MIA KOTILAINEN Dune Stratigraphy As an Indicator of Holocene

Annales Academiae Scientiarum Fennicae

Geologica-Geographica

166

MIA KOTILAINEN

Dune Stratigraphy as an Indicator of Holocene Climatic Change and Human Impact in Northern Lapland, Finland

Academic dissertation

To be presented, with the permission of the Faculty of Science of the University of Helsinki, for public criticism in Auditorium, Arppeanum, Snellmaninkatu 3, University of Helsinki, on November 29th, at 2 pm.

Supervised by: Professor Emeritus Joakim Donner Department of Geology University of Helsinki

and

Professor Veli-Pekka Salonen Department of Geology University of Helsinki

Reviewed by: Professor Juha Pekka Lunkka Department of Geology University of Oulu

and

Docent Peter Johansson Geological Survey of Finland Rovaniemi

Opponent: PhD Reader Philip Gibbard Godwin Institute for Quaternary Research Department of Geography University of Cambridge England

Vammalan Kirjapaino Oy Vammala 2004 Abstract

Kotilainen, Mia Marjut 2004 . Dune stratigraphy as an indicator of Holocene climatic change and human impact in northern Lapland, Finland. Annales Academiae Scientiarum Fennicae, Geologica- Geographica 166, 154 pp. R ECEIVED S EPTEMBER 6 , 2004 .

une reactivation history during the last 11 000 years in northern Finnish Lapland was recon- D structed by studying a total of 75 dune sections from two dune fields, which are located 70 km apart from each other. The study areas were the Muddusjävri dune field between the Lakes Mud- dusjävri and Paatari (in the coniferous forest zone) and the Ijjävri dune field, north from Lake Inari (in the mountain birch forest zone). Study methods included stratigraphical interpretation of the dune sections, sedimentological analyses for grain size distribution and parameters ( 249 samples) and binocular microscopy for grain roundness and surface texture (225 samples). Study methods also included scanning electron microscopy for quartz grain surface texture ( 175 quartz grains), thin sections ( 52) for grain mineralogical composition and microprobe analysis ( 117) for the chemical composition of garnet grains. Con- ventional radiocarbon dating ( 54), AMS dating ( 4 ) and TL/OSL dating ( 15) were used for reconstructing the chronostratigraphy. Three lithostratigraphical units were detected on both dune fields. Palaeoenvironmental model of the Holocene development was based on grain property data and dating results respective for each unit. Unit 1 , the initial dune building period, took place in the early Holocene at circa 10 800 to 8700 yrs before present. Unit 1 is characterised by high wind velocities and transportation rates with effective aeolian processes (sorting and grinding) during the deposition. The morphological frame of the dune fields originates from the deposition of Unit 1 and met only minor modification at the deposition of Units 2 and 3 . Unit 2 represents on both dune fields the longest accumulation series with several stabilization-reactivation cycles and shifting wind directions. Four dune reactivation phases were recognised in connection with the deposition of Unit 2 . The data suggests a dune advance with moderate wind velocities and transportation rates with effective aeolian processes and aeolian maturation of quartz particles during the deposition. Unit 3 accommodates dune reactivation phase R5 (since 500 yrs BP), which is detected at the Ijjävri dune field. The lithostratigraphy is characterised by unstratified, deformed strata. The data suggests moderate to weak wind velocities and transportation rates with effective aeolian processes and aeolian maturation of quartz particles during the deposition of Unit 3 . The data suggests a climate control on the duration of the active phases on both the dune fields during early- and mid-Holocene. However, the climate variability can not explain the reactivation processes during late-Holocene, for the last 4500 yrs before present. Human activity as a promoter for dune reactivation prior 1000 yrs before present is improbable on the studied dune fields. The latest reactivation phenomena at the Ijjävri dune field, from circa 500 yrs BP to the present is promoted by both climate and human impact. The LIA event provided the triggering factor for the beginning of the phase and the continuous aeolian activity has been supported by human activity including reindeer husbandry and off road vehicles. New results provided by this study give an improved picture of the Holocene development of the northern dune areas. The results indicate that stratigraphical interpretation with adequate number of dune sections is vital for understanding the reactivation phenomenon in these areas.

Key words: (GeoRef Thesaurus, AGI): dunes, re activation, stratigraphy, lithofacies, grain size, surface textures, climate change, human activity, absolute age, C- 14, thermoluminescence, Holocene, Inari, Finland.

Mia M. Kotilainen, Department of Geology, P.O. Box 64 (Gustaf Hällströmink. 2 ), 00014 University of Helsinki, Finland Contents

A BSTRACT 5 2.4. Climatic changes during the Holocene in northern Lapland31 C ONTENTS 7 2.5. Present day climate and recorded meteorological data 33 L IST OF ABBREVIATIONS 8 3. M ETHODS 35 1. I NTRODUCTION 9 3.1. Field work 35 1.1.Holocene aeolian activity 9 3.1.1. Mapping and geomor- 1.1.1. Earlier related studies 9 phological studies 35 1.1.2. Dating of the aeolian 3.1.2. Sedimentological events during the observations 36 Holocene 11 3.1.3. Sampling 38 1.2. Definition concepts 14 3.2. Grain size distribution and 1.3. Approach 14 parameters 39 1.4.Objectives17 3.2.1. Winnowing 41 3.2.2. Computing 41 2. S TUDYAREAS 18 3.3. Grain shape; roundness and 2.1. Muddusjävri dune field 18 surface texture41 2.1.1. Bedrock 18 3.3.1. Binocular microscopy 42 2.1.2. Regional deglaciation 3.3.2. Scanning Electron history 21 Microscopy 43 2.1.3. Land uplift 22 3.3.2.1. Preparation of the 2.1.4. Glacial and associated samples 43 sediments 22 3.3.2.2.Producing the 2.1.5. Human impact on the images 43 area; archaeological 3.3.2.3. Classification and historical evidence 24 of the surface 2.1.6. Present vegetation 24 features 44 2.2. Ijjävri dune field24 3.3.2.4.Quantitative 2.2.1. Bedrock 25 analysis 44 2.2.2. Deglaciation history 27 3.4. Mineralogical composition 45 2.2.3. Land uplift 28 3.5. Dating methods 45 2.2.4. Glacial and associated 3.5.1. Conventional and AMS sediments 28 radiocarbon dating 45 2.2.5. Human impact on the 3.5.2. Thermoluminescense area; archaeological and Optically Stimulated and historical evidence 28 Luminescence dating 46 2.2.6. Present vegetation 28 2.3. Holocene fire-frequency and 4. R ESULTS 47 the shift of forest line 30 4.1. Morphology of the dune fields47 7 4.1.1. Muddusjävri 47 5. D ISCUSSION 94 4.1.2. Ijjävri 47 5.1. Morphological features 94 4.2. Grain size distribution 52 5.2. Grain size distribution 97 4.2.1. Muddusjävri dune field 52 5.3. Grain shape103 4.2.2. Ijjävri dune field 52 5.4. Grain mineralogy109 4.3. Grain shape; roundness and 5.5. Dune lithostratigraphy111 surface texture56 5.6. Dating of the aeolian events 4.3.1. Binocular microscope 56 and chronostratigraphy114 4.3.2. Scanning electron 6. P ALAEOENVIRONMENTAL MODEL microscope (SEM) 56 OF THE DUNE FIELDS BASED ON 4.4. Mineralogical composition THE DATA PRESENTED IN THIS STUDY 122 of the dunes 58 6.1. Muddusjävri dune field 122 4.4.1. Muddusjävri dune field 58 6.2. Ijjävri dune field129 4.4.2. Ijjävri dune field 61 6.3. Climate or human impact135 4.4.3. Mineralogical comparison of the dune fields studied 63 7. C ONCLUSIONS AND RECOMMENDATIONS 4.5. Lithostratigraphy of the dunes FOR FUTURE RESEARCH 138 studied 64 4.5.1. Muddusjävri dune field 64 A CKNOWLEDGEMENTS 141 4.5.2. Ijjävri dune field 79 4.6. Dating of the aeolian events 88 R EFERENCES 143 4.6.1. Radiocarbon dates 88 4.6.2. Thermoluminescence (TL) A PPENDIX I 155 and Optically Stimulated Luminescence (OSL) dates 91 A PPENDIX II156

List of abbreviations:

AGIAmerican Geological Institute AMRTApparent Mean Residence Time AMS Accelerator Mass Spectrometry BP Before Present, years before the year 1950 14C yrs BP Radiocarbon years Before Present cal. yrs BP Calibrated radiocarbon date, years Before Present FMI Finnish Meteorological Institute GSF Geological Survey of Finland IRSL Infra Red Stimulated Luminescence LGMLast Glacial Maximum LIA Little Ice Age MWP Medieval Warm Period OSL Optically Stimulated Luminescence SEM Scanning Electron Microscopy TL Thermoluminescence 8 ww windward 1. Introduction

