon, Stanisław Czaja and Andrzej T. Jankowski
to the last 50 years, since the gauging-sta- The last 40 years has have also witnessed tion was only installed in 1955. Due to larger increases in discharges along the smaller catchment area, the observed peak Kłodnica than the Bytomka. This is due to discharges are lower (Figs. 9, 10 and 11). much more major changes occurring in the The analysis points to the Bytomka flood upper Kłodnica as compared to the By- wave usually being part of the Kłodnica tomka. These changes mainly include river flood wave, since its peak is observed over control works resulting in the strengthen- a dozen hours (12—18 hours) before the ing and shortening of the river network and peak on the Kłodnica. The flood of 1997 changes in land use in the catchment area was different because the Kłodnica peak in the direction of urban development, and occurred 12 hours earlier than the Bytom- the resulting development of a storm-water ka one (Fig. 12). drainage network. A similar situation was
Figure 9. Patterns to peak high water on the Figure 11. Patterns to peak high water on the Bytomka and Kłodnica in 1968. Bytomka and Kłodnica in 1985.
Figure 10. Patterns to peak high water on the Figure 12. Patterns to peak high water on the Bytomka and Kłodnica in 1972. Bytomka and Kłodnica in 1997.
KKsisiąążżkka1.indba1.indb 110606 22008-06-26008-06-26 110:56:510:56:51 Factors Influencing Floods in the Urbanized and Industrialized Areas of the Upper Silesia… 107
observed in the Ruhr Area (Ruhrgebiet), and prepared to carry flood waters with where channelling of rivers and streams and a probability of peak discharge of even the building of embankments resulted in 0.3—0.1%. However, most of the USIR ar- larg-scale inundations between the embank- eas are threatened by floods caused by sur- ments and the Emscher Valley escarpment face flow and underground runoff, which (Brüggenmeier, Rommelspacher 1992; Em- leads to flooding not only of river valleys but schergenossenschaft 1910). also of considerable areas located beyond The assessment of the impact of small them. Very often these are highly developed man-made reservoirs on the pattern of areas, such as areas of urban and industrial floods in the Kłodnica catchment is a very development, railway lines, etc. This threat difficult task requiring individual research. overlaps with the location of the ground In the second half of the 20th century two subsidence zone. Because it is impossible to opposing tendencies were in conflict in this restore the natural water retaining capacity catchment. On the one hand, due to the de- of the catchment, and practically also im- terioration of water quality and economic possible to build water storage reservoirs, transformations many fishponds disap- the use of some of the transformed mining peared. On the other hand an unintended areas and in particular water-filled depres- effect of deep mining was the formation of sions forming as a result of ground subsid- subsidence basins, these becoming filled ence should be considered to improve water with water under certain morphological retaining capacity of the catchment. Other and hydrogeological conditions. We are flood prevention activities in the catchment currently unable to define the relative pro- should aim at: portions of these two phenomena, particu- • the afforestation of wastelands, larly because the situation of water-filled • the proper irrigation and drainage of depressions is subject to dynamic change, agricultural lands, these often becoming filled up with waste • limiting the construction of flood em- barren rock from coal mining. In a situation bankments outside densely built-up urban of diminishing natural water retention in areas, the catchment, retaining some of the water- • improving water retention of river- filled depressions should be considered, in beds, e.g. by building multipartite channels. order to improve retention. One of the larg- • limiting the urbanization in river val- est water-filled post-mining depressions in leys. the Kłodnica Valley played a positive role, e.g. during the flood of 1997. REFERENCES
CONCLUSIONS AND FINAL REMARKS Absalon, D., Jankowski, A. T. and Leśniok, M. (2001), Komentarz do Mapy Hydrograficznej The area of the Upper Silesia Indus- Polski w skali 1:50 000, [Comments to Hydro- trial Region has been heavily transformed graphic Map of Poland 1:50 000] Arkusz M- as the result of long-term mining activity, 34-62-A (Gliwice), Główny Geodeta Kraju, urbanization and industrialization. A mi- Warszawa. nor flood risk resulting from natural condi- Brüggenmeier, F. J. and Rommelspacher, T. tions has been multiplied mainly by mining (1992), Blauer Himmel über der Ruhr, Ge- activity and urbanization. The main rivers schichte der Umwelt im Ruhrgebiet 1840–1990, draining the region have been controlled [Blue sky over the Ruhr, History of the
KKsisiąążżkka1.indba1.indb 110707 22008-06-26008-06-26 110:56:520:56:52 108 Damian Absalon, Stanisław Czaja and Andrzej T. Jankowski
environment in the Ruhr Area 1840–1990] Kompleksowy program zabezpieczenia miasta Klartext Verlag, Essen. Gliwice przed powodzią wraz z instrukcją Czaja, S. (1999), Zmiany stosunków wodnych w wa- działania na wypadek zagrożenia powodzio- runkach silnej antropopresji (na przykładzie wego [Complex flood protection programme konurbacji katowickiej), [Changes in wa- of Gliwice and way of action in case of flood ter relations under the conditions of strong emergency] (1998), Unpublished report, Hy- anthropopression (a case study of Katowice droprojekt, Warszawa. conurbation)] Prace Naukowe Uniwersytetu Mann, H. (1905), Hochwasser vom August/Sep- Śląskiego, 1782, Wydawnictwo Uniwersytetu tember 1813, [August/September flood in Śląskiego, Katowice. 1813], Jahrbuch für die Gewässerkunde Nord- Czaja, S. and Jankowski, A. T. (1993), Changes deutschlands, Bd. 1, Nr 2, Königliche Hof- of River Runoff from Urbanized and Indus- buchhandlung, Berlin. trialized Region of Upper Silesia, in Bana- Ocena sytuacji hydrometeorologicznej dla dorzecza sik, K. and Żbikowski, A. (eds.), Runoff and górnej Wisły i Odry w granicach województwa Sediment Yield Modeling (RSY-93), Pro- katowickiego w okresie 6–25 lipca 1997 r. [As- ceedings of the International Symposium held sessment of hydrological and meteorological at Warsaw, Agricultural University—SGGW, conditions in the Upper Vistula and Upper Warszawa. Oder river basins in Katowice region between Czaja, S.and Wach, J. (1999), Ocena zagrożenia 6 and 25 July 1997] (1997), Unpublished re- powodziowego na obszarze gminy Gierałtowi- port, Instytut Metetorologii i Gospodarki ce, [Assessment of flood risk in the Gieral- Wodnej (IMGW) Oddział Katowice. towice commune], Unpublished report, CI- Powódź w roku 1960 [The Flood in 1960] (1967), TEC, Katowice. Instytut Gospodarki Wodnej PIHM, WKiŁ, Emschergenossenschaft [The Emscher coopera- Warszawa. tive] (1910), Das gewerbliche Abwasser im Powódź w sierpniu 1972 r [The Flood in August Emschergebiet, [Industrial sewage in The 1972] (1975), Instytut Meteorologii i Gospo- Emscher Area], Essen. darki Wodnej (IMGW), WKiŁ, Warszawa. Fischer, K. (1915), Niederschlag und Abfluss im Szczypek, T., Wach, J. (1987), Prognoza rozwo- Odergebiet, [Precipitation and Runoff in the ju krajobrazu antropogenicznego na przy- Oder River], Jahrbuch für die Gewässerkun- kładzie rzeki Kłodnicy (Górnośląski Okręg de Norddeutschlands, Bd. 3, Nr 2, Königliche Przemysłowy), [Predicted development of Hofbuchhandlung, Berlin. anthropogenic landscape on the example of Jankowski, A. T. (1987), Wpływ urbanizacji the Klodnica river (USIR)], in: Kulturní kraji- i uprzemysłowienia na zmianę stosunków ny v průmyslovych oblastech, IV, ČSAV, Brno. wodnych w regionie śląskim w świetle dotych- Wach, J., Szczypek, T. (1996), Preobrazovaniya czasowych badań, [Influence of urbanization rel’yefa mestnosti v rayonakh hornodobyvayus- and industrialization on water conditions in hchey promyshlennosti vsledstvie osedanii the Silesian Region as perceived by the up- grunta (na primere Katovitskovo Voyevod- to-date investigations], Geographia, Studia et stva), [Relief transformations as a result of Dissertationes, 10, Prace Naukowe Uniwersyte- terrain subsidence in deep mining areas (on tu Śląskiego, 813, Wydawnictwo Uniwersytetu the example of Katowice region)], Belorusskii Śląskiego, Katowice. Gosudarstvennyi Universitet, Belorusskoye Knothe, H. (1939), Das schlesische Sommerhoch- Geograficheskoye Obshchestvo, Minsk. wasser 1938, [Flood in Silesia in the summer Wieland, I. W. (1736), Principatus Silesiae Ratti- of 1938], Verlag Priebatsch Buchhandlung, boriensis, Wojewódzkie Archiwum Państwo- Breslau. we, Katowice.
KKsisiąążżkka1.indba1.indb 110808 22008-06-26008-06-26 110:56:520:56:52 Factors Influencing Floods in the Urbanized and Industrialized Areas of the Upper Silesia… 109
Żmuda, S. (1973), Antropogeniczne przeobrażenia środowiska przyrodniczego konurbacji górno- śląskiej, [Anthropogenic transformations of natural environment in Upper Silesia Conur- bation], Państwowe Wydawnictwo Naukowe, Warszawa-Kraków.
Paper first received: March 2007 In final form: November 2007
KKsisiąążżkka1.indba1.indb 110909 22008-06-26008-06-26 110:56:530:56:53 STUDIES ON HISTORICAL FLOODS IN GDAŃSK (A METHODOLOGICAL BACKGROUND)
KATARZYNA MAROSZ Institute of Oceanography, University of Gdańsk, Piłsudskiego St. 46, 81-378 Gdynia, Poland E-mail: [email protected]
Abstract: The analysis and reconstruction of historical floods not only enriches historical docu- mentation, but can also be perceived as a useful auxiliary tool in the planning of contemporary flood management techniques. The reconstruction of floods on the basis of historical documents can be carried out with the help of a GIS (Geographical Information System) equipped with the spatial analysis tools allowing the extent of flooding to be mapped. The present study reviews his- torical floods in Gdańsk briefly, before attempting to reconstruct one particular historical flood.
Key words: flood, reconstruction, historical documents, DEM, GIS, Gdańsk, Poland.
INTRODUCTION However, it is not the mere merging of the two disciplines of hydrology and histo- The investigation of floods from the past ry, as could be expected by the term and is an element of paleohydrology or histo- the quoted definition. Rather, the research rical hydrology (depending on the level of utilises input from other disciplines, such as documentation available). Paleofloods are geography, cartography, climatology, geo- defined as: any past or ancient flood events morphology, town-planning and environ- which occurred prior to the time of human mental engineering. observation or direct measurement by mo- dern hydrological procedures, while histori- cal floods are defined as: any flood events FLOODS IN GDAŃSK documented by human observation and re- corded prior to the development of systema- Since the beginning of its existence, Gdańsk tic streamflow measurements (Hirschboeck has been under permanent threat of flood- 2003). Historic hydrology is an emerging ing, due to its location. The immediate pro- branch in the hydrological sciences. One ximity of the Vistula River delta as well as of definition, proposed by Brázdil and Kun- the Baltic Sea augments the risk, and a flood dzewicz (2006) reads: a field situated at the may enter from either side (cf. Fig. 1). interface of hydrology and (environmental) During its multi-century history Gdańsk history, dealing mainly with documenta- has survived many types of flooding caused ry sources and using both hydrological and by such mechanisms as rainfall, snowmelt, historical methodologies. ice jams, and coastal storm surges. Human
KKsisiąążżkka1.indba1.indb 111111 22008-06-26008-06-26 110:56:530:56:53 112 Katarzyna Marosz
Figure 1. Natural causes of floods in Gdańsk.
activity (advertent or inadvertent) was also Catastrophic floods from the past were the cause of so-called anthropogenic floods. commemorated by the flood marks (high- In the past, the improper use of the hydro- water signs), cf. Fig. 2. Three of these still technical devices so numerous in the Żuławy remain in Gdańsk: on the wall of the Maiden area, or insufficient care in the conservation Granary on Ołowianka Island, near the Stą- of levees were causes of frequent floods in giewny Bridge, and on the bridge near the the western part of the Vistula River delta, Stone Flood-gate. the Żuławy of Gdańsk and Gdańsk city it- self (1456, 1512 and 1854). Levees were also deliberately destroyed during the wars in which Gdańsk was often involved as one of the major Hanseatic cities (e.g. that between Poland and Sweden in 1656).
THE SIGNIFICANCE OF DOCUMENTARY SOURCES Dramatic events such as catastrophic floods have always drawn people’s attention. They exert considerable impacts on both the na- tural and built environment and on the po- pulation of the cities. They may also serve as sources of inspiration for painters, poets and chroniclers, who testify to the tragic events Figure 2. High-water mark on the Maiden in question. Granary.
KKsisiąążżkka1.indba1.indb 111212 22008-06-26008-06-26 110:56:530:56:53 Studies on Historical Floods in Gdańsk 113
The key part in the process of reconstru- tion on the catastrophic flood may also be ction of historical floods is played by vario- present in parish and monastery archives. us documents describing the event, and it Other sources are year-books, regulations is crucial to use different sources wherever and accountant registers. The form of in- possible (Barnikel 2004). Such an approach formation constitutes another criterion, e.g. allows the flood event itself to be looked at old prints and manuscripts, or cartographic from many points of view. However, over and graphical materials, such as pictures and 80% of historical events (from the time photographs. before the advent of regular hydrological The search for documents is very time observations) were documented by a single consuming and among the major problema- manuscript only. The rule that appears to tic issues is the homogeneity of the informa- be valid is that, the more recent the event, tion concerning one event, based on several the more complete the documentation ava- different sources (Cyberski et al. 2006). The ilable (Barnikel 2004). Another rule states gathered material has to be interpreted with that “the more extreme the event is, the caution. Some documents may in fact be sub- more widespread and detailed the descrip- jective, while historical descriptions may of- tion that can be found” (Glaser and Stangl ten contain vague information, such as: “the 2004). unparalleled catastrophe so far”. The issue Depending on the assumed criteria, the of authenticity of the sources is also impor- sources of information can be divided into tant. Any relation taking place even within specific groups. The main criterion is the a few years of an event taking place may al- source age (archive vs contemporary sour- ready be subject to distortion and confabula- ces). Numerous libraries and cartographic tion (Barnikel 2004) sources are available for cities that consti- tuted an important part of a country’s eco- nomic and political life. Documentation was AN ATTEMPT TO RECONSTRUCT created by order of a city mayor or a King, A HISTORICAL FLOOD and was usually so expensive that not every settlement could even afford it. In the case The first step in any attempt at the recon- of archival sources, one of the major prob- struction of a flood event is the gathering of lems for the unskilled scientist lies in major the documentation describing it: maps from difficulties with their interpretation. One of the period/year concerned, descriptions of the reasons in the Polish case is that the ter- the flood extent and course thereof. It is not ritory of the land is that is currently Poland always possible to obtain all the necessary was long divided between the three invading material–sometimes it might not even have powers as Prussia, Austria and Russia, with been created. different rules in each being obeyed as re- In order to reconstruct a flood it is ne- gards documentation. Moreover, some ma- cessary to create a DEM–Digital Elevation nuscripts have become partially damaged Model. Unfortunately, the ordinates of the over time, and rendered partly or fully ille- terrain at the time of the selected flood gible. The above-mentioned difficulties are are not available in the case of Gdansk. reinforced by the fact that the handwriting The past elevation data do not correspond of authors of manuscripts may also be dif- with modern elevations due to a change in ficult to read. sea level. Furthermore, homographic maps Another criterion is ownership. Histo- (created by mathematical projections) have rical sources may be a part of private col- been in use since the 17th century only. The lections or national archives. In the case of maps used before that time, described as national institutions, the search comprises anamorphic, are a source of interpretation museums, offices, water management admi- problems. Each map should be assessed nistrations and other institutions. Informa- with the help of several criteria, such as:
KKsisiąążżkka1.indba1.indb 111313 22008-06-26008-06-26 110:56:540:56:54 114 Katarzyna Marosz
precision, faithfulness, cartometrics (cohe- Using the transparency tool one can rence of distances, angles and surfaces with overlay the reconstruction of a historical real values) and legibility. Sometimes the flood on a city map corresponding to the orientation of a map is not straightforward time of the event. The archival map must and should also therefore be identified (Jan- be geo-positioned using the characteristic kowska and Lisiewicz 1998). permanent markers on the contemporary To create a DEM, there is a need to input topographic map. Such points may be stre- the available data on elevation of different ets, bridges or churches. Unfortunately, the points, and subsequently to interpolate in- archival cartographic material, while unque- formation for the area of interest. The final stionable at first glance, often proves to have stage is the presentation of results. The qua- been distorted considerably when applied in lity of the projected surface depends greatly the geo-positioning process. For example, on the number of available data points, their Fig. 3 illustrates an attempt at the geo-posi- spatial distribution and accuracy and the tioning of a map from (as relatively recently interpolation algorithm used (Magnuszew- as) AD 1830. The process started with two ski 1999). Having generated the relief, the corresponding points that seemed to be the analyst may then try to identify the flooded most accurate. At this stage there was no in- area, using the historical information about dication that the map had not been prepa- the flood. red using the necessary mathematical pro-
Figure 3. Illustration of a problem with geo-referencing a historical map (see details in the text).
KKsisiąążżkka1.indba1.indb 111414 22008-06-26008-06-26 110:56:540:56:54 Studies on Historical Floods in Gdańsk 115
jections. The third point was supposed to be SUMMARY the Bear Bastion. With the transparency at 40% it is easy to see how far apart the corre- Research on multiple historical documents is sponding points lay on the archive material a time consuming and arduous process that from AD 1830 (Bear Bastion–dashed line) requires the development and usage of pro- and the contemporary map (Bear Bastion– per methodology (Hohensinner et al. 2003). solid line). Forcing the points overlay results In the case of the search for archive mate- in a significant distortion of the map. rial there can never be final certainty that Glaser and Stangl (2004) use a three-de- everything has been discovered, hence it is gree classification of the intensity of histori- often the case that a subsequent investigator cal floods (after Sturn), cf. Tab. 1. Gdańsk may prove capable of extracting new infor- has suffered from floods of all three degrees mation and interpreting an event of interest of severity, including frequent floods of the in a more reliable and accurate way. Thus, second and third degree. For example, the analysis and reconstruction of historical flood of 1829 was, according to experts, the floods not only enriches the historical docu- most catastrophic in the history of Gdańsk. mentation, but can also improve our under- Flood-gates, ditches and bridges surroun- standing of flood processes, and hence serve ding the city were destroyed. The Long Gar- as a useful auxiliary tool where the planning dens disappeared under water to a depth of of contemporary flood management techni- one metre. The water level in the Lower City ques is concerned. Proper analysis of histo- reached the second floor. The water level rical sources and the corollaries drawn from at Ołowianka Island was 3.36 m above the such analysis may be of use in the process of mean sea level (Makowski 1994). The visu- spatial planning and in the design of natural alization allows for the classification of the disaster protection systems (Tropeano and event as “a great flood”. Turconi 2004).
Table 1. Classification of the intensity of historical floods by Sturn (after Glaser and Stangl 2004)
Level Classification Primary indicators Secondary indicators
1 Smaller regional flood Little damage, e.g. fields and Short-term flooding gardens close to the river, wood supplies that were stored close to the river are moved to another place
2 Above average, or supra-regional Damage to buildings and water- Flood of average duration, severe flood related infrastructure, such as damage to fields and gardens close dams, weirs, footbridges, bridges; to the river, loss of animals and and buildings located close to the sometimes people river (e.g. mills). Water in buildings
3 Above average, or supra-regional Severe damage to buildings and Duration of flood: several days flood of disastrous scale water-related infrastructure, or weeks; severe damage to such as dams, weirs, footbridges, fields and gardens close to the bridges; and buildings located river, extensive loss of animals close to the river (e.g. mills). and people; morpho-dynamic Water in buildings. Some buildings processes like sand sedimentation are completely destroyed or torn cause lasting damage and change away by the flood the surface structure
Source: Glaser and Stangl 2004.
KKsisiąążżkka1.indba1.indb 111515 22008-06-26008-06-26 110:56:550:56:55 116 Katarzyna Marosz
REFERENCES Jankowska M., Lisiewicz S. (1998), Kartograficzne i geodezyjne metody badań zmian środowiska Barnikel, F. (2004), The value of historical docu- [Cartographic and geodetic methods of sur- ments for hazard zone mapping, Natural Haz- veying environmental changes], Wydawnic- ards and Earth System Sciences, 4: 559–613. two Akademii Rolniczej, Poznań. Brázdil R., Kundzewicz Z.W. (2006), Historical Magnuszewski A. (1999), GIS w geografii fizycznej hydrology–Editorial, Hydrological Sciences [GIS in physical geography], Wydawnictwo Journal, 51 (5): 733–738. Naukowe PAN, Warszawa. Cyberski J., Grześ M., Gutry-Korycka M., Nachlik Makowski J. (1994) Największa katastrofalna po- E., Kundzewicz Z. (2006), History of floods wódź w dziejach Gdańska i prawdopodobień- on the River Vistula, Hydrological Sciences stwo jej powtórzenia w obecnych warunkach Journal, 51 (5):799–817. [The largest catastrophic flood in the history Glaser R., Stangl H., (2004), Climate and floods of Gdańsk and probability of its occurrence in in Central Europe since AD 1000: data, present conditions]. Biblioteka naukowa hy- methods, results and consequences, Surveys drotechnika 18, Instytut Budownictwa Wod- in Geophysics, 25: 485–510. nego PAN, Gdańsk. Hirschboeck K. K. (2003), Floods, paleofloods, Tropeano D., Turconi L. (2004), Using histori- and droughts: Insights from the upper tails, cal documents for landslide, debris, flow and CLIVAR/PAGES/IPCC Drought Workshop, stream flood prevention. Applications in November 18–21. Northern Italy, Natural Hazards, 31: 663–679. Hohensinner S., Jungwirth M., Habersack H. (2003), Historical analysis as a reference for Paper first received: March 2007 restoration of the Danube floodplain system In final form: December 2007 in the Austrian Machland, International Con- ference ‘Towards natural flood reduction strat- egies’, Warsaw, 6–13 September.
KKsisiąążżkka1.indba1.indb 111616 22008-06-26008-06-26 110:56:550:56:55 HYDROLOGICAL DROUGHTS IN CENTRAL POLAND—TEMPORAL AND SPATIAL PATTERNS
EDMUND TOMASZEWSKI Faculty of Geographical Sciences, University of Łódź, 90-139 Łódź, ul. Narutowicza 88, Poland E-mail: [email protected]; [email protected]
Abstract: The aim of this contribution has been to identify severe hydrological droughts in central Poland, and to analyse the temporal and spatial patterns they display. The distinguishing of low- flow periods was based on the threshold level method, where the SNQ (mean value of the mini- mum annual runoff) was used as the criterion. Basic calculations were made for daily discharge series at 29 gauging stations situated in the basins of the Rivers Warta, Pilica and Bzura over the time period 1966–1983. Analysis involved such parameters as: mean and maximum low-flow du- ration in half-years, low-flow type index, date of commencement and termination and character- istics connected with minimum runoff: date of occurrence, index of position as well as recession time index. The problems of hydrological drought stability over a multi-annual timeframe, as well as the spatial pattern thereto were also analysed.