he research on Holocene aeolian land- The dune fields of this study locate in the T forms has gone through a develop- tree-line region, where the environment ment typical to many branches in geology response to climate variations has been es- and earth science. The initial descriptive pecially sensitive throughout the Holocene. and qualitative research of the late 19th and In the review of earlier related studies the early 20th century was soon replaced by emphasis is placed on the cold climate ae- more quantitative approach. Until the end olian landforms as defined in Chapter 1 . 2 . of the 19th century, most geologists considered wind transport of sediment to be 1.1.1. Earlier related studies much less important than sediment trans- Some quite comprehensive reviews of the port by water or glaciers (Pye and Tsoar, studies made on cold climate aeolian sands 1990). Early recognition of the significance have been presented (Niessen et al ., 1984 ; of aeolian processes included work by Koster, 1988). They cover the areas of Eu- Ehrenberg ( 1847 ), who described airborne rope (Iceland, England, Belgium, Nether- dust transported fromAfrica to Europe lands, Germany, Poland, Denmark, Swe- and von Richthofen (1882 ), who recog- den, Norway and Finland), United States nized primary aeolian origin of the vast (Wyoming, Colorado, Nebraska, Minneso- loess deposit which blanket much of ta, Maryland, Delaware and Alaska), Can- northern China. A cornerstone of aeolian ada (Labrador, Quebec, Manitoba, Sas- research has been the work by R.A. Bag- katchewan, Alberta, Yukon, Northwest Ter- nold, summarized in “The physics of ritories, Banks Island and Baffin Island) blown sand and desert dunes” (Bagnold, and Antarctica. Filion ( 1984 ; 1987) has 1941 ). He provided an important theoret- studied Holocene development of parabol- ical basis for understanding the mecha- ic dunes and the connection between ae- nism of aeolian transport and dune forma- olian record and fire activity in Quebec, tion which has influenced all subsequent Canada (Filion et al ., 1991). research. In Europe the cold climate aeolian research has been focusing lately on the Eu- 1.1. Holocene aeolian activity ropean “sand belt”. Bateman ( 1994) has Holocene aeolian activity is indicative of studied the chronology of cover sand dep- climate change, especially in the arctic and osition in Britain (South-West Lancashire, sub-arctic regions ( e.g. Smith, 1949; Koster, North Lincolnshire and East Anglia) con- 1988), thus the aeolian research of perigla- cluding that British cover sand deposition cial dune formations has been intensive. coincided with the Younger Cover sand I 9 ( 13 750 to 14 000 14C yrs BP) and Younger was stated that these aeolian formations Cover sand II ( 11 100 to 12 900 14C yrs BP) were barchans. Later they were proved to phases in the European cover sand chro- be parabolic dunes opening to the dune nology (Koster, 1988; Kolstrup et al ., 1990). forming winds (Markov, 1955). Sokolow Castel ( 1991) studied Late Holocene aeo- ( 1894) also describes also dune formations lian drift sands in Drenthe, Netherlands e.g. along The Gulf of Finland, between and concluded that human activity has Vyborg and St. Petersburg (at the villages triggered the beginning of sand accumu- of Murila and Lautaranta), at the Gulf of lation at 1 300 14C yrs BP. Riga around the mouth of the river Dui- Bergqvist ( 1981) produced an assess- na, and a massive 150 km long, 30 km wide ment of the Swedish inland dunes with re- dune field north of the Crimean Peninsu- gard to their value for future research, es- la at the Black Sea. The formation begins pecially on palaeoclimatic studies. He list- at Kachowka – Berislaw and reaches Kin- ed six most significant inland dune areas burn 150 km WSW, with the largest occur- from Sweden. The most extensive inland rence at Alëschki. Shuisky ( 1986) has re- dune areas are in Dalarna. Bonäsfältet is ported aeolian transport rates from the the largest, located north from Mora, be- shores of the Black Sea, the Baltic, and the tween Lake Orsasjön and the River Öster- Sea of Okhotsk. However, detailed strati- dalälven and was originally formed on the graphic investigation of the wind-blown biggest glaciofluvial delta of Sweden, sediments in Russia has only recently be- Morafältet. There are extensive dune are- gun. The dating of some dunes at the Rus- as in northernmost Sweden, near the Finn- sian Plain suggested that they have formed ish border. These dune areas include 12 during Oldest ( 14 404 ± 780 14C yrs BP) dune fields in the vicinity of the River and Older Dryas ( 12 680 ± 520 14C yrs BP, Muonio (some of them described by Sep- laboratory codes not provided) time (A. pälä, 1972) and at Merasjärvi, 25 km SE Drenova, pers. comm., 1994) by westerly from Vittangi (described by Högbom, winds (Drenova, 1994). From West Siberia 1923). The Merasjärvi site belongs to the (Arhipov and Volkov, 1980) and East Sibe- dune formations located near the River ria (Ivanov, 1966) there is data that suggest Lainio (8 sites). similar ages for periods of dune forming More recently a lot of research has been as at the Russian Plain. However, the con- carried out in Iceland, where the total area crete numerical results of the dating are of sandy deserts covers about 21% (Arnalds lacking in both references. On the map of et al ., 2001). In terms of areal distribution, Quaternary deposits of Finland and north- there are very few additions to make since western part of the Russian Federation the reviews by Niessen et al . ( 1984 ) and (scale 1 : 1000 000), the continuation of the Koster ( 1988). However, studies of Russian Salpausselkä -formations can be seen in origin, as well as more recent ones of the Karelia (Niemelä et al ., 1993). Actually, on areas of Russia were missing from these the map several dune fields can be located reviews, because they lacked a summary or especially in connection of the Salpaus- abstract in English, French or German. selkä-complex formation. There are many Thus, some of them are described here in dune fields associated with eskers, for ex- more detail. ample at the area 10 km NW from the Lake In Russia continental dune formations Rukajärvi (Heujpthj), at Lake Jyskyjärvi were described from The Russian Plain al- ( ?irjpthj) and another quite extensive ready a hundred years ago (Sokolow, 1894; one 10– 30 km west of Lake Kuittijärvi 10 Tutkovskiy, 1909). In the early studies it ( Chtlyttreqnj). On the southern shore of the White Sea there is also a dune field (Leiviskä, 1929) as well as at various plac- (along Bay Äänislahti, Jyt;crfzux,f), as es along the Hämeenkangas – Pohjankan- well as on the northern shore of the White gas esker chain (Herlin, 1896; Sauramo, Sea near Varzuga ( Dfhpeuf) area. There also 1924 ). In Central-Finland postglacial dunes are dune fields along eskers at the north- have also been found in connection with ern and the southern end of Lake Pääjärvi eskers, like at Kyyjärvi, Kinnula (Brander, ( Gzjpthj), near Lake Kuolajärvi (Re jkfzhdb) 1934), Vierämä and Rokuanvaara (Leiviskä, and 10 km north of both Puitsitunturi and 1907 ; 1929 ; Aartolahti 1973 ). Tanner ( 1915) Korvatunturi on the Russian side of the mentions many postglacial dune forma- border. tions in Lapland (in connection with raised In Estonia the formation of coastal beaches: at Sammalkangas near Sodankylä, dunes is successive to marine transgression at the northern part of Rovaniemi- parish, but dune building itself is related to peri- at Helppi near Kittilä, at Martinkylä near ods of constant sea level (Martin, 1988). Savukoski and to the north of the town of The biggest dunes in Estonia were formed Sodankylä). He also mentions dune areas during the transgressive phases of the An- near the Lutto-river and near Tieva- cylus Lake and Litorina Sea (Raukas, 1966). selkäjärvi, Inari. Another set of dune ob- Zeeberg ( 1993) has summarized the stud- servations is located by Tanner ( 1915) to ies made in Estonia and the Baltic region the areas of Enontekiö and Muonio. Lin- and thus has studied the aerial photo- doos (1972) studied the dunes of Joensuu, graphs and geomorphological and geolog- Liperi and Tohmajärvi in eastern Finland, ical maps of these areas in order to define while Seppälä ( 1971) made his study on the the extent of the wind-blown sediments. dunes of the Kaamasjoki-Kiellajoki river Earlier studies suggested a continuation of basin in northern Lapland. the European sand belt into the Baltic States ( e.g. Maarleveld, 1960; Nowaczyk, 1 . 1 . 2 .Dating of the aeolian events 1976; Catt, 1977; Koster, 1978; 1988; Kol- during the Holocene strup, 1982 ; Ruegg, 1983 ; Schwan, 1988; Ko- The early Holocene has been characterized ster et al ., 1993 ). However, Zeeberg ( 1993 ) by rapid climatic fluctuations, intervals of concluded that such continuation was not colder and warmer periods. In cold climate indicated by the aerial photographs. In- regions aeolian deposition processes and stead, the geomorpological data indicates increased aeolian activity have been con- a continuation of the zone with large per- nected with periglacial environment and iglacial aeolian deposits to the Kiev region early ice-free landforms. The initial forma- of Ukraine. The dune fields in the Baltic tion of European aeolian sediments dates region can be related to similar areas in back to the Last Glacial Maximum (LGM), Denmark and Fennoscandia (Zeeberg, when the loess deposits were formed and 1993 ). to the end of Weichselian glaciation, which In Finland pioneering studies on dune coincides with the initial formation of cov- evolution were made by Seppälä ( 1971) and er sand sheets and peri- and postglacial Lindroos ( 1972). There were, however, ear- dunes. lier observations of dunes and dune fields Dating of aeolian events has derived as summarized by Lumme ( 1934). Postgla- great benefit from the development of cial dunes occur in connection with both Thermoluminescence (TL) and Optically of the Salpausselkä formations (Okko, Stimulated Luminescence (OSL) dating 1954), at municipalities of Hollola, Savi- methods. Aeolian material is ideal for both taipale, Kiihtelysvaara and at Lemo-esker of the methods, because quartz and feld- 11 spar grains are easily well-bleached during area are shore dunes, which date back to the sedimentation process. Despite the pre- about 8800 14C yrs BP. These datings are eminent suitability of aeolian sediments, based on four radiocarbon dates obtained there are still many problems with the from peat samples taken from interdune techniques. bogs. At Rokuanvaara, the dune building Another means of dating aeolian events was most effective during the Boreal chron, has been radiocarbon dating. This dating while the onset of paludification and the method has offered an indirect way of dat- rise of the ground water table took place ing aeolian phenomena. However, great during the Atlantic chron. Secondary de- caution should be paid to the interpreta- flation is almost entirely confined to the tion of the actual 14C-results. Interpretation slopes of the highest dunes (Aartolahti, should always consider the nature of the 1973). original material. Suitable materials for Seppälä ( 1971 ; 1981; 1995) has discussed radiocarbon dating in aeolian environ- the initial formation and reactivation of ment include: the dune fields in northernmost Finland. a. Peat deposits overlain by aeolian In his thesis on the dunes of Kaamasjoki- sedimentation ( e.g. cover sand Kiellajoki area Seppälä ( 1971) concludes sheets, dunes) that the area became ice free and the ini- b. Interdune peat deposits tial dune formation took place circa 9700 c. Buried charcoal horizons as mark- 14C yrs BP, based on dating of plant remains ers of forest fires ( e.g. reactivated taken from the bottom of a palsa bog. The dunes) dunes were anchored by vegetation ap- d. Buried soil horizons proximately 1000 yrs after that. He based e. Vegetation remnants buried in aeo- this conclusion on a radiocarbon date 7160 lian sediments ± 200 14C yrs BP (HEL- 31) from the low- The general picture of the aeolian pheno- ermost buried soil horizon. Furthermore, ma in Finland during the Holocene time he concluded at a rough estimate that ap- was once considered clear: the dunes orig- proximately 1500 yrs are needed for pod- inated in periglacial environment soon af- sol profiles in the dunes of northern Lap- ter the land emerged from the water and land to develop, thus it may be assumed the initial active phase came to an end that the dunes arrived at their present po- when the vegetation stabilized the land- sition at least 8600– 8700 14C yrs ago. De- scape ( e.g. Okko, 1954). This impression of flation has continued throughout almost clarity has later been challenged as it has the whole of the time the dunes have ex- become evident, that these periglacial dune isted: it has been weakest during the At- fields have been exposed to wind erosion lantic chron some 6000 years ago as a re- and deflation at intervals during the whole sult of increasing humidity which caused Holocene ( e.g. Kotilainen, 1991; Seppälä, the ground water to rise in the deflation 1995; Käyhkö, 1997 ). basins. Deflation action increased greatly In Urjala dune development started after forest fires. According to Seppälä 8200 14C yrs BP and lasted only a few hun- ( 1971) two forest fires occurred, but the dred years (Aartolahti, 1967 ). Lindroos fires did not cause any dune development, ( 1972) has stated that the dunes on the dis- only small scale re-deposition of sand in tal side of Salpausselkä II originated un- connection with the deflation. Seppälä der periglacial conditions approximately ( 1971) continues that recently deflation ac- 10 000 14C yrs BP. Furthermore, he con- tion has again increased in strength. In his 12 cludes that the youngest dunes in his study table (Table 18) he places this recent phase between 0 – 500 14C yrs BP. Seppälä ( 1981) dunes in northern Finland were active for compared buried charcoal horizons from several millennia after the deglaciation, a kettle hole to those found in a nearby and a diachronous stabilisation between dune in Kuttanen area. The kettle hole sec- 7000 and 4300 appears likely. The latest tion had several charcoal layers, whereas phase of aeolian activity dates back to only one buried charcoal horizon was 1100– 1650 AD across the area Käyhkö found at the dune blowout, with 14C –date ( 1997 ) had studied. He was not able to in- of 1290 ± 130 14C yrs BP. He concludes that dentify any unique deflation triggering fac- the missing charcoal horizons at the dune tor, but came to the conclusion that sev- section can indicate either that the fire nev- eral agents – such as LIA event, reindeer er reached the dune area or that those lay- trampling and forest fires – acted in com- ers have been eroded due to deflation. Sep- bination. pälä ( 1995) has summarized the published In Finland we have more recent exam- work on aeolian activity studies in north- ples of dune activity on a recently emerged ern Finland and comes to the conclusion land. On Hailuoto, a number of parallel lit- that there were three main activity phases toral dune ridges have been formed with- namely: 1 .The post-glacial dune forming in an area of rapid land uplift since the episode which ended before 7000 14C yrs emergence of the island from the sea about BP. 2 .An Early Sub-boreal deflation phase 1800– 1900 yrs BP. Some of the ridges have that started approximately 4800 14C yrs BP moved over a short distance and are de- as a result of climatic change. The defla- positing a net of parabolic dunes further tion continued in several phases driven by inland. The highest parts of the island like forest fires. 3 . From 1300 14C yrs BP on- Hyypänmäki 31. 6 m a.s.l. are old parabol- wards strong deflation, which has de- ic dunes covered by pine forest. The ridg- stroyed the original shape of dunes. How- es of some dunes have been reshaped by ever, Seppälä ( 1995) finds very little evi- wind and some are still migrating. The old- dence of regional correlation for this date. est shore ridge formed about 1700 yrs BP Käyhkö ( 1997) criticizes the way the dat- (Alestalo, 1979). Hellemaa ( 1998) has dis- ing material from different sources has cussed dune formation and processes at been pulled together in Seppälä ( 1995). the coastal areas of Finland. The focus of Apparently the correlation between TL/ her work has been the dune development OSL and 14C- dates has been done without during the past 100 years. One of the con- calibration of the radiocarbon dates. clusions in her study based on aerial pho- The connection between climate and tographs is that the coastal dune areas are aeolian activity seems to be complicated. continuously diminishing. Initially grazing According to Aartolahti ( 1976; 1980) the at the beginning of the 20th century caused old stabilized dunes in Finland were acti- wide deflation surfaces and transgressive vated during the climatically distinct Lit- dune building, but later forests have been tle Ice Age (LIA) 1550– 1850 AD ( 400– 100 invading the coastal dune meadows and yrs BP). Vliet-Lanoë et al . ( 1993 ) limits the heaths. Most recent human impact (tour- Holocene aeolian activity at Hietatievat, ism, trampling) has led to impoverishment Enontekiö to three periods: 1 . older than of the vegetation and erosion. The dune Mid-Atlantic ( circa 7000 yrs BP), 2 . near areas constitute a unique environment the Middle-Ages climatic optimum both biologically and geologically. The (known also as Medieval Warm Period, Geological Survey of Finland carried out MWP) 1000– 700 yrs BP and 3 . the present a study where the geologically most signif- day event. Accorging to Käyhkö ( 1997 ) the icant raised beaches, aeolian and morainic 13 landforms in Finland were identified and environment refers to the modern day classified. The description of the localities temperature, although it is obvious that all also expressed a view to their endanger- periglacial aeolian landforms were origi- ment. According to this study the most se- nally deposited under cold climate condi- rious threat to the natural state concerns tions ( Fig. 1 ). the aeolian deposits, mainly due to rein- Subarctic aeolian activity refers also to deer overgrazing, off-road traffic and tour- the cold environment. Subarctic means ism (Johansson et al ., 2000). basically the zone between the arctic and A summary of the aeolian activity cy- the temperate (mid latitude) zones ( e.g. cles in the northern hemisphere during the Käyhkö, 1997 ). However, the southern limit Holocene is presented in Table 1 . The are- of the arctic zone has several definitions: as represented include sites from: Sahara, the area north of a. the Arctic Circle, b. the Netherlands, Britain and European cover tree line, c. the + 10 C July isotherm and sand chronology, Denmark, North-Amer- d. the limit of permafrost. The study areas ica and Russia. Activity phases suggested in Inari are situated just below and beyond by Seppälä ( 1995) and Käyhkö ( 1997) in the Scots pine (Pinus sylvestris) forest limit northern Lapland, Finland are indicated by (see Chapter 2 ). columns. Cold climate aeolian sand deposits are described as follows: (partly according to 1.2. Definition concepts Smith, 1964 ; in Koster, 1988) 1 . interfinger- The terminology used in the studies on ing of conformable relations with cold climate aeolian sand deposits varies (fluvio)glacial deposits; 2 . evidence for greatly according to the approach of the snow meltwater activity or denivation study. Geomorphological, stratigraphical forms; 3 . derivation from source areas and pedological concepts often are used which can be attributed to glacial margins, indiscriminately while describing aeolian ice-dammed lakes or to eustatically low- sand deposits and related relief forms ( e.g. ered sea level; 4 . inclusion of plant, pollen Castel et al . 1989). Hence, it is important or insect remains indicative of cold envi- to summarise here, how and in which con- ronments; 5 . associated structures or forms text are particular terms being used in this of permafrost of seasonal frost origin; or thesis. Koster (1988), in his review article 6 . relief forms and orientations recording of cold climate aeolian deposition, deter- palaeowind directions divergent from that mined cold climate environment general- of the present and consistent with that ly as areas where the mean annual temper- which might occur in formerly glaciated ature is less than – 3 ° C (French, 1976 ) in- environments, e.g. of katabaltic origin cluding also all areas with snow cover, (Koster, 1988). According to these defini- where the coldest mean monthly temper- tions, both dune areas in the present study ature is below – 3 ° C. In the present study can be described as cold climate aeolian area, the mean annual temperature is sand deposits formed in subarctic environ- – 1.3˚C (The Finnish Meteorological Insti- ment. tute, FMI, 1991, referring to the data 1961– 90). The coldest mean monthly tempera- 1.3.Approach ture (January) in the study areas is – 19.6˚C This study focuses on aeolian process- (FMI, 1991, referring to the data 1961 –90) es in the northern Finnish Lapland in two and the mean January temperature ( 1961– distinct dune fields. The present study con- 90) is – 14.4˚ C at the Toivoniemi weather cerns the time span of the Holocene, ap- 14 station, Inari (FMI, 1991). Cold climate proximately the last 11 000 yrs. Table 1 . Some Holocene aeolian activity cycles in the northern hemisphere. Columns Käyhkö, 1997 and Seppälä, 1995. BP refers to a radiocarbon date, years ago to an OSL/TL date.

15 Figure 1. Mean annual temperature (˚C) 1971–2000 (Drebs et al., 2002) and the dune areas of northern Finland according to Tikkanen and Heikkinen (1995) and as interpreted from the Quaternary deposit maps (scale 1: 400 000) of the area (Geological Survey of Finland). The white circles indicate vegetated dune fields and the black ones dune areas affected by deflation. Numbers and large circles indicate the locations of Muddusjävri dune field (1) and the Ijjävri dune field (2).

An earlier study at Muddusjävri sug- present study, the approach is mainly sed- gested that dune stratigraphy in northern imentological. In order to avoid coinciden- Finland was to be a potential palaeoenvi- tal observations, both dune fields were well ronmental archive (Kotilainen, 1991). The covered by a net of sampling sites. Each idea of comparing two dune fields, origi- sampled dune was studied at three loca- nally formed in similar conditions, but tions: the windward, the top and the lee having undergone different environmen- side. In earlier studies, a number of ran- tal development due to physical location, dom 14 C-dates (in fact all that were ever took shape. There are some previous quite published) had been combined in attempt comprehensive and well conducted stud- to understand the reactivation history of ies on aeolian landforms in northern Fin- the whole northern dune areas (Seppälä, land ( e.g. by Seppälä, 1971; 1995; Käyhkö 1995; Käyhkö, 1997 ). In these attempts no 1997). However, in the previous studies attention was paid on the nature or strati- there are a limited amount of observations graphic position of the dated samples. It 16 on internal dune structure. Here, in the seemed worthwhile to conduct systematic study on only some of the fields. The fo- sults of these analyses have been processed cus of the present study is vertical rather using quantitative, statistical and modeling than horizontal. The working hypotheses methods. An emphasis in this thesis has of the study can be outlined as follows: been laid on dating results. The sampling 1 .Dune fields in northernmost Finland technique for the dating material has been have experienced continuous stabiliza- crucial in order to get uncontaminated, tion/ reactivation cycles during the reliable samples. Postglacial time. 2 .Sedimentological interpretation in-  .4. Objectives volves methods like grain texture anal- 1 .The primary aim of the study is to pro- ysis, which in this environment are ex- vide a new insight of the dune based pected to correlate with the transpor- palaeoenvironmental modeling by tation history of a single grain. It is pos- combining different data sets and find- sible to detect and separate features im- ing links between them. printed by glacial, glaciofluvial and eo- 2 .In order to clarify the first objective, the lian environments. following questions can be phrased: 3 .It was expected that the combination of a. is the variation in dune stratigraphy, a sufficient amount of both radiocar- grain compositon and surface tex- bon and TL/OSL dates would provide ture traceable over a “dune front “; a comprehensive picture of aeolian ac- i.e. as a factor of transportation dis- tivation/stabilization phases on the tance ? dune areas. Furthermore it would be b. how do the litofacies units correlate possible to establish whether dune on a dune-to-dune level ? stratigraphy constitutes a climatic sig- c. how much information can be ob- nal and/or indicator of human impact tained from the re-deposited sand in the area. units indicating reactivation? The methodological approach adopted 3 .And finally: to contribute to the discus- in this study consists of a combination of sion of the deflation-reactivation phe- traditional and specific methods including nomena in the dune fields of northern field observations for dune morphology Finland during the Holocene. The and stratigraphy, dry sieving for grain-size amount of published radiocarbon evi- analysis, binocular microscopy and scan- dence related to dune reactivation will ning electron microscopy for grain texture be increased from 34 to 89 and the and thin section microscopy for minera- number of available TL/OSL dates from logical composition of the sand. The re- 20 to 35.

17 2. Study areas

The two dune fields are situated 70 km years. Secondly, the dune fields represent apart, the Muddusjävri (in Lappish, the different vegetation zones at present, which Finnish forms Muddusjärvi and Mutus- allows the comparison of the role of their järvi are also used in literature) dune field vegetation to their development. The shift- (lat. 68˚ 52'–68˚ 58', long. 26˚ 40' – 26˚ 45') ing of the vegetation zones is fairly well is located between Lake Paatari and Lake known in these areas (see Chapter 2 . 3 ). Fi- Muddusjävri about 12 km west of the town nally the preliminary fieldwork in these of Inari. The Ijjävri (in Lappish, the Finn- areas revealed several palaeosols with over- ish form Iijärvi and another Lappish form lying charcoal layers buried in the lee side Idjajavri are also used in literature) dune of the dunes. This suggested several reac- field (lat. 69˚ 22'–69˚ 25', long. 27˚ 35'–27˚ tivation phases had occurred during post- 50') is 70 km further northeast of Muddus- glacial time. jävri (Fig. 2 ). These dune fields were formed in association with a single esker 2.1.Muddusjävri dune field chain, which runs from the Lemmenjoki The Muddusjävri dune field (also known valley northeast to the Lake Opukasjärvi, as Tirro dune field, e.g. Johansson et al., a total length of 156 km. According to Mik- 2000) has an average width of 3 . 2 kilome- kola ( 1932) this esker chain is considered tres and a length of 9 . 2 kilometres, the to- to be the longest and most remarkable gla- tal areal coverage being circa 30 km 2 . Lo- ciofluvial formation in northern Finland. cation of the dune field and the studied In places, such as the Muddusjävri region, dune sites are presented in Fig. 3 . The area the esker deposits were overlain by other was first recognised as a dune field by Tan- sediments formed during the deglaciation. ner ( 1915), although no detailed studies of The locations of the study areas provid- the dunes have ever been published. ed a good basis for the aims of this study. First of all their association with glacial 2 . 1 . 1 .Bedrock deposits indicates that they were formed The study area is underlain by a granulite shortly after the deglaciation, because that complex. The rocks of this complex are period was most suitable for their forma- mostly garnet gneisses that were formed by tion with strong winds and lack of vege- metamorphism from acid and intermedi- tation. In this case, they have probably ex- ate supracrustal rocks and also from in- isted for the whole of postglacial time and fracrustal rocks under the granulite facies may provide a record of environmental conditions (Meriläinen, 1976). The main 18 changes over the time span of nearly 11 000 minerals of the granulite are plagioclase, Figure 2 . Locations of the Muddusjävri (1 ) and the Ijjävri (2 ) dune fields in relation to the present forest zones (according toAario, 1960). The esker chain is indicated with a dash line. quartz, garnet and occasionally cordierite stratigraphy of the rocks is particularly (Meriläinen, 1970). Between Lake Muddus- complicated in the northeastern part of the jävri and Lake Paatari the bedrock consists complex, at Lake Inari and southeast and of garnet-cordierite gneisses. In texture the northwest of it. Intense vertical movements garnet-cordierite gneisses are granoblast- following granulitisation have probably ic, medium-to coarse-grained rocks, with uplifted the granulite complex proper for a grain diameter of 1 – 5 mm. The main several kilometres relative to the granite minerals are potassium feldspar, quartz, gneiss complex (Meriläinen, 1976), to plagioclase (An 25– 40), garnet, biotite and which Ijjävri region belongs ( Fig . 4 ). cordierite (Meriläinen, 1976). The granu- The whole granulite complex is divid- lite complex also includes fine- and coarse- ed by many fault and fracture zones (Meri- grained garnet-quartz-feldspar gneisses läinen, 1970). The Geological Survey of which are present in the area surrounding Finland has carried out several seismic the dune field. Some small areas of quartz profiles across the Muddusjävri dune field, diorites and granodiorites containing hy- which reveal thick sediment deposits un- persthene are also found near the dune der the dune field. Quaternary sediments field ( Fig. 4 ). overlying the bedrock are 70 metres thick Mikkola ( 1932 ) described the granulite at the northern end of Lake Kotkajärvi, complex as a 300 km long curve shaped while the maximum depth for these layers formation, which can be divided into two is 73 m (K. Mäkinen, pers. comm., 1989; main zones. The boundary between these 2004 ), in the middle of the dune field. zones lies between Lake Muddusjävri and Faults and fracture zones are dominant in the river Vaskojoki (Meriläinen, 1970). The the area and may have been deepened by 19 Figure 3 . Muddusjävri dune field. Numbers 1 – 21 indicate the locations of the studied dunes (number 23 is a delta section). Basic maps 1 : 20 000 (3823 07, 08, 10 and 11). Copyright National 20 Land Survey of Finland, permission no. 49/MYY/04. Figure 4 . Bedrock of the northernmost Finland. Locations of the Muddusjävri dune field and the Ijjävri dune field are indicated with numbers 1 and 2 , respectively. Bedrock map of Finland, 1 : 1000 000, Geological Survey of Finland (Korsman et al., 1997 ). glacial erosion, because they may have act- the Tromsø-Lyngen end moraine was ed like preglacial valleys directing the melt formed as a part of the Younger Dryas end water flow under the glacier. moraines (Sollid et al., 1973) represented by the Salpausselkä moraines in southern 2 . 1 . 2 . Regional deglaciation history Finland. In the Inari area the ice retreated The knowledge of the deglaciation of the generally from northeast to southwest to- Inari region is largely based on the early wards the last ice divide area in central studies by Tanner ( 1915; 1930) and later by Finnish Lapland. When the ice sheet re- Synge ( 1969 ). However, the dating of the treated from the Main Substage, the Inari events has greatly changed since these basin became ice free some 10 000 – 9000 studies. Moreover, the early theories by 14C yrs BP (Hyvärinen, 1973). Saarnisto Tanner ( 1915; 1930) of the so called Port- ( 1973) has concluded from radiocarbon landia (f - line) transgression, which ex- and shoreline evidence that the Lutto riv- tended as far inland as south of Lake In- er valley, SE from Inari basin, was degla- ari, has been later criticized (Marthinus- ciated closer to 9000 14C yrs BP. The old- sen, 1974). The Tromsø-Lyngen end mo- est radiocarbon dates from Inari area are raine (Main Substage, also known as Tana two samples (B-510 and Hel-33) taken from -formation) was thought to mark the stag- the base of palsa bogs that were dated at nation line of the ice from 12 000 to 10 000 9700– 9800 14C yrs BP (Oescher and 14C yrs BP on the basis of radiocarbon Riesen, 1965; Seppälä, 1971). Seppälä ( 1980) dates from the organic material in till has calculated, on the basis of the date (Marthinussen, 1962 ; Andersen, 1965 ; 1968). 9740 ± 220 14C yrs BP (Hel-33), that the Subsequent studies suggested however, that retreating rate from the Tromsø-Lyngen 21 end moraines to the Kaamasjoki-Kiellajoki the Inari basin (Tanner, 1915, p. 276 ), which region (circa 40 km North of Muddusjävri) could have slowed the deglaciation. On the was approximately 200 m/year. This leads basis of the varve chronology and the stud- to the conclusion that the area was degla- ies by Kujansuu et al . ( 1998) the Muddus- ciated by 9800 14C yrs BP. Dating the de- jävri region was ice free at circa 10 800 cal- glaciation on the basis of radiocarbon dat- endar years ago. There are no radiocarbon ed lowest organic deposits is uncertain, be- dates from the bottommost organic sedi- cause the time difference between degla- ments from Lake Muddusjävri, but in the ciation and organic deposition is un- dune field, the onset of dune formation known. The beginning of the organic dep- marks the end of the deglaciation phase osition probably lasted several hundred with the formation of a delta. The land had years (Saarnisto, 1973). The date Seppälä become ice free and dry, yet the vegetation has obtained is remarkably old in compar- was not there. The luminescence date of ison to those dates from Virtaniemi: 8900 the lowest dune material at Muddusjävri ± 200 14C yrs BP (B-624 ; Sorsa, 1965 ), Kau- dune field is presented in Chapter 4 . 6 . nispää: 8530 ± 120 14C yrs BP (B-566; Sor- sa, 1965 ) and Lutto: 8160 ± 190 14C yrs BP 2 . 1 . 3 . Land uplift (Hel-221 ; Saarnisto, 1973). On the other Recent results from the ice sheet minimum hand, dates of over 10 000 14C yrs BP have and maximum models suggest that ice been obtained from the bottommost or- thickness over the Gulf of Bothnia was be- ganic deposits, althougth they have been tween 2 – 2 . 75 km during the LGM (Siegert thought to be too old (Salmi, 1963 ; Sorsa, et al ., 2001). The land uplift had already 1965 ). According to Saarnisto ( 1973), the ice begun during the deglaciation, and was retreat from the upper Lutto region lasted strongest in those areas that were most de- more than 1000 years. The difference be- pressed under the thickest part of the ice tween the dates probably reflects the for- sheet. Glacio-isostatic land uplift was ex- mation of ice-dammed lakes, which occu- tremely rapid at the end and immediately pied the regions immeaditely following after the deglaciation, decreased signifi- deglaciation. cantly circa 8500– 8000 BP (Eronen et al., The Vaskojoki valley was connected to 1995; Ristaniemi et al ., 1997; Miettinen, Lake Muddusjävri and Lake Paatari at the 2002 ) and has gradually decreased to the end of the deglaciation by the formation present rate. The annual land uplift in the of an ice-dammed lake that was connect- Muddusjävri area is approximately 3 . 6 mm ed to Lake Inari (Tanner, 1930, p. 461 ; Mik- (Saaranen and Mäkinen, 2002 ). kola, 1932 , p. 39). This also covered the area now occupied by the dune field. The ice- 2 . 1 . 4 . Glacial and associated sediments dammed lake located in the Kaamasjoki- During the deglaciation, a long esker chain Kiellajoki region was probably formed ear- was formed running from the Lemmenjoki lier according to Seppälä’s ( 1980) theory of valley to Lake Opukasjärvi, NE of Lake early deglaciation there. Ijjävri (Johansson and Kujansuu, in press). Using the same rate of deglaciation ( 200 The esker chain is indicated with a dotted m/a; Seppälä, 1980) the Muddusjävri re- line in Fig . 2 . The esker is not visible in the gion must have become ice free about 200 Muddusjävri area, between the Lakes Paa- years later, i.e. around 9600 14C yrs BP tari and Muddusjävri, but is buried be- ( 11 000 cal. yrs BP). It has been suggested neath younger sediments. However, it can that there was a deceleration in the move- be traced on seismic profiles and ground 22 ment of the ice sheet as it moved towards penetrating radar profiles made by the Figure 5 . The Quaternary sediments of the northernmost Finland. Locations of the dune fields are indicated with numbers 1 = Muddusjävri and 2 = Ijjävri dune field. Quaternary Deposits 1 : 400000, Sheets n:o 38 + 48 INARI, Geological Survey of Finland (Lahtinen and Mäkinen, 1985 ).

Geological Survey of Finland. Three sedi- The actively flowing ice sheet formed ment layers with different acoustic veloci- also a rather remarkable drumlin field in ties can be distinguished in the seismic the Inari lake area (Heikkinen and profiles on the top of the bedrock. On the Tikkanen, 1979). This phase is coincident surface the thickness of the layer of dry with the formation of the Salpausselkä for- sand (velocity of 500– 700 m/s) ranges mations in the southern Finland as the ice from 2 – 6 meters. The thickness of the sec- was flowing from SSW to NNE towards the ond layer with acoustic velocity of 1400– Tromsø-Lyngen end moraines in northern 1500 m/s indicating wet sand varies be- Norway (Hirvas, 1991). The drumlin field tween 54 and 73 m. In the lowermost part include drumlins and flutings. In the Lake the profile there are pockets of till (acous- Muddusjävri area there are some drumlins tic velocity of 2200 m/s) at the depth of that are almost 25 m high. The drumlin over 100 m in contact with the bedrock field in Lake Inari region is one of the larg- (K. Mäkinen, pers.comm., 2004 ). The est in Finland (Johansson et al ., 2000). The ground penetrating radar profiles reveal Quaternary sediments of the northern- dome shaped esker layers above the bed- most Finland are shown in Fig . 5 . rock followed by horizontal delta sedi- During the invasion of the Inari ice-lake ments capped by tilted sediment layers at into the area, the nearby edge of the ice the very top of the profile indicating dune sheet, especially the glaciofluvial channel, formations ( Chapter 4 . 1 ). was feeding finer sediments into the lake 23 forming a delta in the narrow bay between though most of them were probably the Innihvaara and Ristivaara hills. The caused by lightning. delta sediments are distinctive on the People also lived in the area in prehis- ground penetrating radar profiles. The toric times. On the beach of Lake Paatari, Muddusjärvi delta has been thought to a Stone Age dwelling site has been found have formed in two separate accumulation and 5 stone age finds have been recorded stages (Tanner, 1915). This assumption was along the river Kettujoki (C. Carpelan, based on the distinctive boundary ob- pers. comm., 1990). Later, in historic times, served under the shallow water, which was the area was inhabited by a few farmers recognised on aerial photographs. and the forests became privately owned. During recent years more holiday homes 2.1.5. Human impact on the area; have been built in the area. Human impact archaeological and historical is therefore increasing. evidence The archaeological record indicates that 2.1.6. Present vegetation the Inari region has been inhabitated since The Muddusjävri dune field area lies com- the early Inari Stone Age (beginning 8000 pletely within the coniferous forest zone. 14C yrs BP) continuously to the present The well sorted sand forms a fairly poor, (Carpelan, 1988, p. 54). Here, as everywhere dry site for the vegetation to grow. The else in Finland, most of the finds are con- present day vegetation in the Muddusjävri centrated around the modern inhabitated dune field is characterized by tall pine for- areas, where the land use is most intensive est ( Pinus sylvestris ). In a study of ecolog- and the building of houses and roads has ical classification of the forest soils in revealed prehistoric finds. Items from the Northern Finland, the proportion of pine Inari Stone Age ( 8000– 3000 14C yrs BP) in the tree stand was 97% of the basal area and Older Iron Age ( 3000– 1700 14C yrs in stands which had developed on aeolian BP) are found from nearly 200 sites around soils (Sepponen, 1985 ). The ground vege- Inari Commune (Carpelan, 1988, p. 51). tation is typical of dry sites where the most The Muddusjävri dune field is bordered common type of plants are lichens, Cla- to the north and south by the lakes Mud- donia stellaris forming the major cover. C. dusjävri and Paatari and the rivers Kettu- rangiferina and C. sylvatica are also well joki and Vaskojoki in the east and west re- represented. The lichen stand is very dense, spectively. The most remarkable set of pre- which provides good grazing ground for historic finds in the area occur in the reindeers. The typical composition of the northwestern corner of the dune field and vegetation cover is presented in Table 2 . consists of 10 trapping pits (Näkkäläjärvi, The human impact on the vegetation is 1964 , Fig. 1 ), which were used to hunt wild increasing at present, but is not severe. deer. The narrow, flat sandy isthmuses be- Overgrazing of reindeer has not occurred tween the lakes were especially useful for and the cutting of the forest has not caused trapping deer with pits (Luho, 1948, p. 115), any reactivation or deflation of the dunes. because they were easy to dig and fence. The hunting of the wild deer in the dune 2.2. IJjävri dune field areas is an interesting assumption, because The average width of the dune field in according to one hypothesis, fire was used Ijjävri is about 2 kilometres, the length 9 to force the animals in the required direc- kilometres and the areal coverage 18 km 2 . tion (Tolonen, 1983). This could be consid- The dunes do not form such a uniform 24 ered as a possible cause of forest fires, al- body as the Muddusjävri dune field does, Table 2 . The plant species composition on the Muddusjävri dune field (unpublished data from Dr. Pentti Sepponen, The Finnish Forest Research Institute, Rovaniemi).

SpeciesMean coverage (%)StdDev

Pinus sylvestris 0.42 0.71 Arctostaphylos uva-ursi 0.260.95 Diphasiastrum complanatum 0.02 0.06 Empetrum nigrum 11.97 13.08 Vaccinium uliginosum 0.01 0.03 V. vitis-idaea 5.89 5.33 Deschampsia flexuosa 0.01 0.03 Festuca ovina 0.63 1.67 Solidago virgaurea 0.01 0.04 Dicranum scoparium 0.73 2.57 Hepatica spp. 0.10 0.33 Pleurozium schreberi 0.17 0.64 Pohlia nutans 0.01 0.03 Polytrichum commune 0.23 0.36 P. juniperinum 0.13 0.38 P. piliferum 2.00 3.34 Ptilium crista-castrensis 0.00 0.00 Cetraria crispa 0.02 0.04 C. islandica 0.01 0.04 C. nivalis 1.10 1.76 Cladonia stellaris (C.alpestris) 38.58 14.92 C. botrytes 0.01 0.03 C. cenotea 0.04 0.05 C. cornuta 1.81 1.82 C. crispata 0.12 0.32 C. deformis 0.36 0.70 C. gracilis 0.85 1.36 C. rangiferina 12.664.37 C. sylvatica 8.83 4.15 C. uncialis 3.99 3.19 Pletigera aphthosa 0.01 0.03 Stereocaulon spp. 4.87 9.94 • Number of the study sites: 15 but the dunes are separated from each oth- huge blowouts. Location of the Iijjävri er over longer distances. The dunes are also dune field and the studied dune sites are much larger compared to those at Mud- presented in Fig. 6 . dusjävri. This results partially from their occurrence along or on the top of other 2.2.1. Bedrock high relief formations, such as moraine The Ijjävri region belongs to the granite ridges (see Chapter 4 . 1 . 2 ). The preliminary gneiss complex (Meriläinen, 1976, Fig. 1 ), study of the aerial photographs revealed it which is marked by the monotony of its to be a very impressive dune field with kil- rock types, which are mainly migmatic and ometres long massive dunes scattered with veined granite gneisses and gneissose gran- 25 Figure 6 . Ijjävri dune field. Numbers 52– 80 indicate the locations of the studied dunes. Location of “The big deflation basin” is indicated with a square. Basic maps 1 : 20 000 (3842 09, 12 and 3931 07, 10). Copyright National Land Survey of Finland, permission no. 49/MYY/04.