Key words: hydrological drought, low flows, central Poland
INTRODUCTION The terms: hydrological drought and low-flow period are well-known in the field The low-flow problem attracts many research- of hydrology. However, various methods de- ers because of its connection with significant fine these phenomena in different ways. One elements of the hydrological cycle, as well as approach is based on a threshold level. A pe- its impact on the management of water re- riod for which the runoff attains values be- sources. However, as an extreme event, low low the established limit is selected as runoff flows have been investigated to a much lesser deficit. Its two basic parameters are: low-flow extent than flood waves, which develop fast, duration and deficit volume (Fig. 1). There and may have immediate destructive conse- are two methodological approaches allowing quences. However, in the long run, drought analysts to select the threshold: conventional and runoff deficit may reduce water resources (connected with water management) or sta- to a level jeopardizing development of local tistical. The former approach assumes that areas or even whole countries. For this rea- the threshold should be derived from a flow
son the identification of low flows and study duration curve such as the percentile Q70
of various aspects to their development have or Q90 (Hisdal et al. 2004). The latter uses both practical and scientific importance. the minimum annual daily discharge in the
KKsisiąążżkka1.indba1.indb 111717 22008-06-26008-06-26 110:56:560:56:56 118 Edmund Tomaszewski
Figure 1. Parameters of a hydrological drought
DEF—runoff deficit, t—duration, tR—recession time, Qmin—minimum low flow during hydrological drought
calculation of SNQ (mean minimum runoff), interval of 91%–99% on the flow duration WNQ (maximum runoff out of the minima) curve (Fig. 3). or ZNQ (median minimum runoff), cf Ozga- Two conditions other than SNQ level Zielińska (1990). were also taken into consideration. A sepa- rate low-flow period had to be longer than or or equal to 5 days, and if the interval be- STUDY AREA AND DATA tween two consecutive periods was of less than 3 days, both were treated as one pe- A set of 29 water-gauges situated in the ba- riod. The described procedure led to iden- sins of the Warta, the Pilica, and the Bzura tification of the low-flow periods and an (Fig. 2) was selected for analysis. It is worth estimation of their basic parameters: date noting that their location reflects simple re- of commencement and termination, dura- gimes of small rivers in homogeneous basins, tion, runoff deficit, minimum runoff and its as well as a more complex alimentation re- date of appearance, as well as recession time gime in larger basins of heterogeneous water (Fig. 1). Basic computations were carried courses. Basic calculations were made for out using NIZOWKA 2003 software (Ja- daily discharge series for the observation pe- kubowski 2004). The resulting parameters riod 1966–1983, as published by the IMGW and characteristics, which were constructed (Institute of Meteorology and Water Man- on this basis, allowed the author to analyse agement). hydrological droughts in both their temporal One of the main purposes of this study and spatial aspects. was to distinguish the very major hydrologi- cal droughts appearing after severe mete- orological droughts. Therefore, the most TYPE AND DURATION OF HYDROLOGICAL extreme option—SNQ was established as DROUGHT a threshold level for estimation. A compari- son with other criteria showed that, in most Basic analyses were made on the basis of sea- cases, the estimated SNQ level reached the sonal characteristics. Firstly, the low-flow
KKsisiąążżkka1.indba1.indb 111818 22008-06-26008-06-26 110:56:570:56:57 Hydrological droughts in central Poland—temporal and spatial patterns 119
Figure 2. Locations of the studied water-gauges Basin areas (km2): River Warta – Poraj (390), Działoszyn (4089), Sieradz (8140), Konin (13351), Poznań (25911), River Liswarta – Niwki (218), River Widawka – Rogoźno (1269), Podgórze (2355), R. Ner – Dąbie (1713), R. Kiełbaska – Kościelec (476), R. Czarna Struga – Trąbczyn (423), R. Wrześnica – Samarzewo (361), R. Prosna – Mirków (1255), Piwonice (2938), Bogusław (4304), R. Lutynia – Raszewy (534), R. Mogilnica – Konojad (663), R. Kopel – Głuszyna (369), R. Pilica – Przedbórz (2536), Spała (5955), Białobrzegi (8664), R. Czarna Maleniecka – Dąbrowa (941), R. Luciąża – Kłudzice (506), R. Wolbórka – Zawada (616), R. Drzewiczka – Odrzywół (1004), R. Bzura – Sochaczew (6281), R. Mroga – Bielawy (467), R. Rawka – Kęszyce (1191), R. Utrata – Krubice (715).
type index (LFT) was constructed (Kasprzyk NW—number of days with low flow in and Kupczyk 1998, Tokarczyk 2001). Its val- winter half-year (XI–IV) ue shows the ratio of number of days with Summer low flows were clearly dominat- low flow in the summer (May—October) ing at all investigated stations. In some rivers, and winter (November—April) half-years. winter hydrological droughts did not exist at all. This refers to the zone which crosses N the studied area in the central part—except LFT = S N for the middle and lower course of the Warta W (1) (Fig. 4). Winter hydrological droughts occur extremely rarely in this area and are very short. where: The high LFT level seems to be determined LFT—low-flow type index by summer vegetation and rain conditions, as
NS—number of days with low flow in well as mild winters without hard frost. The summer half-year (V–X) mean summer low-flow duration in this zone
KKsisiąążżkka1.indba1.indb 111919 22008-06-26008-06-26 110:56:580:56:58 120 Edmund Tomaszewski
does not vary much, though its maximum value is much higher (up to 171 days) in small catchments, which are prone to strong water resources recession because of the shallow dissection of groundwater reservoirs by river channels. To the north and south of this zone, the predominance of summer low flows is rather less marked and, in some cases, mean winter low-flow duration can be even longer than mean summer low-flow. This means that winter conditions determining hydrological droughts are more stable, while during the summer half-year a runoff recession curve is very often broken by rainfall-based alimenta- tion. In the middle and lower courses of the Warta, low flows are determined more by its tributaries than by local and regional condi- tions connected with hydrogeology, precipi- Figure 3. Distribution of estimated SNQ tation distribution and physiography. This results in a reduction of the summer low-flow and WNQ values on the flow duration curve predomination and a gradual lengthening for 29 gauges of low-flow duration downstream.
Figure 4. Spatial variation of the low-flow type index (LFT) and length of the hydrological droughts in half-years
KKsisiąążżkka1.indba1.indb 112020 22008-06-26008-06-26 110:56:590:56:59 Hydrological droughts in central Poland—temporal and spatial patterns 121
Due to the relatively more limited impor- tions are connected with the major influence tance of winter hydrological droughts in the of local conditions like hydrogeology, water investigated area, in both the quantitative management or agriculture. and temporal contexts, subsequent analyses The stability of occurrence of low-flow have been restricted to low flows in the sum- periods over a multiannual horizon was mer half-year. measured by reference to a variation coef- ficient calculated in a statistical sense, on the basis of the quotient of the mean value DATES OF OCCURRENCE and the standard deviation. Its value is di- rectly proportional to the height of the The mean date of commencement of low boxes on the map according to the sca- flow in summer is found to occur between le showed in the legend. The commence- 11th July and 28th August (Fig. 5). The first ment date of summer hydrological drought
reaction to limited alimentation is noticeable (CVDS) shows considerable stability in the in small catchments (the Lutynia, Czarna investigated area (Fig. 5). However, there Struga, Luciąża and Mroga). In most rivers, is a group of small catchments in which the the commencement date for hydrological commencement date does vary quite signifi- drought occurrs in the second half of July. It cantly, and in which there is a proneness to is worth noting that, in the Warta River sys- the marked variability in weather conditions tem, the commencement of low flow is grad- (the Wrześnica, Liswarta, Ner and Czarna ually delayed downstream, due to increasing Struga). Hydrological droughts in the larger water resources and upper-course alimenta- (especially the transit) rivers start at a very tion. Certain exceptions at a few gauging sta- similar time.
Figure 5. Spatial variation of summer hydrological drought commencement date and its variation
DS—date of commencement of summer hydrological drought
CVDs—variation coefficient of DS
KKsisiąążżkka1.indba1.indb 112121 22008-06-26008-06-26 110:56:590:56:59 122 Edmund Tomaszewski
T he mean date of occurrence of minimum (tR) depends on stationary factors mainly summer low-flow differs much less than the connected with the hydrogeological condi- starting date (Fig. 6). This parameter tends tions of groundwater reservoirs, depicted to lag downstream. The variation coefficient by the master recession curve, which may
(CVDM) for this characteristic is also very be modified by local and temporary factors, similar, and not too high. Such spatial and such a s weat her, temp erat u re, a nd veget at ion. temporal stability results from the major role The mean time between the commencement played by hydrogeological conditions, which of low flow and the lowest discharge during determine the recession rate of groundwater, the summer hydrological drought fluctuates especially where levels of their resources are between 4 and 40 days (Fig. 7). In general, low. Notwithstanding significant spatial vari- the upper courses of rivers attain minimum ability (Kożuchowski, Wibig 1988), climate flow much earlier than the lower ones. How- conditions, especially pluvial, seem to mod- ever, it is worth noting that some small tribu- ify the present state only. taries along the lower course of the Warta (the Kopel, Mogilnica, Lutynia, Czarna Struga and Wrześnica) also display a long RECESSION TIME AND MINIMUM POSITION run-in time to minimum discharge—over 15 days, something that probably results One of the important characteristics of the from the predomination of autumn pre- development of hydrological drought is the cipitation in these areas (Kożuchowski and time after which runoff reaches its minimum Wibig 1988). As a result, the discharge re- level (see Fig. 1). The length of this period cession rate is significantly slowed down.
Figure 6. Spatial variation of summer low-flow minimum date and its variation
DM—minimum date of the summer low flow
CVDM—variation coefficient of DM
KKsisiąążżkka1.indba1.indb 112222 22008-06-26008-06-26 110:56:590:56:59 Hydrological droughts in central Poland—temporal and spatial patterns 123
Figure 7. Recession time and the minimum low-flow position of summer hydrological drought
tR—recession time, IP—index of minimum low-flow position, CVIp—variation coefficient of IP
When it comes to the time structure position is reached after the midpoint as fre- of a hydrological drought, the position quently as before (Fig. 7). There does not of the minimum flow plays a significant role. seem to be any dominating order to the spa- A characteristic which quantifies this prop- tial distribution of this parameter. This may erty should be relative to the commence- indicate that, in that case, local conditions ment and termination dates of the low-flow play a very important role. In general, many
period: tributaries have a tendency to reach an IP value equal to or less than 0.5, whereas main t rivers remain above that level. Moreover, I R P = a high IP level is very often connected with t (2) considerable stability of this index multian-
nually (CVIp). This results in a slow response where: to rapid recession of groundwater resources,
IP—index of minimum flow position and in the reaching of a long-term ground-
tR—time between start point and mini- water flow limit at a very late phase of the mum flow point during hydrological drought hydrological drought. t —duration of hydrological drought The above parameter assumes values in the range between 0 and 1. Its interpretation CONCLUSIONS is quite easy because, if the calculated value is equal to 0.5, the minimum point is situated The analyses presented allow us to make exactly in the middle of the low-flow period. a few general statements pertaining to con- In the investigated rivers, the minimum flow tent, as well as to the method applied:
KKsisiąążżkka1.indba1.indb 112323 22008-06-26008-06-26 110:57:000:57:00 124 Edmund Tomaszewski
• the SNQ level seems the most appro- wanie niżówek o znacznym deficycie odpływu priate threshold in estimating major hydro- [Vulnerability Estimation of River Catchment logical drought, System to Severe Hydrological Droughts Ap- • hydrological droughts in the winter pearance], in Magnuszewski, A. and Soczyń- half-year appear very seldom in central Po- ska, U. (eds.), Hydrologia u progu XXI wieku, land, or even do not occur at all, Warszawa, 157–165. • the temporal characteristics of hydro- Kożuchowski K., Wibig J., (1988), Kontynenta- logical droughts demonstrate a spatial order lizm pluwialny w Polsce: zróżnicowanie geo- quite precisely (some zones appear or some graficzne i zmiany wieloletnie [Pluvial Con- directions of spatial variability are notice- tinentality in Poland: Geographic Variations able), and are determined by local as well as and Long-Term Variability], Acta Geographi- regional conditions, ca Lodz., 55, Ossolineum, Łódź: 1–102. • information about recession time as Ozga-Zielińska, M. (1990), Niżówki i wezbrania well as minimum flow position may be useful – ich definiowanie i modelowanie [Droughts in describing the regime of low-flow forma- and Floods—their Definition and Modeling], tion, because it shows relationships between Przegląd Geofizyczny, 1-2, Warszawa: 33–34. recession and the renewal rate of groundwa- Tokarczyk, T. (2001), Zmienność przepływów ter resources and their multiannual variabil- niskich na obszarze Kotliny Kłodzkiej [Vari- ity. ability of Low Flow in Kotlina Kłodzka Re- gion], Zeszyty Naukowe Akademii Rolniczej we Wrocławiu, Seria Inżynieria Środowiska, ACKNOWLEDGEMENTS XII, 413, Wrocław: 105–127.
The research in the present article has been Paper first received: November 2007 carried out within the framework of the re- In final form: February 2007 search project entitled “Extreme meteoro- logical and hydrological events in Poland”, financed by the Ministry of Science and Higher Education of Poland (PBZ-KBN- 086/P04/2003).
REFERENCES
Hisdal, H., Tallaksen, L.M., Clausen, B., Peters, E. and Gustard, A. (2004), Hydrological Drought Characteristics, in Tallaksen, L.M. and van Lanen, H.A.J. (eds.), Hydrological Drought. Processes and Estimation Methods for Streamflow and Groundwater, Develop- ments in Water Sciences, 48, Elsevier, Amster- dam: 139–198 Jakubowski, W. (2004), NIZOWKA 2003 soft- ware, in Tallaksen, L.M. and van Lanen, H.A.J. (eds.), Hydrological Drought. Process- es and Estimation Methods for Streamflow and Groundwater, Developments in Water Scienc- es, 48, CD-Appendix, Elsevier, Amsterdam. Kasprzyk, A. and Kupczyk, E. (1998), Ocena po- datności systemu zlewni rzecznej na występo-
KKsisiąążżkka1.indba1.indb 112424 22008-06-26008-06-26 110:57:010:57:01 A DESCRIPTION OF HYDROLOGICAL DROUGHTS IN THE BIAŁOWIEŻA PRIMEVAL FOREST IN THE YEARS 2003—2005
ANDRZEJ CIEPIELOWSKI, EWA KAZNOWSKA Faculty of Engineering and Environmental Science, Warsaw Agricultural University, Ul. Nowoursynowska 159, 02-787 Warsaw, Poland E-mails: [email protected]; [email protected]
Abstract: The Białowieża National Park is located in northeastern Poland, in the Narewka River Basin upstream of the Narewka gauge profile. Discharge records at this gauge were investigated and streamflow drought parameters, such as minimum and average discharges occurring during the drought, drought durations and deficit volumes were determined. These parameters define hydrological droughts and can serve in an indirect way to assess the degree of deformation of forest site types. The investigations covered the extremely dry 2003–2005 period, during which hydrological droughts occurred in each year. In order to check whether these droughts were more intense than those observed previously, characteristics were compared with those corre- sponding to the earlier period 1951–2002. The characteristics of intensity, minimum and average streamflow drought in the recent years were not found to have been more extreme than in the last multi-decade period. However, the streamflow droughts of 2003 and 2004 were extremely protracted, lasting 134 and 67 days. The drought of 2003 was in the nature of a “disaster”, though was still not an event more extreme than any noted in past records.
Key words: streamflow drought, hydrological drought, Białowieża Forest
INTRODUCTION aim of the study described here was to find out whether the droughts occurring at the Poland is a country exposed to the periodic site in the period 2003–2005 were more in- occurrence of hydrological droughts defi- tense than those observed previously. ned as deficits of surface water and groun- Investigations by Mager et al. (2000), dwater. The Podlasie Lowland situated in based on daily water levels in the years the northeastern part of the country is one 1891–1950 as well as river runoff in the pe- of the regions in which droughts occur most riod 1951–1995, showed that the mean fre- frequently (Farat et al. 1995, Kaznowska and quency of hydrological droughts in Poland Ciepielowski 2006). The hydrological drou- in the last 45 years was greater (at one dry ghts occurring there are a threat, especially year in each 3.2-year period on average) to the hydrogenic forest sites of the Biało- than in the whole period under analysis (one wieża Forest, which represents natural and dry year per 44.0-year period). This result- cultural heritage on the global scale. The ed from frequency analyses of streamflow
KKsisiąążżkka1.indba1.indb 112525 22008-06-26008-06-26 110:57:010:57:01 126 Andrzej Ciepielowski, Ewa Kaznowska
drought occurrence since the year 988, zones (Kundzewicz 2003). The projections based on historical notes in the “chronicles of global climatic change also foresee a much of elementary disasters” and records from greater climatic threat to forest ecosystems water level measuring gauges as presented than can actually be seen today (Sadowski by Fal (2004), that years with hydrologi- and Galinski 1998). More frequent drought cal droughts built groups within multi-year periods and high air temperatures could cycles; with 2-year and 3-year cycles being adversely affect the stability of forests and most common. The longest dry period oc- increase their vulnerability to industrial pol- curred in the 20th century, when years with lution and to the stress imposed by insect streamflow droughts occurred one after an- pests and fungal pathogens (MICE Team other in the period 1947–1954. 2005). Long-lived forest trees might prove However, the frequency analysis of the incapable of adaptation to changing site con- occurrence of hydrological droughts in the ditions. country in successive decades starting from The analysis of hydrological droughts in 1951 makes it clear that the most years with the Białowieża Forest area between 2003 streamflow drought were noted in the period and 2005 included an estimation of the se- 1951–1960 (Table 1). Recently, the period verity and intensity of streamflow droughts 2001–2006 brought four years with stream- along the Narewka River. That river togeth- flow droughts(Kaznowska and Ciepielowski er with its tributaries drains the central and 2006). The results of investigations of the northern parts of the Białowieża Forest. It causes of the occurrence of dry year series is a tributary of the Narew situated in the pointed to the presence of climatic circula- Vistula River Basin. In order to examine tion features that enhance the occurrence whether the recent streamflow droughts and severity of droughts. These investiga- were more intensive than earlier ones, a com- tions also revealed that the last 25–30 years parison of parameters of droughts from the have witnessed a reorientation of circulation 1951–2002 and 2003/2005 periods was types in Central Europe, not only in sum- made. mer periods. An increase in the occurrence of blocking masses (high air pressure) was noted, this favouring meridional inflows THE RESEARCH AREA of air masses linked with a lack of precipita- tion over Europe (Lorenc 2006). The Narewka River Basin is situated in the Poland is a country located in a region northeastern part of Poland in the Białowie- of marked sensitivity to possible changes in ża Forest area, this being a large forest tract climatic relations (Kaczmarek 1996), so the with a prevalence of natural forests exten- warming reported by IPCC (2001) enhanced ding along both sides of the Polish-Belarus- the formation of streamflow droughts. Thus, sian border. The Białowieża Forest consti- events in recent years (the spring drought tutes 73% of the catchment, while meadows of 2000 and the summer droughts of 1992 and marshes of the Narewka River lowland and 2003) might be regarded as harbingers form the majority of the remaining part. The of the more frequent dry years predicted Białowieża National Park of 10,502 ha is lo- by the IPCC for the temperate latitude cated in the central part of the catchment, in
Table 1. The frequency (numbers) of streamflow droughts in successive decades in the later 20th century in Poland.
Decade 1951–1960 1961–1970 1971–1980 1981–1990 1991–2000
Number of years with drought 65043
KKsisiąążżkka1.indba1.indb 112626 22008-06-26008-06-26 110:57:010:57:01 A Description of Hydrological Droughts in the Białowieża Primeval Forest in the Years 2003–2005 127
the interfluve between the Narewka and the The Narewka River springs are situated Hwozna. This is the most pristine forest any- in the Belarussian part of the Forest (there where in Europe, in which broadleaved trees referred to as Belovezhskaya) at an altitude predominate. It has been on the UNESCO of 159 m a.s.l. The catchment area up to the List of World Heritage Sites since 1979 Narewka River gauge profile, in both riparian (Fig. 1) and among the European Heritage countries, Poland and Belarus, is 635.4 km2, sites (in line with a status bestowed by the 346.2 km2 of this being on the Polish side. Council of Europe) since 1998. Unique, rare The river bed slope is 0.4‰, and its length and endemic, species of flora and fauna oc- 50.1 km. The upper reach of the river in Be- cur here larus is regulated and the catchment man-
Figure 1. Location of the investigated area.
The average stand age in the Białowieża aged. Downstream from the Poland-Belarus National Park is 130 years, i.e. almost twice frontier the Narewka River flows through the that of the same Forest’s managed part. Al- Torfowisko Wysokie forest reserve, before most all forest site types found on the central passing through the hamlet of Białowieża it- European lowland are present. Fresh soil self. In that reach the Narewka was regulated sites dominate, accounting for 59.6% of the as early as in the 18th century, but the final total. Moist represent 25.3%, and marshy straightening of its channel was done in the sites 15.1% of the area. Moist-soil broad- early 1960s. Two ponds established by dam- leaved and ash-alder forest sites account for ming the Narewka at the end of the 19th cen- the greatest share among moist and marshy tury are situated on the Białowieża Glade; forest sites. The share of bog and bog mixed a weir dams the water for the ponds. The coniferous forest is smallest. Moist and bog Braszcza, Hwozna and Lutownia rivulets are forest sites bear tree stands of considerable the main tributaries. species diversity, while pine, spruce, and al- Groundwater is generally present close der predominate (Chojnacki 2004). to the surface, due to the lowland character
KKsisiąążżkka1.indba1.indb 112727 22008-06-26008-06-26 110:57:010:57:01 128 Andrzej Ciepielowski, Ewa Kaznowska
of the catchment and to the less-pervious flow in the channel. Notwithstanding an soil formations spread out at shallow depth. abundance of Polish and foreign litera- Boggy areas along river valleys and peat- ture devoted to this issue, there is a lack lands situated below the 151 m a.s.l contour of a genetically-based and univocal defini- line have groundwater at a shallow 0–1 m tion of this feature. The definition of stre- depth linked hydrostatically with the water amflow drought that is adopted and used level in the Narewka River and its tributaries most often entails a recognition of the pe- (Głogowska 2005). riod over which discharges are less than or The climate of the area is subcontinental, equal to the assumed threshold level dis- moderately warm and moderately humid. charge. The method involving extraction The mean annual air temperature is +6.5°C, of streamflow drought periods from daily the mean annual amplitude being 22.5°C. discharge hydrographs using the assumed The average air temperature of the cold- threshold value occurs in the literature est month (February) is -5.2°C, that of the under the TLM (threshold level method) warmest (July) +17.4°C (Prusinkiewicz and name, having been first adopted in hydrolo- Michalczuk 1998). The average annual to- gy by Yevjevich (1967). The choice of thres- tal for precipitation is 617 mm (1951–2002) hold level depends on the research area and in the case of the meteorological station on the threshold definition criteria, e.g. at Białowieża; the lowest precipitation total hydrological or economic (Ozga-Zielińska noted was for 1954 (400 mm), as compared 1990), or hydrobiological –connected with with the peak of 900 mm noted in 1970. the survival capabilities of aquatic orga- It is the catchment areas situated in the nisms (Kaznowska 2006). lowlands along the Narewka River and its A group of hydrologists proposed a uni- tributary the Hwozna, as well as the adja- fication of methods and techniques for ana- cent sandy areas with a superficial water lysing the streamflow drought phenomenon. table that are most exposed to transforma- They cooperate within the international tions of vegetation associations due to the FRIEND (Flow Regimes for International adverse effects of streamflow droughts. The Experimental and Network Data) project groundwater resources in biotopes are sub- within the IHP (International Hydrol- ject to depletion as a result of hydrological ogy Programme) affiliated to UNESCO. droughts. Forest sites begin to suffer from A comprehensive description of probabil- disturbances in water balance and this can istic and stochastic methods of researching be the cause of far-reaching changes in the streamflow drought parameters is contained species compositions of forest associations. in the monograph edited by Tallaksen and Norway spruce, the dominant species in the van Lanen (2004) produced in the context Białowieża Forest, has a shallow root system of the aforementioned project. The authors and so is one of the species more sensitive to of the monograph suggest that the stream- drought (Ciepielowski and Głogowska 2005). flow drought threshold discharge be defined
An analysis of the occurrence of streamflow as a the Q70% discharge determined from the drought in forest catchments can provide curve of discharge duration time sums. representative indices characterising water However, the separation of features conditions in those forest sites whose func- of streamflow drought from among the more tioning depends on the groundwater level. common low-flow periods occurring every year as a natural feature of a river’s hydro- logical river requires the adoption of a lower DEFINITION OF THE FEATURE value of the threshold level. The use of the
Q90% discharge threshold gives such a pos- Streamflow drought is a commonly used sibility (Zelenhasic and Salvai 1987; Hisdal concept serving in the interpretation of the and Tallaksen 2000; FRIEND 2002; Fleig et state of a river as this relates to the water al. 2005). The minimum discharge duration
KKsisiąążżkka1.indba1.indb 112828 22008-06-26008-06-26 110:57:020:57:02 A Description of Hydrological Droughts in the Białowieża Primeval Forest in the Years 2003–2005 129
below the threshold value is used as an addi- taking into account half-year periods dur- tional criterion by which to identify stream- ing which those droughts occurred. Daily flow droughts. The range of 10–20 days is discharge hydrographs of hydrological years typically adopted as the minimum stream- (running from 1 November to 31 October) flow drought duration in research work car- served as the input data for the evaluation ried out in Poland (Byczkowski 1999). of hydrological droughts. The streamflow droughts were distinguished from daily dis- charge hydrographs, and were described THE METHOD OF RESEARCH by reference to the quantitative param- eters minimum discharge, mean discharge, A sequence of discharges of at least 10-day drought duration, and water deficit vol- duration with values equal to or less than ume expressed by the space between the 3 -1 Q90% = 0.79 m s was recognized as consti- hydrograph line and the threshold level tuting a streamflow drought. Such stream- (Fig. 2). These parameters were identified flow droughts separated from each other by using the Nizowka 2003 model (Jakubowski less than 5 days were treated as one and the and Radczuk 2004). same feature; in order to avoid connections The following characteristics were ap- between neighbouring droughts. Summer plied in assessing drought severity: and winter streamflow droughts are cau- sed by different mechanisms. The summer Tn T = ∑ i droughts result from long-lasting deficits n of precipitation, while the winter droughts ∑ i (1) are caused by a lack of surface and subsurfa- ce alimentation of rivers. This can be due to where: negative air temperatures, resulting in, not ⎯T is the average duration of stream- only an accumulation of precipitation in the flow drought per year [days]; form of snow cover, but also in a break in the Σ ni is the occurrence of a number subterranean water supply of a river due to of streamflow droughts in N years of the pe- deep freezing of the bedrock. riod under research;
Summer and winter droughts were ana- Σ Tni is the summed number of days on lyzed separately in order to preserve the which streamflow droughts occurred in an character of the feature under analysis, N-year period: and
Figure 2. Parameters of streamflow drought.
KKsisiąążżkka1.indba1.indb 112929 22008-06-26008-06-26 110:57:030:57:03 130 Andrzej Ciepielowski, Ewa Kaznowska
N is the number of years in the re- droughts occurring there. The beginning search period; of streamflow drought occurrence in the summer half-year usually falls in July and V June, while August, September and July are ∑ n i V = the months in which such droughts occur n ∑ i (2) most often. Winter droughts occur most fre- quently in January and February (Table 2). where: The average duration of summer and winter ⎯ V is the average volume of streamflow streamflow droughts exceeds 40 days, and is drought deficit per year [th. m3]; about 30 days, respectively (Tab. 3).