odiorites, quartz diorites and granites. The The tectonic structure of the granite area of the dune field lies almost complete- gneiss complex is complicated, because the ly in the area mainly underlain by gneis- rocks of the complex have been deformed sose granites. Only the northernmost cor- several times (Meriläinen, 1976, p.21). The ner of the dune field is related to the area typical granite gneisses are often intensive- of migmatitic granitic gneisses (Meriläi- ly folded: both their foliation and bedding nen, 1965). are folded. The general trend of the folia- Mineralogically the gneissose granites tion is approximately aligned N-S and the consist of quartz-plagioclase-biotite rocks dip of the foliation at Ijjävri varies from ( Table 3 ). Potassium feldspar is sparse, al- 25– 40 degrees to 45– 90 degrees (Meriläi- though where there are potassium feldspar nen, 1976, p.22– 23). The deep foliation -rich veins, it is more abundant. Some epi- zone between Ijjävri and Paudijärvi (also dote is also often found and occasionally known as Pautujärvi) represents the hornblende. Accessory minerals include boundary between tectonic blocks. The apatite, zircon, titanite, allanite and narrow schist zone at the end of Paudijärvi 26 opaques (Meriläinen, 1976, p.12). is a local relict of a syncline compressed Table 3 . Mineralogical composition of granite ation or as fracture zones (Meriläinen, gneisses (after Meriläinen, 1976). The site 2 , 1976, p.23). No seismic profiles has been Petsijärvi is located nearest to Ijjävri. undertaken across the isthmus between Ijjävri and Paudijärvi, but the location of 123 the fracture zone can be seen. This zone Quartz ...... 34.9 32.2 23.7 probably acted as a preglacial valley direct- K-feldspar ...... 8.6 10.1 7.7 ing the melt waters during the deglacia- Plagioclase...... 48.5 53.7 62.3 tion. Biotite...... 0.6 2.3 2.5 Muscovite...... 3.1 –2.4 2.2.2. Deglaciation history Chlorite ...... 1.9 ––The Ijjävri region should have been degla- Hypersthene ...... –0.8 – ciated in geographical terms earlier than Accessories ...... 2.4 0.9 1.4 the Muddusjävri region, because of its lo- 1. Inari, Surnuvuono, Mustavaara cation 70 km northeast of Muddusjävri. 2. Inari, Petsijärvi According to Hirvas ( 1991) both of the 3. Inari, Virtaniemi, Tervavuono study areas were affected by local glacial lobes, which resulted in the deposition of till bed I. Underlying this unit is till bed II, which was laid down during the second into shape between the dome-shaped an- youngest glacier flow, called flow stage II, ticlines of the granite gneiss complex. that extended over the whole of Finnish These synclines usually continue even in Lapland ( Fig . 7 ). The oscillations of the ice the country rocks either as trends of foli- margin or the local ice lobes have not been

Figure 7 .The main ice flow directions (arrows) during the Flow stage II of the last glaciation, which extended over the whole of Finnish Lapland (Hirvas, 1991). 27 studied in detail in either of the study ar- deposits. Gravel and sand deposits are not eas. Considering the studies mentioned very thick and they are overlay basal till. above by Hyvärinen ( 1973), Saarnisto There is, however, a small delta (approxi- ( 1973) and Seppälä ( 1981), the Ijjävri area mately 1 km 2 ) in the western corner of the was probably deglaciated around 10 000 sandy area. The water level of the early 14C yrs BP (see Chapter 2 . 1 . 2 ). The activity Lake Ijjävri has been higher than at of the local ice lobe may however delay the present, resulting in a raised beach 3 . 4 m final deglaciation by a few hundred years. above the present lake level (co-ordinates Kujansuu et al. ( 1998) concluded that the of the ancient shoreline x = 7701, 17; y = Ijjävri area became ice free at circa 10 900 527 , 20). The whole area of the Ijjävri dune calendar years ago. field is characterized by the presence of other than wind-blown material on the 2.2.3. Land uplift surface. Moraine hummocks are common. The land uplift has followed the same gen- Their orientation is the same as that of the eral trend as that described in the Mud- esker chain and the lake itself, i.e. SSW- dusjävri dune field, the annual land uplift NNE. being about 3 . 4 mm at present (Saaranen and Mäkinen, 2002 ). The outlet of the 2.2.5. Human impact on the area; Ijjävri basin is NE to the river Näätämö- archaeological and historical evidence joki, which discharges eventually into the There is evidence of prehistoric human Barents Sea. The tilting of the basin has not activity in the Ijjävri dune field. Four stone affected the surface flow system of the wa- age dwelling sites have been found along terway, because Ijjävri is situated in a de- the beaches of Lake Ijjävri (Kotivuori, 1986, pression which collects the waters from the p. 32– 40). According to Carpelan ( 1988) the surrounding areas. The lake levels on the Inari Stone Age was between 8000 and southwestern side of the lake are much 3000 14C yrs BP. The area has been used higher ( 205, 199 and 198 m a.s.l) than the to trap wild reindeer in the similar way as Ijjävri level, 193 m a.s.l. The level of the in the Muddusjävri dune field. Eleven sep- Lake Opukasjärvi further NE down the arate systems of trapping pits have been waterway from Ijjävri is 126. 7 m a.s.l. found, all together 332 pits, around the south-eastern shores of Lake Ijjävri (Näk- 2.2.4. Glacial and associated sediments käläjärvi, 1964 , p. 231). The theory men- The area of Ijjävri consists mainly of cov- tioned above of using fire to force animals er moraine. Morainic landforms such as in the desired direction (Tolonen, 1983 and drumlins are common especially at the the references therein) is the only way the northern part of the lake and between the stone age people could have caused reac- lakes Syysjärvi and Ijjävri at the SW side tivation on the dune fields. Nevertheless, of the dune field. The hill Sammuttivaara a far more likely explanation for the fires is a 30 m high drumlin, which has a 3 km is natural thunderstorms (see Chapter 2 . 3 ). long morainic tail (Johansson et al ., 2000). The esker chain, described above (Figs 5 2.2.6. Present vegetation and 6 ), runs on the SE side of the dune The vegetation of the Ijjävri dune field rep- field. All but a few of the dunes are situat- resents the transitional zone from the birch ed on the western side of the esker. The and pine forests to the zone of birch scrub main source area of the dune material lies and woodland. Occasionally, some pines in the southern part of Lake Ijjävri, where are growing, but the landscape is dominat- 28 there are quite extensive gravel and sand ed by Betula nana.Although low growing, the vegetation is quite dense. Ericales (e.g. separated in Arctic Canada: 1 . Pioneer stage Arctostaphylos uva-ursi, Empetrum nigrum, (typical species: Polytrichum piliferum, Calluna vulgaris, Vaccinium uliginosum Stereocaulon paschale, Rhacomitrium canes- and V. vitis-idaea) forms the main plant cens and Lecidea granulosa), 2 . Cladina mi- groups. tis -stage and 3 . Cladina alpestris -stage. There is an ongoing process of deflation The pioneer stage lasts from 1 to 20 years and re-vegetation of the basins along the in those conditions. In Ijjävri most of the dune crest. In the basins one can see, at deflation basins represent the pioneer stage what stage the process is in. At present, and probably are younger than 100 years. most of the deflation basins are vegetated In one of the basins in Ijjävri (co-ordinates by lichens such as Stereocaulon paschale. x = 76 99 15, y = 52493 , z = 200 m) there The re-vegetation process follows certain was a quite tall pine growing in the lowest succession pattern. There have been sev- part of the deflation basin. The pine was eral studies of vegetation succession on sampled with a 5 mm tree corer. The tree dune material, mainly made in Arctic Can- ring counting indicated that the deflation ada (Filion and Payette, 1989; Carroll and basin had existed in the same place at least Bliss, 1982 ) also studying the post fire de- 44 years. velopment (Maikawa and Kershaw, 1976; On the leeside of the dune the edge of Foster, 1983 ). These authors distinguish the deflation basin is often eroding ( Fig. 8 ). between three to seven phases in the re- This is the case especially when the defla- vegetation succession. In general, the fol- tion basin is deep and the sides of it form lowing synthesis of these studies can be two parallel ridges (see Chapter 4 . 1 . 2 ). made and the following three stages can be

Figure 8 .Collapsing Betula pubescens ssp. tortuosa (syn. ssp. czerepanovii) in the Ijjävri dune field. 29 2.3. Holocene fire-frequency In the areas of well sorted, aeolian land- and the shift of forest line forms in northern Lapland the density of Fire has always constituted an important the forest must have been important fac- factor in the boreal forest zone (Terasmae tor contributing to the extent of the fires. and Weeks, 1979). The recurrence of for- However, destruction of the sparse, sensi- est-fires in boreal forests of North Ameri- tive vegetation could have led to increased ca has been estimated as being 95– 100 years aeolian activity, even if the fire was not during the last 9000 years (Terasmae and widespread. Under the climatic conditions Weeks, 1979). They have also suggested that of northern Finland the post fire recovery the forest-fire frequency might have in- of vegetation is a very slow process. When creased to one fire every 48 to 50 years dur- discussing the role of the forest fires as pro- ing the pine period (about 7000– 4000 14C moters of the aeolian phenomena, one yrs BP). On the other hand, Seppälä ( 1981) should consider where the actual pine for- has found the most favorable period for est tree line in the study areas has been lo- forest fires in the Kuttanen region between cated, regarding to 6000 and 3500 BP. Tolonen ( 1983 ) has a. climatic changes and the pine advance studied fire-frequencies from lake sedi- during the climatic optimum and ments in southern and eastern Finland and b. land uplift? What has been the influence found significant differences depending on of a lower altitude? the vegetation, soil and geomorphology of Latest reconstructions of the pine tree limit the sites: dry, pine forests show a fire-fre- in the northernmost Finland suggest that quency of 78 ± 26 years, while it was only during the climatic optimum, at around 6 158 ± 12 years in spruce forest on morain- 000 cal. yrs BP, the Inari area was covered ic soil. In northern Sweden, the mean in- with pine trees (Kultti et al., in press). At terval between fires during the last 600 yrs present, the pine tree limit is about 30 km has been 80 years (Zackrisson, 1977). Sim- north from the actual pine forest limit, ilar results ( 80– 120 yrs) have been report- leaving Ijjävri area with no pine forest. ed in studies made in Ulvinsalo, Kuhmo, During the warmer periods it might have eastern Finland (Haapanen and Siitonen, been forested, as seems to be the case even 1978 ). with + 0 . 55 ˚C temperature rise. This rise Based on the exceptionally low amount in temperature has been suggested to ap- of charcoal particles in lake sediment from ply to Medieval Warm Period (MWP) at NW Lapland, Sarmaja-Korjonen ( 1995) around 1000 cal. yrs BP ( Fig. 9 ; Kultti et suggested that fire frequency and the sig- al., in press). nificance of fires has been low in the north- Forest fires have not been recorded in ern birch-pine forests of Finland during the Muddusjävri dune field area this cen- the Holocene. Actually, this indicates that, tury. The tree rings of one pine fallen by a since fires have been very rare in the area, storm in the area were calculated and they the role of fire may not have been impor- indicated well over 200 years of record tant in the birch-pine forests above the (M. Helminen, pers. comm.). present coniferous tree line (Sarmaja-Kor- Indeed, Tikkanen and Heikkinen ( 1993 , jonen, 1995). However, there have been sev- 1995) have presented the idea of forest fires eral fires in Ijjävri, which seem to have had being most abundant during the time an impact on the environment by trigger- when the pine forests were at their dens- ing dune deflation. Hence, aeolian activi- est and most extensive, according to the ty is reported to increase greatly after a for- latest pine subfossil record 6800– 4500 cal. 30 est fire (Seppälä, 1971 ). yrs BP (Eronen et al ., 2002 ). However, the Figure 9 . The reconstructed pine tree limit at different temperature conditions. The + 0 . 55 ˚C refers to MWP (around 1000 cal. yrs BP) and the + 2 . 55 ˚C indicates the climatic optimum (around 6000 cal. yrs BP) situation. The latter reconstruction caters for the influence of land uplift (Kultti et al., in press).

actual significance of forest fires as aeolian plicated the picture seems to become. The initiator has been controversial. Forest fires subject has been clarified by tree-ring re- were seen as the main trigger which pro- search along with numerous proxy stud- moted the reactivation at Enontekiö ies ( e.g. pollen, diatom, and cladocera - (Vliet-Lanoë et al ., 1993 ), while Tikkanen analyses). The main results are being sum- and Heikkinen ( 1993, 1995) are of the opin- marized here. ion that fires do not trigger aeolian activi- The tree-ring chronology from subfos- ty to any significant extent. sil Scots pine in Northern Fennoscandia ( Pinus sylvestris, L. ) is well studied and 2.4. Climatic changes during spans over 7500 years in Northern Fen- the Holocene in northern noscandia ( e.g. Eronen and Zetterberg, Lapland 1996; Zetterberg et al., 1995; Eronen et al ., The interaction between climate and aeo- 2002 ). There were severe growing condi- lian activity is sensitive even to small tions around 330 BC ( 2330 cal. yrs BP), changes in precipitation (Stetler and Gay- which are likely to have been caused by in- lord, 1996). Temperature and humidity creased wetness (Eronen et al., 2002 ), be- changes during the postglacial time in cause no exceptional features in pollen northern Lapland have fortunately re- data was found, nor was there any tephra ceived a fair amount of scientific attention, layer detected (Saastamoinen et al., 2000). alas the more data we have the more com- The humidity increased inline with the 31 cooling trend after the warm mid-Holo- terberg et al ., 1994). There is evidence for cene ( Hyvärinen and Alhonen, 1994; Ero- general lowering of summer temperatures nen et al., 1999a). by about 2 ˚C since the mid-Holocene ( e.g. Pine migration began from the Varang- Donner, 1995). This either, is not discerni- er-Nordkinn area, where the species was ble in the pine tree-ring data (Eronen et present probably around 10 000 cal. yrs BP. al., 2002 ). Pine arrived in the Inari basin area from The development of the vegetation and the north at around 8500– 8000 cal. yr BP thus the climate in eastern Lapland is well (Seppä, 1996). According to Eronen et al. established. The traditional model ( e.g. ( 2002 ) the evidence of the megafossils and Donner, 1971; Hyvärinen, 1975; 1976 ) dis- pollen data are mutually supportive in in- tinguishes 3 phases, as follows: I Birch dicating that the time of maximum extent -phase ( circa 9000– 7800 14C yrs BP) with of pine in Finnish Lapland was between 80– 90% of Betula, regularly with some Pi- 6800 and 4500 cal. yrs BP. After this pe- nus; II Pine -phase ( circa 7900– 2500 14C riod, the northernmost pine tree-line and yrs BP) with Pinus as the most common the limit of forest gradually retreated, as tree, regularly with some Betula together indicated by pollen and megafossils (Hy- with occasional Alnus , the pollen of Quer- värinen, 1975; Eronen and Huttunen, 1987). cus, Corylus, Ulmus and Tilia are rare, Pi- However, there is an interesting point in cea appears in the upper part, but the per- the data provided by tree-ring chronolo- centage is still 1 – 2 %; III Pine-Picea -phase gy: it seems that the pine tree-line retreat- ( circa 2500–) with the increase in the per- ed in north-western Finnish Lapland about centage of Picea to 10%, Alnus is weaker 70 km and at the same time moved 100 m than in phase II. Several reconstructions lower on the mountain slopes (Eronen and for the climatic development and the tem- Huttunen, 1987; Eronen et al ., 1999a; perature pattern during Holocene from the 1999b), but the shift was less pronounced Finnish Lapland are given in Fig. 10. in more easterly parts of Lapland (Eronen Hyvärinen and Alhonen ( 1994) studied et al ., 2002 ). two lakes from western Lapland and found Helama et al. (2002 ) have been inter- a dry period between 8000– 6000 14C yrs preting the climatic signal from the tree- BP when one of the lakes, Lake Jierstivaara ring master curve (July temperatures). The may have been almost dry. This was shown warmest 30-yr periods were 1471– 1442, both by the relationship of cladoceran 3192– 3163 and 461 – 432 calendar yrs BP. plankton/littoral and the diatom compo- The coldest 30-yr periods were 7242– 7213, sition. Thus the pollen composition of the 7152– 7123 and 5712– 5683 calendar yrs BP. same lake indicated a dry period at the If the data was tuned to show 100-yr peri- same time (Mäkelä et al., 1994 ). ods, the picture was slightly different. The According to Korhola et al. (2002 ) warmest 100-yr periods were 501 – 402 , abrupt climatic changes were particularly 2602– 2503 and 2302– 2203 calendar yrs characteristic for the early Holocene peri- BP and the coldest 100-yr periods were od. Based on Bayesian modeling of chi- 7202– 7103, 4502– 4403 and 3502– 3403 ronomid data they have detected cooling calendar yrs BP. episodes around 9200, 8600 and 8200 The tree-ring data, available at present, cal. yrs BP. Their data indicates also that demonstrates that the recent climatic var- the interval from circa 8000 to 5800 cal. iations such as the MWP and the LIA do yr BP appeared to be warm and stable, but not stand out as distinct mild and cool pe- was followed by a pronounced climatic 32 riods respectively (Briffa et al ., 1990; Zet- cooling around 5800 cal. yrs BP leading to a relatively long-lasting mid–Holocene 2.5. Present day climate and temperature minimum between circa 5800 recorded meteorological and 4000 cal. yrs BP. They have not ob- data served any particular changes in the latter At present the snow cover melts from May part of the Holocene, which may be part- 25th– 31st (1955– 73) and the arrival of the ly due to gradual sedimentation and thus permanent snow cover takes place from shallowing of the lake. On the other hand, October 25th– 31st (1954/ 55– 1972/ 73). The based on diatom data Korhola et al. (2000) annual rainfall is 500 mm at Ijjävri and have suggested that a major cold episode 500 – 550 mm at Muddusjävri (1931– 60) ac- occurred over a period of several hundred cording to the homogenized data (Atlas of years, centered at around 7200 cal. yrs BP. Finland, folio 131, climate, 1987). The rain- Subsequently, Helama et al. (2002 ) have fall during the summer months is present- detected the coolest centennial period of ed in Table 4 . The Inari, Toivoniemi weath- the tree-ring data to be between 7210– 7111 er station is located between the two dune calendar yrs BP. Closer to the modern cli- fields, at the northern end of Lake Mud- mate dilemma: response to global warm- dusjävri (lat. 69˚ 41’, long. 27˚ 07’). AtToivo- ing has been recorded in high-arctic eco- niemi ( 1961– 1990) the mean annual rain- systems in the post eighteen century sedi- fall was 432 . 9 mm and relative humidity ments (Douglas et al. , 1994). 72% (at 2 pm, annual mean).

Figure 10. Summary of the Holocene mean July temperature reconstructions from Finnish Lapland (according to Kultti, 2004) . 33 Table 4 . Rainfall during the summer months ing at the monthly data, it is apparent, that 1931–60, homogenized data. the wind system changes seasonally. The difference can be seen as the quite constant Month Muddusjävri Ijjävri summer month data is compared with the mean annual data (Fig. 11). Northern and June 45–50 mm 40–45 mm July 60–65 mm 55–60 mm northeastern winds prevail during the August 60–65 mm 55–60 mm summer months (with mean velocities of 4 . 1 to 3 . 7 m/s for N winds and 3 . 4 to 3 . 0 m/s for NW winds), while SW winds prevail during the winter and autumn. In the study area the mean annual tem- The wind rose from the meteorological perature is – 1 . 3 ˚C (FMI, 1991, referring to station of Inari, Ivalo airport is presented the data 1961– 90). The coldest mean in Fig. 11. The airport lies 46 km SSE from monthly temperature (January) at the Muddusjävri dune field and 86 km SSE study areas is – 19. 6 ˚ C (FMI, 1991, referring from Ijjävri.Wind data is presented for the to the data 1961– 90) and the mean Janu- summer months (June, July and August) ary temperature ( 1961– 90) is –14. 4 ˚C at the with the mean annual wind data (FMI, Toivoniemi weather station, Inari (FMI, 1991, referring to the data 1961– 90). Look- 1991).

34 Figure 11.Wind rose data from Inari, Ivalo airport (1961– 1990). See text for discussion. 3. Methods

3.1. Field work order to allow a larger scale for the inter- Field work for this study was carried out pretation. The available topographical during the summers of 1988– 1993 . Allto- maps, at a scale 1 : 20 000 (3823 07, 08, 10, gether, 21 localities ( 44 sections) in the 11) from the Muddusjävri dune field, were Muddusjävri dune field and 26 localities not very accurate for the purposes of this ( 31 sections) in the Ijjävri dune field were study because their contour interval is 5 m. studied. From the Ijjävri dune field, a topographi- Both dune fields were studied in detail. cal map, scale 1 : 50 000, ( 3931 02 and An attempt was made to reach every limit 3842 02) was available, but again with a of a dune field; the dune sections studied contour interval of 5 m. For this reason, were selected in order to achieve both spa- stereographs were found to be most appro- tially and temporally representative cover- priate. age of the area. Temporal aspect was The morphology of the dunes was stud- achieved by digging the dune excavations ied in the field for those dunes that were deep enough into the underlying glacioflu- also stratigraphically studied in detail. For vial or delta sediments. These older sedi- this purpose, levelling was carried out and ments were exposed, sampled and dated. cross-sections (where available) were also Spatial coverage was ensured by examina- measured. The area of the Muddusjävri tion of the aerial photographs and by field dune field is privately owned. A number studies. of tracks and roads have been built across the dunes. Often it was possible to use a 3.1.1. Mapping and geomorphological road cutting as a base for the cross-section studies of a dune. Selection of the dune fields was made us- Mapping of the Quaternary deposits of ing the 1 : 400 000 map of the Quaternary the areas was undertaken to determine the deposits in Northern Lapland (sheets n:o full extension of the dune fields and to 38+ 48 Inari and n:o 39+ 49 Utsjoki). Mon- study the local deglaciation history, mainly ochrome black and white aerial photo- to provide a context for the source area of graphs at a scale of 1 : 31 000 from the Mud- the dunes. Some till fabric analysis were dusjävri dune field and 1 : 60 000 from the done from the surrounding till areas. In Ijjävri dune field were used as stereographs addition, at Muddusjävri, where the dune to study the morphology of the dune fields. field overlies a delta, some palaeocurrent Enlargements of the aerial photographs measurements were made from the delta from Ijjävri dune field were also used in sediments. 35 3.1.2. Sedimentological observations 3 .tractional deposition producing sub- Relatively few comprehensive studies have critically climbing translatent stratifica- focused on detailed lithofacies characteri- tion, supercritically climbing stratifica- zation of modern dunes, although the in- tion, ripple foreset cross-lamination, ternal geometry of ancient aeolian sand rippleform lamination or planebed bodies has been extensively investigated. lamination. This largely reflects the difficulty of obtain- Steeply dipping ( 32– 34˚) foresets are pro- ing suitable sections in dry, unlithified duced mainly by grainflow deposition on dune sands (Pye and Tsoar, 1990). Actual- the slip face. They correspond to the en- ly, most work on modern dunes has been croachment deposits recognized by Bag- based on excavation of shallow pits and nold ( 1938 ). Grainflow deposits formed in trenches following drenching of the sand dry sand consist of a series of tongue- with large quantities of water ( e.g. McKee shaped bodies, typically 2 – 10 cm thick and and Tibbits, 1964), and is concentrated on 5 – 30 cm wide. They often truncate under- small dunes (height less than 10 m). Re- lying foreset laminae at low angles. In the cently, ground penetrating radar has been upper part of the flow the grain size of in- used to produce three-dimensional imag- dividual sand tongues coarsens upwards es of sedimentary structures of linear since larger grains are moved towards the dunes in the Namib sand sea (Bristow et top of the flow owing to disperse stress al., 2000). Internal structure of stagnant (Bagnold, 1954; Sallenger, 1979). The grain dunes provide a means of reconstructing size may also increase in the downflow di- palaeowind directions and testing global rection since the coarser surface grains roll circulation models for earlier periods in faster and further than the finer grains be- Earth history ( e.g. Bigarella and Salamu- neath (Kocurek and Dott, 1981; Pye, 1982 ; ni, 1961; Parrish and Peterson, 1988). The Pye and Tsoar, 1990). Individual foreset postglacial dune fields in northern Lapland layers, which are usually 2 – 5 cm thick, are offer moist, well preserved sections for grouped together into crossbed sets which studies of dune stratigraphy and as such are defined by bounding surfaces. The lat- they are rarely described. ter may be planar or curved, while cross- Sedimentological observations were bed sets may be either tabular, wedge- collected according to Hunter ( 1977). He shaped, or trough-shaped in cross-sec- recognized three groups of processes which tion. are responsible for the formation of pri- Especially in small dunes grainflow mary aeolian sedimentary structures: laminae frequently alternate with inclined 1 .grainflow deposition producing (sand) grainfall laminae. Grainfall deposition is grainflow cross-stratification (renamed typical in zones of flow separation, as the by Kocurek and Dott, 1981), character- saltating grains which pass over the dune ized by high angle (typically 28– 34˚), crest accumulate on the lee side zone of thick strata ( 2 – 5 cm) and sharp contact reduced wind velocity (Hunter, 1977). (erosional or nonerosional). These deposits lie concordantly on the ex- 2 .grainfall deposition producing grainfall isting dune lee slope topography. Grainfall lamination, characterized by interme- lamination includes deposits formed by a diate angle (typically 20– 30˚, variation complex of processes including pure grain- 0 – 40˚), individual thin ( 1 – 3 mm) lam- fall, i.e. grainfall includes any mode of dep- inae, sets of intermediate ( 1 – 10 cm) lay- osition excluding grainflow and climbing ers and sharp or gradational contact, ripple migration. Grainfall cross-strata can 36 nonerosional. only be preserved if the angle of the lee slope is insufficient to cause complete re- syn- and post-depositionally. For the working by avalanching, and if the lower stratigraphical description of the studied lee slope deposits are not reworked by rip- dune sections the most relevant secondary ples migrating across the dune slope (Pye features are denivation structures and and Tsoar, 1990). forms produced by niveo-aeolian processes Primary sedimentary structures formed and melt waters. Presence of niveo-aeolian by tractional deposition like climbing rip- processes has affected the sedimentation of ple strata are formed by the migration of the dune fields from the periglacial time ripples under conditions of net deposition and is still a significant factor. Niveo-aeo- ( e.g. Hunter, 1977). The angle of ripple lian deposits are mixed deposits of wind- climb varies depending on the rate of rip- blown snow and sand, silt or other detri- ple migration and sedimentation. The pre- tus which form in climates that are cold fixes sub- and super refer to the angle of at least seasonally (Koster, 1988; Koster and ripple climb. Climbing ripple structure Dijkmans, 1988; Dijkmans, 1990). The phe- may be composed of wavy layering paral- nomenon is annual, with all the snow dis- lel to successive rippled depositional sur- sipating in summer, but in exceptionally faces (rippleform laminae) or of even lay- cold environments, such as Antarctica, per- ering parallel to the vector of ripple climb ennial niveo-aeolian deposits also occur (translatent strata) (Hunter, 1977). Subcrit- (Calkin and Rutford, 1974). The surfaces of ical climbing translatent strata are the most fresh niveo-aeolian deposits are frequent- commonly found type in aeolian sedi- ly covered by mixed sand-snow ripples. In ments, whereas cross-laminated ripple summer, denivation forms are produced by foresets are rarely visible unless they are melting or sublimation of the snow. Fea- abnormally thick (Pye and Tsoar, 1990). tures which are produced include sand pel- Topsets forming at the windward side of a lets and sand rolls, surface cones and hill- dune are mainly climbing ripple strata. ocks with a surface of cracked sand, and In addition to top- and foresets, bot- sink holes surrounded by tensional cracks tomsets are part of the aeolian environ- (Koster and Dijkmans, 1988). After comment. They are produced by tractional plete melting of the snow, the sand cover deposition and formed in connection with dries out and is largely reworked by the very strong winds affecting the dune crest. wind. If the advance rate of the dunes is Some of the finer grains are blown beyond low, the preservation potential of the deni- the downwind limit of the slip face and vation features is low (Dijkmans, 1990). accumulate as a low-angle sand platform This applies especially to the re-deposited several metres wide. As the dune advanc- surface unit at Ijjävri dune field. It has been es, these deposits are buried, forming fine- observed that the deposition of niveo- grained, horizontal to gently leeward-dip- aeolian beds occurs mainly in early win- ping bottomsets. Actually, as planebed ter, before a continuous snow cover devel- lamination forms at wind velocities too ops (Dijkmans, 1990). Dijkmans ( 1990) has high for ripple formation, they are found studied the preservation potential of deni- rarely in aeolian dune deposits (Hunter, vation forms with field experiments and 1977). Horizontal or low-angle planar lam- micromorphological analyses and con- inated sand are common in interdune area cludes that the intercalation of snow dur- and sand sheets, but not all are true ing sedimentation of sand and loess does planebed deposits (Pye and Tsoar, 1990). not in itself induce a laminated structure. In addition to primary structures, there It is also concluded that periglacial niveo- are secondary structures which develop aeolian transport has been a more impor- 37 tant process than is suggested by sedimen- order to avoid contamination. In most cas- tological record of the Weichselian cover es the sections were moist and this made sands. However, the presence of snow-re- it easy to work on them. Basically two types lated deformation structures like, dimpled of samples were taken: sand samples and surfaces, infilled cracks, strongly faulted charcoal samples. zones, chaotic bedding and rotated brec- Samples of sand were taken first at three cias in the sediments indicates cold-climate uniform depths in the Muddusjävri dune aeolian sand formation. field: 100– 110 cm, 60– 70 cm and 30– 40 cm. The internal structure and stratigraphy Later as the study continued, in addition of each dune was initially studied at three to these “bottom, middle and top” samples, locations: the windward side, the crest samples of special interest were also col- (top) and the lee side. At the Ijjävri dune lected. The latter were especially those field, in particular, it was not always pos- from sand units either above or beneath sible, however, to study the dune at all erosion surfaces, as well as those of spe- three locations because of the frequency of cial grain composition. Particularly at blow outs. Thus, it was considered that the Ijjävri, where the sections were fairly high material of a dune did not vary very much and their stratigraphy rather complicated, from one side to another and that the most up to 8 sand samples were taken from the complete series of charcoal horizons were same section for various analyses. Samples to be found on the original lee side of the of approximately 1 kg were taken from a dune (Kotilainen, 1990). In view of these square of circa 10 x 10 cm using a trowel. observations, the dunes in the second The dip of the strata was taken to account study area, the Ijjävri dune field, as well as when sampling in order to have the sam- in further studies on the Muddusjävri ple of a certain sand unit and the sample dune field, were mainly studied at only one was placed in a plastic bag. location: the lee side. To investigate the Samples for thermoluminescence (TL) sediments excavations were dug to 1 . 8 m and optical luminescence (OSL) dating depth in the Muddusjävri dune field and were taken using a 50 cm long black plas- to 2 . 0 m depth in the Ijjävri. Photographs, tic tube with a diameter of 5 cm. The tube drawings and notes of the sections were was hammered into the section carefully, taken. The dip of the strata was also meas- parallel to the dip of the strata, in order ured in three-dimensions. to avoid contamination across an erosion Interdunes are an integral part of the surface. The tube was then removed from aeolian bed form system and must be tak- the section, sealed and placed in an opaque en into account when attempting to char- bag. Both TL and OSL dates were made acterize, or determine the nature of the sys- from each sample. Later additional sam- tem (Ahlbrandt and Fryberger, 1981; Kocu- ples were collected in a more effective way, rek, 1981). The interdune areas and defla- according the instructions of The Dating tion basins especially in the Ijjävri dune Laboratory in the University of Helsinki. field were described and studied. In Mud- This was to collect the samples directly to dusjävri the dune field forms a net of par- an opaque bag with a trowel carefully re- abolic dunes (see Chapter 4 . 1 . 1 ), and the role moving the outside layers and acting of interdune areas is much more limited. quickly to avoid exposure to light. The size of each sample was approximately 1 kg. 3.1.3. Sampling Samples of the charcoal layers for radi- Sampling of the sections was made from ocarbon dating were collected from sub- 38 the bottom to the top of the exposure in horizontal bedding planes. Where a sequence of charcoal layers occurred, the Allen, 1981). However, four main methods uppermost layer was exposed first over an are a. sieving, b. settling tube analysis, c. area of approximately 1 m2 . The pieces of electro-optical methods, including laser charcoal (approx. 5 x 5 mm) were then col- granulometry and d. computerized image lected with tweezers and placed in a plas- analysis. All methods of grain size deter- tic bag. Later, because the samples tended mination have disadvantages and the to be too small, a sieve was found to be choice of the most appropriate method is useful for separating the sand from the governed by the nature of the sample and charcoal pieces. It was then possible to col- the usage to which the data are intended lect the whole layer of charcoal with a (Pye and Tsoar, 1990). The definition used trowel and separate charcoal from sand in this study is called d A , sieve diameter, the and roots afterwards in the field base when width of minimum square aperture the sample was dry. through which the particle will pass. Dry sieving is the most widely used method 3.2. Grain size distribution and most appropriate for the sand samples and parameters of this study. The disadvantages attached The movement of grains in the sedimen- to this method are not valid for the study tation process is governed partly by size, material, since the sand grains originate shape, and density of the grains and part- from crystalline bedrock and their shape ly by physical properties of the transpor- can be compared to a sphere. In addition tation medium. Entrainment of grains to dry sieving, for three samples from the from the bed is influenced not only by the fine deposits beneath the dunes, the grain characteristics of individual grains, but size distribution for the fraction of finer also by bulk sediment properties which than Φ 3. 99 (0 . 063 mm) was studied us- include the grain distribution (sorting), ing a sedigraph at the Geological Survey orientation, packing arrangement, poros- of Finland. ity and cohesion. During transportation, The grain size distribution of a sample grains are sorted according to size, shape, is continuous within a certain range. There and density, and may undergo changes in are four basic graphic methods to describe shape due to inter-particle collisions or the distribution; a histogram, a frequency contact with the bed. An understanding of distribution curve, a cumulative curve and the physical characteristics of sand grains a logarithmic probability curve. The log- and the manner in which the characteris- probability curve has been widely used for tics of grain populations are modified dur- aeolian material: the original grain popu- ing aeolian transport is therefore essential lations produced by the sorting processes for correct palaeoenvironmental interpre- in entrainment and transport can be dis- tation of aeolian sediments (Pye and Tsoar, tinguished as separate segments of the 1990). curve (e.g. Visher, 1969 ; Middleton, 1976 ). Grain size can be specified and meas- In order to simplify the graphical pres- ured in several different ways. Indications entation and statistical manipulation of of size can be obtained by measuring the grain size frequency data, Krumbein ( 1934) calibre dimensions of a particle, by deter- proposed that the grade boundaries should mining its volume or its mass, or by de- be logarithmically transformed into phi termining its settling velocity (Pye and ( Φ ) values, using the expression:

Tsoar, 1990). In fact, there are numerous Φ = –log2 d , where d is the diameter in different definitions of particle size and millimeters. Later the expression was re- thus formulas for determining them ( e.g. vised to the form: 39 Φ = –log2 (d/d0 ), where d 0 is the stand- , where Φ 95, Φ 5 , Φ 75 and Φ 25 are the phi size ard grain size of 1 mm (McManus, 1963 ). values corresponding to the percentiles, re- The dry sieving divides the material spectively, read from the cumulative fre- into certain grain size populations, conse- quency curve. Classifications platykurtic, quently enabling the use of parameters like mesokurtic (approximate Gaussian pro- sorting, mean size, kurtosis and skewness file) and leptokurtic are being used to de- to describe the material. The grain size pa- scribe the “peakedness”. Platykurtic distri- rameters used in this study are widely ac- bution is flatter than normal probability cepted and used for aeolian material (ac- curve and leptokurtic distribution is cording to Folk and Ward, 1957 ): strongly peaked ( Table 5 ).