Σ Vni is the sum of volumes of streamflow In the period 2003–2005, streamflow drought deficits in N years [th. m3]. droughts occurred every summer in the Nare- Characteristics describing either maxi- wka profile. Their characteristics were then mum or minimum values of analyzed pa- compared with those of summer droughts rameters of individual streamflow droughts occurring from 1951 to 2002. The intensity occurring in the analyzed multi-decade pe- plus mean and minimum values for stream- riod were additionally defined as: flow drought discharges of recent years were
Tmax – maximum duration of stream- not found to be the most extreme on record flow drought recorded in N years [days] in the multi-annual period. However, the
Vmax – maximum volume of streamflow streamflow droughts of 2003 and 2004 were drought recorded in N years [th. m3] very prolonged: at 134 and 67 days respective-
NQmin – lowest discharge during stream- ly (Table 3), while the nine 1951–2002 sum- flow droughts observed in N years [m3s-1] mer droughts lasting over 60 days constituted
SQav – average discharge during stream- 26.5% of all summer droughts. Taking into flow droughts observed in N years [m3s-1] account the recent years, the mean duration T he mean intensit y of streamf low drought of a drought calculated from the multi-year in multi-year periods as given by Kasprzyk period increased from 44 to 47 days (Fig. 3). (2002) was the next streamflow drought char- The water deficits characterizing the 2003 acteristic taken into account. It was calcu- and 2004 droughts can be classified as high. lated as the ratio of the discharge deficit size When recent years are taken into account in and the number of days with low discharges the process of calculating the mean deficit, in the multi-year period. This characteristic this increases from 682,660 m3 for 1951–2002 measuring the size of the discharge deficit to 738,420 m3 for 1951–2005. per day of drought was recalculated in terms The summer of 2003 merits special at- of percentage of mean annual discharge per tention; an intensive and major hydrological day of drought: drought following an exceptionally dry end of spring. This was noted over the prevail- V ing part of the country. A detailed analysis Σ ni Iar = of the causes and course of the 2003 drought T Σ ni (3) was carried out by Mierkiewicz and Sasim (2005), who showed that the drought as- where: sumed disastrous dimensions for many re- the legend is as for equations (1) and (2). gions of the country. The most unfavourable conditions occurred in August, when levels below the absolute minima were noted at 60 RESULTS river-level gauges located mainly in south- western and southern Poland. However, the Summer streamflow droughts predominate streamflow drought of 2003 was not yet the at the Narewka River gauge profile of the most extreme observed in the Narewka gauge Narewka catchment, constituting 80% of all profile. The longest and most severe stream-
KKsisiąążżkka1.indba1.indb 113030 22008-06-26008-06-26 110:57:040:57:04 A Description of Hydrological Droughts in the Białowieża Primeval Forest in the Years 2003–2005 131
Table 2. Streamflow drought occurrence by decades in the Narewka river (Narewka) in 1951–2005
Lata XI XII I II III IV V VI VII VIII IX X 1951 1952 1954 1955 1958 1959 1960 1961 1962 1963 1964 1965 1966 1969 1971 1983 1984 1989 1991 1992 1993 1994 1995 1996 1997 1999 2000 2001 2002 2003 2004 2005
flow drought lasted 194 days. It began on wka River Basin in the period 1950–2003. 12 July 1951 and ended on 23 January In recent years that increase has mainly 1952, and its discharge deficit amounted to occurred in winter (November–April) half 5,002,560. m3. The lowest discharge of the years, these being warmer by 1.5°C than the multi-annual period was recorded during multi-year average. that streamflow drought; featuring 0.30 m3s-1 In the period 1989–2003, the mean tem- on 25 November 1951 (Table 3). perature of the winter half-year was near The streamflow drought of 2003 was to 0°C, while for the period 1951–2003 the 3 -1 shorter and shallower (Qmin = 0.35 m s ), warmest winter was observed in 1990, when as well as less intensive (0.021% < 0.027%). the mean was 4.3°C (Czerepko et al. 2005). Not a single winter drought has been ob- It also results from the comparison between served in recent years (since 1997). The lack the two periods 1966–1982 and 1983–2000 of winter droughts can be attributed to the for the Białowieża observation station that climatic warming that is to be observed now. the mean temperature in the winter half- An increase in average annual air tempera- year in recent years was higher (0.76°C) than ture of 0.9°C was recorded at the Białowieża in the earlier period (-0.24°C) (Maksymiuk meteorological station located in the Nare- et al. 2004).
KKsisiąążżkka1.indba1.indb 113131 22008-06-26008-06-26 110:57:040:57:04 132 Andrzej Ciepielowski, Ewa Kaznowska
Table 3. Characteristics of droughts in the Narewka river; S—summer drought, W—winter drought.
Characteristics of droughts Type of Periods Drou- Σ ni Σ Tni ⎯TTmax,n ⎯VVmax,n SQav NQmin Iav ghts / days th. m3 m3s-1 %
1951–2002 S 34 1505 44 194 682.66 5002.56 0.66 0.30 0.016 W 9 261 29 73 267.36 775.01 0.69 0.39 0.010
2003 S 1 134 134 134 2711.23 2711.23 0.56 0.35 0.021 W 0
2004 S 1 67 67 67 838.08 838.08 0.65 0.52 0.013 W 0 2005 S 1 16 16 16 154.66 154.66 0.68 0.58 0.010 W 0 1951–1971 S 25 1050 42 194 677.55 5002.56 0.66 0.30 0.018 W 6 171 29 73 281.23 775.01 0.69 0.39 0.010
1983–2005 S 21 933 44 134 589.62 2711.23 0.67 0.35 0.014 W 3 90 30 38 239.62 260.93 0.68 0.52 0.008
Symbols according to the formulae (1) to (3)
Two periods 1951–1971 and 1983–2005 precipitation totals to mean multi-annual to- are distinctly noted when days of droughts tals adopted as a norm. and drought deficits are summed up, year- by-year, within the period 1951–2005. Hy- drological droughts occurred almost every CONCLUSIONS year in those periods (Figs. 3 and 4). These periods are separated from each other by The Narewka profile experienced hydro- an interval of 11 years without streamflow logical droughts every year in the period droughts. The characteristics of drought se- 2003–2005. In the years 2003 and 2004 the verity described with formulae 1, 2, and 3 for scale of the phenomenon was major, and the both periods were compared with each other streamflow drought of 2003 being disastrous (Table 3). Mean drought duration, mean in nature. However, this was not the greatest deficit, intensity, and mean and minimum one on record. The drought of 2003 lasted discharges of both periods were not found to for 134 days, with a deficit equal to 2.82% have differed significantly from one other. of mean annual discharge in the multi-annu- The occurrence of two periods with al period. Research on streamflow droughts streamflow droughts appearing almost in Poland shows that years with hydrological every year in the multi-year period can be droughts form clusters in multi-annual cyc- compared with the variability to annual pre- les. Droughts in the 2003–2005 period can cipitation totals recorded at the Białowieża be regarded as a fragment of a longer cycle, meteorological station. During the periods the beginning of which occurred in 1991. 1951–1971 and 1983–2004, dry and average The tendency towards the annual occur- years predominated over moist ones, while rence of droughts visible in recent years is in the period 1972–1982 moist and very moist related to prolonged deficits of precipitation. years predominated over average and dry Since 1982, permanent droughts have been ones. This feature was found by reference observed in two 4–5-year periods, though to Kaczorowska’s criterion (1962), through atmospheric droughts did not occur over a comparison of the proportion of annual the greater part of the country. Consider-
KKsisiąążżkka1.indba1.indb 113232 22008-06-26008-06-26 110:57:070:57:07 A Description of Hydrological Droughts in the Białowieża Primeval Forest in the Years 2003–2005 133
Figure 3. Number of days with streamflow droughts in the multi-annual 1951–2005 period in the Narewka river down to the Narewka gauging station.
Figure 4. Deficit of streamflow droughts in the multi-annual 1951–2005 periods in the Narewka river down to the Narewka gauging station.
able and long-lasting precipitation deficits 235 mm) than in the period 1966–1982 (at have been observed, especially during the 262 mm) (Maksymiuk et al. 2004). last 15–20 snowless winters (Lorenc 2006). The long-lasting streamflow droughts In the period 1983–2000, the mean precipi- noted in the Narewka River Basin in re- tation total for the winter half-year at the cent years pose a threat to riparian forest Białowieża observing station was lower (at ecosystems in lowland areas and adjacent
KKsisiąążżkka1.indba1.indb 113333 22008-06-26008-06-26 110:57:070:57:07 134 Andrzej Ciepielowski, Ewa Kaznowska
territories with relatively shallow groundwa- Chojnacki, T. (2004), Gospodarka leśna na siedli- ter. Research carried out in the Białowieża skach wilgotnych i bagiennych w Puszczy Bia- Forest on especially valuable sites proved łowieskiej [Forest Management on Moist and that the groundwater level in forest wetland Boggy Sites in the Białowieża Forest], in Ma- biotopes decreased by 10 cm in the years teriały seminarium naukowo-technicznego, 1985–2004 (Czerepko et al. 2005a). The Zagrożenia leśnych siedlisk hydrogenicznych lowering of groundwater levels caused by w Puszczy Białowieskiej, Białowieża, 21 maja the occurrence of streamflow droughts in 2004 [Papers of the Scientific and Technical rivers normally derived from groundwater Seminar on ‘Threats to Forest Hydrogenic discharge, influences the dynamics of forest Sites in the Białowieża Forest’, Białowieża, associations. 21.05.2004]. Comparison of the affected percentag- Ciepielowski, A., Głogowska, E. [Kaznowska, E.] es of Białowieża Forest sites from the years (2005), The Analysis of Hydrological Drou- 1998 and 2002 reveals a decrease in the ghts in the Narewka Stream Area within The percentage of moist and marshy sites, and Białowieża Primeval Forest—One of the Gre- an increase in fresh-soil sites. In 1998 the atest Natural Forest Tracts in the Central Eu- fresh-soil, moist and marshy sites together ropean Lowland, in International Conference accounted for 55.2%, 28.2%, and 16.6% on Forest Impact on Hydrological Processes of the total respectively (BULiGL 1998), and Soil Erosion, University of Forestry, Yun- while in 2002 the figures were: 59.6%, dola Bulgaria, 5–8 October, pp. 43–51. 25.3%, and 15.1% (Chojnacki 2004). These Czerepko, J., Boczoń, A., Nikitin, A. (2005), Threats results, and the tendency towards the an- to alder swamp forests in a changing environ- nual occurrence of streamflow droughts ment, in Kotowski, W. (ed.) Anthropogenic indicate that an increase in retention and Influence on Wetlands Biodiversity and Sustain- a decrease in water discharge are needed. able Management of Wetlands, Center of Excel- The analysis of streamflow droughts can lence in Wetlands Hydrology (WETHYDRO), establish measures characterising the state Wydawnictwo Szkoły Głównej Gospodarstwa of water conditions for the functioning Wiejskiego (Warsaw Agricultural University of forest sites. Press), Warszawa, pp. 21–33. The problem of decline in the number Czerepko, J., Boczon, A., Pierzgalski, E., Soko- of moist sites in the Białowieża Forest should lowski, A.W., Wróbel, M. (2007), Habitat be of particular interest and importance for diversity and spontaneous succession of for- the forest service due to the need to protect est wetlands in Bialowieza primeval forest, in the natural diversity of forest sites as estab- Okruszko T. et al (eds.), Wetlands: Modeling, lished in the rules of forest management in Monitoring and Management, Taylor&Francis, the State Forests (Rozwałka 2003). London, 37-43. Fal, B. (2004), Czy niżówki ostatnich lat są zja- wiskiem wyjątkowym? [Are the Streamflow REFERENCES Droughts of Recent Years an exceptional phenomenon?], Gazeta Obserwatora IMGW, BULiGL (1998), Opracowania glebowo- 3:16–18. siedliskowe Nadleśnictw: Białowieża, Browsk, Farat, R., Kepińska-Kasprzak, M., Kowalczak, Hajnówka, Białystok [Soil-site Masterplans P., Mager, P. (1995), Susze na obszarze Pol- of the Forest Districts of Białowieża, Browsk, ski w latach 1951–1990 [Droughts on the Area Hajnówka and Białystok]. of Poland in the Years 1951–1990], Materiały Byczkowski, A. (1999), Hydrologia tom II, (wyda- badawcze IMGW, Seria: Gospodarka Wodna nie II poprawione) [Hydrology, vol. II, edition i Ochrona Wód, Warszawa, IMGW, 16. II corrected], Wydawnictwo Szkoły Głównej Fleig, A.K., Tallaksen, L.M., Hisdal, H., De- Gospodarstwa Wiejskiego (Warsaw Agricul- muth, S. (2005), A Global Evaluation of Stre- tural University Press), Warszawa. amflow Drought Characteristics, Hydrology
KKsisiąążżkka1.indba1.indb 113434 22008-06-26008-06-26 110:57:070:57:07 A Description of Hydrological Droughts in the Białowieża Primeval Forest in the Years 2003–2005 135
and Earth System Sciences Discussions, 2: Kaznowska, E. (2006), Charakterystyka susz 2427–2464. hydrologicznych na przykładzie wybranych FRIEND Report 2002 (2002), FRIEND—a Glo- rzek północno-wschodniej części Polski [The bal Perspective 1998–2002, in Gustard A. and Characteristics of Hydrological Droughts Cole G. A (eds.), Walingford, UK, Published as Exemplified by Selected Rivers in the by the Centre for Ecology and Hydrology North-East Poland], Infrastruktura i Ekolo-
KKsisiąążżkka1.indba1.indb 113535 22008-06-26008-06-26 110:57:070:57:07 136 Andrzej Ciepielowski, Ewa Kaznowska
Poland], Roczniki Akademii Rolniczej w Poz- Groundwater, Developments in Water Science, naniu, Melioracje i Inżynieria Środowiska, 48, Amsterdam Elsevier. Wyd. Akademii Rolniczej im A. Cieszkow- Yevjevich, V.M. (1967), An Objective Approach skiego Poznań, CCCLVII, 25: 355–362. to Definition and Investigations of Continen- MICE Team (2005), MICE Summary of Final tal Hydrological Droughts, Hydrology Papers Report, Modelling the Impact of Climate Ex- 23, Colorado State University, Fort Collins. tremes—Project financed by the European Zarządzenie Nr 11A Dyrektora Generalnego La- Community within the Fifth Framework Pro- sów Państwowych z dnia 11 maja 1999 r. [Or- gramme. der No 11A by the Director General of State Mierkiewicz, M., Sasim, M. (2004), Susza w Pol- Forests of 11 May 1999]
KKsisiąążżkka1.indba1.indb 113636 22008-06-26008-06-26 110:57:080:57:08 GEOMORPHIC ACTIVITY OF DEBRIS FLOWS IN THE TATRA MTS AND IN OTHER EUROPEAN MOUNTAINS
ADAM KOTARBA Institute of Geography and Spatial Organization, Polish Academy of Sciences, św.Jana 22, 31-018 Kraków, Poland E-mail: [email protected]
Abstract: Debris flows constitute the dominant high-energy slope processes in the high-mountain belt of the Tatra Mountains, the Alps and other European mountain massifs. Rainfall inten- sities responsible for triggering recent flows include that of ca 35–40 mm in one hour. Under such a condition, whole talus slopes several hundred meters long are affected by rapid flow in the High Tatras, on both the Polish and Slovak sides, and the maximum volume of debris re- moved and accumulated by such events is ca. 25,000 m3 . Debris flows with a maximum volume of ca. 500,000 m3 are triggered by rainstorms of similar totals and intensities in the Alps.
Key words: debris flows, extreme rainfall events, Tatra Mountains, European geomorphic hazards.
INTRODUCTION circulation of water and mineral substances occurs. In the highest European massifs The present-day development of the relief this circulation is controlled to a significant of the Tatras, like other mountains, is de- extent by the presence of mountain glaciers. termined by active and passive elements In studies on the evolution of the present- of the natural environment. Relief dyna- day high-mountain relief, two vertical zo- mics are conditioned by geological setting, nes are distinguished in the Alps: haute especially by lithology, while relief energy is montagne alpine–the zone subjected to controlled by relative and absolute heights, glacial morphogenesis–and haute montag- gradients and dissection of slopes. Howe- ne pyrénéene–the zone without glaciers, yet ver, relief dynamics are simultaneously with all other environmental features of the conditioned by hydrometeorological factors high-mountain landscape (Galibert 1960). which vary over time and space. In high- The studies carried out in the highest parts mountain areas, the considerable elevation of the glaciated Alps, rising above 4,000 m above sea level leads to the development of a.s.l., support a belief that the haute mon- systems of vertical circulation of heat, water tagne alpine zone experiences a slower evo- and loose rocks. Vegetation, superimposed lution than the haute montagne pyrénéene on relief elements and Quaternary covers zone. The highest zone in the Alps is ‘pro- of diversified origin, determines geoecolo- tected’ by a snow-ice cover cemented to the gical altitudinal belts, wherein a permanent substratum. Due to numerous days with
KKsisiąążżkka1.indba1.indb 113737 22008-06-26008-06-26 110:57:080:57:08 138 Adam Kotarba
frost (ca. 250 days a year) this cover never is no glacial-nival climate in the Tatras no- melts in the hottest months of a year. wadays, the mountains are affected by influ- ences of nival, nival-pluvial and pluvial-nival climates, depending on elevation above sea AIM AND METHODOLOGY OF THE STUDY level (Hess 1965). It denotes that at present IN THE POLISH TATRAS in the areas of high-mountain relief, which were formed by the Pleistocene glaciers, long- The purpose of the study is to analyse the lasting snow retention is sustained, perennial imprint of debris flows on the present-day snow patches occur, and an abundant amount relief in the Tatras in relation to the Alps of water is supplied by rainfall, snowfall and and other European mountains. All compa- snowmelt. In a renewal of the water resources risons have to refer to the same high-moun- and water circulation, rainwater contribution tain zone–namely to the non-glaciarised is the most important and amounts to 75% haute montagne pyrénéene zone, because of recorded annual precipitation (Łajczak this zone has developed in all high mountain 1996). Rainfall totals as well as rainfall in- massifs of Europe, from the Scandinavian tensity reach their maximum in the summer mountains to the Sierra Nevada in southern months, i.e. from June to August. Spain. Consequently, such comparisons can The eastern part of the mountains, cal- be made for fast mass movements triggered led the High Tatras, has a classic alpine re- by hydrometeorological factors. It has to be lief characterized by the presence of glacial remembered that, in the highest parts of the cirques and glacial troughs formed by, at le- Alps the non-glaciarised haute montagne ast, triple valley glaciation. Glacial erosional pyrénéene zone is found below the glaciated forms are filled with moraine and glaciflu- alpine zone and vertical transfer of mineral vial covers originating from the last glacia- substances and water, as a medium, takes tion (Würm). The forms were transformed place between these zones. by periglacial processes during the Holo- Several methods have been used in the cene. In the granodiorite High Tatras the Polish Tatras (Kotarba 1997, Kotarba et al. glacial cirques lie one over the other, their 1983), among them: floors, excluding that of the topmost cirque, • an inventory of debris flows using are usually overdeepened and filled with airphoto interpretation, scale 1:10 000, lakes. Rocky sills separate the cirques from • periodic photogrammetry of selected the glacial troughs. The glacial cirques form sites, scale 1:1 000–1:5 000 for field experi- autonomous morphodynamic systems with mental slopes located relatively close to me- a local erosional level to which present-day teorological stations (the Hala Gąsienicowa geomorphic processes respond. High rocky Valley), ridges with steep slopes rise among glacial • field experiments on rockfall talus, al- valleys. The upper parts of the slopes were luvial talus and debris-mantled slopes in the not glaciated and have periglacial relief. High Tatras, ongoing since 1975 They are dissected by systems of rocky tro- ughs which had been forming since the end of the last glaciation, during a retreat of the GEOMORPHOLOGICAL AND GEOLOGICAL glaciers. At the rocky troughs’ outlets late SETTING OF THE TATRAS glacial heaps and alluvial talus cones formed (Klimaszewski 1996). During the Holocene The Tatras are non-glaciated high-moun- evolution of glacial and periglacial relief of tains (Fig. 1). The outstanding feature of the High Tatras, hillslope debris flows play- such mountains is a presence of glacial and ed an important role. nival forms, although the system of the forms In the Western Tatras, composed of less is of a relict nature, i.e. it was formed during resistant gneisses and shales, glaciation was the Pleistocene glacial cycle. Although, there not extensive and covered only the upper
KKsisiąążżkka1.indba1.indb 113838 22008-06-26008-06-26 110:57:090:57:09 Geomorphic Activity of Debris Flows in the Tatra Mts … 139
Figure 1. Tatra Mountains sketch and vertical distribution of debris flow tracks (%%) in the High (HT) and Western Tatras (WT), Slovakia. Source: Midriak, R. (1984)
KKsisiąążżkka1.indba1.indb 113939 22008-06-26008-06-26 110:57:090:57:09 140 Adam Kotarba
sections of the valleys. In the limestones DEBRIS FLOWS IN HIGH MOUNTAINS and dolomites of Czerwone Wierchy massif, only two glaciers formed definite systems of Fast mass movements comprise debris flows, cirques separated from troughs by high ro- mud-debris flows and landslides. Their oc- cky sills. Here, the slopes of glacial cirques currence is most often, yet not always, asso- are, in the majority, gentler when compared ciated with catastrophic precipitation and to their equivalents in the High Tatras and floods (Photo 1). At the lower border of the the valley bottoms were not overdeepened by glaciarised alpine zone, the magnitude and glacial erosion. Longitudinal valley profiles dynamics of mass movements depends also are levelled out. In the Holocene relief evo- on substrate properties of a ‘paraglacial’ lution of the glaciarised parts of the Western environment. The term paraglacial environ- Tatras, the role of valley-confined debris ment refers to terrain which is at present be- flows was particularly important. ing exposed from beneath melting glaciers. The paper focuses on high-energy geo- This terrain is rich in various-grained loose morphic events namely debris flows, which sediments. As the material is not consoli- operate on these non-glaciated high moun- dated, it is easily mobilized in the form of tains, mainly above the upper timber line. fast debris flows. Along with the retreat of
Photo 1. Debris flows on alluvial cone in Dolina Zeleného Plesa (Slovakia) triggered in 2001 and 2002, and photo centred on rockwall Maly Kežmarsky štit after 6-hour lasting rainfall (17.07. 2001, photo taken by G. Bugár)
KKsisiąążżkka1.indba1.indb 114040 22008-06-26008-06-26 110:57:090:57:09 Geomorphic Activity of Debris Flows in the Tatra Mts … 141
glaciers to higher locations, permafrost These are slope and channel subsystems wastes and areas prone to accelerated de- (Kotarba et al. 1987). Under such conditions, gradation by denudation processes expand, there is no link between the two subsystems, because the covers are being soaked with so products of weathering and of denudation water originating from permafrost thawing. as broadly understood are not transferred During the last several dozen years, under from one to the other subsystem. Changes in conditions of accelerated melting of glaciers slope relief are local, and material triggered in all mountains, paraglacial areas have been during slope formation stays in the given sub- expanding. This denotes a territorial expan- system. A similar situation occurs in channel sion of accelerated degradation of mountains subsystems on valley floors. The above is ob- by fast mass movements, including debris served in the Tatras when short-lasting rains flows, in a transitional zone between glacia- fall with an intensity over 1 mm/min and re- rised and non-glaciarised zones (haute mon- ach 40 mm in the course of an hour. The de- tagne alpine and haute montagne pyrénéene bris flows which form under such conditions zones). An example of a catastrophic debris on scree slopes and rock-debris mantled slo- flow in the paraglacial area of the Alps is the pes of the High and Western Tatras do not event which took place in the Aosta valley on descend to the valley bottoms and have no the Mont Blanc massif in July 2003. During influence on the dynamics of channels (si- a dry period without precipitation, at the di- tuation A). Long-lasting summer precipita- rect foreland of the Frébourge glacier (the tion of duration above 3 days and intensity Ferret valley) a debris flow was triggered by << 1 mm/min cause a linear erosion and mud an outburst flood from an ice avalanched flows, mainly in a timber vertical zone of the dammed lake formed in the zone of the ret- Tatras, while a lateral erosion by flood water reating glacier. In the proglacial alluvial cone in the valley channels results in erosional un- a volume of 30,000 m3 of debris was transfer- dercuts, landslides and scars formed near the red at a speed of 5–8 m/s. The whole event bases of slopes, at the contact part between occurred in 3 surges (Deline et al. 2004). slope and channel subsystems (situation B). Phenomena of this type are not rare in the The most intensive geomorphological work paraglacial zone of the Alps. The lack of is done in the mountains during rapid flo- glaciers in the Tatras explains why such pro- ods, when a precipitation of a few day’s dura- cesses are unknown in the environment of tion (situation B) is boosted by precipitation our mountains, nevertheless, on the micro- characteristic of situation A. Then, both sub- scale, paraglacial processes do occur in the systems join together and threshold values highest parts of the High Tatras, particularly determining the stability of slopes and valley in the most immediate vicinity of perennial floors are exceeded in the entire system (Re- snow patches (Rączkowska 2005). The geo- betz et al. 1997, Kotarba 2002). After the we- morphological hazards of debris flows in athering-soil covers on the slopes have been European mountains have been discussed, saturated with water, they become liquefied, and results of different studies published and earth-debris flows and landslides occur (e.g. Pasuto and Soldati 2004, Glade 2005, in particular geoecological belts. During the Jakob and Hungr 2005, Luino 2005, Hürli- catastrophic flood of July 1997 in the Tatras, mann et al. 2006). an unusual phenomenon was observed. The phenomenon might be called a ‘wash-out’ of entire slope covers, together with trees. DEBRIS FLOWS IN THE TATRA MOUNTAINS In the floors of small valleys, which dissect the slopes in the timber vertical zones, bare A system analysis of mountain relief evo- bedrock has exposed locally. The wood-mi- lution assumes two morphodynamic subsy- neral material was deposited as cones at the stems which function independently under outlets of side valleys, while on the floors average hydrometeorological conditions. of main valleys this material formed dams.