Mean size, M z = (Φ 16 + Φ 50 + Φ 84) / 3 (1 ) Inclusive Graphic Skewness, Sk1 : Sk = [( Φ + Φ 2 Φ ) / 2 ( Φ – Φ )] + , where Φ , Φ and Φ are the phi size val- 1 16 84– 50 84 16 16 50 84 [(Φ + Φ – 2 Φ ) / 2 ( Φ – Φ )] ( 4 ) ues corresponding to the percentiles, re- 5 95 50 95 5

spectively, read from the cumulative fre- , where Φ 95, Φ 5 , Φ 75 and Φ 25 are the phi size quency curve. values corresponding to the percentiles, respectively, read from the cumulative fre- Inclusive Graphic Standard Deviation quency curve. Sk describes the asymmetry (IGSD), 1 of skewness of the distribution. Positive σ 1 = [(Φ 84– Φ 16) / 4 ] + [(Φ 95– Φ 5 ) / 6 . 6 ](2 ) values of Sk1 indicate that the distribution

, where Φ 84, Φ 16, Φ 95 and Φ 5 are the phi size has more pronounced tail of fine material values corresponding to the percentiles, re- compared with a log normal distribution.

spectively, read from the cumulative fre- Conversely, negative values of Sk1 indicate quency curve. Better sorted sediments have a tail of coarser particles or a deficiency of

lower values of σ 1 (Table 5 ). fine particles compared with the log-normal distribution.The classification of the Inclusive Graphic Kurtosis: graphical grain size parameter values is K = ( Φ – Φ ) / 2 . 44 (Φ – Φ )(3 ) G 95 5 75 25 summarized in Table 5 .

Table 5 . Terminology applied to graphical statistical parameter values (originally Folk and Ward, 1957 ; modified by Pye and Tsoar, 1990).

Inclusive graphic standardInclusive graphic Inclusive graphic

deviation ( σ 1 )skewness (Sk1 )kurtosis ( K G ) Very well sorted< 0.35 Very positively Very platykurtic <0.67 skewed +0.3 to +1.0 Well sorted0.35–0.50 Positively skewed +0.1 to +0.3 Platykurtic 0.67–0.90 Moderately wellSymmetrical +0.1 to –0.1 Mesokurtic 0.90–1.11 sorted 0.50–0.70 Moderately sorted0.70–1.00 Negatively skewed –0.1 to –0.3 Leptokurtic 1.11–1.50 Poorly sorted 1.00–2.00 Very negatively Very leptokurtic 1.50–3.00 skewed –0.3 to –1.0 Very poorly sorted2.00–4.00 40 The moment parameters are analogous tion, skewness and kurtosis, based on the to the graphical statistical parameters. The formulas by Folk and Ward (1957 ). Grain use of moment statistics (Friedman, 1961; size distribution was presented as a cumu- McBride, 1971 ) is widely considered to be lative curve, a logarithmic probability more representative, because they take into curve and a frequency curve. account the entire grain size distribution rather than just the part between the fifth 3.3. Grain shape; roundness and ninety-fifth percentiles. However, in and surface texture this study, the graphical statistical param- The terms grain “shape” and “form” have eters are being used because of the nature been used in literature in a variety of ways, of the material and the aims of the study. but here, in this study, the term“shape” The sorting is very good (grain popula- includes all aspects of external morphol- tions in the first and the last fifth are very ogy of a grain, including the gross form small) and the results can be correlated to (sphericity), the roundness (sharpness of previously attained results from Finland edges and corners), and the surface texture and elsewhere in subarctic environments. (roughness or smoothness and further on: imprints of different transportation 3.2.1. Winnowing agents). Methods for determining the grain Sand samples were dried in a incubator at shape are numerous. They include quali- 100 ˚C overnight and allowed to cool to tatively descriptions in terms of resem- room temperature. In most cases it was not blance to recognizable geometric shapes or necessary to pre-treat the samples to re- organic analogues, using terms like cubic move organic material. ASTM series sieves or conical. The weakness of such methods 1 with an interval of approximately / 2 Φ is their subjective nature. To avoid this, “a units were used for electric dry sieving of training set” of grains, classification for the samples. The sieves used were: Φ – 0 . 49 grain roundness was introduced by Pow- ( 1 . 400 mm), Φ 0 . 00 (1 . 00 mm), Φ 0 . 49 ers ( 1953). This visual comparison of grain ( 0 . 71 mm), Φ 1 . 00 (0 . 500 mm), Φ 1 . 74 morphology allows quick, repeatable and ( 0 . 300 mm), Φ 2 . 00 (0 . 250 mm), Φ 2 . 47 reliable classification of grains and has ( 0 . 180 mm), Φ 3 . 00 (0 . 125 mm), Φ 3 . 74 widely been used. Furthermore, a statisti- ( 0 . 075 mm) and Φ 3 . 99 (0 . 063 mm). This cally operable roundness scale was adopt- series of sieves was found to be satisfacto- ed to the visual classification. In this study, ry in order to achieve a reliable grain-size the visual classification for the grain distribution (Folk, 1966; Simola, 1979). The roundness was undertaken using the Pow- sample size selected was 400 g.All togeth- ers index (Powers, 1953; Shepard, 1963 ; er 249 samples were dry sieved. In addi- Fig. 12). tion, for 3 samples from the fine deposit A number of numerical parameters present beneath the dunes, the grain size have been developed to enable quantitative fraction of less than 3 . 99 (0 . 063 mm) was descriptions of grain shape. These meth- studied using a sedigraph at the Geologi- ods require measurements of grain axes cal Survey of Finland. and e.g. curvature radius of individual grain corners (Wadell, 1933; 1935 ). Meas- 3.2.2. Computing urements of this type are time consuming To investigate the graphical parameters of and thus subjective, as long as the meas- the grain size data a computer program uring is hand made. There have been many was used, which determined mean size, attempts to avoid human error, for in- mean diameter, sorting, standard devia- stance by determining the grain roundness 41 on dividing the average of the radii of the corners of the grain image by the radius of the maximum inscribed circle (Wadell, 1935). Also the method evolved for this work, Scanning Electron Microscopy (SEM) -analysis of quartz grains, empha- sises the idea of combination. Randomly selected areas of individual grains are visually divided into areas of different imprint classes and the areas are digitally transformed to quantitative proportions of the whole area.

3.3.1. Binocular microscopy Only the quartz grains were used for both the grain roundness and surface texture (glossy/matt) analyses with a binocular microscope. The grains were visually classified into six categories; very angular, angular, sub-angular, sub-rounded, rounded Figure 12. Chart for visual estimation of roundness according to Powers ( 1953) and and well rounded (Powers, 1953; Shepard, Shepard (1963 ). Classes A = very angular 1963 ). The classification is presented in Fig. (class intervals 0 . 12– 0 . 17, geometric mean 12. In addition, the quartz grains were di- 0 . 14), B = angular (class intervals 0 . 17– 0 . 25, vided in two groups: glossy and matt par- geometric mean 0 . 21), C = sub-angular (class ticles. In each analysis one hundred grains intervals 0 . 25– 0 . 35, geometric mean 0 . 30), were counted. All the quartz grains in a D= sub-rounded (class intervals 0 . 35– 0 . 49, field of view were counted each time to geometric mean 0 . 41), E = rounded (class avoid selection. intervals 0 . 49– 0 . 70, geometric mean 0 . 59) and A total amount of 225 samples were an- F = well rounded (class intervals 0 . 70– 1 . 00, geometric mean 0 . 84). alysed, 136 from Muddusjävri dune field and 89 from Ijjävri. Most of the samples were dune sand, but also samples from sed- by its rolling ability. This was measured iments associated with the dune field; es- with a graniformameter (Krygowski, 1964 ; ker, delta, fluvial and littoral deposits were Seppälä, 1971). Also the applications of analysed. image analyzing systems incorporated into Roundness and surface texture of the microscopy work are aiming to avoid sub- sand grains were investigated using a low jective errors. However, these automatic power, binocular microscope. The fraction methods have complications of their own, used for these analyses was Φ 0 . 49 (0 . 710 and those are very difficult to control. mm) – Φ 0 . 00 (1 . 00 mm). This fraction Consequently, it appears that most success- was found to be most suitable for this ful method would be a controlled subjec- roundness and surface texture studies, be- tive approach demonstrated by a combi- cause grains are known to become less nation of visual and measured methods. rounded as the grain size decreases (Car- Actually the Powers index is a combination roll, 1939, p.16– 17). The effects of both of methods because class intervals and ge- wind and water are also most visible in 42 ometric means on Powers index are based grains of 0 . 7 mm in diameter (Cailleux, 1952 , p. 18 ). On the other hand, in this grain mate and potassium permaganate dis- size the grains tend to be represented by a solved in 15 ml of concentrated sulphuric single mineral although there still are some acid (McIntyre and Be, 1967 ). The sample rock fragments among them. In addition, was then again washed with distilled wa- a series of all the grain size fractions from ter and dried in the same way, without us- one sample was studied. ing ultrasonic vibration, which might damage the surface (Porter, 1962 ). The 3.3.2. Scanning Electron Microscopy sample was then stored in a petri dish. Scanning Electron Microscopy (SEM) has A total amount of 15 to 20 grains were been found to be a useful method for grain then selected from the sample at random surface texture studies. Krinsley and under a binocular microscope for SEM Doornkamp ( 1973) demonstrated that viewing. Fifteen is usually an adequate quartz sand surface features reflect varia- number in a study of the variability present tion in sedimentation processes and envi- in a single sample (Krinsley and Doorn- ronments. Each sedimentation cycle leaves kamp, 1973). The line method (Galehouse, imprints on the surface of a single grain. 1971) was used to ensure random selection More recently Mahaney ( 2002 ) has updat- of the grains. The grains were mounted on ed the “Atlas of sand grain surface textures” a metal specimen plug, which was coated with a large variety of examples. SEM anal- with double sticky tape. The grains were ysis was applied to this study in order to then placed in rows on the plug and the gain more data of the sedimentary units first was marked by cutting a small notch in addition to grain size parameters for into the tape. This was done to ensure the litofacies analysis and correlation. possibility of returning to a single grain of SEM work was carried out in two parts. interest later if needed. Some of the analyses were undertaken at The plugs were then coated with gold the Department of Electron Microscopy, in a vacuum evaporator using a Jeol FINE University of Helsinki and later the work COAT JFC-1100 sputtering device. Gold is was completed at the Department of Earth found to give marginally better resolution Sciences, University of Cambridge. In both than coal or platinum-palladium (Krins- places the same type of a microscope was ley and Doornkamp, 1973). Coating is done used, a Jeol JSEM-820 scanning electron to make the sample to emit more second- microscope. ary electrons, to ensure low ionisation of the sample and to protect the sample from 3.3.2.1. Preparation of the samples the electron beam. The vacuum evapora- The fraction used was the same that for the tor coats the sample with a gold coat of 10– binocular microscopy: Φ 0 . 49 (0 . 710 mm) 50 nm. This coating does not mask the – Φ 0 . 00 (1 . 00 mm). Only the quartz viewing of the samples because the gold grains were studied. The samples were pre- particles can only be seen at a magnifica- treated as follows (Krinsley and Doorn- tion of x 50 000. kamp, 1973): 5 g of the sample were boiled in concentrated HCl for 10 minutes and 3.3.2.2. Producing the images rinsed with distilled H 2 O. The samples The grains were studied in detail under the were then allowed to dry slowly in an in- SEM microscope. An observation form for cubator at 30 ˚C overnight. If the sample each grain was used during the viewing to contained organic matter, the HCl treat- make notes of some features, which had to ment was followed by treatment with a be studied with a greater magnification mixture of 1 . 5 g each of potassium dichro- than x 250 in order to be identified. For 43 of 250 x. However, the magnification of the negative of the image was 0 . 6 x and because the paper print was 3 . 1 x the magnification of the negative, the final magnification of the paper print was 465 x the original part of the grain. The size of the paper prints used was 19. 3 cm x 14. 2 cm in Helsinki and 20. 4 cm x 15. 2 cm in Cam- bridge. The visual analysis was carried out using a magnifying glass and the outlines of areas of different features were drowned on transparent foil. Areas not in focus were excluded. All together 15 codes were used to describe the detected quartz grain sur- Figure 13. The photographing system for the face features. grain assemblages. Numbers indicate the parts enlarged and interpreted from each grain 1 – 15 of a sample. 3.3.2.4. Quantitative analysis The quantitative analysis was carried out using a computer program, written by Si- mon Crowhurst (Sub-department of Qua- quantitative analysis, the grains were pho- ternary Research, Godwin Laboratory, tographed at the magnification of x 250. Cambridge). The program is written in This provided a view of only approximate- Microsoft Quickbasic to count the areas ly 1 / 9 of the whole grain. To ensure that associated with different surface patterns the photographs were not taken selective- on grains of sand. Procedurally, there are ly from surface textures that caught the eye two main steps involved: collecting the in the first place, an order of photograph- outlines of the shapes using a “bit pad”, and ing the grains was developed ( Fig. 13). The counting the areas within the outlines, whole grain was also photographed at x 60 once these have been fully classified into magnification. 10– 15 grains were studied texture types using a color code. The x y and photographed from each sample. and color data were written intoASCII files and stored on diskette for later analysis. 3.3.2.3. Classification of the surface The area counting routine reads the col- features lected outlines sequentially and draws The surface features encountered were them onto a defined area of a high resolu- classified according to Krinsley and tion (VGA) screen. Then it records the ar- Doornkamp ( 1973) and Al-Saleh and Kha- eas of different colours as the number of laf (1982 ). The features present of the pixels of each particular colour present quartz surface textures were: 1 . Precipitat- within the drawing area of the screen, and ed upturned silica plates, 2 . Conchoidal expresses the proportions as percentages of fracture, 3 . Mechanical V-forms, 4 . Straight the total grain area of the image. The or slightly curved grooves, 5 . Dish-shaped counting routine expressly excludes areas concavities, 6 . Deep surface solution, 7 . Flat that are recorded as background (white cleavage face, 8 . Cleavage planes (semi-par- colour) as well as the outlines of the allel lines), 9 . Adhering particles. shapes. The latter may, however, cause a The visual analysis was done using black systematic tendency to under estimate the 44 and white SEM images, at a magnification size of the very small or narrow shapes. The collected counts are simultaneously study, that the variations of the mineralog- printed onto paper and onto text files. Pro- ical composition within any individual cedurally the technique is quite similar to dune were small, whereas the real differ- cutting and weighing cardboard shapes; ences were regional (Kotilainen, 1990). however, in practice it turns out to be less In addition, some special sediment time-consuming and laborious. It does, structures were sampled by pressing a plas- however still take time – each sand grain tic container on the polished strata. Air was map requiring some 30 minutes in total to slowly pumped out and displaced with a collect on the bit pad and process. fixing resin, so that the structure was hardened. Thin sections of such fixed strata 3.4. Mineralogical were prepared and photographed. composition The chemical composition of garnet The bedrock of Muddusjävri dune field grains was found to be of special interest. belongs to the Paleoproterozoic Lapland In order to determine the composition, Granulite Belt and the Ijjävri area is a part four (two from each field) polished, un- of the Granite Gneiss Complex. A detailed coated thin sections were prepared for mi- description of the rocks in the vicinity of croprobe analysis. These analyses were un- the dune fields is given in Chapters 2 . 1 . 1 dertaken at the Geological Survey of and 2 . 2 . 1 . The approach to the petrograph- Finland. Chemical composition of 117 gar- ic analysis of the sediments was four-fold: net grains was analysed. 1 . to establish the mineralogical composition of the dune sediments at different 3.5. Dating methods parts of the fields, possibly in relation to The dating methods adopted in this study the transportation distance (Muddusjävri) consist of conventional and accelerator and in relation to the parent materials; 2 .to mass spectrometry (AMS) radiocarbon analyse the mineralogical composition at dating, optically stimulated luminesence different parts over a single dune; wind- (OSL) and thermoluminesence dating ward, top (crest) and lee side; 3 . to deter- (TL). All of them except the AMS analy- mine the possible vertical variation of the ses were carried out at the Dating Labora- mineralogy at different sampling depths of tory, University of Helsinki. Pre-treatment a single dune; and 4 . to define the miner- of the samples for AMS analysis was done alogical composition of separate grain size in The Dating Laboratory, University of fractions of a dune sediment sample. Helsinki and the AMS analyses were per- Thin sections of the sand samples were formed in University of Uppsala, Sweden. prepared at the University of Helsinki, Department of Geology. The thin sections 3.5.1. Conventional and AMS were studied using a petrographic micro- radiocarbon dating scope. A total of 100 grains on each slide The two detectors used for conventional were identified and point counted. All to- radiocarbon dating in the Dating Labora- gether 52 thin sections were studied. Ini- tory, University of Helsinki are Östlund- tially samples from all sections were stud- Engstrand type detectors ( 1 . 0 l and 0 . 5 l of ied from two dunes: the windward side, the volume) with small modifications. Typical crest and the lee side. Samples were taken values for these detectors are for back- at two uniform depths from each part. ground 2 . 4 and 1 . 5 cpm and for modern Subsequently, one thin section per dune 13. 2 and 6 . 8 cpm respectively (Jungner, was prepared and analysed. This was be- 1979). The pretreatment for conventional cause it was concluded earlier in the M.Sc. radiocarbon dating of wood, peat, charcoal 45 and sediment samples is a standard alka- eration -method (Banerjee et al ., 1999). li-acid procedure using NaOH and HCl at Twenty-four aliquots of circa 10 mg each elevated temperature, the concentrations were measured for every sample. Six dif- varying in the NaOH treatment from 0 . 5 – ferent preheat temperatures ranging from 2 % and in the HCl treatment from 2 – 5 %. 180 to 280 ˚C for 20 s were used. This was Dates reported are based on 95% of the followed by 5 s of infrared and 100 s of activity of NBS oxalic acid and the Libby green-plus-blue stimulation at elevated half-life 5568 + 30 a (Jungner, 1979). ( 125˚C) temperature. For sensitivy correc- Pretreatment for the AMS analyses fol- tions a test dose of 500 mGy followed by lowed a standard procedure (Slota et al., preheat to 160˚C for 10 s before shine down 1987 ). AMS radiocarbon dating analysis (at 125˚C) was used. Three different regen- was performed using the Uppsala EN-tan- eration doses were used in increasing or- dem accelerator as an ultra sensitive mass der to bracket the natural dose. spectrometer (Possnert, 1984; 1990). For determation of the paleodose the dating laboratory used the OSL signal ob- 3.5.2. Thermoluminescence and Optically tained during the first 20 s of the shine Stimulated Luminescence dating down curve and subtracted the back- Thermoluminescence (TL-) and optically ground signal from final 20 s of shine stimulated dating (OSL-) results were ob- down. The paleodose obtained at the dif- tained at the Dating Laboratory, Univer- ferent preheat temperatures showed good sity of Helsinki. The method used there agreement and the final result was taken was an earlier, yet congruent version of the as the arithmetic mean of all these six val- method described in (Jungner et al ., 2001). ues. For both TL and OSL dating, feldspar The beta activity of the sand samples grains of 100– 200 µ m diameter were used. was measured on a RISØ GM-5 – 25 beta Normal pre-treatment including heavy liq- multicounter (Bøtter-Jensen and Mejdahl, uid separation and HF etching were ap- 1988) and annual doses were obtained by plied. comparing the beta activity to our own ref- All measurements were made on a Risø erence material assuming water content of TL-DA-12 reader (Bøtter-Jensen and Dull- 20%. The reference material consists main- er, 1992). For OSL measurements an exci- ly of Finnish glacigenic sediments and tation wavelength band between 420 and dune sands. The cosmic-ray dose rate was 560 nm and a detection window with two assumed to be 0 . 15 mGy/a for all sampling U 340 filters were used. Paleodoses were depths. determined using a single-aliquot-regen-

46 4. Results

4.1. Morphology of the dune er, no real peat deposited in these bogs, fields which could have been studied for pollen and macrofossils. Traces of later aeolian 4.1.1. Muddusjävri activity in a form of sandy interlayer were The dunes at the Muddusjävri dune field noticed, but the whole sequence was too are mainly parabolic dunes which have watery to be sampled by a Russian corer developed from the short transverse dunes (see Chapter 5 for discussion of the changes to increasingly bended ones. The apparent at the ground water table). presence of vegetation during their form- Not only the form of the dune itself is ing process has resulted in their deeply par- indicating the direction of the dune form- abolic shapes ( Fig . 14) at the NE-end of the ing wind. The profile of a dune is charac- dune field. The orientation of the dunes terized by a gentle windward (ww-) side can be seen in Fig. 15. The reworking of the and a steeper lee side (Dune 4 , Fig. 16 ). The material can also be seen here as bending top of a dune is often fairly rounded ( Fig. of the nose of the dune ( Fig. 25). The orig- 16), which is caused by the melt waters and inal dune building wind has been blowing erosion. from SW and the reworked orientation is W-E. The dunes are forming a fairly com- 4.1.2. Ijjävri plex parabolic net at the NE end of the The areal distribution of the dunes at dune field. Their eastern arms often are Ijjävri dune field is clearly different of that elongated and joined with other dunes. at Muddusjävri. The dunes are sparse and The simplified model of the dune form much bigger. They are characterized by transition presented by Lindroos ( 1972; huge blowouts scattering the ridge of the based on the models by Kádár 1938, Lands- dune. The role of vegetation in the defla- berg 1956 and Galon 1959) is adaptable to tion process is evident: Muddusjävri area Muddusjävri dune field as well. However, was covered with pine forest whereas in the development of the dune forms ends Ijjävri the vegetation belongs to the birch at the phase C since the longitudinal dunes scrub and woodland zone. The underlying are absent ( Fig. 23). till cover controls the morphology of the In the interdune area the most distin- dune field. There is no continuous dune guished features are the deep deflation ba- field at all, but the dunes are separated by sins, which are filled with water. Most of till areas ( Figs 17 and 18). them have been paludified after the rise of However, the morphology of the dunes the ground water level. There was, howev- themselves shows some the key elements 47 Figure 14.Aerial photograph of the Muddusjävri dune field. A typical dune with a parabolic shape is indicated with an arrow in the middle of the picture. Copyright Finnish Defence Forces Topographic Service, authorization no. 46/ 2004 .

Figure 15. The orientation of the dunes at the Muddusjävri dune field. Dune crests are indicated with lines. Original dune building wind direction is indicated with an arrow. Figure 16. Examples of the dune profiles in the Muddusjävri dune field. Co-ordinates according to the Finnish system, z = m a.s.l.)

similar to dunes in the Muddusjävri dune On long continuous dune ridges the field: the dunes are parabolic in nature, al- blowouts scatter the dune forming a string though many of them are still represent- of blowouts. The points, where the course ing phase A (transverse dune, see Fig. 23) of the dune alters, are characterised by veg- and the dune profile is characteristic with etated ridges ( Fig. 20). They have the form very steep leeside and gentle windward of the original dune top overlain by some (ww)-side. At Ijjävri the over- steepening accumulation sand, which has been trans- of the leeside is very pronounced (Fig. 19). ported along the blow out. Even if the A dip of the lee side is often as steep as 60° , main wind direction still was from the while the natural angle of repose of sand original windward side over to the lee side is 34° . of the dune, the deep blowouts have their 49 Figure 17.Aerial photograph of the Ijjävri dune field. Copyright Finnish Defence Forces Topographic Service, authorization no. 46/ 2004 .

Figure 18. The orientation of the dunes at the Ijjävri dune field. Dune crests are indicated with lines. own surface wind system with reverse air- flow. The situation compares with a dune/ interdune pattern on a larger scaled dune system. Thus here the separated dune sides act as individual linear dune ridges. On some occasions, the blow out has developed into a deep gully that has no ridge at the turning point. As the wind velocity slows down at the turn, sand is accumulating forming small hummocks ( Fig. 21). At the outlets of the deflation basins the sand is often accumulating in a tongue-like formation ( Fig. 22). However, here this phenomenon is enhanced and started by the influence of reindeer, not wind action. The outlets are originally trailheads made by the animals. Figure 19. Over-steepening dune lee side at the Ijjävri dune field.

51 Figure 20.Aerial photograph of a dune ridge at Ijjävri dune field. The dune is scattered by blow outs (white areas, the continuous one referred in the text as “the big deflation basin”). Notice the ridge at the turning point (indicated by an arrow). Copyright Finnish Defence Forces Topographic Service, authorization no. 46/ 2004 . Figure 21. Small accumulation hummocks at a turning point of a dune blowout at Ijjävri dune field. Photo: A. Kotilainen.

4.2. Grain size distribution 4.2.1. Muddusjävri dune field A total amount of 141 samples for grain size analysis, mainly from 22 dune sites ( 50 sections), were obtained from the Muddus- jävri dune field. In addition grain size for sediment types associated with the dune field; esker, delta, fluvial and littoral deposits were analysed. The average values of graphic grain size parameters for each dune are presented in Table 6 and for the whole data in Table 7 . The average values of grain size parameters for the whole data on different vertical sections (windward side, top and leeside) are presented in Ta- ble 8 .

4.2.2. Ijjävri dune field A total amount of 108 samples for grain size analysis, mainly from 26 dune sites ( 30 sections), were obtained from the Ijjävri Figure 22. Tongue formation at a dune dune field. Also samples from adjoining 52 blowout at Ijjävri dune field. fluvial sediments and deflation basins were Table 6 . Muddusjävri dune field, average graphic grain size parameters: Mean grain size, Mz , ( Φ ), Inclusive Graphic Standard Deviation for sorting, σ 1 ,

Inclusive Graphic Kurtosis, KG and Inclusive Graphic Skewness, Sk 1 from each dune (numbered 1–21 ). For the dune samples AVG is average and STD standard deviation, n is the number of samples from each dune.

Dune no M z ( Φ ) σ 1 K G Sk1 n (comment) AVG 1 2,82 0,40 1,12 0,13 6 AVG 2 1,94 0,74 1,01 0,01 7 AVG 3 2,30 0,57 1,02 0,05 6 AVG 4 1,91 0,77 1,01 0,03 7 AVG 5 2,23 0,67 1,00 0,10 9 AVG 6 2,01 0,64 1,10 0,19 9 AVG 7 2,54 0,74 1,06 0.00 9 AVG 8 2.00 0,64 1,09 0,17 10 AVG 11 2,30 0,60 1,00 0,10 1 AVG 12 1,88 0,70 1,15 0,18 4 AVG 13 1,83 0,73 1,07 0,13 3 (interdune) AVG 14 1,92 0,65 1,01 0,08 10 AVG 15 2,27 0,63 1.00 0,10 3 AVG 16 1,51 0,86 1,01 0,21 9 AVG 17 2,63 0,46 1,08 0,04 9 AVG 19 1,70 0,50 1,30 0,20 1 (interdune) AVG 20 2,56 0,54 1,03 0,02 9 AVG 21 1,84 0,69 1,04 0,14 14

STD 1 0,13 0,00 0,11 0,17 6 STD 2 0,18 0,12 0,06 0,03 7 STD 3 0,14 0,05 0,07 0,10 6 STD 4 0,08 0,09 0,03 0,09 7 STD 5 0,18 0,05 0,05 0,07 9 STD 6 0,16 0,05 0,09 0,06 9 STD 7 0,18 0,13 0,13 0,13 9 STD 8 0,08 0,08 0,08 0,05 10 STD 11 0,00 0,00 0,00 0,00 1 STD 12 0,18 0,14 0,18 0,08 4 STD 13 0,05 0,05 0,05 0,05 3 (interdune) STD 14 0,11 0,07 0,03 0,04 10 STD 15 0,09 0,05 0,00 0,00 3 STD 16 0,17 0,07 0,06 0,06 9 STD 17 0,09 0,05 0,04 0,05 9 STD 19 0,00 0,00 0,00 0,00 1 (interdune) STD 20 0,07 0,05 0,05 0,04 9 STD 21 0,22 0,10 0,10 0,08 14

53 Table 7 . Muddusjävri dune field, all the dune samples, graphic grain size

parameters: Mean grain size, Mz ,(Φ ), Inclusive Graphic Standard Deviation

for sorting, σ 1 , Inclusive Graphic Kurtosis, KG and Inclusive Graphic Skewness,

Sk1 . For the data AVG is average, MAX and MIN maximum and minimum values, STD standard deviation and VAR variability, n is the number of samples from the dune field.

: All dunes M z ( Φ ) σ 1 K G Sk1 n AVG 2,13 0,65 1,05 0,10 126 Muddusj. MAX 1,30 1,00 1,40 0,50 126 Muddusj. MIN 3,00 0,40 0,70 -0,30 126 Muddusj. STD 0,38 0,14 0,09 0,10 126 Muddusj. VAR 0,14 0,02 0,01 0,01 126 Muddusj.

Table 8. Muddusjävri dune field, graphic grain size parameters sorted by

sampling sections. Mww, Mtop and Mlee are samples from ww-side, top and leeside sections. AVG is average, STD standard deviation, VAR variability of the data and n number of samples from each section.

section M z ( " ) " 1 K G S k 1 n AVG Mww 2,17 0,66 1,05 0,06 37 AVG Mtop 2,05 0,62 1,02 0,14 41 AVG Mlee 2,14 0,66 1,04 0,09 44

STD Mww 0,40 0,14 0,09 0,11 37 STD Mtop 0,51 0,17 0,18 0,10 41 STD Mlee 0,33 0,13 0,08 0,08 44

VAR M ww 0,16 0,02 0,01 0,01 37 VAR Mtop 0,26 0,03 0,03 0,01 41 VAR Mlee 0,11 0,02 0,01 0,01 44

analysed. The average grain size parame- at the Muddusjärvi field, because on the ter values for each dune are presented in basis of preliminary work (Kotilainen, Table 9 and for the whole dune data in Ta- 1990), the emphasis at the Ijjävri dune field ble 10. The average values for the whole was placed on leeside sediments. Conse- dune data on different sampling sections; quently, a large number of samples ( 20) windward side, top and leeside, are pre- were collected from sections dug into de- sented in Table 11. The number of the sam- flation basins for grain size analysis. 54 ples from each section is not uniform as Table 9. Ijjävri dune field, graphic grain size parameters: Mean grain size, Mz ,

( Φ ), Inclusive Graphic Standard Deviation for sorting, σ 1 , Inclusive Graphic

Kurtosis, KG and Inclusive Graphic Skewness, Sk 1 from each dune (numbered 52–75). For the dune samples AVG is average and STD standard deviation, n is the number of samples from each dune.