KKsisiąążżkka1.indba1.indb 114141 22008-06-26008-06-26 110:57:100:57:10 142 Adam Kotarba
In consequence, the river/stream channels to trigger flows affecting entire slopes. The were altered, those which had earlier been probability of such precipitation at Hala Gą- stable for tens of years were dissected, and sienicowa is 10% (Niedźwiedź 2003). Howe- river load (including boulders up to 2 m in ver, tremendous debris flows can sometimes diameter) was transported over a distance of happen at lower threshold values. Lukniš a 30 m while finer material was transported (1973), after Zaruba and Mencel (1954), ci- out of the Tatras altogether (Kotarba 1998). tes the description of the enormous debris A cartographic analysis of the trails of flow at Slavkovsky štít in the Slovak Tatras debris flows in the Slovak High and Western which formed during a 26 mm hourly preci- Tatras attests to a height differentiation of pitation event taking place on 15 July 1933. the forms (Fig. 1). They are most numerous The corresponding diurnal total was ‘only’ in the height interval of 1400–1900 m a.s.l. 62 mm. The aforementioned threshold value This is the alpine meadow zone and, simul- might be underestimated because the pre- taneously, the zone in which precipitation cipitation attributed to the discussed event totals are highest. has been measured from the meteorological In geomorphological studies in the Ta- station in Starý Smokovec, which is rather tras, the determination of the conditions ne- distant from the site at which the flow was cessary to trigger fast mass movements has triggered. Midriak (1984) in his synthesis been attempted for a long time. The thres- for the Tatras and the neighbouring areas hold values which should be determined are concluded that the largest volumes of debris the lowest precipitation totals for events of masses transported by single flows are up to brief duration that are directly responsible 25,000 m3. Janáčik (1971) presented the de- for the fast transfer of rock masses in the scription of the huge debris flow of volume form of debris- and mud-debris flows on 22,000 m3 formed at the Osobitá summit in the slopes, with a sufficient amount of loose the Western Tatras in August 1970. It can weathered material, and which give rise to be accepted that the flows in the Polish part new forms that modify the circulation sy- of the Tatras transport smaller amounts of stem of water and mineral substances that debris. In the Polish part of the High Tatras formerly existed on the slope. The threshold the material transferred in individual hillslo- values have to differ in various mountain pe flows reached only a few thousands cubic areas because the same precipitation does meters in total volume (Kotarba 1992), whi- not always generate similar geomorpholo- le in the Western Tatras the figure is smal- gical results. Substratum resistance to an ler–often of an order of a few hundred cubic external impulse is controlled by lithology meters (Krzemień 1988). The collected data and tectonics, morphometric parameters on recorded precipitation and masses of ma- and morphographic features of a given area terial transferred in debris flows triggered (i.e. the gradient and length of slopes, the by such precipitation suggest that the sizes of network density of slope dissection) and forms generated by debris flows are mainly by types of vegetation cover (coverage with conditioned by local topography, sometimes high-mountain meadows and dwarf pine pa- called ‘relief energy’. Precipitation of similar tches). Observations of talus slopes carried intensity and efficiency can generate flows out since the 1960s in the Sucha Woda and differing significantly as to size. The length Pańszczyca Valleys, by the Research Station of slopes, their dissection by the valley sy- of the Institute of Geography and Spatial stems of the first and second orders, the Organization, Polish Academy of Sciences spatial extent of precipitation, type of slope (IGSO PAS) at Hala Gąsienicowa allowed it covers and, especially, their consolidation to be assumed that hourly precipitation to- decides on sizes of generated forms. tals exceeding 25 mm are the threshold valu- The analysis of the tracks of the debris es that can give rise to a debris flow. Hourly flows in the Slovak High and Western Ta- precipitation of just 35–40 mm is sufficient tras, performed by Midriak (1984) showed,
KKsisiąążżkka1.indba1.indb 114242 22008-06-26008-06-26 110:57:110:57:11 Geomorphic Activity of Debris Flows in the Tatra Mts … 143
that despite similar hydrometeorological Floods in the Alps and Tatras are often conditions generating the valley confined characterised by hydrometeorological pa- debris flows, the length of these forms differ rameter which are alike, yet their outcomes (Table 1). In the Western Tatras 250–500 m are not comparable. The last huge flood in long flows predominate (51.4%), while in the Alps which occurred on 21–23 August the Western Tatras those 500–1000 m long 2005, affected an area of 6,500 km2. Pre- are the most common (55.2%). The relief cipitation of 200 mm falling in 48 hours, type decides upon lengths of the flow tracks. with an instantaneous intensity reaching In the relief of the High Tatras, in the areas 20 mm/hour resulted in floods in central located above the upper timber line, glacial Switzerland, western Austria and southern cirques predominate, these being arranged Bavaria, as well as partially in the Roma- in a stairway pattern, separated by rocky sills nian eastern Carpathians and in Transyl- and, most often, each cirque has a bipartite vania. According to calculations, floods of slope/slope sequence; rockwall–talus slope. such a territorial extent occur once every Therefore, debris flows which form on talus 100 years (Tropeano and Turconi 2005). slopes, flow down an open hillslope, and are In 2000, a huge flood of smaller extent yet not topographically constrained (hillslope higher precipitation occurred in the Pen- flows according to Brunsden classification, nine Alps. On 11–15 October, the Swiss 1979). In contrast to such topographic con- meteorological service recorded maximum ditions, in the Western Tatras glacial cirques precipitation totals of 700–800 mm during are not overdeepened, and continuous valley 12 hours in Piemonte-Aosta valley at the confined flows are the most common. Thus Swiss-Italian boundary. In the Alps, extre- their length and vertical distribution is gre- me diurnal precipitation reaching 400 mm ater. They often follow gullies cut in drift, can happen. For example, in the Locarno- talus or regolith by previous flow events. Magadino region precipitation of 414 mm was recorded on 10 September 1983. Much higher values (840 mm/23hrs) were obser- Table 1. Length of valley-confined debris flow ved in the Eastern Pyrenees during the ca- tracks in the High and Western Tatra, Slovakia tastrophic flood in October 1940 (Soutadé 1969). Fig. 2 presents the recurrence of ma- Length of debris flow tracks High Tatras Western Tatras ximum diurnal precipitation in the Tatras (in meters) (%) (%) and the Alps. During the last tremendous flood in the ≤ 250 2.8 3.9 Tatras on 4–8 July 1997 the sum of precipi- 250–500 51.4 35.9 tation at Hala Gąsienicowa was 330.3 mm, 500–1,000 36.5 55.2 while diurnal precipitation of 223.5 mm ≥ 1,000 9.3 5.0 and an hourly intensity of 39 mm were re- Source: R.Midriak 1984 corded on 8 July. According to the data compiled by Niedźwiedź (2003), it is belie- ved that the extreme precipitation at Hala Gąsienicowa recorded in 1927–2002 were similar to those recorded in the Alps: one- COMPARISON OF DEBRIS FLOWS IN THE day precipitation—300.0 mm (30.06.1973), TATRA MOUNTAINS AND OTHER EUROPEAN three-day precipitation—422.4 mm (16–18. MOUNTAINS 07.1934) and five-day precipitation— 462.3 mm (14–18.07.1934). The above pre- Comparison of geomorphologic events of cipitation is comparable with the alpine, types A + B in the Tatras to similar situa- although the geomorphologic effects were tions in other European mountains is diffi- small, on the scale of particular Tatra dra- cult, due to the small size of our mountains. inage basins.
KKsisiąążżkka1.indba1.indb 114343 22008-06-26008-06-26 110:57:110:57:11 144 Adam Kotarba
Figure 2. Recurrence intervals (in years) of annual maximum daily rainfall for high-mountain part of the Tatra Mts compared with selected Alpine stations. Source: Zeller, J., Geiger, H. and Röthlisberger, F. (1976–1984) and Anselmo,V. (1979)
Examination of air circulation patterns precipitation reaching 40 mm occurred with in the Tatras has offered a background for a 10% probability (once per 10 years) (Ce- determination of the genesis and intensity bulak et al. 1986, Niedźwiedź 1992). On the of atmospheric precipitation in various syn- other hand, a number of extreme precipita- optic situations (Niedźwiedź 1981, Cebulak tion events increased during the last 7 years 1983). The origin and magnitude of pre- when compared with the period of 76 years cipitation triggering fast mass movements analysed by Niedźwiedź (2003). Since 1995 on the Tatras slopes have been identified. a transition to a moister climatic phase has Knowing hydrometeorological parameters been observed. This finding is crucial for for the Alps and the Tatras, one can attempt considering the present-day evolution of the to determine their role in the present-day slope relief under the influence of fast mass modelling and transformation of the high- movements above the upper timber line both mountain slopes of these mountains, howe- in the Tatras and the Alps. ver, with respect to the slope subsystem only An intensified activity of debris flows is and only under the influence of short-lasting also observed in many mountain areas of storm and intensive convection precipitation Europe. Both the magnitude and freque- (situation A). Under the influence of such ncy of precipitation have increased in the events, the most spectacular changes in slope recent decades. The above is substantiated relief take place. Following statistical calcu- by the records from historical documents lations, maximum hourly precipitation in the combined with other methods (Jomelli et Tatras, reaching 60 –80 mm, occurred with al. 2004, Marchi and Tecca 2006). Jomelli a 1% probability (once per 100 years) while et al. (2004) have evidenced that the influ-
KKsisiąążżkka1.indba1.indb 114444 22008-06-26008-06-26 110:57:110:57:11 Geomorphic Activity of Debris Flows in the Tatra Mts … 145
ence of climatic changes on the dynamics of composition was concerned, there was a pre- debris flows in the French Alps is visible in dominance of fine particles (150,000 m3), the scale of the recent 50 years. Twelve sets fine particles with coarser chunks of 0.25 m3 of aerial photographs taken in two test areas (80,000 m3) and rocky blocks exceeding 1 m3 (Massif des Ecrin and Massif du Dévoluy), (70,000 m3). located at 1,600 (1,900 m) to 2,400 m a.s.l., Measurements of the volume of mate- in 1948–2000 have been analysed. In these rial transferred by debris flows has been study areas, there are talus slopes not cove- made in other massifs of the Alps, and in red by glaciers since the Little Ice Age. The the mountains of Scotland, Scandinavia and importance of spatial and temporal overlap Spitsbergen. The results of the above exami- between heavy rainfall and debris flow acti- nation, compiled by Van Steijn (1996), are vity was analysed, making use of data from shown in Fig. 3. The Tatra data refer to all meteorological stations located in these are- forms of debris flows mapped and described as. The increase in temperature at high al- in the Slovak and Polish Tatras. The Tatra titudes and an upward shift of the isotherm flows occupy a central position with respe- 0°C provoke a degradation of the permafrost ct to other mountain groups and are closest in rockwalls, as well as reducing snow cover to the forms described in the French Alps duration. The slopes elevated above 2,200 m in the Bachelard massif belonging to the are nowadays exposed to greater tempera- Southern Alps which are not glaciarized ture variations and intensified physical we- nowadays and in which the highest summits athering. A significant increase in summer reach 2,600–2,800 m (le Grand Cheval de rain, higher than 30 mm/day, has also been Bois—2,839 m). In terms of the sizes of rid- observed in high elevated areas since 1975. ges and valleys they are similar to the Tatras. As a consequence, a shift of triggering zo- However, the highest non-glaciarized alpine nes of debris flows towards higher elevations ridges are subjected to intensive modelling has occurred during the period 1976–2000 by debris flows (e.g.. Ariège, Arve, Zillertal (Jomelli et al. 2004). Similar tendencies have and Zell on Fig. 3). been recognized in other areas in the Alps The fast mass movements in the moun- (Zimmermann and Haeberli 1992). tains of Scotland and Spitsbergen produce The above finding is confirmed when forms smaller by an order of magnitude than we compare mountains differing in their si- those generated in the Tatras, although the zes–the Tatras to the Alps (Fig. 3). Similar threshold values are much smaller there. diurnal and instantaneous precipitation can According to Larsson (1982), precipitation of result in the debris and mud-debris flows of intensity 2 mm/hr can generate debris flows the volume of 400,000 m3 (Aulitzky 1970), or in northern Scandinavia, if preceded by diur- even of 500,000 m3 in the Alps (Govi 1984). nal precipitation of 30–50 mm. According to The debris flow in Tirol, described in details Rapp and Nyberg (1981), the weathering co- by Aulitzky, was formed in Inzing by convec- vers in these mountains are thin and perm- tion precipitation amounting to 39.4 mm/24 afrost is present there. The active layer is hrs, on 26.07.1969. The recorded instanta- thin in summer and saturated with water neous intensities reached 2.1–2.5 mm/min. originating from thawing of permafrost, This precipitation fell on a summer day in so the debris flows are triggered even by a 12.2 km2 tributary valley joining the Inn a tiny precipitation impulse and affect only Valley. The material triggered in a hanging a thin active layer. In the case of the Scot- valley was transferred to the Inn Valley and tish mountains it is evidenced that the mag- resulted in catastrophic damage in Inzing nitude of relief changes is controlled by the village located on the alluvial fan. The vo- amount of loose weathered material which lume of material mobilised in the hanging can be transferred. In numerous areas sheep valley and deposited on the fan in the Inn pasturage and deer husbandry which have valley was 404,000 m3. Where the grain size lasted for at least 200 years , as well as grass
KKsisiąążżkka1.indba1.indb 114545 22008-06-26008-06-26 110:57:120:57:12 146 Adam Kotarba
Figure 3. Debris flow magnitude-frequency relationshops in European mountains. Source: reprinted from Van Steijn, Geomorphology, (1996), vol. 15, p. 267.
burning, have resulted in accelerated degra- Wales. Innes (1985) compiled the features dation of slope covers in the past few cen- of over 900 debris flows in the mountains of turies. The present-day debris flows are of Scotland and Norway. The average volume limited extent and size, as they form within (arithmetic mean) of the forms in Scotland those thin, degraded covers (Innes 1985). is 10–50 m3, compared with 100–350 m3 in The form sizes are thus small, notwithstan- Norway. ding the Scottish climate favouring convec- Icelandic debris flows are of small to me- tional precipitation. Addison (1987) provi- dium size (from less than 100 m3 to about ded evidence that precipitation of 40 mm 3,000 m3, and are triggered mainly by rapid during an hour is the threshold value trig- snowmelt (27%) or snowmelt associated with gering debris flows in Snowdonia in northern rain (21%), long-lasting rainfall (27%) and
KKsisiąążżkka1.indba1.indb 114646 22008-06-26008-06-26 110:57:120:57:12 Geomorphic Activity of Debris Flows in the Tatra Mts … 147
intense rainfall (only 13%) (Decaulne and fer of volcanic material in non-European Saemundsson 2006). The above data sug- regions and do not transport material of gest that the grading of hydrometeorological block sizes. factors triggering debris flows is different in Based on the extensive array of results Iceland. Under conditions of a subpolar-oce- from field studies performed in non-Euro- anic climate rapid snowmelt is responsible pean mountains, it is unambiguously con- for transforming both the fine and coarse cluded that substrate properties and topo- material on hillslopes, while intense rainfall graphy determine erosion and type of slope is of minor importance. transformation. According to Hovius et al. (2004), finding a decisive factor remains a challenge. The Tatra massif is characteri- FINAL CONCLUSIONS sed by a moderate rate, yet the results are catastrophic for people in exceptional cases Debris flows are among geomorphological only and on a small scale. The alpine debris processes occurring commonly in different flows result in dramatic, catastrophic dama- climatic zones, though, they play a particu- ge to the infrastructure of towns and villa- lar relief-forming role in mountain areas ges, and are hazardous for people dwelling due to the large amount of water originating in those mountains. In this domain, the Ta- from rain, snow melting and the thawing of tras are definitely different from the Alps, glaciers. Simultaneously, water is a factor though since the early 1980s high-magni- which triggers a flow process, and is a car- tude debris-flow activity has become more rier of mineral and organic matter during frequent in the Tatras also. This tenden- a flow event. Rainwater, as the factor ini- cy is visible in many localities, both in the tiating the process, has to be supplied fast, Polish and Slovak parts of the mountains. even rapidly. Water of other origins can only This circumstance supports the need for a add to the relief-forming effects. An abun- comprehensive study of debris flow activity, dance of loose weathered material is also a even if the debris flow hazard is very low and necessary condition for debris flow forma- limited to the local scale only. Systematic tion. Therefore, high-mountain areas and precipitation records, mainly regarding the polar regions are prone to strong modelling, frequency of heavy rain, and the triggering yet the scale of impacts on the existing re- of debris flows would seem very appropria- lief is conditioned, to a significant degree, te from the geomorphological and practical by a series of other environmental factors. points of view. The analysis of such data will As these factors can overlap or cancel each provide a better understanding of debris other out their combined effect is a tremen- flow initiation in the context of global clima- dous differentiation of the manner and rate te change also. of slope relief transformation on the scale of the European mountains. Considering the size of the debris flows ACKNOWLEDGEMENTS on the global scale, the European debris flows are among the small and medium The study in the present paper has been car- ones. L. Innes (1983) graded debris flows in ried out within the framework of the research four sizes; large-scale (>105 m3), medium- project entitled “Extreme meteorological scale (103–105 m3), small-scale (1–103 m3) and hydrological events in Poland”, financed and micro-scale (<1 m3). In the 10-grade by the Ministry of Science and Higher Edu- classification of M. Jakob (2005) the Tatra cation of Poland (PBZ-KBN-086/P04/2003, debris flows belong to grade 4 (104–105 m3 ), Task 5.1). while the alpine flows are assigned to gra- de 5 (105–106 m3). The largest debris flows of grade 10 (> 109 m3) refer to the trans-
KKsisiąążżkka1.indba1.indb 114747 22008-06-26008-06-26 110:57:130:57:13 148 Adam Kotarba
REFERENCES Govi, M. (1984), Les phenomenes catastrophi- ques d’origine exogene, in Geomorphologie Addison, K. (1987), Debris Flow during Intense et risques naturels dans les Alpes, 25e Congrès Rainfall in Snowdonia, North Wales: a Pre- International de Géographie, Paris-Alpes. liminary Survey, Earth Surface Processes and Hess, M. (1965), Piętra klimatyczne w polskich Kar- Landforms, 12: 561–566. patach Zachodnich [Vertical Climatic Zones in Anselmo, V. (1979), Il nubifragio del 7 Agosto the Polish Western Carpathians], Zeszyty Na- 1978 nel bacino del Toce [The Flood of 7 Au- ukowe UJ, Prace Geograficzne, 44: 1–267. gust 1978 in the Toce Basin, Bolletino dell’ As- Hovius, N., Lague, D. and Datson, S. (2004), Pro- sociazione Mineraria Subalpina, 16, 2. cesses, Rates and Patterns of Mountain-Belt Aulitzky, H. (1970), Der Enterbach (Inzing in Ti- Erosion, in Owens, P.N. and Slaymaker, O. rol) am 26. Juli 1969: Versus der Analyse eines (eds.), Mountain Geomorphology, Edward Murganges als Grundlage für die Neuerstel- Arnold (Publishers) Limited: 109–131. lung einer zerstörten Wildbachverbauung. Wil- Hürlimann, M., Copons, R. and Altimir, J. (2006), bach- und Lawinenverbau, Zeistchrift des Ver- Detailed Debris Flow Hazard Assessment in eins der Diplomingenieure der Wildbach- und Andorra: a Multidisciplinary Approach, Geo- Lawinenverbauung Österreichs, 34, 2: 31–66. morphology, 78: 359–372. Brunsden, D. (1979), Mass Movements, in Em- Innes, J.L. (1983), Debris Flows, Progress in Phys- bleton, C. and Thornes, J. (eds.), Process in ical Geography, 7, 4: 469–501. Geomorphology, London, Edward Arnold, Innes, J.L. (1985), Magnitude-Frequency Rela- 130–186. tions of Debris Flows in Northern Europe, Cebulak, E. (1983), Maximum Daily Rainfalls Geografiska Annaler, 67A: 23–32. in the Tatra Mountains and Podhale Basin, Jacob, M. (2005), A Size Classification of Debris Zeszyty Naukowe UJ, Prace Geograficzne, 57: Flows, Engineering Geology, 79: 151–161. 337–343. Jacob, M. and Hungr, O. (2005), Debris-Flow Cebulak, E., Farat, R., Koczorowska R., Niedź- Hazards and Related Phenomena, Springer, wiedź T. and Plenzler, W. (1986), Deszcze Berlin. ulewne [Heavy Rainfalls], in: Atlas Hydrolog- Janáčik, P. (1971), Nektoré poznatky z inventari- iczny Polski [Hydrological Atlas of Poland], začného vyskumu v chránenej krajinnej ob- Vol. 2, book 1, (a joint publication under di- lasti Malá Fatra [Some pieces of knowledge rection of Stachỳ, J.), Wydawnictwa Geolo- of the inventory research and investigation in giczne, Warszawa, 31–35. the protected landscape region of Mala Fa- Decaulne, A. and Saemundsson, T. (2006), Geo- tra], Geograficky Časopis, 23, 2: 186–191. morphic Evidence for Present-Day Snow- Jomelli, V., Pech, P., Chochillon, C. and Bruns- Avalanche and Debris-Flow Impact in the den D. (2004), Geomorphic Variation of De- Icelandic Westfjords, Geomorphology, 80: bris Flows and Recent Climatic Change in the 80–93. French Alps, Climatic Change, 64: 77–102. Deline, P., Chiarle, M. and Mortara, G. (2004), Kotarba, A. (1997), Formation of High-Mountain The July 2003 Frebouge Debris Flows (Mont Talus Slopes Related to Debris-Flow Activity Blanc Massif, Valley of Aosta, Italy): Water in the High Tatra Mountains, Permafrost and Pocket Outburst Flood and Ice Avalanche Periglacial Processes, 8: 191–204. Damming, Geografia Fisica e Dinamica Qua- Kotarba, A. (1998), Morfogenetyczna rola opa- ternaria, 27, 2: 107–111. dów deszczowych w modelowaniu rzeźby Galibert, G. (1960), L’évolution actuelle des „fa- Tatr podczas letniej powodzi w roku 1997 ces Nord” de la haute montagne alpine dans [The Morphological Role of Rainfalls in the le massif de Zermatt, Revue Géographique des Modelling of Tatra Relief during the Summer Pyrénées et du Sud-Ouest, 31. Flood of 1997], Z badań fizycznogeograficz- Glade, T. (2005), Linking Debris-Flow Hazard nych w Tatrach–III Dokumentacja Geogra- Assessment with Geomorphology, Geomor- ficzna 12, Instytut Geografii i Przestrzennego phology, 66: 189–213. Zagospodarowania, PAN, Warszawa: 9–23.
KKsisiąążżkka1.indba1.indb 114848 22008-06-26008-06-26 110:57:130:57:13 Geomorphic Activity of Debris Flows in the Tatra Mts … 149
Kotarba, A. (2002), Współczesne przemiany elements in the upper Vistula basin], Rozpra- przyrody nieożywionej w Tatrzańskim Par- wy habilitacyjne UJ, 58, Kraków. ku Narodowym [Recent Changes of Abiotic Niedźwiedź, T. (1992), Climate of the Tatra Components of Natural Environment of the Mountains, Mountain Research and Develop- Tatra National Park], in Borowiec, W., et al ment, 12, 2: 131–146. (eds.), Przemiany środowiska przyrodniczego Niedźwiedź, T. (2003), Extreme Precipitation Tatr, Kraków-Zakopane:13–19. Events on the Northern Side of the Tatra Moun- Kotarba, A., Kaszowski, L. and Krzemień K. tains, Geographia Polonica, 76, 2: 15–23. (1987), The High-Mountain Denudational Rączkowska, Z. (2005), Morfodynamiczne cechy System of the Polish Tatra Mountains, Geo- obszarów z wieloletnią zmarzliną w Tatrach graphical Studies, Spec. Issue 3, Polish Acad- [Morphodynamic characteristics of areas of emy of Sciences, Wrocław. permafrost in the Tatras], VII Zjazd Geo- Kotarba, A., Kłapa, M. and Rączkowska, Z. morfologów Polskich, Kraków, 385–388. (1983), [Present-Day Transformation of Al- Pasuto, A. and Soldati, M. (2004), An Integrated pine Granite Slopes in the Polish Tatra Moun- Approach for Hazard Assessment and Miti- tains] Dokumentacja Geograficzna 1, Instytut gation of Debris Flows in the Italian Dolomi- Geografii i Przestrzennego Zagospodarowa- tes, Geomorphology, 61: 59–70. nia (IGiPZ) PAN, Warszawa. Rapp, A. and Nyberg, R. (1981), Alpine Debris Krzemień, K. (1988), The Dynamics of Debris Flows in Northern Scandinavia, Geografiska Flows in the Upper Part of the Staroro- Annaler, 58A, 3–4: 193–200. bociańska Valley (Western Tatra Mts), Studia Rebetz, M., Hugon, R. and Baeriswyl, P.-A. Geomorphologica Carpatho-Balcanica, XXII: (1997), Climatic Change and Debris Flows in 123–144. High Mountain Regions: The Case Study of Larsson, S. (1982), Geomorphological Effects on the Ritigraben Torrent (Swiss Alps), Climatic the Slopes of Longyear Valley, Spitsbergen, Change, 36: 371–389. after a Heavy Rainstorm in July 1972, Geo- Soutadé, G. (1969), Un milieu sub-alpin de glyp- grafiska Annaler, 64A, 3–4: 105–125. togénèse, Revue Géographique des Pyrénées et Luino, F. (2005), Sequence of Instability Process- du Sud-Ouest, 40, 4: 353–370. es Triggered by Heavy Rainfall in Northern Tropeano, D. and Turconi, L. (2005), Effetti del Italy, Geomorphology, 66: 13–39. nubifragio del 21–23 Agosto 2005 nelle Alpi Lukniš, M. (1973), Relief Vysokych Tatier a ich Austro-Svizzere: osservazioni preliminari [Ef- predpolia [Geomorphology of the High Ta- fects of the Flood of 21–23 August, 2005 in tra Mountains and Foreland], Vydavatel’stvo the Austrian-Swiss Alps] , GEAM Territorio SAV, Bratislava. e Difesa del Suolo, Torino, 77–85. Łajczak, A. (1996), Hydrologia [Hydrology], in Van Steijn, H. (1996), Debris-Flow Magnitude- Mirek Zb. (ed.) Przyroda Tatrzańskiego Parku Frequency Relationships for Mountainous Narodowego, Zakopane-Kraków, 169–196. Regions of Central and Northwest Europe, Marchi, L. and Tecca, P.R. (2006), Some Obser- Geomorphology, 15: 159–273. vations on the Use of Data from Historical Wit-Jóźwik, K. (1974), Hydrografia Tatr Wysokich Documents in Debris-Flow Studies, Natural [Hydrography of the High Tatra Mountains], Hazards: 38, 301–320. Objaśnienia do mapy hydrograficznej „Tatry Midriak, R. (1984), Debris Flows and their Oc- Wysokie” 1:50 000, Dokumentacja Geogra- currence in the Czechoslovak Carpathians, ficzna 5, Instytut Geografii i Przestrzennego Studia Geomorphologica Carpatho-Balca- Zagospodarowania, PAN, Warszawa. nica, XVIII: 135–149. Zaruba, Q. and Mencl, V. (1954), Inženyrská geo- Niedźwiedź, T. (1981), Sytuacje synoptyczne i ich logie [Engineering Geology], Praha: 425p. wpływ na zróżnicowanie przestrzenne wybra- Zeller, J., Geiger, H. and Röthlisberger, F. nych elementów klimatu w dorzeczu górnej (1976–1984), Starkniederschläge des schwei- Wisły [Synoptic situations and their influence zerischen Alpen- und Alpenrandgebietes, Bir- on the spatial differentiation of some climatic mensdorf.
KKsisiąążżkka1.indba1.indb 114949 22008-06-26008-06-26 110:57:140:57:14 150 Adam Kotarba
Zimmerman, M. and Haeberli, W. (1992), Cli- matic Change and Debris Flow Activity in High Mountain Areas. A Case Study in the Swiss Alps, Catena (suppl.) 22: 59–72.
Paper first received: March 2006 In final form: November 2007
KKsisiąążżkka1.indba1.indb 115050 22008-06-26008-06-26 110:57:140:57:14 EXTREME RAINFALLS AND THEIR IMPACT ON SLOPES —EVALUATION BASED ON SOIL EROSION MEASUREMENTS (AS EXEMPLIFIED BY THE SUWAŁKI LAKELAND, POLAND)
EWA SMOLSKA Faculty of Geography and Regional Studies, Warsaw University, Krakowskie Przedmieście 30, 00-927 Warszawa, Poland E-mail: [email protected]
Abstract: Monitoring of soil erosion on selected slopes of the Suwałki Lakeland (NE Poland) was conducted in the years 1987–1989 and 1998–1999. The extreme rainfall was characterised by an efficiency of 35.7 mm and an average intensity of 0.5 mm per minute. This rainfall caused erosion along the entire length of the slopes, and its volume was equal to the average annual value. Almost 75% of the material deposited in the lower, concave section of the slopes during the 5-year period of measurements was transferred beyond the slope base. Some of the slopes of length over 100 m was cut by networks of rills up to 50 cm deep, and the rate of soil loss was 30 t ha-1. This rainstorm was most important in respect to the intensity and transfer of eroded soil material, and was a decisive factor in soil loss and in the redistribution of soils on the slopes over the entire period of measurement.
Key words: rainfall, extreme events, soil loss, soil erosion and deposition, NE Poland.
INTRODUCTION faciuk (1991), such areas are threatened by erosion of the soil due to water to a limited, In research on soil erosion by water, lake- locally a moderate degree. A similar situation lands play a special role as compared with applies in Germany, where areas of post-gla- other lowland areas, on account of their cial relief are threatened by water-induced complex relief. Where the share of cultivated soil erosion to a lesser degree than loess or slopes inclined over 9° is large a high inten- upland terrain (Auerswald 2006). During sity of soil erosion may take place (Niewia- rainfall of average intensity, local environ- domski and Skrodzki 1964, Chudecki and mental conditions are more important where Niedźwiecki 1983, Uggla et al. 1968, 1998). soil erosion is concerned. However, rainfall The important role of brief storms occur- of high intensity sees threshold values for ring in June and July has been emphasized overland flow exceeded, erosion then de- by several researchers (Uggla et al. 1967, pending mainly on rainfall intensity and run- 1968, Niewiadomski 1968, 1998, Smolska et off magnitude. Runoff increases from a di- al. 1995, Smolska 2002). According to Józe- vide toward lower parts of slopes, providing
KKsisiąążżkka1.indba1.indb 115151 22008-06-26008-06-26 110:57:150:57:15 152 Ewa Smolska
material for valley floors and river channels a straight shapes. Occasional dry bowl-shaped (Carson and Kirbky 1972, Froehlich 1982, valleys and gullies add variety to the slopes. Selby 1982, Teisseyre 1994, Bryan 2000). Soil erosion was studied on slopes typical The aim of research carried out in the for the area and lying within the basin of the periods 1987–1989 and 1998–1999 was thus upper Szeszupa (Fig. 1). A detailed descrip- to assess the intensity of soil erosion on se- tion of the studied slopes is presented in lected slopes in the Suwałki Lakeland (Smol- previous papers of Smolska et al. (1995) and ska 2002, 2005). The work described in this Smolska (2002, 2005). Selected parameters paper focused on the role of extreme rainfall of the slopes are as shown in the Table 1, on soil erosion and redeposition. while longitudinal profiles of the slopes are presented in Fig. 2. Measurements of interrill erosion were conducted on short slopes which STUDY AREA were several dozen meters to about 100 meters long, and 10–15 meters high. Rill erosion was The relief of the Suwałki Lakeland is typical observed sporadically on short slopes dur- of the last glaciation. It lies within the range ing field survey (Smolska et al. 1995), so this of the Pomeranian phase of the Leszczyńsko- process was measured on longer slopes of the Pomorski stadial (Ber 2000). A landscape Szeszupa depression, which were 125–280 m of hilly and undulating morainic plain prevails long with a high elevation (30–70 m). in this area. There are also hills and mounds of end moraine or of dead-ice moraine ori- gin. Other typical relief features are exten- CLIMATE OF THE STUDY AREA sive glacial depressions with forms of aerial AND METEOROLOGICAL CONDITIONS deglaciation on their floors. There are nu- OF THE STUDY PERIOD merous tunnel valleys in the area. Elevations are limited, usually of 15–30 m, only locally The temperate climate of the Suwałki Lake- exceeding 50 m. Slopes are usually inclined by land also has continental features. Mean 6° to 18° and have a convex-concave or annual air temperature is 6.1°C and mean
Table 1. Selected characteristics of investigated slope and slopewash rates
Average annual soil erosion or deposition Extreme soil loss (+) [kg ha-1] [kg ha-1]
Convex Convex Wash and and
] straight Concave straight Concave o Lithology Slope Lendth [m] Hight [m] Inclination [ segment segment segment segment
Udziejek I 67 15 2–6 sandy loam 49.4 35.4 38.9 55.1 Udziejek II 46 11 4–22 loam 23.7 +69.82 29.6 121.3 Udziejek Górny 105 10 3–7 silty sand 390.0 +79.4 488.9 819.2 Interrill Łopuchowo 66 12 3–11 sand 7,255.6 +720.7 10 320.0 16 512.5 Smolniki 280 45 4–17 sand and clay 3,775.0 0 4 729.5 0 Krejwelek 155 32 3–12 loamy sand 2,356.9 +9 247.9 31 097.7 18 742.3
Rill Snołda 120 27 2–10 loamy sand 470.2 +2 044.4 9 611.6 6 805.0 Gulbin 175 45 4–15 sandy loam 601.5 +3 624.4 6 458.1 8 922.5
KKsisiąążżkka1.indba1.indb 115252 22008-06-26008-06-26 110:57:150:57:15 Extreme Rainfalls and their Impact on Slopes—Evaluation Based on Soil Erosion Measurements 153
Figure 1. Geomorphological sketch (according to Ber 2000) of study area and location of soil erosion measurements: 1—morainic plateau, 2—end moraines, 3—dead-ice moraines, 4—outwash plains, 5—eskers, 6—kames, 7—tunnel valleys, 8—kettle holes, 9—bottom of depression, 10—dray valleys, 11—gullies and fans, 12—ice-contact slopes, 13—rivers and lakes, 14—slopes of interrill erosion measurements, 15—slopes of rill erosion measurements, 16—precipitation station.
annual rainfall 576 mm. Rainfall in excess ally, one of the lowest in Poland (Banasik of 30 mm occurs once every 2 or 3 years, and and Górski 1993). rainfall events exceeding 70 mm occur ap- Both periods of field research (1987–89 proximately once every 50 years (Stopa-Bo- and 1998–99) were relatively warm and hu- ryczka and Martyn 1985). Rainfall intensity mid in comparison with the multi-year pe- is limited. In Suwałki, the rainfall erosiv- riod 1951–65 analysed by Stopa-Boryczka
ity (Re) (according to the formula used in and Martyn (1985). Mean annual air tem- USLE) is 43Je (N h-1 yr-1) on average annu- perature was between 6.8°C and 7.5°C, and
KKsisiąążżkka1.indba1.indb 115353 22008-06-26008-06-26 110:57:150:57:15 Figure 2. Profiles of investigated slope with marked mean and extreme soil redistribution: 1—forest, 2—meadows and pastures, 3—arable land, 4—measurement sites, 5—mean annual values for erosion or accumulation, 6—extreme values for erosion or accumulation
KKsisiąążżkka1.indba1.indb 115454 22008-06-26008-06-26 110:57:150:57:15 Extreme Rainfalls and their Impact on Slopes—Evaluation Based on Soil Erosion Measurements 155
annual rainfall was between 541 mm and shape and inclination. Each collector collect- 716 mm (the Sidory-Kleszczówek meteoro- ed water and eroded soil for a strip of ground logical station). There were two rainfall epi- stretching from the water divide to the meas- sodes of considerable intensity, one in each urement stand. Depending on slope morphol- research period. This events were extreme ogy, measurements were based around 3 or 4 due to rainfall intensity, exceptional for the stands, the lowermost being located at the Suwałki Lakeland region. This is clearly vis- base of the slope. Every stand consisted of ible when values are set against the intensity 3 collectors, each with a 50 cm-wide en- of other rainfall events (Fig. 3). The heavi- trance. The collectors were emptied once est rainfalls occur in June and July. They every 4 to 6 weeks, and additionally after usually amount to only 25–30 mm and are each intensive rainfall. only of significant intensity during first Determination of the magnitude of rill 10 to 15 minutes. Their erosivity according to erosion was based on the volume of rills as the USLE formulae, is within a dozen or so calculated after Klimczak (1988). Similarly, erosivity units (Je). The heavy rainfall with the amount of accumulated material was an efficiency of 35.7 mm detected on June, determined as the volume of fans or cov- 22, 1999 was exceptional as regards its very ers deposited at slope bases. Measurements high intensity of 0.5 mmmin-1. Its erosivity were conducted seasonally with regard to amounted to 35.2 Je, which equalled the av- changing meteorological conditions, usually erage erosivity of all summer rainfall. 4 time per year.
Figure 3. Daily precipitation and its erosivity in the study periods 1987–1989 and 1998–1999.
METHODS RESULTS
Measurements of interrill erosion were con- SOIL EROSION ducted using Słupik’s collectors (Słupik 1973) During rainfall of moderate intensity, sheet modified slightly (Smolska 1993). Measure- wash causes erosion of upper and middle ment stand design was after Gerlach (1976), segments of short slopes, and deposition i.e. in places where the slope changed its takes place in lower segments of straight
KKsisiąążżkka1.indba1.indb 115555 22008-06-26008-06-26 110:57:150:57:15 156 Ewa Smolska
and convex-concave slopes (in their concave the entire research period. Seasonal evalua- segments)(Fig. 2). Only a limited amount tions of interrill erosion on the short loamy- of material is transported beyond slopes and sandy slope of Udziejek I (Fig. 4) illustrate accumulated near their bases. Many stud- the aforementioned event well. During the ies of the lakeland area indicate a similar rainstorm, erosion in the upper and middle course for this process (Niewiadomski and parts of the slope reached the mean annual Skrodzki 1964, Niewiadomski 1968, Uggla value. Strong sheet wash occurred in the et al. 1968, 1998, Chudecki and Niedźwiecki lower part of the slope. Rills developed in 1983, Klimczak 1993, Niewiarowski the middle and lower parts of the slope. Al- et al. 1993, Szpikowski 2002). In the Suwałki most 75% of the mass of soil material that Lakeland, erosion at the top of hills ranges had accumulated on the concave segment from 24 to 400 kg ha-1, while in the convex of the slope over the 5-year period of meas- and middle segments it is of 40 kg to over urement was rapidly transferred beyond the 7 t ha-1 (Smolska 2005). base of the slope. The heavy rainfall of June 22,1999 was Depending on the effectiveness and especially significant, due to the magnitude force of rainfall, the mechanics of the proc- of the erosion it caused relative to that over ess and its intensity, and in consequence,
Figure 4. Seasonal rates of erosion and deposition due to interrill and rill wash on the Udziejek I slope in the study period: st. 1–4—measurement sites.
KKsisiąążżkka1.indba1.indb 115656 22008-06-26008-06-26 110:57:160:57:16 Extreme Rainfalls and their Impact on Slopes—Evaluation Based on Soil Erosion Measurements 157
soil re-deposition differed on particular Table 2. Parameters of selected rainfall events slope fragments. This differentiation is calculated using USLE formulae well illustrated by the wash on the Udzie- jek I slope during 3 rainfall events of dif- Parameter 18.07.1988 12.06.1998 22.06.1999 fering erosivity (Table 2). Erosion in the upper part of the slope and deposition in P [mm] 11.1 18.5 35.7 I [mm min-1] 0.11 0.17 0.51 the middle and lower parts occurred dur- avg I [mm min-1] 0.31 0.41 0.66 ing brief (30 min) storms of significant in- 30 I [mm min-1] 0.56 0.65 0.82 tensity but a low efficiency of 8 to 15 mm 15 (Figure 5—18.07.1988). Strong splash oc- Re [Je] 3.90 9.50 35.20 curred during brief storms. While intensity of the rainfall exceeded infiltration capac- ity of the ground, the short duration made convex and straight segments of the slope, it possible for almost all the water to be with deposition taking place in the concave retained in the soil’s upper layers (Gil and segment (Fig. 5—12.06.1998). Only heavy Słupik 1972, Froehlich and Słupik 1980, rainfall exceeding 30 mm caused erosion Słupik 1981, Bryan 2000, Parsons and Stone on the entire slope length, with deposition 2006 ). Overland flow diminishes during its of material occurring on the Szeszupa river initial stage of formation or slightly later a flood plain (Fig. 5—22.06.1999). (Froehlich 1982, Parsons et al. 1998, Gil On many longer (> 100 m) slopes, a net- 1999). Much more material becomes mo- work of rills has been created. These were bile due to splash than to overland flow up to 30 cm deep, locally 50 cm, and erosion competency. Transport of material was in them amounted to 10–30 tha-1. Distinct pulsating, and deposition occurred directly accumulation cover developed at the slope beyond a zone of strongest erosion, which base, though the average thickness was only is beyond the convex segment of the slope. 2–3 cm, with 5–7 cm occurring locally only. Rainfall of similar intensity but greater per- The effectiveness of rill erosion was several sistence resulted in erosion along the entire times greater than that of interrill wash,
Figure 5. Sheet wash erosion and accumulation on the Udziejek I slope for selected rainstorms: st. 1–4—measurement sites; location of sites as in Figure 4.
KKsisiąążżkka1.indba1.indb 115757 22008-06-26008-06-26 110:57:160:57:16 158 Ewa Smolska 7—gullies, 8—roads, 9—scarps. 8—roads, 7—gullies, Figure 6. Rill networks and fans on the Gulbin slope formed during heavy rainfall in July and 1989 June 1999 (A) (B): 1—rills cm up to 10 deep, 2—rills of 10–20 cm deep, 3—rills deeper than 20 cm, 4—fans, 5—arable land, 6—pastures,
KKsisiąążżkka1.indba1.indb 115858 22008-06-26008-06-26 110:57:170:57:17 Extreme Rainfalls and their Impact on Slopes—Evaluation Based on Soil Erosion Measurements 159
but the spatial range was relatively limited. 100 kg ha-1 to 7 t ha-1. From 50% to 75% The rills developed in lower parts of slopes, of colluvium accumulated over 5 years was clearly occurring where vegetation cover is transported away. poorly developed. The impact of extreme Eroded soil material transferred beyond rainfall events on slopes with different the slope bases as a results of occasional vegetation is well illustrated by the net- extreme rainfalls events caused significant work of rills and fans on the Gulbin slope changes in soil redistribution on the Suwałki (Fig. 6). Only single rills were observed af- Lakeland slopes. This process is necessary ter extreme rainfall taking place in the mid- to preserve the steepness of slopes. Only dle of the vegetation period (Fig. 6A). How- such extreme events result in intensive ero- ever, when vegetation cover was less dense, sion along the entire length of slopes, and es- rills created a network of density up to pecially in their middle and lower parts. The 180 m per 100 m2 (Fig. 6B). role of barrier limiting material transport for longer distances beyond the slope base is as- sumed by vegetation growing on the valley THE SIGNIFICANCE OF EXTREME EVENTS floor and in depressions. The influence of extreme events on soil In terms of both intensity and the trans- redeposition is as yet only poorly recognized, portation of eroded material, the extreme most of all because data from experimental rainfall of June 22, 1999 had a decisive im- fields are insufficient (Boardman 2006). pact on total soil loss during the research Measurements made on plots or on whole- period, and on soil loss calculated per slope scale are limited. It often happens that unit area. Its importance for soil redepo- rainfall of maximum intensity falls not on sition on cultivated slopes is clearly visible an experimental plot but on neighbouring when compared with average annual values slopes (e.g. Niewiadomski 1968, Kostrzewski of soil loss on the slopes studied (Table 1). et al. 1989, Evans 2005). The detection and The event was responsible for a soil loss temporal and spatial understanding of heavy of from 300 kg ha-1 to 10 t ha-1 in the upper rain episodes therefore demands further re- and lower parts of the slope due to interrill search (Boardman 2006). erosion, and up to 30 t ha-1 due to rill ero- Mean and extreme values for soil erosion sion, which equaled 25% to 40% of total on cultivated slopes in selected Polish lake- soil loss during the 5-year period of meas- lands are as shown in Table 3. In the lake- urements. land belt, continentality of climate increases Slope cover studies show that maximal in an easterly direction. Despite differentia- thickness is present on the lower, concave tion of local conditions such as slope incli- segment of the slope (Smolska 2005). An nation and lithology, soil erosion is generally important role of topographic and vegeta- of lesser intensity in climates of lesser hu- tion barriers in the downward transporta- midity. Extreme rainfall and soil losses are tion of material is emphasized by Froeh- thus much rarer further to the east. lich (1982), Teisseyre (1994), Govers et al. Soil erosion in the area researched, as (1996) and Święchowicz (2002). During located in the eastern part of the lakeland the research period, the concave segments belt, amounts to 0.2–12.9 t ha-1yr-1. In the of slopes played an significant role as a top- western part (in northern Germany), mean ographic barrier. Measurements showed soil erosion on cultivated slopes is at twice that nearly all soil eroded in the upper and the rate, ranging from 0.4 up to about middle segments of slopes was deposited 20 t ha-1yr-1 or even 35.2 t ha-1yr-1 (Auerswald here (Smolska 2003). During the extreme 2006). Mean annual precipitation in north- rainfall under analysis, erosion also oc- ern Germany and north-west Poland (at curred in lower, concave, aggradational 500–800 mm) is only slightly greater than parts of slopes. Soil loss ranged from ca. in north-eastern Poland (500–750 mm),
KKsisiąążżkka1.indba1.indb 115959 22008-06-26008-06-26 110:57:170:57:17 160 Ewa Smolska
Table 3. Mean annual and maximum values for soil loss on selected lakeland areas in Poland
Rainfall Location Average Relief energy [mm yr-1] of and (max. lakeland (max.) soil Inclina- daily and study loss Elevation tion rainfall) period [t ha-1] [m] [°] Methods [mm] References
Western Pomerania 1987–89 0.2–10.3 6–12 <16 plots, Gerlach’s method 574–823 Klimczak (1993) 1994–96 1.1–19.2 6 4–5 plots 561–801 Szpikowski (1998, (80) volume of rills (130) 2002) Kostrzewski 1995–99 0.5–2. 2 volume of rills and et al. (1992) (51.6 ) deposits (28) Koćmit et al. (2006) Eastern Pomerania 1967–70 0.5–24.8 5–30 8–20 Gerlach’s method 443–739 Lankauf (1975) (50.5) volume of rills Mazury 1955–64 1.6–11.5 12 <16 plots 623 Niewiadomski, Skrodz- (50.3) volume of rills (72) ki (1964) Niewiadomski (1968) Suwałki 1987–89 0.2–12.9 7–50 <15 Gerlach’s method, 541–717 Smolska et al. (1995), 1998–99 (30.0) volume of rills (43) Smolska (2002)
but the erosivity is markedly different. The exceeding 20 mm are characteristic of the mean annual erosivity index in the western Suwałki Lakeland. Their 15-minute inten- part of the lakeland belt is 40–60 Je (Au- sity is significant; but, as the average inten- erswald 2006, Szpikowski 1998, Koćmit sity is rather low, the erosivity of such rain- et al. 2006), reaching even 100 Je in some fall is limited, and generally ranges from 6 years. Such high values have not so far been to 16 erosivity units (Je), using the USLE detected in the area studied (Banasik and formula. Such rainfall is responsible for ero- Górski 1993), where mean annual rainfall sion in the upper and middle parts of slopes, erosivity is of only 35–50 Je, with a maxi- and for the deposition of nearly all eroded mum of 75 Je (at the Suwałki and Kleszczó- soil in concave parts of slopes. Only 30 mm wek stations). This explains the short dis- rainfall with an average intensity of 0,5 mm tances of soil translocation and deposition min-1 and relatively high erosivity (>30 Je), on lower parts of slopes. give rise to erosion along the entire length of the short slopes, and to the develop- ment of rills along longer slopes (> 100 m). CONCLUSIONS One extreme event during the study period played a significant role in redistributing soil Conducted over 5 years, measurements on the cultivated slopes. A significant share of interrill and rill erosion on cultivated of the colluvium (50% to 75%) accumulated slopes reveal the dependency of the process in the concave parts of these slopes over 5 upon the intensity and efficiency of rainfall years was transferred away from the slopes events. Short storms of efficiency seldom during that storm.
KKsisiąążżkka1.indba1.indb 116060 22008-06-26008-06-26 110:57:170:57:17 Extreme Rainfalls and their Impact on Slopes—Evaluation Based on Soil Erosion Measurements 161
ACKNOWLEDGEMENTS Froehlich, W. (1982), Mechanizm transportu fluwialnego i dostawy zwietrzelin do kory- The research in the present article has been ta w górskiej zlewni fliszowej [The Mecha- carried out within the framework of the re- nism of Fluvial Transport and Waste Supply search project entitled “Extreme meteoro- into the Stream Channel in a Mountainous logical and hydrological events in Poland”, Flysch Catchment], Prace Geograficzne 143, financed by the Ministry of Science and Instytut Geografii i Przestrzennego Zago- Higher Education of Poland (PBZ-KBN- spodarowania (IGiPZ), PAN, Warszawa. 086/P04/2003) and partially within the Froehlich, W., Słupik J. (1980), Importance framework of the project “Environmental of Splash in Erosion Process within a Small and anthropogenic conditions of slopewash Flysch Catchment Basin, Studia Geomorpho- in Poland as exemplified by chosen area” logica Carpatho-Balcanica, 14: 77–112. (KBN2 PO4E 05330). Gerlach, T., (1976), Współczesny rozwój stoków w polskich Karpatach fliszowych [Present- day Slope Development in the Polish Flysch REFERENCES Carpatians], Prace Geograficzne 122, Instytut Geografii i Przestrzennego Zagospodarowa- Auerswald, K. (2006), Germany, in: Boardman, nia (IGiPZ), PAN, Warszawa. J. and Poesen, J. (eds.), Soil Erosion in Eu- Gil. E. (1999), Obieg wody i spłukiwanie na fli- rope, Wiley and Sons, 213–230. szowych stokach użytkowanych rolniczo w la- Banasik, K. and Górski, D. (1993), Evaluation tach 1980–1990 [Water Circulation and Wash of rainfall erosivity for east Poland, in: Bana- Down on Flysch slopes used for Farming Pur- sik, K. and Żbikowski, A. (eds.), Runoff and poses in 1980–1990], Zeszyty IGiPZ PAN 60, sediment yield modelling (RSY-93), SGGW Warszawa. Warszawa, 129–134. Gil, E., Słupik, J. (1972), The Influence of Plant Ber, A. (2000), Plejstocen Polski północno-wschod- Cover and Land Use on the Surface Runoff niej w nawiązaniu do głębszego podłoża i ob- and Wash Down During Heavy Rain. Studia szarów sąsiednich [The Pleistocene of Northe- Geomorphologica Carpatho-Balcanica, 6: astern Poland and Neighbouring Areas Against 181–190. Crystalline and Sedimentary Basement], Prace Govers, G., Quine, T.A., Desmet, P,J.J., Walling Państwowego Instytutu Geologicznego, 170. D.E. (1996), The relative contribution of soil Boardman, J. (2006), Soil Erosion Science: Re- tillage and overland flow erosion to soil re- flections on the Limitations of Current Ap- distribution on agricultural land, Earth Sur- proaches, Catena, 68, 73–86. face Processes and Landforms, 21: 929–946. Bryan, R.B. (2000), Soil Erodibility and Proces- Józefaciuk, Cz. (1991), Procesy spłukiwania i ero- ses of Water Erosion on Hillslope, Geomor- zji wąwozowej [Soil Erosion and Gully Ero- phology, 32: 385–415. sion], in Starkel, L., (ed.), Geografia Polski. Carson, M.A., Kirbky, M.J. (1972), Hillslope Form Środowisko przyrodnicze. Wydawnictwo Na- and Process, University Press, Cambridge. ukowe PWN, Warszawa, 420–425. Chudecki, Z., Niedźwiecki, E. (1983), Nasilanie Klimczak, R. (1988), Metoda obliczania objęto- się erozji wodnej na obszarach słabo urzeź- ści form wklęsłych i jej zastosowanie w geo- bionych Pomorza Zachodniego [Intensifica- logii dynamicznej [Calculation Method of the tion of Water Erosion on Poorly Rolled Areas Concave Form Volume and its Application to of Western Pomerania], Zeszyty Problemowe Dynamic Geology], Czasopismo Geograficz- Postępu Nauk Rolniczych 272, Państwowe ne, 59 (2): 201–208. Wydawnictwo Naukowe, Warszawa: 7–18. Klimczak, R. (1993), Spłukiwanie na obszarach Evans, R. (2005), Monitoring Water Erosion in o zróżnicowanym użytkowaniu – przebieg Lowland England and Wales—A Personal i rola we współczesnym środowisku morfo- View of its History and Outcomes, Catena, genetycznym (zlewnia Młyńskiego Potoku, 64: 142–161. Pomorze Zachodnie) [Slope wash in ares with
KKsisiąążżkka1.indba1.indb 116161 22008-06-26008-06-26 110:57:170:57:17 162 Ewa Smolska
diversified land-use patterns—their role in Selby, M.J. (1982), Hillslope Materials and Proc- a contemporary morphogenetic environment esses, Oxford University Press, Oxford. (The Młyński Potok Catchment, West Po- Słupik, J. (1973), Zróżnicowanie spływu powierzch- merania)], Zeszyty Naukowe PAN „Człowiek niowego na fliszowych stokach górskich [Dif- i środowisko” 6: 61–77. ferentiation of the Surface Runoff on Flysch Koćmit, A., Podlasiński, M., Roy, M., Tomasze- Mountain Slopes], Dokumentacja Geograficz- wicz, T., Chudecka, J. (2006), Water Erosion na, 2, Instytut Geografii i Przestrzennego Za- in the Catchment Basin of the Jeleni Brook, gospodarowania (IGiPZ), PAN, Warszawa. Journal of Water and Land Development, 10: Smolska, E. (2002), The Intensity of Soil Erosion 121–131. in Agricultural Areas in North-Eastern Po- Kostrzewski, A., Klimczak, R., Stach, A., Zwoliń- land. Landform Analysis, 3: 25–33. ski, Z. (1989), Morphologic effects of heavy Smolska E. (2005), Znaczenie spłukiwania w mo- rainfall (24 May 1983) over relief features delowaniu stoków młodoglacjalnych [Slope of scarpland in the middle Parsęta valley, Wash Processes in Late Glacial Area on Ex- West Pomerania, Poland, Quaestiones Geo- ample of Suwałki Lake District], Wyd. WGSR graphicae, Special Issue 2: 101–110. UW, Warszawa, (English summary): 1–146 Niewiadomski, W., Skrodzki, M. (1964), Nasile- Smolska, E. (1993), Rola spłukiwania w dosta- nie spływów i zmywów a system rolniczego wie materiału do transportu fluwialnego zagospodarowania stoku [Intensification w obszarze młodoglacjalnym (na przykładzie of Runoff and Soil Losses on Agricultural górnej Szeszupy) [The Role of Wash in Su- Slopes], Zeszyty Naukowe WSR Olsztyn, 17 pply of Material in Fluvial Transport on the (2): 269–291. Young Glacial Area (Examplified by the Sze- Niewiadomski, W. (1968), Badania nad erozją szupa Waterbasin)], Zeszyty Naukowe PAN gleb na północy Polski (okres 1950–1976) [Re- ‘Człowiek i środowisko’ 6: 159–165. search on Soil Erosion in North Poland], in Smolska, E., Mazurek, Z., Wojcik, J. (1995), Dy- Ziemnicki, S. (ed.), Procesy erozyjne i problem namika procesów geomorfologicznych na sto- ochrony gleby w Polsce, Wyższa Szkoła Rolni- ku pojeziernym jak czynnik środowiskotwór- cza (WSR) Lublin i Państwowe Wydawnictwo czy [Dynamics of Geomorfological Processes Rolnicze i Leśne (PWRiL) Warszawa: 29–49. on Slopes as a Factor of Habitat Formation Niewiarowski, W., Celmer, T., Marciniak, K., Pie- in Lake Landscape], Zeszyty Naukowe PAN trucień, C., Proszek, P., Sinkiewicz M. (1992), ’Człowiek i środowisko’ 12: 205–220. Przebieg współczesnych procesów denudacyj- Stopa-Boryczka, M., Martyn D. (1985), Klimat nych na młodoglacjalnej wysoczyźnie more- [Climate], in: Województwo Suwalskie – studia nowej intensywnie użytkowanej rolniczo, na i materiały, 1, Oddział Badań Naukowych Bia- przykładzie okolic Koniczynki, na północny łystok and IGiPZ PAN, Warszawa, 81–118. wschód od Torunia [Contemporary Denuda- Szpikowski J., (2002), Contemporary Processes tion on Young Glacial (Vistulian) Moraine Pla- of Soil Erosion and the Transformation of the teau Subject to Intensive Farming as Exempli- Morphology of Slopes in Agricultural Use in fied by a Study of the Environs of Koniczynka, the Postglacial Catchment of the Chwalimski NE of Toruń], Prace Geograficzne, 155: 47–67, Potok (upper Parseta, Drawskie Lakeland), Instytut Geografii i Przestrzennego Zagospo- Quaestiones Geographicae 22, 79–90. darowania (IGiPZ), PAN, Warszawa. Święchowicz, J. (2000), The Influence of the Parsons, A.J., Stone P.M. (2006), Effects of In- Plant Cover and Land Use on Slope-Channel tra-Storm Variations in Rainfall Intensity on Decoupling in Foothill Catchment. A Case Interrill Runoff and Erosion, Catena, 67: Study from the Carpatian Foothill, Southern 68–78. Poland, Earth Surface Processes and Land- Parsons, A., Stromberg, S.G., Greener, M. (1998), forms, 27: 463–479. Sediment Transport Competence of Rain-Im- Teisseyre, A. K. (1994), Spływ stokowy i współ- pact Overland Flow, Earth Surface Processes czesne osady deluwialne w lessowym rejonie and Landforms, 23: 365–375. Henrykowa na Dolnym Śląsku [Overland
KKsisiąążżkka1.indba1.indb 116262 22008-06-26008-06-26 110:57:180:57:18 Extreme Rainfalls and their Impact on Slopes—Evaluation Based on Soil Erosion Measurements 163
Flow and Present-Day Deluvial Deposits in Uggla, H., Solarski, H., Rytelewski, J., Mirow- the Loessial Area of Henryków, Lower Sile- ski, Z., Nożyński, A., Grabarczyk, S., (1998), sia, Southern Poland], Acta Universitas Wrati- Problematyka erozji wodnej gleb północno- slaviensis, Prace Geologiczno-Mineralogiczne wschodniej Polski [Soil Erosion in Northea- 63. stern Poland]. Bibliotheca Fragmenta Agro- Uggla, H., Mirowski, Z., Garbarczyk, S., Nożyń- nomica, 4B: 179–197. ski, A., Rytelewski, J., Solarski, H. (1968), Procesy erozji wodnej w terenach pagórkowa- Paper first received: March 2007 tych północno-wschodniej części Polski [Wa- In final form: November 2007 ter Erosion Processes in Hilly Area in North- eastern Poland], Roczniki Gleboznawcze, 28 (2): 415–447.
KKsisiąążżkka1.indba1.indb 116363 22008-06-26008-06-26 110:57:180:57:18 MODELS OF IMPACTS OF HYDROMETEOROLOGICAL EXTREMES
ZBIGNIEW W. KUNDZEWICZ, ROMAN MAŃCZAK, IWONA PIŃSKWAR and MACIEJ RADZIEJEWSKI Research Centre for Agricultural and Forest Environment, Polish Academy of Sciences, Bukowska 19, 60-809 Poznań, Poland E-mails: [email protected]; [email protected]; [email protected]; [email protected]
Abstract: Mathematical modelling of hydrometeorological extremes and their impacts was dis- cussed. An introduction to the notion of modelling is proposed. Annual extremes in temperature records were examined, also on the basis of qualitative indices, when quantitative data are not available due to cost restrictions. Trends in long time series of records were studied. Concepts as regards projections of risk of extreme events (such as floods) are also discussed; including syn- thesis of projection models for a range of climate impacts. The facets of uncertainty are also dealt with. A conceptualization of the risk in the load-resistance framework is proposed.
Key words: mathematical modelling; climate impact; hydrometeorological extremes; curve fitting; trend.
INTRODUCTION TO MODELLING mathematical constructs to describe featu- res of systems or processes. Virtually every The term “modelling” is typically interpre- use of a mathematical equation to represent ted as the replacing of a (possibly very com- links between variables, or to mimic a tem- plex) real object of interest by a simpler, and/ poral or spatial structure of variable(s), can or more tractable, model, in order that in- be called mathematical modelling. formation about the object might be drawn One can distinguish a number of stages by examining the model. A model is a wor- to the modelling process, beginning with king analogy of the real object that imitates the selection of the model structure (i.e. (mimics) selected aspects thereof which are mathematical equations governing links deemed important to the study at hand, whi- between variables describing the object) for le omitting aspects deemed non-essential. the problem at hand, and progressing via However, since the model can be regarded the development of an efficient algorithm, as similar, but not identical, to the object, it writing of a code, and testing. If a well-te- may give a distorted view, and lead to false sted model is available, it is necessary to conclusions. identify values of model parameters (para- The mathematical modelling this paper meters of mathematical equations) for the deals with can be understood as the use of system being modelled. This can be difficult
KKsisiąążżkka1.indba1.indb 116565 22008-06-26008-06-26 110:57:180:57:18 166 Zbigniew W. Kundzewicz, Roman Mańczak, Iwona Pińskwar and Maciej Radziejewski
(a mathematically ill-posed, inverse prob- ries model) and nonlinear models, such as lem, where by a small inaccuracy in input artificial neural networks (ANNs) and fuzzy data may lead to a large parameter identi- sets; capable of representing a wide range fication error). If a large number of para- of complex relationships. However, results meters need to be identified, problems with of black-box models are valid only under over-parameterization may occur, hence the conditions for which they have been va- parsimony in parameters is important. Sim- lidated. The fuzzy sets approach provides pler, approximate models have fewer para- a framework for the interpretation of any meters than more complete models. Once existing theoretical knowledge (also expert model parameters have been identified, the knowledge elicitation) and adding to it a le- stage of validation is proceeded to. This im- arning process, in which way the approach plies the checking of how the selected model becomes less of a black-box. with identified parameter values performs There may be a non-uniqueness, in the on independent data. Once the model has sense that several models that represent the been verified, it can be used (for simulation, system of interest may exist, being different hindcasting, or forecasting). If the valida- in model structures. tion fails and model performance is rated There are a number of factors influen- non-satisfactory, another modelling appro- cing the process of model selection, such as: ach (e.g. one based on a different structure) • the general modelling objective; is necessary. • representation of the most relevant Typically, there is a trade-off between mo- processes (e.g. for operational forecasting, del accuracy and complexity. Simple models a model should contain an updating compo- produce results which are usually less accura- nent, reacting to the so-far forecast errors, te than those from the complex (hence more which are likely to be correlated, hence there costly) and physically-based models (provided is potentially useful information in the time that appropriate data necessary to run such series of forecast errors that can be tapped); models are available). However, at times, sim- • the type and size of the system to be ple models may produce results of sufficient modelled; accuracy and this is a clue that a simpler mo- • the variables to be modelled; del can be chosen in this particular case. • the characteristics of the system There are many criteria for model clas- (e.g. climate, physiography, land use); sification; one of these being the degree of • data (availability, type, length of physical justification. Here one can distin- record and quality/accuracy); guish at least three classes: • the possibility of transposing model • mechanistic, process-based, techniqu- parameters for ungauged systems, i.e. trans- es involving (possibly complex) systems of posing from a smaller object to a larger obje- equations expressing rigorous physical laws ct; or from one set of conditions to another; and theoretical concepts; • model availability (whether free, i.e. in • conceptual models consisting of simple the public domain, or at a cost; user-friendli- elements, which may represent, in an appro- ness, available know-how); ximate way, processes occurring in the basin. • required model simplicity (ease of ap- [Conceptual models may have physically-ba- plication to match the available manpower); sed parameters (cf. Kundzewicz, 1986)]; • the availability and power of compu- • the ‘systems’ approach capable of be- ters—a criterion for both model develop- ing understood as purely empirical, black- ment and operation that has largely lost its box techniques, which match the input and importance with the advent of more and output signals of the system, without mimi- more powerful and inexpensive personal cking the internal structure. computers, (however, exceptions continue to This last class of models includes linear exist, e. g. the „curse of dimensionality” still models (e.g. integral operator, or a time-se- making it difficult to optimize the opera-
KKsisiąążżkka1.indba1.indb 116666 22008-06-26008-06-26 110:57:180:57:18 Models of Impacts of Hydrometeorological Extremes 167
tion of multiple water storage reservoirs, by available, the data are better approximated dynamic programming). by an exponential function (Fig. 1b):
GT ~ a + b exp(c year) (2) 1. CLIMATE EXTREMES AND THEIR IMPACTS Indeed, the fitting is slightly improved It is natural that geophysical variables (and (residual error 0.1243). Another approach, meteorological and hydrological variables the broken-line regression, is given by the therein) are subject to strong variability, at formula: times attaining extreme values. Some of the- se extreme values, such as high wind speeds, GT ~ a + b year + c |year–d| (3) extremely high (or low) temperatures, extre- mely high precipitation, and extremely high where: (or low) river discharge, may lead to substan- d is the break-point year. tial human and material damage. For this If d is fixed and the remaining parame- reason, the modelling of extremes and their ters fitted, the residual standard error as impacts is of considerable importance to a function of d is obtained (Fig. 1c). This fun- a better understanding of the risk past and ction has a global minimum, corresponding future events may pose. to the fitted model, and another, local, mi- The series of global temperature anoma- nimum, only slightly higher. Such behaviour lies available since 1850 is known to exhibit may indicate that the model is not well suited a statistically significant increase. Here, ex- for this data, but the model performance (in tremes can be understood as very warm or terms of standard error) is similar to that of very cold years, with mean annual tempera- the previous two models, with the residual ture departing greatly from the long-term error of 0.1265 (Fig. 1d). central value (e.g. mean or median). The The models presented above are very shape of the visible trend is clearly non-li- different in their characteristics, especial- near: a notable increase in recent decades ly in the rate of change outside the obser- is to be observed, compared with little (if ved interval (bounded change, exponential any) change in the earlier decades. Nume- change, linear change, oscillations only). rous models that reflect such behaviour may This augments the well known fact that be fitted to the available global temperature a model based on observed data alone is use- data. Here we make no claims as to the vali- less for long-term forecasting. It also shows dity of the fitted models, and do not consider that discriminating between different mo- the possibility of long-term correlations in dels may be difficult or impossible when the the residuals, treating them as independent sole basis is exploration of the data, rather Gaussian variables. The series of global tem- than a good understanding of the underlying perature anomalies is denoted GT below. system. While a phenomenon remains po- Fig. 1a shows the results of a least-squares orly understood, simple, robust conclusions fitting of the logistic curve with the restric- (e.g. the existence of an upward trend) may tion that the saddle-point be put on the last still be drawn from the characteristics sha- year with available data: red by different matching models, altho- ugh the significance associated with such GT ~ a + b/(1 + exp(c(year—2006))) (1) findings must be studied using appropriate methods allowing for long-term correlations The residual error is 0.1263. When the in the data. restriction on the saddle-point is relaxed, Long time series (1954–2005) of ave- closer fits are obtained when the saddle po- rage values of annual mean temperatures int is placed in a distant future. This suggests at 16 stations in Poland (data from several that, for the time interval for which data are Polish statistical yearbooks published by
KKsisiąążżkka1.indba1.indb 116767 22008-06-26008-06-26 110:57:190:57:19 168 Zbigniew W. Kundzewicz, Roman Mańczak, Iwona Pińskwar and Maciej Radziejewski
Figure 1. Least-squares fitting of different functions to global temperature data. Points: global temperature anomaly. Line: fitted function. (a) Logistic function; (b) Exponential function; (c) Residual standard error of a fitted broken line model as a function of the break-point year; (d) Broken line. Points: global temperature anomaly. Solid line: fitted model. Dashed line: model corresponding to local minimum.
GUS—Main Statistical Office), show a clear only, it is tempting to use free (qualitative) upward trend (cf. Fig. 2). Table 1, presenting data on the thermal classification of months, the 10 coldest and 10 warmest years in the prepared and made available on the internet 52-year period of records, demonstrates that by Professor Halina Lorenc (2007) on the many entries in the list of warmest years oc- website of the Polish hydrometeorological curred recently (six of the ten entries have service (Institute of Meteorology and Wa- been since the 1990s and only one before ter Management). The series of information 1970). Only one entry in the list of 11 coldest on ranges of monthly temperature for War- years occurred after 1990 (in 1996) and six saw spans the 36-year period (1971–2006). entries before 1970. For every month, the mean temperature is Since daily and monthly numerical data classified to one of 11 categories (extreme- are available at a (rather substantial) cost ly cold, anomalously cold, very cold, cold,
KKsisiąążżkka1.indba1.indb 116868 22008-06-26008-06-26 110:57:200:57:20 Models of Impacts of Hydrometeorological Extremes 169
slightly cold, normal, slightly warm, warm, that the mass of the pdf shifts in time towar- very warm, anomalously warm and extreme- ds higher temperatures. However, the cold ly warm). The range marked „normal” spans extremes do not tend to disappear. Actually, the interval (m-0.5sd, m+0.5sd) where the peak of pdf flattens, and the variability m and sd represent mean and standard de- grows with time. viation, respectively. Hence, the width of this As can be seen in Fig. 4, the number of interval is 1 sd, while the width of each of the warm monthly extremes has been increasing 10 remaining categories is 0.5 sd. with time. For example, the interpretation The 36-year period of available monthly for 2006 reads: four warm extremes (July, data (1971–2006) has been divided into four September, October, December) and two consecutive nine-year sub-periods. Fig. 3 cold extremes (January, March). The num- compares qualitative monthly temperature ber of cold and warm extremes for monthly indices for these sub-periods. It is evident data are shown in Table 2. 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005
Figure 2. Average values of mean annual temperatures (in °C) at 16 stations in Poland (data source: GUS); y = 0.0235x + 6.7438, R2 = 0.1941. The dashed line represents the long-term mean.
Table 1. List of the 10 coldest and 10 warmest years in Poland in the period 1954–2005 (mean for 16 stations).
Ten coldest years in 1954–2005 Ten warmest years in 1954–2005 (mean for 16 stations) (mean for 16 stations)
Mean annual temperature Mean annual temperature Year [°C] Year [°C]
1956 5.66 2000 9.04 1980 5.96 1989 8.72 1987 6.07 2002 8.68 1985 6.15 1990 8.55 1996 6.15 1994 8.49 1965 6.24 1999 8.43 1969 6.42 1983 8.35 1962 6.43 1992 8.34 1963 6.44 1967 8.17 1954, 1976 6.53 1975 8.16
Data source: GUS.
KKsisiąążżkka1.indba1.indb 116969 22008-06-26008-06-26 110:57:220:57:22 170 Zbigniew W. Kundzewicz, Roman Mańczak, Iwona Pińskwar and Maciej Radziejewski
1971-1979 1980-1988
0 10 20 30 40 50 60 0 10 20 30 40 50 60 1989-1997 1998-2006
0 10 20 30 40 50 60 0 10 20 30 40 50 60
Figure 3. Comparison of qualitative temperature indices for Warsaw (Okęcie) in four nine-year periods (1971–1979, 1980–1988, 1989–1997, 1998–2006). Number of months in a particular temperature class is given on the x-axis, qualitative classes given on the y-axis. Data source: Lorenc (2007).
1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005
Figure 4. Cold and warm extremes in the monthly temperature data for Warsaw, 1971–2006. Notation: green denotes non-extreme months (categories: cold, slightly cold, normal, slightly warm, warm), orange—warm extremes (very warm, anomalously warm, extremely warm) and blue—cold extremes (very cold, anomalously cold, extremely cold). Data source: Lorenc (2007).
KKsisiąążżkka1.indba1.indb 117070 22008-06-26008-06-26 110:57:220:57:22 Models of Impacts of Hydrometeorological Extremes 171
Table 2. Number of cold and warm extremes for Fig. 5b presents changes in standard devia- monthly temperature data from Warsaw in four tions. The interpretation of these changes is: nine-year sub-periods. Interpretation of cold and towards warming and increasing variability. warm extremes—see caption to Fig. 4. Quantitative classes were used to represent the categorical, qualitative data. Number of cold Number of hot It is difficult to determine the frequency Sub-period extremes extremes of meteorological or hydrological extremes, 1 (1971–1979) 4 4 especially for events of a high return pe- 2 (1980–1988) 10 2 riod. Under conditions of stationarity, it can 3 (1989–1997) 8 7 be assumed that a 100-year record of high 4 (1998–2006) 6 15 quality data should be available in order to determine, in a rigorous and reliable way, Data source: Lorenc (2007). the magnitude of a 10-year event. However, attempts to draw conclusions based on very Changes in the distribution of quan- rare events (e.g. a 1,000-year event) are of- titative temperature data from Warsaw ten made on the basis of a meagre data re- (cf. Figs 3–4 and Table 2) are illustrated in cord (e.g. of 30 years or less, with possible Fig. 5. Fig. 5a shows changes in means, while gaps and missing values, and questionable
8
7.5
7
6.5
6
5.5
5 y = 0.0253x + 5.6142 R2 = 0.1847 4.5 mean 4 1 3 5 7 9 11131517192123252729313335
3
2.5
2
1.5
1 y = 0.0181x + 1.2141 R2 = 0.2175 0.5
std
0 1 3 5 7 9 11131517192123252729313335
Figure 5. Changing annual temperature distributions in Warsaw; (a) change in means—towards warming; (b) change in standard deviations—towards increasing variability. Time (in years after 1970 onwards is represented in the x-axis; i.e. x=1 means year 1971). Data source: Lorenc (2007).
KKsisiąążżkka1.indba1.indb 117171 22008-06-26008-06-26 110:57:220:57:22 172 Zbigniew W. Kundzewicz, Roman Mańczak, Iwona Pińskwar and Maciej Radziejewski
quality, especially in the range of extremes, fication is needed. Fig. 6 illustrates a simpli- where error may be very high). Moreover, fied structure of interconnected components the stationarity assumption is not likely to of the global super-system. Only first-order hold. impacts are noted in the simplified scheme, There is an obvious tautology that rare, but in fact, Fig. 6 could contain many more truly extreme events do not occur frequen- arrows linking individual components, in tly. Hence, there is an endemic small-sam- both directions. One can postulate math- ple syndrome—even a long time series may ematical forms of relationship between com- contain very few (or none) really destructive ponents. extreme events. Therefore, recourse to exte- Data scarcity is a commonplace. If rior information is indispensable. A single data are collected, they typically refer to event (the only extreme on record), such as a small number of variables, e.g. precipita- the summer 1997 flood on the Odra, the Vi- tion and water stage, while understanding of stula, and their upper tributaries—cannot the flood phenomena and their impacts in guide the development of a rigorous dama- a holistic context would require collection of ge-discharge relationship. Even estimation a comprehensive dataset embracing long of the return period of this flood has become time series of environmental and socio- a subject of national debate (Kundzewicz et economic variables (cf. Kundzewicz and al., 1999). Schellnhuber, 2004). Even in the developed Analysis of climate extremes, such as world, such data are not readily available. temperature, is less difficult than study of the impacts of climate extremes. This is so for several reasons, such as: the complex 2. PROJECTIONS FOR THE FUTURE structure of the climate-impact system and the lack of necessary data. In the global sys- The convenient assumption of stationarity tem, everything is connected with everything does not hold in a changing world. Many else, so that a pragmatic decision of simpli- aspects of the system change over time,
Figure 6. Simplified structure of the global super-system.
KKsisiąążżkka1.indba1.indb 117272 22008-06-26008-06-26 110:57:220:57:22 Models of Impacts of Hydrometeorological Extremes 173
either for natural reasons (such as changes
in solar activity, Earth orbit, volcanic activi- Pt+T (Q ≥ QI) ≠ Pt (Q ≥ QI) (4) ty) or anthropogenic reasons (e.g. land use, land cover). There is increasingly strong evi- where:
dence to support the hypothesis that recent Pt is the probability in the time instant t; climate changes are largely of anthropogenic Q is the river discharge;
origin. We are indeed observing an induced QI is the minimum river discharge causing global warming caused by the enhanced gre- flood damage. enhouse effect, and resulting from increa- Projection for a long time horizon, e.g. sing emissions of greenhouse gases into the T = 50 years allows the flood preparedness atmosphere (burning fossil fuels and metha- system to be adapted, including by imple- ne emission from agriculture and animal menting structural measures (reservoirs, husbandry), as well as a reduction of carbon polders, dikes and relief channels). sequestration via land-use change (e.g. de- In order to understand a system, it is forestation and urbanization). Hence, the important to observe it, by measuring the properties of the system observed in the past relevant variables over a long time and to may change considerably in the future. carry out interpretations of collected data Human societies adapt to the changing and search for trends. Yet, in order to proje- environment—climate changes trigger the ct the reaction of the system to an arbitrary need for a (possibly costly) change in adap- forcing (e.g. a future forcing corresponding tation. One needs models to project the fu- to a changed climate), one needs models. ture situation in the changing world, in order Mathematical models (calibrated for past- to plan appropriate adaptation measures, to-present conditions) are the principal way whose implementation may take a long time. to study future situations. Although purely In the case of structural water-engineering empirical and black-box relationships con- measures, the adaptation (from concept to tinue to prove beneficial under certain cir- the start of operations) may take several de- cumstances in global change studies (with cades. validity typically limited to the range of si- Projections for the near future, called tuations encompassed in the validation pro- forecasts, are among the routine activities cess), they may be subject to serious errors of meteorological and hydrological services. when applied under conditions not previo- For instance, forecasting high river discharge usly experienced. Hence, physically-based and stage allows for the issuance of warnings models, founded on theoretical background, early enough for a response (e.g. evacuation) and conceptual models, whose parameters to be mounted. An early flood warning is in- have a quasi-physical sense are expected to formation that water is likely to rise (a flood be more trustworthy under such conditions. is likely to occur) in a defined time point in They are based on a more plausible assump- the near future. Hence, the forecast of river tion that the physical laws are not changing,
discharge, Qforecast(t + T | t) is made in the i.e. the present laws would also hold in the current time instant, t, for the forecast hori- future. zon (small time scale), T, i. e. up to the time The process of modelling of the past and instant t + T. Loss reduction depends on the present behaviour of systems of interest is forecast horizon (T) and accuracy (error in different from modelling for future climate. amplitude and timing). In the former, validation is an indispensable In the present paper, a projection is un- phase, while in the latter, it is not possible. In derstood in a long time scale. For example, the classical approach, the available data re- in reference to floods, it could be called cord is split into two parts (the split-sample a flood risk warning; indicating that the approach). One is used for the identification flood risk is likely to change (typically rise) of parameters and the other—for valida- on the long time scale T. That is: tion. They check how the model performs on
KKsisiąążżkka1.indba1.indb 117373 22008-06-26008-06-26 110:57:230:57:23 174 Zbigniew W. Kundzewicz, Roman Mańczak, Iwona Pińskwar and Maciej Radziejewski
independent observation material. In future approach can be applied to ungauged basins, projection studies, direct validation is not whose physical characteristics can be deter- possible. One can assume that a (physically- mined (at least approximately). This holds based) model that worked well for past data for prediction (e.g. flood frequency analysis– is likely to perform satisfactorily for future regionalization of annual maximum flood, conditions. In particular, if a model can re- and then determination of floods with the construct the past, it is expected to hold pro- return period of interest, such as a 100-year
mise for future projections. A known input flood, Q100, etc): signal (past observation data) and known
output signal (past observation data) ser- Q100 = f(p1, t) (5) ve in the determination and identification of parameters of the climate impact model where:
(e.g. from climate, via water, extreme indi- p1 is a vector of parameters. ces to environmental and socio-economic The projecting of future flood damage, impacts). flood risk, and vulnerability requires evalu- If there are no observation data (e.g. be- ation of future load (river flow / stage / inun- cause ungauged areas or future conditions dation area) and damage potential (Fig. 7). are under study), use of methods not requi- ring the availability of a long time series is 2.1 SYNTHESIS OF PROJECTION MODELS necessitated. If among many similar and As stated in Stern (2007), physical and adjacent areas, some are gauged, and others biological principles indicate that impacts in are not, one can try to establish regionally many sectors will become disproportionate- valid laws. Models for gauged catchments ly more severe with rising temperatures, but can be developed and their parameters lin- there is little empirical support for specific ked to physical characteristics (e.g. by linear quantitative relationships, which remain rat- regression). Once this is done, the regional her speculative (largely based on common
Figure 7. A conceptual sketch linking factors controlling future flood damage.
KKsisiąążżkka1.indba1.indb 117474 22008-06-26008-06-26 110:57:230:57:23 Models of Impacts of Hydrometeorological Extremes 175
sense or expert judgment). Hitz and Smith regions may already have passed the peak. (2004) reviewed results of published studies The shape and location of the “hill” curve examining the nature of the relationship be- depends on the crop. However, regionally, tween global impacts and the amplitude of e.g. in Poland, temperature rise itself is not global warming. They found increasingly the only important factor—impacts on agri- adverse impacts for several climate-sensiti- cultural production systems will depend on ve sectors, though others gave no consistent water availability (soil moisture, as result relationships with temperature. A few illu- of precipitation), which may become more strating examples are given below, for orien- scarce in the growing season. tation. Available data show that the link be- The Clausius-Clapeyron equation shows tween the heat-related human mortality that the water holding capacity of the atmo- and temperature is U-shaped (the bath- sphere increases exponentially with tempe- tub pattern). A sharp increase in mortality rature, in line with the equation: can be observed once human temperature tolerances are exceeded, both on the side 2 des(T) / es(T) = L dT / (R T ) (6) of the minimum and maximum thresholds (cold spells and heat waves). Rising tem- where: peratures cause a decrease in cold-related
es(T) is the saturation vapor pressure at tem- mortality and increase in heat-related mor- perature T, tality. In some temperate countries, such as L the latent heat of vaporization, Poland, cold spells are still the weather ex- R the gas constant. tremes causing the most fatalities, exceed- When linearizing the above equation for ing 200 during colder winters in the (warm) the present conditions, one may find that ev- 2000s. ery 1°C increase in temperature corresponds Storm damage is over-proportionally to a 6–7% increase in the mean water holding related to wind speed and a cubic relation capacity of the atmosphere. This means that (infrastructure damage increases as a cube the water cycle will intensify, leading to more of wind-speed) is often postulated. Mills severe intense precipitation (and, in conse- et al. (2001) proposed a quadratic relation- quence, floods). Mills et al. (2001) estimated ship whereby a doubling of wind speed is that a 25% increase in 30 min. precipitation translated into a four-fold increase in da- may reduce the flooding return period from mage. Grace (2007) cited Insurance Au- 100-year to 17-year. There will be more en- stralia Group’s experience, showing that an ergy to drive storms and hurricanes. Accord- increase in peak wind gust strength from ing to an approximate assessment reported 40–50 to 50–60 knots can generate a 6.5-fold in Mills et al. (2001), a 2.2°C mean tempe- increase in building claims. rature increase leads to a 5–10% increase Hitz and Smith (2004) report on para- in hurricane wind speed, while a 1°C mean bolic relations found for temperature rising summer temperature increase may corre- impacts on sectoral damage in agriculture, spond to a 17–28% increase in wildfires. terrestrial ecosystems, productivity, and The agricultural production systems fol- forestry. Increasingly adverse effects are low the inverse parabolic (“hill function”) expected in coastal and marine ecosystems, relation with temperature. In cooler regions, biodiversity as well as health. A parabolic re- low levels of warming may improve condi- lation with temperature was postulated for tions for crop growth (lengthening the grow- additional numbers of people in the coastal ing season and increasing the area under ag- hazard zone (e.g. living below the 1000-year ricultural production), but further warming storm surge elevation), under the assump- will have increasingly negative impacts as tion that adaptation is based on observed critical temperature thresholds are crossed practice. However, by adaptation one can more often. It is likely that the tropical considerably improve the situation.
KKsisiąążżkka1.indba1.indb 117575 22008-06-26008-06-26 110:57:230:57:23 176 Zbigniew W. Kundzewicz, Roman Mańczak, Iwona Pińskwar and Maciej Radziejewski
The insurance industry requires capital ter uncertainty, and uncertainty involved in for an extreme year (now ca $120bn). Cli- the input data (due to measurement errors, mate change is likely to shift the distribution a lack of representativeness of the measu- towards higher values. Extreme losses are rement site, or problems in aggregating or likely to increase more than average losses, disaggregating data, in order to cover areas cf. Stern (2007). Storm intensity growth by of concern). Thorough sensitivity studies
6% (for doubling CO2 and 3°C warming) help in assessing uncertainty. may increase the need for insurance capital Even if our understanding of uncertain- in Japan and the US by 80–90%. ties and their interpretation has improved, and new methods (e.g. ensemble-based 2.2 UNCERTAINTY approaches) are being developed for their Future projections cannot be deterministic. characterization, quantitative projections Indeed, they are only possible in a statistical of changes in precipitation, river flows and sense, loaded by strong uncertainty. Uncer- water levels at the river-basin scale remain tainty of projections cannot be eliminated. uncertain (Kundzewicz et al., 2007). Among sources are different possible sce- Uncertainties to climate change projec- narios for the future (socio-economic devel- tions increase with the length of the time ho- opment and new technologies; emissions of rizon. In the near-term (e.g. 2020s), climate greenhouse gases and land use, adaptation model uncertainties play the most important and mitigation). Uncertainty in regard to role, while over longer time horizons, un- the models is also marked; there is a scar- certainties due to the selection of emission city of reliable, rigorously tested models. scenarios become increasingly significant Model uncertainty (in structure, parameters (Jenkins and Lowe, 2003). and data) results from the simplification of Most GCMs have difficulty in producing land-surface processes, and mismatches in consistent precipitation simulations, while spatial and temporal scales, between the cli- temperature simulations are well correlated mate model (large grid cells and daily data with observations. These uncertainties, in made commonly available) and the hydro- turn, induce biases in the simulation of river logical (catchment and event) scale. Hence, flows when using direct GCM-output repre- disaggregation of information from climate sentative of a current time-horizon. models is needed. Since different models For precipitation changes to the end of produce largely varying results, one may the 21st century, the multi-model ensemble consider an ensemble of model-based pro- mean exceeds the inter-model standard de- jections, allowing one to grasp the range of viation at high latitudes only. Over several possible futures. regions, e.g. in Poland, models disagree on Future input signals to hydrological mo- the sign of the seasonal precipitation change dels (climatical input—temperature and pre- (forecasting decreased or increased future cipitation) differ very much for various cli- summer precipitation). mate models and various assumptions about Uncertainties in climate change im- future socio-economic developments dri- pacts on water resources are mainly due to ving greenhouse-gas emissions. It is not un- the uncertainty in precipitation inputs, only common that different climate scenarios or to a lesser extent uncertainties over gre- models produce projections differing in the enhouse gas emissions (Döll et al., 2003), sign of changes for a given variable, area and climate sensitivities (Prudhomme et al., time horizon. Projections may range from 2003), or hydrological models themselves a substantial increase according to one mo- (Kaspar, 2003). Comparison of different del to a substantial decrease for another. This sources of uncertainty in flood statistics strong uncertainty augments the “traditional in two UK catchments (Kay et al., 2006) uncertainty” in hydrological models, such as led to the conclusion that GCM structure model structure uncertainty, model parame- is the largest source of uncertainty, larger
KKsisiąążżkka1.indba1.indb 117676 22008-06-26008-06-26 110:57:240:57:24 Models of Impacts of Hydrometeorological Extremes 177
than emission scenarios and hydrological In ungauged basins, where precipitation or modelling. river flow, or both, are not measured, the It is difficult to evaluate the credibility models have to be developed without access of individual scenarios. Multi-model proba- to a long time series of gauge records. Yet, bilistic approaches are preferable to using urgent practical problems need to be solved the output of one climate model only, when in both gauged and ungauged basins. This assessing uncertainty in the climate -change drives the search for universal hydrological impacts on water resources. laws, utilisable in ungauged basins. The large range for different climate mo- It is not necessary to measure the mass, del-based scenarios suggests that adaptive force and acceleration of every moving ob- planning should not be based on only a few ject, since the formulation of the general scenarios, since there is no guarantee that Newtonian law allows us to understand any the range simulated represents the full one. motion. Does one have general hydrological laws of comparably universal validity, which could be of use in ungauged basins or for fu- 3. CONCLUDING REMARKS ture climate? One could say that drainage basins are so very different from each other. If informed decisions on building the prepa- Yet so are the objects obeying Newton’s laws redness system for meteorological and hydro- of dynamics. Certainly, one obvious and logical extremes are to be made, mathemati- essential law ruling hydrological systems is cal models are needed, enabling the analyst the principle of the conservation of mass to convert the measured (or postulated) (the continuity rule), valid over spatial and values of some variables into values of other temporal scale (Kundzewicz, 2007). variables of interest. In order to simulate sy- Fig. 8 illustrates the impacts of heat stemic behaviour, time series of observations wave effects on the Polish population in the (e.g. precipitation over a basin area and ri- load-resistance framework, stemming from ver flow in a terminating cross-section) are mechanics, but useful in multiple system first used to identify the system’s model (an studies. This conceptual sketch shows that impulse response in the linear case). Once the load (summer heat) is likely to grow with this has been done and the system’s respon- time, while the resistance is likely to dec- se is identified, one can model the response rease with time in the ageing society, despite of the system corresponding to any input. possible developments in public healthcare.
load resistance
time
probability grows
System change Climate change
load, resistance
Figure 8. Increasing risk of adverse heat wave effects on the Polish population in the load-resistance framework. A conceptual sketch.
KKsisiąążżkka1.indba1.indb 117777 22008-06-26008-06-26 110:57:240:57:24 178 Zbigniew W. Kundzewicz, Roman Mańczak, Iwona Pińskwar and Maciej Radziejewski
Since life span is likely to grow, there will be IPCC (Intergovernmental Panel on Climate Chan- a larger proportion of older people, sensitive ge) (2001), Climate Change 2001: The Scienti- to excess summer heat. Heat wave impacts fic Basis (edited by Houghton, J. T., Ding, Y., depend on age structure, the health status Griggs, D. J., Nouger, M., van der Linden, P. J., of the nation, and adaptation (e.g. air con- Dai, X., Maskell, K. and Johnson, C. A.). ditioning, establishing cooling areas in old- Contribution of the Working Group I to the people’s homes and hospitals). Third Assessment Report of the Intergovern- Difficulties in modelling future impacts mental Panel on Climate Change, Cambridge of climate extremes are immense. Rather University Press, Cambridge, U.K. (Also ava- than a general theory, there are fragmented ilable at
KKsisiąążżkka1.indba1.indb 117878 22008-06-26008-06-26 110:57:240:57:24 Models of Impacts of Hydrometeorological Extremes 179
Lorenc, H. (2007),
KKsisiąążżkka1.indba1.indb 117979 22008-06-26008-06-26 110:57:250:57:25 APPLICATION OF HYDRODYNAMIC MODEL OF THE BALTIC SEA TO STORM SURGE REPRESENTATION ALONG THE POLISH BALTIC COAST
HALINA KOWALEWSKA-KALKOWSKA*, MAREK KOWALEWSKI** and BERNARD WIŚNIEWSKI* *Institute of Marine Sciences, University of Szczecin, Wąska 13, 71-415 Szczecin, Poland E-mails: [email protected]; [email protected] **Institute of Oceanography, University of Gdańsk, Al. Marszałka J. Piłsudskiego 46, 81-378 Gdynia, Poland E-mail: [email protected]
Abstract: A hydrodynamic model of the Baltic Sea based on the Princeton Ocean Model was ap- plied in analyses of extreme storm surges along the Polish Baltic coast. When the applicability of the model in cases of high-amplitude and rapid water level fluctuations, such as those observed at the beginning of November 2006, was tested a good fit was obtained between observed and computed data. The model correctly predicted the hydrodynamic situation; it also generated relatively good simulations of water-level variations. The best fit between the numerical calcula- tions and readings from the sea-level gauges was obtained for Gdańsk Nowy Port, while only slightly worse agreement characterized Świnoujscie and Ustka.
Key words: numerical modelling, storm surges, low-pressure systems, southern Baltic Sea.
INTRODUCTION is anticipated to increase in the future as Extreme storm surges along the Polish Bal- a result of sea-level rise and an increased tic coast occur as a result of the passage of frequency of severe storms. a deep and intensive depression over the In the Baltic Sea, the water level varies southern Baltic Sea. They produce flooding substantially as a result of an overlap of vari- of coastal areas, polders, and areas adjacent ous types of periodic and non-periodic os- to rivers; impact upon shore and beach sta- cillations. It is also influenced by a number bility; result in coastal erosion; affect port of meteorological and hydrological factors. operations and navigation negatively; and The main ones affecting sea level are wind impinge upon coastal zone infrastructure. and sea-level pressure patterns over the It should be mentioned that the flood risk Baltic Sea, weather systems over the North
KKsisiąążżkka1.indba1.indb 118181 22008-06-26008-06-26 110:57:260:57:26 182 Halina Kowalewska-Kalkowska, Marek Kowalewski and Bernard Wiśniewski
Atlantic, the water exchange between the Over recent years, numerical modelling North and Baltic Seas and the inflow of has become an essential tool in water level fresh river water. Precipitation, evapora- forecasting within the Baltic Sea region. The tion, seasonal water density changes, seiches High Resolution Operational Model for the (standing waves) as well as bathymetry of the Baltic Sea (HIROMB) was developed at body of water are other factors influencing the Bundesamt für Seeschifffahrt und Hy- the sea level (Dziadziuszko and Jednorał drographie (BSH) in Hamburg (Germany) 1987, Wiśniewski 1978, Heyen et. al 1996, and subsequently extended in cooperation Carlsson 1998, Wiśniewski et al. 2000, Le- with the Swedish Meteorological and Hy- hmann and Hinrichsen 2001). As reported drological Institute (SMHI) in Norrköping by Jasińska and Massel (2007), tidal effects (Sweden) (Kleine 1994, Eigenheer 1999, on sea level are negligible along the Bal- Funkquist 2001). It is now run operation- tic Sea coast (tidal amplitudes range from ally by the SMHI (http://www.smhi.se/ocea- 0.02 m to 0.1 m). nografi/oce_info_data/models/hiromb.htm). Prolonged stationary weather systems The Maritime Institute in Gdańsk (Poland) over the Baltic Sea are important factors verified the model for the Polish zone of the determining its water volume and sea level. Baltic (Kałas and Szefler 1999, Kowalska et During prolonged periods of domination by al. 2001). At present, the Maritime Branch westerly winds an inflow from the North Sea of the Institute of Meteorology and Water into the Baltic Sea takes place, causing an Management (IMWM) in Gdynia (Poland) increase in sea level. On the other hand, the issues, i.a., an online daily water level fore- prolonged domination of easterlies reduces cast for the southern Baltic Sea (http://bal- in sea level, due to the outflow from the Bal- tyk.imgw.gdynia.pl/hiromb/). In Denmark, tic Sea into the North Sea (Łomniewski et al. the Danish Meteorological Institute (DMI) 1975, Dziadziuszko and Jednorał 1987). (http://ocean.dmi.dk/models/index.html) is- A storm surge, as a type of non-peri- sues storm surge predictions using Mike 21 odic oscillation, is a rapid increase in sea (supplied to the DMI by the Danish Hydrau- level associated with the passage of a deep lic Institute for Water and Environment, i.e. and intense low-pressure weather system. DHI Water & Environment). Marine fore- Fluctuations in sea level result from the casts concerning, i.a., water level, produced combined effects of persistent wind over by the DHI Water & Environment are avail- a shallow body of water and changes in at- able at http://www.waterforecast.com and mospheric pressure on the sea surface; such http://www.bsh.de/. effects generate a sea surface deformation, This paper discusses the applicability of the so-called baric wave, with its positive a 3D pre-operational hydrodynamic model phase inside the low and its negative phases of the Baltic Sea (M3D_UG), developed at outside it. The wave is like a cushion of wa- the Institute of Oceanography, University ter under a depression that moves together of Gdańsk to analyses of storm surges along with it. Effects of the wind and the baric the southern Baltic coast. wave may be additive, i.e., both factors act in concert to increase or to decrease the sea level at the shore, or they may be non-addi- MODEL DESCRIPTION tive, i.e., one factor increases the sea level and the other decreases it (Wiśniewski et The three-dimensional, pre-operational al. 2005, Wiśniewski and Kowalewska-Ka- hydrodynamic model of the Baltic Sea lkowska 2007). (M3D_UG), developed at the Institute of
KKsisiąążżkka1.indba1.indb 118282 22008-06-26008-06-26 110:57:260:57:26 Application of Hydrodynamic Model of the Baltic Sea… 183
Oceanography, University of Gdańsk, is climatic distributions based on the Baltic a baroclinic model that describes water cir- Environmental Database (http://data.ecol- culation, with due consideration given to ad- ogy.su.se/models/bed.htm). vection and diffusion processes. The model The model is valid for three areas with is based on the Princeton Ocean Model different spatial grid resolutions: 5 nauti- (POM), described in detail by Blumberg cal miles (NM) for the Baltic Sea, 1 NM for and Mellor (1987). Adapting the model to the Gulf of Gdańsk, and 0.5 NM for the Po- the Baltic Sea required certain changes meranian Bay and Szczecin Lagoon (Fig. 1). in the numerical calculation algorithm, as A numerical grid with sigma-transformation described in detail by Kowalewski (1997). allows for a vertical profile at any point in The open boundary was located between the sea, irrespective of depth, to be divid- the Kattegat and the Skagerrak to param- ed into 18 layers (Kowalewska-Kalkowska eterise water exchange between the North and Kowalewski 2005, Kowalewski and and Baltic Seas. A radiation boundary con- Ostrowski 2005). dition for the vertically averaged flows was Evaluation of the model’s performance applied. A monthly averaged vertical dis- for both the eastern and western parts of tribution of salinity and the temperature the southern Baltic coast showed the re- gradient normal to the border equal to zero sults of the model calculations to be of the were assumed at the open boundary as well same quality. The model correctly reflected (Ołdakowski et al. 2005). The solar energy hydrodynamic conditions and seasonal vari- input was calculated based on astronomi- ability in sea level; it also generated rela- cal data and meteorological conditions tively good simulations. The results showed (Krężel 1997). Other components of the a significant correlation between calculated heat budget at the sea surface were derived and measured distributions of water levels. from meteorological data and simulated sea As described by Kowalewski (2002), cor- surface temperatures (Jędrasik 1997). The relation coefficients for the relationships model takes 153 riverine discharge events between modelled water levels and those into account. Because of wind-driven water measured by the IMWM gauges in August back-flow in the River Odra’s downstream 1999 in Władysławowo, Hel, and Gdańsk- reaches, a simplified operational model of Port Północny were 0.91, 0.89 and 0.87 re- the Odra discharge based on water budget spectively. In the model validation reported in a stream channel was developed (Kow- by Jędrasik (2005), correlation coefficients alewska-Kalkowska and Kowalewski 2006). for relationships between numerical simula- Meteorological data (wind field, air tem- tions and readings from tide-gauges located perature, atmospheric pressure, and water along the southern Baltic coast in 1995 and vapour pressure) were taken from the Uni- 2000 ranged from 0.69 in Ustka to 0.85 in fied Model for Poland Area (UMPL) mes- Władysławowo. Finally, a good fit was ob- oscale operational weather forecast model tained for the Pomeranian Bay and Szc- that generates 60-h weather forecasts with zecin Lagoon stations, with correlation co- 1 hour steps (Herman-Iżycki et al. 2002). efficients ranging from 0.81 for Świnoujście Contiguous tensions related to wind and (a 60-h forecast) to 0.93 for Trzebież (a 0-h the direct stress of atmospheric pressure forecast), as reported by Kowalewska-Ka- on the sea surface (the baric wave) are fac- lkowska and Kowalewski (2005, 2006). tored into the hydrodynamic model as well. It should be mentioned that the numeri- The initial conditions for the hydrodynamic cal data are relative only, that is, due to the fields were adopted in accordance with their imperfection of the boundary conditions
KKsisiąążżkka1.indba1.indb 118383 22008-06-26008-06-26 110:57:260:57:26 184 Halina Kowalewska-Kalkowska, Marek Kowalewski and Bernard Wiśniewski
Figure 1. Regions modelled: Baltic Sea (1), Gulf of Gdańsk (2), Pomeranian Bay and Szczecin Lagoon (3).
applied to the open border in the Skagerrak mated by the model, may be visualised for and Kattegat, it is difficult to render them a case involving a heavy storm surge that oc- comparable to the average sea level, and to curred in the beginning of November 2006. calculate the absolute sea levels (Kowalews- This was the result of the passage of a deep ki 2002, Kowalewska-Kalkowska and Kow- and intensive low-pressure system over the alewski 2005). Hence, when the modelled Baltic Sea. Initially, a decrease in water level and observed values are compared, a certain at coastal stations was observed on 31 Octo- constant is added to render the initial fore- ber, as a result of the low centre’s fast shift cast and observed values identical. over the North Sea (983 hPa in the centre) (Fig. 2a). The lowest sea level of 467 cm, i.e. 33 cm below the mean sea level, was record- APPLICATION OF THE MODEL TO STORM ed in Świnoujście at 11 a.m. as a result of the SURGE REPRESENTATION negative phase of the baric wave and south- westerly winds. The agreement between that The good fit between simulations and read- empirical minimum and the forecast for ings of sea level from gauges located on 31 October in the timing and extent of the the southern Baltic coast was an incentive drop was very good (Fig. 3a). Subsequently, for testing the applicability of the model to the water level along the southern Baltic storm surge representation. Temporal sea- coast was observed to increase rapidly as level variations in the region, as approxi- a result of the low centre’s shift over the
KKsisiąążżkka1.indba1.indb 118484 22008-06-26008-06-26 110:57:270:57:27 Application of Hydrodynamic Model of the Baltic Sea… 185 B
C D
Figure 2. The synoptic situation (after http://www.wetterzentrale.de/topkarten/fsfaxsem.html) during the November 2006 storm-surge event.
southern part of Sweden and then eastward and Gdańsk. Over the following days, the over the central Baltic (Fig. 2b). Along the low’s centre passed over the Gulf of Fin- southern Baltic, the maximum levels were land and then moved north, resulting in the observed on 1 November, resulting from ensuing drop in sea level along the south- the overlap of the positive phase of the baric ern Baltic coast (Figs. 2c,d). The tempo- wave and northerly winds. In Świnoujście ral variations in sea level occurring at that the maximum level of 647 cm (147 cm time were reproduced fairly accurately by above the mean sea level) was observed the model. During those days, the best fit at 3 p.m.; in Ustka the observed level (at between the modelled and observed sea 6 p.m.) was 630 cm (130 cm above the mean level fluctuations was obtained for Gdańsk level), while the level of 645 cm (145 above (Fig. 3c). The sea level variations in Świno- the mean level) was observed in Gdańsk at ujście were also correctly predicted by the 10 p.m. (Fig. 3). The rapid increase in sea model; however some underestimates were level was accurately approximated by the produced on 2 November from the 1 No- model. The timing and extent of maximum vember forecast and some overestimates on values as calculated by 31 October and 4 November from the 2 November prediction 1 November were also predicted with high (Fig. 3a). On the other hand, the continuing accuracy. It was only in Świnoujście that drop in sea level on 2 November at Ustka was the maximum level was overestimated by underestimated by the 1 November forecast, the 31 October forecast and predicted to whereas the 2 November prediction gener- occur 3 h after the real maximum. On the ated some overestimates (Fig. 3b). Figs 4, 5 other hand, the 31 October forecast un- and 6 illustrate all the storm phases as pre- derestimated the maximum levels in Ustka dicted by the 3D numerical model.
KKsisiąążżkka1.indba1.indb 118585 22008-06-26008-06-26 110:57:290:57:29 Figure 3. Observed and predicted water levels in Świnoujście (A), Ustka (B), and Gdańsk (C) during the November 2006 storm-surge event (in cm, relative to mean sea level). Legend: forecasts from 31 October, 1, 2, 3, 4, 5, and 6 November.
KKsisiąążżkka1.indba1.indb 118686 22008-06-26008-06-26 110:57:290:57:29 Figure 4. Spatial distribution of water levels (in cm, relative to mean water level) in the southern Baltic during the November 2006 storm-surge event, as simulated by the M3D_UG model and placed on the
Figure 5. Spatial distribution of water levels (in cm, relative to mean water level) in the Pomeranian Bay and Szczecin Lagoon during the November 2006 storm-surge event, as simulated by the M3D_UG model and placed on the
KKsisiąążżkka1.indba1.indb 118787 22008-06-26008-06-26 110:57:290:57:29 188 Halina Kowalewska-Kalkowska, Marek Kowalewski and Bernard Wiśniewski
Figure 6. Spatial distribution of water levels (in cm, relative to mean water level) in the Gulf of Gdańsk during the November 2006 storm-surge event, as simulated by the M3D_UG model and placed on the
CONCLUSIONS Continuously updated results of the model’s application are placed daily on the Univer- Evaluation of the M3D_UG model’s perform- sity of Gdańsk website (http://model.ocean. ance showed a good fit between the modelled univ.gda.pl) as maps of 60-hour forecasts of and observed distributions of water levels. In water level, currents, water temperature, and the cases of high-amplitude and rapid water salinity for the southern Baltic Sea and the level fluctuations, such as those observed at Gulf of Gdańsk, as well as for the Pomera- the beginning of November 2006, the model nian Bay and the Szczecin Lagoon. correctly predicted the hydrodynamic situa- Fast online access to the hydrographic tion; it also generated relatively good simu- forecast allows potential users to predict the lations of water level variations. During the day-by-day development of processes that storm analysed, the best fit between the nu- may affect different areas of human life and merical calculations and readings from the activities, e.g. navigation, port operations or sea-level gauges was obtained for Gdańsk flood protection in coastal areas. In addition, Nowy Port, only a slightly worse agreement the model may prove of assistance in studies being shown for Świnoujscie and Ustka. on extreme sea level variations. Hence it is The adequate approximation of sea-level intended to fine-tune the model to improve variations along the southern Baltic Sea its prognostic reliability so that a better fit coast by the model makes it a reliable tool between the observed and computed data is for analysing and forecasting storm surges. obtained.
KKsisiąążżkka1.indba1.indb 118888 22008-06-26008-06-26 110:57:300:57:30 Application of Hydrodynamic Model of the Baltic Sea… 189
ACKNOWLEDGEMENTS Atlantic Air-Pressure to Sea-Level Anomalies in the Baltic Sea, Tellus A, 48 (2): 312–323. Thanks are due to Dr Teresa Radziejewska Herman-Iżycki, L., Jakubiak, B., Nowiński, K., of Institute of Marine Science, University Niezgódka, B. (2002), UMPL—Numerical We- of Szczecin for the assistance in preparing ather Prediction System for Operational Appli- the English version of the manuscript. The cations, in Jakubiak, B. (ed.), Research Works routinely collected water level data from Based on the ICM’s UMPL Numerical Weather Świnoujście were made available by the Szc- Prediction System Results, Wydawnictwo ICM, zecin-Świnoujście Harbour Master’s Office. Warszawa, 14–27. The sea level data for Ustka and Gdańsk Jasińska, E., Massel, S.R. (2007), Water Dynam- were drawn from
KKsisiąążżkka1.indba1.indb 118989 22008-06-26008-06-26 110:57:310:57:31 190 Halina Kowalewska-Kalkowska, Marek Kowalewski and Bernard Wiśniewski
system Results, Wydawnictwo ICM, Warszawa, Long-Term Fluctuations of Sea Level in the 109–119. Baltic Sea], Wydawnictwo Wyższej Szkoły Kowalewski, M., Ostrowski, M. (2005), Coastal Morskiej, Szczecin. up- and Downwelling in the Southern Baltic, Wiśniewski, B., Kowalewska-Kalkowska, H. Oceanologia, 47 (4): 453–475. (2007), Water Level Fluctuations in the Odra Kowalska, B., Stanisławczyk, I., Mykita, M. mouth Area in Relation to Passages of Deep (2001), Quality Assessment of the Hiromb low-Pressure Systems, Oceanological and Hy- Water Level Forecast for the Polish Costal drobiological Studies, 36 (1): 69–82. Zone, Bulletin of the Maritime Institute in Wiśniewski, B., Wolski, T., Kowalewska-Ka- Gdańsk, 28(2): 65–70. lkowska, H. (2000), The Fluctuations of Wa- Krężel, A. (1997), A Model of Solar Energy Input ter Level in the Odra Estuary, Baltic Costal to the Sea Surface, Oceanological Studies, 26 Zone, 4: 5–13. (4): 21–34. Wiśniewski, B., Wolski, T., Kowalewska-Kal- Lehmann, A., Hinrichsen H. -H. (2001), The kowska, H. (2005), Wahania poziomu morza Importance of Water Storage Variations for w Świnoujściu i poziomów wód w rejo- Water Balance Studies of the Baltic Sea, Phys- nie ujścia Odry [Sea Level Fluctuations in ics and Chemistry of the Earth (B), 26(5–6): Świnoujście and Water Level Variations in 383–389. the Odra Mouth Area], in Borówka, R.K. and Łomniewski, K., Mańkowski, W., Zalewski, J. Musielak, S. (eds.), Środowisko przyrodnicze (1975), Morze Bałtyckie [The Baltic Sea], wybrzeży Zatoki Pomorskiej i Zalewu Szcze- PWN, Warszawa. cińskiego [Natural Environment of Coasts of Ołdakowski, B., Kowalewski, M., Jędrasik, J., the Pomeranian Bay and Szczecin Lagoon], Szymelfenig, M. (2005), Ecohydrodynamic Wyd. Oficyna IN PLUS, Szczecin, 113–125. Model of the Baltic Sea, Part I: Description of the ProDeMo model, Oceanologia, 47 (4): Paper first received: February 2007 477–516. In final form: November 2007 Wiśniewski, B. (1978), Sezonowe i wieloletnie wa- hania wód Morza Bałtyckiego [Seasonal and
KKsisiąążżkka1.indba1.indb 119090 22008-06-26008-06-26 110:57:310:57:31 PROJECTIONS OF CLIMATE EXTREMES IN POLAND
MAŁGORZATA SZWED, DARIUSZ GRACZYK, IWONA PIŃSKWAR and ZBIGNIEW W. KUNDZEWICZ Research Centre for Agricultural and Forest Environment, Polish Academy of Sciences, Bukowska 19, 60-809 Poznań, Poland E-mails: [email protected]; [email protected] [email protected]; [email protected]
Abstract: The climate change projections for Poland are consistent in foreseeing overall tempera- ture increase in the coming decades. Precipitation is projected to decrease in summer (though this finding is not robust, being model-dependent) and to increase in winter. It is expected that the occurrence of climate extremes over Poland may change in the future, warmer climate. In this study, daily temperature and precipitation data from the Hadley Centre HadRM3- PRECIS regional model simulations (for the SRES A2 scenario in three model experiments) in Poland were used to study temperature and precipitation extremes defined according to the specification made in the Integrated Project entitled “Extreme meteorological and hydrological events”. Climate extremes in the control period, 1961–1990, were compared with those in the projection period, 2071–2100.
Key words: climate change, climate model, extremes, precipitation, temperature, Poland.
INTRODUCTION and 2006 caused a major fall in harvests and challenged water supply systems. A large Several episodes of temperature extremes number of wildfires was also observed. The (heat waves) and precipitation extremes heat wave in 2006 with a record hot July (intense and/or long-lasting precipitation or caused considerable damage. In this context, long dry spells, often coincident with high it is natural to ask such questions as: Is this air temperature) have occurred worldwi- a natural climate fluctuation superimposed de, and in Europe, in recent years. These on socio-economic factors, such as increas- events, leading to high-impact floods, drou- ing human impact. Do these events signal ghts, and heat waves did not omit the terri- climate change, and are such events likely to tory of Poland. The Odra River flood in July be more severe and to occur more frequently 1997 was the most destructive natural event in the future? There have been numerous in Polish history, though floods in 1998 and studies in Europe tackling this issue. For 2001 also led to fatalities and severe mate- example, Klein Tank and Koennen (2003) rial damage. The summer droughts in 1992 analyzed a large set of station data for daily
KKsisiąążżkka1.indba1.indb 119191 22008-06-26008-06-26 110:57:310:57:31 192 Małgorzata Szwed, Dariusz Graczyk, Iwona Pińskwar and Zbigniew W. Kundzewicz
temperature and precipitation in Europe, for direction to changes in precipitation. Most the years 1946–99. Indices of temperature models foresee a decrease in summer pre- extremes indicate that both cold and warm cipitation and an increase in winter precipi- tails to the distributions of minimum and tation (Parry, 2000), but there are some pro- maximum temperature have shifted towards jecting contrary changes (Fig. 1). Based on higher values. Klein Tank and Koennen HadRM3-PRECIS regional model simula- (2003) also found that Europe-average indi- tions, this paper aims to illustrate the way in ces of precipitation extremes had increased, which the projected changes will influence even though the spatial trends were not co- the characteristics of climate extremes in the herent. The extreme heat wave in Europe in future. the summer of 2003 (not as severe as that in 2006 in Poland) was examined by Beniston (2004), on the basis of Swiss climatological DATA data and model simulations. The observed event resembles what regional climate mod- In this paper, it is the temperature and pre- els are projecting for summers in the latter cipitation extremes defined according to decades of the 21st century. Hence, one can the specification made in the Integrated interpret the 2003 heat wave as a proxy for Project entitled ‘Extreme meteorological a hotter future climate. and hydrological events’ that have been The climate-change projections for Po- analyzed. Daily temperature and precipi- land are consistent in foreseeing an overall tation data from the HadRM3-PRECIS increase in temperature in the coming de- regional model simulations (for the SRES cades. However, the models are not con- A2 scenario in three model experiments) in sistent as regards their projections of the Poland have been used to compare climate
Figure 1. Scenarios for changes in mean seasonal air temperature and precipitation for Poland in winter (DJF = December, January, February) and in summer (JJA = June, July, August); results obtained by Tim Carter as a part of the ACACIA project for time horizon of the 2080s, as compared to control period 1961–1990. Levels of one and two standard deviations are also marked (crosses passing through the centre of the coordinate systems). Symbols linked refer to results of single climate model for different SRES emission scenarios (A1, A2, B1, B2). Source: Parry, 2000.
KKsisiąążżkka1.indba1.indb 119292 22008-06-26008-06-26 110:57:320:57:32 Projections of Climate Extremes in Poland 193
extremes in the 30-year periods 1961–1990 The modelled data reconstruct the and 2071–2100, where the period 1961–1990 monthly distribution of mean temperature is the control one, while 2071–2100 is the correctly, yet the values in the model are projection one. The 1961–1990 values/cha- slightly overestimated, especially in the racteristics are based on model simulations summer months. The present value for an- and not on observations. nual precipitation is also reconstructed qui- This Regional Climate Model (RCM) te correctly, but the distribution and values has been developed by the Hadley Centre, for the monthly precipitation for modeled to help address issues where resolution data differ significantly from the observed higher than that of the General Circulation station data. The highest precipitation in Models (GCMs) is necessary. The RCM the model occurs in June and not in July, model covers Europe with a spatial resolu- as in the observation records. Winter pre- tion 0.44° by 0.44° (about 50 by 50 km), with cipitation is too high as compared with the 149 grid-cells over the territory of Poland. summer one. If a model cannot cope with The model has been developed in the U.K. reconstructing the details of the present cli- and is most accurate over Britain. Some of mate, it is not a trustworthy tool to make the inaccuracies in the RCM model have projections for the future, yet, it has been been analyzed by various authors, e.g. Holt reported in several studies that models et al. (2005). do not perform very well in reproducing In order to verify model simulations and observational data over Eastern Europe. to assess the usefulness of the HadRCM mo- Fig. 2 presents, as an example, a compari- del for this research, the control period data son of the monthly distribution for mean (1961–1990) from the model have been com- temperature and precipitation, based on pared with those from the European Climate the climate model and on the observations Archive (ECA) for six Polish stations (Biały- contained in the ECA holding, for the stok, Hel, Poznań, Szczecin, Warszawa and city of Poznań and for the cell containing Wrocław) for the same period. Poznań.
Figure 2. Mean monthly temperature and precipitation in the present (1961–1990) in Poznań. Source: Authors’ own elaboration based on ECA data and in cell with the city of Poznań, HadRM3 results.
KKsisiąążżkka1.indba1.indb 119393 22008-06-26008-06-26 110:57:320:57:32 194 Małgorzata Szwed, Dariusz Graczyk, Iwona Pińskwar and Zbigniew W. Kundzewicz
PRECIPITATION EXTREMES Projections of maximum one-month precipitation in a given year for the future Exceptionally intense and/or long-lasting (2071–2100) range from below 100 mm at precipitation or long dry spells are among the Baltic coast to more than 400 mm for the the categories of extremes defined in the higher mountains. In general, this precipita- specification made for the Integrated Proj- tion characteristic is projected to decrease in ect “Extreme meteorological and hydrologi- the western and southern parts of Poland, and cal events”. Among extremes related to at- to increase in the east. The biggest decrease in mospheric precipitation, the characteristics the future as compared to the control period studied are: is expected for high mountains, about 30 mm • maximum one-month precipitation in and more. It is projected to exceed 20 mm in a given year, the central part of the country (Fig. 3). • maximum one-day precipitation, For maximum one-day precipitation, the • daily precipitation with a probability greatest increases are projected for central of exceedence of 10%, Poland, while the biggest decrease is likely • number of days with 24-h precipitation along the Odra River, in the mountainous in excess of 10, 30, 50 and 100 mm, catchments of its tributaries (Fig. 4). The • meteorological drought defined as maximum one-day precipitation in a single lack of precipitation extending over at least year is projected to feature strong regional 15 days. variations in the future (2071–2100), from less than 30 mm in Pomerania to more than
Figure 3. Difference in maximum one-month precipitation between the future (2071–2100) and the present (1961–1990) Source: Authors’ own elaboration based on HadRM3 results [in mm].
KKsisiąążżkka1.indba1.indb 119494 22008-06-26008-06-26 110:57:340:57:34 Projections of Climate Extremes in Poland 195
Figure 4. Difference in maximum one-day precipitation between the future (2071–2100) and the present (1961–1990) Source: Authors’ own elaboration based on HadRM3 results [in mm].
100 on the South. Appropriately, daily pre- 50 and 100 mm per day), it is difficult to ana- cipitation with a probability of exceedence lyze them on the basis of climatic models. of 10% is expected in the future to range That is because the values apply to the area from 2 mm to more than 20 mm, while for of a large grid cell (50x50 km), rather than most of Poland it is projected to be about to a point. As a result, the climate model 4–6 mm. flattens extreme values. The only conclusion The projections for the future indicate that can be drawn on this basis is the pro- a decrease in the number of days with in- jected decrease in the number of such days tense precipitation (defined as 10 mm over in mountainous areas. one day or more). The greatest decrease in The precipitation deficit can be as seve- the future (2071–2100) in comparison with re as (or even more severe than) the surp- the control period is projected for the moun- lus, especially if it occurs in the growing tains; over 10 days. A decrease in the West is season. In this study, drought is defined as also projected, while in the Eastern part and a period without precipitation extending at the Baltic coast an increase in the number over at least 15 days. The future mean drou- of such a days is foreseen. Yet the projected ght is projected to last less time in relation increases in the number of days with preci- to the control period, approximately by two pitation above 10 mm are not as large as the days, in the North-East and in the moun- decreases, and they are to last up to 4 days tains. In the other areas of Poland, the du- (Fig. 5). ration of the mean drought will be longer in If higher thresholds of intense precipita- the future (2071–2100) by more than four tion are adopted in analysis (e.g. above 30, days (Fig. 6).
KKsisiąążżkka1.indba1.indb 119595 22008-06-26008-06-26 110:57:350:57:35 Figure 5. Difference in the number of days with precipitation P ≥10 mm between the future (2071–2100) and the present (1961–1990) Source: Authors’ own elaboration based on HadRM3 results [in days].
Figure 6. Difference in mean duration of dry spell (at least 15 days) in a year between the future (2071–2100) and the present (1961–1990) Source: Authors’ own elaboration based on HadRM3 results [in days].
KKsisiąążżkka1.indba1.indb 119696 22008-06-26008-06-26 110:57:370:57:37 Projections of Climate Extremes in Poland 197
The analysis of the longest dry spell in the TEMPERATURE EXTREMES year yields similar results. In the future, the longest dry spell in a year is projected to last Climate projections indicate that the from 20 days on the Baltic coast and Pomera- increase in mean temperature across Poland nia to more than 35 days in the Sandomierz is likely to continue. A question may arise as Valley. Thus, only for the north-eastern part to the behaviour of temperature extremes. of Poland and for the higher mountains does In the category of extremes related to air the climate projection indicate a shorte- temperature, the characteristics studied are: ning of the longest dry spell in a year. In the • absolute minimum and maximum of other areas of Poland, the duration of drou- temperature, ght is projected to be longer in the future • the number of extremely warm days
(2071–2100), even by more than 20 days. (tmax ≥ 35.0°C),
The HadRM3 model draws just one po- • the number of tropical nights (tmin ≥ ssible future (for a specific emission scena- 20.0°C), rio, A2). Other models are likely to produce • the length of the frost-free period with different results. As seen in Fig. 7 (IPCC a probability of 10%, WG1-AR4, 2007), models do not agree even • the number of days with severe frost in
on the sign of precipitation change over much May (tmin ≤ -2.0°C) and of Europe, including Poland. In summer, less • the number of extremely cold days
than 8 out of 12 models considered agree (tmin ≤ - 20.0°C). on the sign. In winter most models project Absolute maximum daily temperatu- increased precipitation over the territory of re is projected to increase in the future Poland. Best agreement of models (11 out (2071–2100) by 2 to even 6°C in the North- of 12 models agree on the sign of significant East. In absolute terms, the modeled ma- precipitation change) can be observed in the ximum is projected to range from 30°C on north of Europe (winter precipitation incre- the Baltic coast to more than 50°C in the ases) and in the Mediterranean (summer Bug and Narew River basins (Fig. 8). The precipitation decreases). extreme temperature (with a probability of
Figure 7. Projected patterns of precipitation changes. Relative changes in precipitation (in percent) for the period 2090–2099, relative to 1980 –1999. Values are multi-model averages based on the SRES A1B scenario for December to February (left) and June to August (right). White areas are where less than 66% of the models agree in the sign of the change and stippled areas are where more than 90% of the models agree in the sign of the change. Source: IPCC, 2007.
KKsisiąążżkka1.indba1.indb 119797 22008-06-26008-06-26 110:57:390:57:39 198 Małgorzata Szwed, Dariusz Graczyk, Iwona Pińskwar and Zbigniew W. Kundzewicz
Figure 8. The increase of absolute maximum daily temperature in the future (2071–2100) versus present (1961–1990) Source: Authors’ own elaboration based on HadRM3 results [in °C].
exceedence of 10%) is also projected to be nights is foreseen for the Sandomierz Basin higher in the future than at present. This (Fig. 10). temperature is expected to range from 22°C There is evidence that a high and prolon- on the Baltic coast to more than 34°C for the ged period of raised temperature may have Lublin and Małopolska Uplands in the futu- a dramatic impact on different fields of hu- re projection horizon (2071–2100). man activity, in particular on human health. In the future horizon, the number of Heat waves often turn fatal when the night- days with extreme temperature (days and/or time temperature does not drop considerab- nights) is projected to increase. The num- ly below the high day-time temperature. For ber of extremely hot days, with maximum this reason, the numbers of days with both temperature higher than or equal to 35°C maximum temperature higher than or equal is projected to increase by 5 to 30 days. In to 35°C and minimum temperature higher absolute terms, it is projected to range from than or equal to 20°C have been analyzed. 10 days on the Baltic coast to more than This number is projected to increase in the 40 days in the south, except for the high future. In the upper and middle Odra River mountain areas (Fig. 9). The number of basin, the Sandomierz Basin and the Lub- tropical nights, with minimum temperature lin Upland, the longest spell of this thermal higher than or equal to 20°C is also projec- characteristic may increase by up to 10 days ted to increase in comparison to the control for the projection period 2071–2100, in com- period. It varies from 5 nights at the Baltic parison with the control period. The number coast to 45 and more in the south-east and of such days in an average year may rise in along the upper and middle Odra River. the future to more than 20 in the South of The greatest increase in the number of hot Poland.
KKsisiąążżkka1.indba1.indb 119898 22008-06-26008-06-26 110:57:400:57:40 Figure 9. The increase in the number of extremely hot days (with Tmax ≥35°C) in the future (2071–2100) versus present (1961–1990) Source: Authors’ own elaboration based on HadRM3 results [in days].
Figure 10. The increase in the number of tropical nights (with Tmin ≥20°C) in the future (2071–2100) versus present (1961–1990) Source: Authors’ own elaboration based on HadRM3 results [in days].
KKsisiąążżkka1.indba1.indb 119999 22008-06-26008-06-26 110:57:420:57:42 200 Małgorzata Szwed, Dariusz Graczyk, Iwona Pińskwar and Zbigniew W. Kundzewicz
Based on climate projections, one can by 50–60 days compared with in the pre- conclude that frost is expected to retreat sent climate of Poland. The greatest in- from the territory of Poland in the future crease is projected for the central part of (2071–2100). The length of the frost-free Poland. The length of such a period with period will increase markedly, and the frosts a probability of exceedence of 10%, which will probably be uncommon. means that such a long frost-free period is Absolute minimum daily temperature projected to occur in only 1 out of 10 years, is projected to range from -5°C at the coast will vary from 210 days in the high moun- and along the Odra River to below -20°C tains to about 270 days in Western Pome- in the Tatra Mountains, so it will higher by rania and along the Lower and Middle 6 to even 12 degrees in the East, in relation Odra River. It follows that the frost-free to the present (Fig. 11). The minimum tem- period is projected to increase by 40 to 75 perature with a probability of exceedence days in the future (2071–2100) (Fig. 12). of 10% is also projected to increase in the A temperature below 0°C in May is not li- future. It will probably range from to -4°C kely to occur in an average year in the fu- to 0°C in the future, as compared with the ture projection horizon. Similarly, climate range -10°C to 0°C at present. In the Lower projections indicate a lack of very cold days Odra Valley, only temperatures above 0°C (defined as those with minimum tempera- are expected for the future (2071–2100). ture below -30°C) in the future. Even the The longest frost-free period, with mi- minimum temperatures of -20°C are not nimum temperature above 0°C, is projec- likely to occur in the prediction period, ted to last longer in the period 2071–2100 2071–2100.
Figure 11. The increase of absolute minimum daily temperature in the future (2071–2100) versus present (1961–1990) Source: Authors’ own elaboration based on HadRM3 results [in °C].
KKsisiąążżkka1.indba1.indb 220000 22008-06-26008-06-26 110:57:440:57:44 Projections of Climate Extremes in Poland 201
Figure 12. The increase in the length of frost-free period with probability of exceedence 10% in the future (2071–2100) versus present (1961–1990) Source: Authors’ own elaboration based on HadRM3 results [in days].
FINAL REMARKS less remote future, while for long-term effe- cts the importance of scenarios comes into The projections of climate extremes for play. Poland, obtained for 2071–2100 with the help of the Hadley Centre Regional Clima- te Model (HadRM3-P) show considerable ACKNOWLEDGEMENTS changes in comparison to the control period (1961–1990), which can be summarized as: The climate-model data produced by the • an increase in mean temperature and Hadley Centre have been provided by the such temperature-related indices as mini- UK Department of the Environment, Food mum and maximum temperature, percen- and Rural Affairs. The financial support of tiles of temperature, length of hot period, the Ministry of Science and Higher Edu- duration of a suite of tropical nights and the cation of Poland provided to the authors frost-free interval, through the research project “Extreme • intensification of the hydrological cyc- meteorological and hydrological events in le; with intense precipitation and droughts Poland” (PBZ-KBN-086/P04/2003) is also becoming more extreme. gratefully acknowledged. Unfortunately, there is a marked un- certainty to future projections. Results are strongly scenario- and model-dependent. Model dependence is important even for the
KKsisiąążżkka1.indba1.indb 220101 22008-06-26008-06-26 110:57:460:57:46 202 Małgorzata Szwed, Dariusz Graczyk, Iwona Pińskwar and Zbigniew W. Kundzewicz
REFERENCES of the Intergovernmental Panel on Climate Change, Cambridge University Press, Cam- Beniston, M. (2004), The 2003 Heat Wave in Eu- bridge, UK. rope: A Shape of Things to Come? An Analy- Klein Tank, A.M.G. and Koennen, G.P. (2003), sis based on Swiss Climatological Data and Trends in Indices of Daily Temperature and Model Simulations, Geophysical Research. Precipitation Extremes in Europe 1946–99, Letters, 31, L02202. Journal of Climatology, 16: 3665–3680. Holt, T., Bärring, L., Kundzewicz, Z.W. and Palu- Parry, M.L. (ed.) (2000), Assessment of Potential tikof, J.P. (2005), Comparison of Extremes Effects and Adaptations for Climate Change in from Global and Regional Climate Models Europe. The Europe ACACIA Project, Jack- with NCEP Reanalysis Data, manuscript. son Environmental Institute, University of IPCC (2007), Summary for Policymakers, in East Anglia, Norwich, UK, 324 pp. Solomon, S., Qin, D., Manning, M.,Chen, Z., Marquis, M., Averyt, K.B., Tignor, M. and Paper first received: March 2007 Miller, H.L. (eds.), Climate Change 2007: The In final form: November 2007 Physical Science Basis. Contribution of Work- ing Group I to the Fourth Assessment Report
KKsisiąążżkka1.indba1.indb 220202 22008-06-26008-06-26 110:57:470:57:47