Dune no M z ( Φ ) σ 1 K G S k 1 n (comment) AVG 52 2,60 0,53 1,10 -0,03 3 AVG 54 2,54 0,49 1,02 0,04 8 AVG 55 2,30 0,50 1,00 0,05 2 AVG 57 2,50 0,57 1,07 0,07 3 AVG 60 2,64 0,64 1,18 -0,12 5 (defl.basin) AVG 61 2,58 0,86 1,22 -0,22 5 AVG 62 2,40 0,60 1,10 0,00 1 (acc.hill) AVG 63 2,95 0,50 1,03 0,15 3 AVG 65 2,60 0,62 1,01 0,01 11 AVG 66 2,93 0,72 1,02 -0,05 6 AVG 67 2,37 0,70 1,03 -0,02 6 AVG 68 2,60 0,55 1,10 -0,05 2 AVG 69 2,15 0,65 1,00 0,05 2 AVG 70 2,20 0,80 1,13 -0,13 6 AVG 71 2,31 0,75 1,00 0,00 8 AVG 72 2,50 0,60 1,03 0,03 4 AVG 73 2,20 0,82 1,04 -0,08 5 AVG 74 2,46 0,75 1,07 -0,06 10 AVG 75 2,35 0,65 1,10 -0,05 2

STD 52 0,16 0,05 0,08 0,05 3 STD 54 0,12 0,03 0,04 0,07 8 STD 55 0,00 0,00 0,00 0,05 2 STD 57 0,08 0,09 0,05 0,05 3 STD 60 0,14 0,05 0,04 0,04 5 (defl.basin) STD 61 0,29 0,10 0,25 0,07 5 STD 62 0,00 0,00 0,00 0,00 1 (accum.hill) STD 63 0,11 0,00 0,04 0,05 3 STD 65 0,14 0,07 0,03 0,05 11 STD 66 0,31 0,13 0,11 0,11 6 STD 67 0,35 0,15 0,07 0,11 6 STD 68 0,10 0,05 0,00 0,05 2 STD 69 0,15 0,05 0,00 0,05 2 STD 70 0,36 0,16 0,20 0,11 6 STD 71 0,35 0,05 0,05 0,00 8 STD 72 0,14 0,07 0,08 0,08 4 STD 73 0,11 0,07 0,05 0,04 5 STD 74 0,34 0,11 0,05 0,07 10 STD 75 0,05 0,05 0,00 0,05 2

55 Table 10. Ijjävri dune field, all dune samples, graphic grain size parameters:

Mean grain size, Mz , ( Φ ), Inclusive Graphic Standard Deviation for sorting, σ 1 ,

Inclusive Graphic Kurtosis, KG and Inclusive Graphic Skewness, Sk 1 . For the data AVG is average, MAX and MIN maximum and minimum values, STD standard deviation and VAR variability, n is the number of samples from the dune field.

All dunes: M z ( Φ ) σ 1 K G S k 1 n AVG 2,47 0,65 1,04 -0,01 76 Ijjävri MAX 1,50 1,10 1,40 0,20 76 Ijjävri MIN Ijjävri3,40 0,40 0,80 -0,30 76 STD Ijjävri0,32 0,14 0,09 0,09 76 VAR Ijjävri0,10 0,02 0,01 0,01 76

Table 11. Ijjävri dune field, graphic grain size parameters sorted by

sampling sections. Iww, Itop and I lee are samples from ww-side, top and leeside sections. AVG is average, STD standard deviation, VAR variability of the data and n number of samples from each section.

section M z ( " ) " 1 K G S k 1 n AVG Iww 2,61 0,55 1,05 0,02 15 AVG Itop 2,24 0,64 1.00 0,03 12 AVG Ilee 2,45 0,68 1,05 -0,02 51

STD Iww 0,17 0,09 0,06 0,08 15 STD I top 0,53 0,18 0,04 0,07 12 STD I lee 0,37 0,14 0,10 0,10 51

VAR I ww 0,03 0,01 0,00 0,01 15 VAR Itop 0,29 0,03 0,00 0,01 12 VAR Ilee 0,14 0,02 0,01 0,01 51

4.3. Grain shape; roundness vertical sections (windward side, top and and surface texture leeside) are presented in Table 15. Average, maximum and minimum values for all 4.3.1. Binocular microscopy dune data from Muddusjävri and Ijjävri The average values for percentage of eve- are presented in Table 16 with standard ry roundness class for each dune are pre- deviation and variance. sented in Table 12 from Muddusjävri and in Table 13 for the parent materials from 4.3.2. Scanning electron microscopy Muddusjävri. In Table 14 the average val- (SEM) ues for roundness classes are reported from The procedure for SEM-study of the the dunes of Ijjävri. The average values for quartz grains ( Chapter 3 . 4 ) was time con- 56 the Muddusjävri dune data on different suming and laborious. The number of Table 12. Muddusjävri dune field: Proportions of quartz grains classified into roundness classes A (very angular) to F (well rounded) according to Powers (1953) & Shepard (1963 ). Percentages of glossy and matt particles are also presented. For the data AVG is average and STD standard deviation, while n is the number of analyses from each individual dune.

Dune A% B% C% D% E% F% Glossy% Matt% n AVG 1 0.8 35 46 17.8 0.4 0 --5 AVG 2 0 13.9 60.9 24.9 0.9 0 31.1 68.9 7 AVG 3 0 23.5 60.3 16.2 0 0 48.7 51.3 6 AVG 4 0 22.4 63.7 13.9 0 0 39.3 60.7 7 AVG 5 0 14.4 70.9 14.7 0 0 34.6 65.4 9 AVG 6 0 7.8 60 27.8 4.1 0.3 33.6 66.4 9 AVG 7 7.6 29.3 43 18 2.1 0 30.2 69.8 9 AVG 8 6 35.6 41.3 16.6 0.5 0 48.4 51.6 10 AVG 10 16 41 23 20 0 0 64 36 1 AVG 11 18 44 29 9 0 0 64 36 1 AVG 12 7 30.5 42.3 17 3.3 0 43.3 56.8 4 AVG 13 2.0 35.3 52.3 10 0.3 0 36.3 63.7 3 AVG 14 0.4 30.3 49 17.2 3.2 0 42.4 57.6 11 AVG 15 0.3 28.7 46.7 20.3 4 0 62.7 37.3 3 AVG 16 0.2 28.1 47 21.9 2.8 0 46 54 9 AVG 17 0 18.2 47.1 33.4 1.2 0 42.6 57.4 9 AVG 19 1 44 44 11 0 0 33 67 1 AVG 20 0.1 21.7 47.3 29.6 1.3 0 46 54 9 AVG 21 0 23.7 48.8 24.4 3 0.1 57.3 42.7 13

STD1 1.2 7.8 8.7 11.6 0.5 0 --5 STD2 0 12.5 9.5 13.2 1.0 0 26.7 26.7 7 STD3 0 6.3 6.7 6.3 0 0 19.1 19.1 6 STD4 0 15.9 9.5 9.2 0 0 14.5 14.5 7 STD5 0 10.1 11.2 8.6 0 0 19.7 19.8 9 STD6 0 4.1 11.2 10.4 3.8 0.7 14.9 14.9 9 STD7 7.2 9.9 8.3 7.8 1.9 0 16.3 16.3 9 STD8 7.7 9.7 9.7 8.9 1.2 0 14.4 14.4 10 STD12 5.1 16.1 9.1 10 5.6 0 22.7 22.7 4 STD13 0.8 17.5 13.5 6.5 0.5 0 21.2 21.2 3 STD14 0.6 10.8 8.1 6.8 3.7 0 17.8 17.8 11 STD15 0.5 3.8 3.4 0.5 0.8 0 3.3 3.3 3 STD16 0.6 11.7 7.5 7.0 2.6 0 27.9 27.9 9 STD17 0 10.1 11.8 9.0 1.7 0 17.3 17.3 9 STD20 0.3 6.6 5.8 5.9 1.6 0 17.6 17.6 9 STD21 0 10.8 8.0 7.3 1.8 0.3 17.6 17.6 13

samples had to be limited. Consequently alysed from each sample. The average val- the selection of the samples had to be con- ues for each sample in Table 17. sidered carefully.As 11 samples from 9 sites Grain 10 from the sample 3 lee/100cm were studied, the number of quartz grains was digitised four times in order to test re- analysed with SEM-technique mounted to peatability of the method. Experiment re- 175 . Samples 3 /ww/100 cm, 3 /top/100cm, 3 / sulted in a very good congruence between lee/100 cm, 9 /delta/90cm, 12/lee/90cm and the sessions. In the Table 17 the values for 16/lee/80cm were studied from Muddus- each sample represent a single series of jävri dune field, and samples 63/delta/ grains. Some grains were actually digitised 50cm, 69/lee/100cm, 70/lee/60cm, 71/lee/ twice because there were two layers of sur- 160cm and 75/lee/140 cm were selected face textures visible, while the uppermost from Ijjävri. Ten to fifteen grains were an- surface texture was determinant. 57 Table 13. Roundness analyses from parent materials adjoining Muddusjävri dune field. Roundness classes A–F according to Powers (1953) & Shepard (1963 ). Percentages of glossy and matt particles are also presented. Site indicates number/sampling depth in centimetres.

Site A% B% C% D% E% F% Glossy% Matt% comment

9/120 5 41 33 21 0 0 58 42 delta

9/80 27 47 21 5 0 0 75 25 delta

9/45 20 50 20 10 0 0 70 30 delta 18/top 17 45 29 9 0 0 77 23 re-deposited aeolian

18/bottom3 32 44 21 0 0 72 28 fluvial

22/100 1 21 52 25 1 0 43 57 esker

22/70 0 27 50 20 3 0 75 25 esker

22/50 2 28 46 24 0 0 68 32 esker

23 0 34 45 21 0 0 61 39 delta

24 2 24 50 24 0 0 39 61 esker

4.4. Mineralogical 6 .Heavy minerals 2 (Hm2 ), grains of zir- composition of the dunes con, apatite and titanite A total of 52 thin sections of the sediments 7 .Oxides (Ox), grains of magnetite and il- were studied and point-counted. After pre- menite, mainly liminary viewing the minerals were divid- 8 .Rock fragments (RF), grains with sev- ed into 9 groups: eral minerals 1 .Quartz (Q), grains of quartz (mono- 9 .Talc (T), grains of talc chrystalline and polychrystalline) 2 .Feldspars (Fsp), grains of plagioclase 4.4.1. Muddusjävri dune field and potassium-feldspar At Muddusjävri dune field a total of 19 sites 3 .Amphiboles and pyroxenes (AP), grains were studied. Mineralogical composition of amphiboles (mainly hornblende) and of the sites is given in Tables 18 and 19. pyroxenes The mineralogical composition of the 4 .Mica (M), grains of biotite and musco- analysed parent sediments is presented in vite Table 19. The sites include Sites 9 and 10 5 .Heavy minerals 1 (Hm1 ), grains of gar- at the erosion bank of Lake Muddusjävri 58 net and epidote at the northern end of the dune field, Site Table 14. Ijjävri dune field: Proportions of quartz grains classified into roundness classes A (very angular) to F (well rounded) according to Powers (1953) & Shepard (1963 ). Percentages of glossy and matt particles are also presented. For the data AVG is average and STD standard deviation of the analyses derived from each dune, while n indicates the number of analyses from each individual dune.

Dune A% B% C% D% E% F% Glossy Matt% n % AVG 52 0 27.5 51 21.5 0 0 50.3 49.7 3 AVG 54 0.6 46.9 38.1 12.8 1.6 0 61.3 38.8 8 AVG 55 0 16.5 51 27 5.5 0 41.5 58.5 2 AVG 57 0 16 52.5 30.5 1 0 40 60 2 AVG 60 0 2.6 50 42.4 5 0 50.6 49.4 5 AVG 61 0 3.4 48.2 42.2 6.2 0 40.4 59.6 5 AVG 62 0 8 56 29 7 0 59 41 1 AVG 63 0 1 67.8 29 2.3 0 35 65 4 AVG 64 0 2 32 52 14 0 1 99 1 AVG 65 8.5 12.1 37.9 32.8 8.8 0 47.9 52.1 8 AVG 66 5.6 19.4 41 29.2 4.8 0 57.8 42.2 5 AVG 67 0 14.7 46.7 37 1.7 0 47 53 3 AVG 68 0 1.5 43.5 46.5 8.5 0 38.5 61.5 2 AVG 69 0 8 63 28 1 0 14.5 85.5 2 AVG 70 0.2 14 47.8 33.7 4.3 0 54.5 45.5 6 AVG 71 0.1 13.7 39.6 43.4 3.1 0 15.4 84.6 7 AVG72 1.3 7.5 39.8 47.3 4.3 0 48.3 51.8 4 AVG 73 0 10.8 34.8 48 6.4 0 22.6 77.4 5 AVG 74 1.2 18.5 41.1 33.3 5.9 0 32.5 67.5 10 AVG 75 0 14.5 42.5 40.5 2.5 0 44.5 55.5 2 AVG 76 1 11 33 54 1 0 35 65 1

STD 52 0 3.3 3.0 6.5 0 0 16.0 16.0 3 STD 54 1.7 15.4 6.6 8.5 3.2 0 23.4 23.4 8 STD 55 0 14.5 2 11 5.5 0 25.5 25.5 2 STD 57 0 2 3.5 2.5 1 0 0 0 2 STD 60 0 1.2 12.3 10.7 3.2 0 5.6 5.5 5 STD 61 0 1.6 9.4 9.9 2.9 0 6.2 6.2 5 STD 63 0 1.7 19.8 17.4 3.9 0 30.1 30.1 4 STD 65 22.1 9.1 11.4 15.7 7.7 0 19.5 19.5 8 STD 66 5.2 11.2 7.9 7.2 5.2 0 18.1 18.1 5 STD 67 0 6.2 8.8 5.7 0.9 0 12.7 12.7 3 STD 68 0 1.5 1.5 4.5 4.5 0 6.5 6.5 2 STD 69 0 3 14 11 0 0 8.5 8.5 2 STD 70 0.4 5.3 4.9 5.3 5.1 0 7.7 7.7 6 STD 71 0.3 12.6 12.7 19.9 2.8 0 13.3 13.3 7 STD 72 2.2 5.9 4.8 10.3 1.1 0 9.8 9.8 4 STD 73 0 5.3 6.6 5.8 3.2 0 5.6 5.6 5 STD 74 1.3 7.0 5.5 9.1 5.9 0 12.0 12.0 10 STD 75 0 11.5 2.5 9.5 0.5 0 5.5 5.5 2

13, which is an inter dune section between tance from the western border of the dune Dunes 2 and 5 and Site 18, a fluvial sand at field to a dune site), as presented in Fig. the bank of River Vaskojoki. 23. For dunes with multiple studied thin At Muddusjävri the Quartz/Feldspars sections (Dune 1 ; 6 analyses, Dune 2 ; 7 ratio was correlated with the transporta- analyses and Dune 3 ; 3 analyses), all the Q/ tion distance (defined as the shortest dis- Fds relations were used in the diagram. 59 Table 15. Muddusjävri dune field, average values for the roundness classes sorted by sampling sections and sampling depth, 100 cm. Scale from A (very angular) to F (well rounded) according to Powers

( 1953) & Shepard (1963). Percentages of glossy and matt particles are also presented. M ww, Mtop and

M lee are samples from ww-side, top and leeside sections, M100 indicates all the samples from the sampling depth of 100 cm. AVG is average, STD standard deviation, VAR variance of the data and n number of samples from each section.

-$4/'*) "+ " + #+ %+ &+ ,+ .(*--1+ 23//+ "

:;% 7 55 '-$+2-.+2-2++-02-+ 6-332-632-6 2. :;% 7 1*, 2-2+3-&++-2+6-++-) 6-2+0-6+0-6 '+ :;% 7 (## '-6++-0+3-$)-6 3-$6-3 +)-$+)-$ '+ :;% 7 +66 '-)+3-0+'-++6-3+-$ 6+0-0+0-0 3&

<"9 755 32-.+)2-$+$+-2+2.-))-$ 6-6&'&-$&'&-$ 2. <"9 71*, ++-6+&)-)+2+-$+6'-02-$ 6-+302-0302-0 '+ <"9 7(## +0-$+2$-&+0$-2$3-)$-+ 6-6'62-+'62-+ '+ <"9 7+66 3& +0'-+360-$+6$-)2-' 63$$-03$$-0 3&

Table 16. Muddusjävri and Ijjävri dune fields, average values of all dune samples for the roundness classes A (very angular) to F (well rounded) according to Powers (1953) & Shepard (1963). Percentages of glossy and matt particles are also presented. For the data AVG is average, MAX and MIN maximum and minimum values, STD standard deviation and VAR variance, n is the number of samples from the dune field.

"(( 70)$-6 "+ 5+ #+ %+ &+ ,+ .(*--1+ 23//+ " "

These can be seen in Fig. 23 as multiple Q/ sented in Fig. 24. In addition, Dune 3 was Fds values on a single transportation dis- analysed at the depth of 100 cm in all three tance value. sections. The results of this are presented Two Dunes, 1 and 2 were studied at 60 in Fig. 25. cm and 100 cm sampling depths in all three Thin sections for the grain size fractions sections; windward, top and lee side. The of Φ 1 . 00 (0 . 500 mm), Φ 2 . 00 (0 . 250 mm), 60 mineralogical composition of these is pre- Φ 3 . 00 (0 . 125 mm) and Φ 3 . 74 (0 . 075 mm) Table 17. Results of the SEM-analyses. Averagepercentagevalues for the analysed quartz grains from each sample. Site code implies number, sampling section and sampling depth in centimetres. N indicates the number of analysed grains.

1. 2. 3. 4. 5. 6. 7. 8. 9. n Upturned C onchoidal Mech. Straight Dish- Deep Flat Cleavage Ad. Site silica fracture V- grooves shaped surface cleavage p lanes Par t. plates f orms c onc. soluti on f ace

3/ww/100 85.7 6.1 0 0.3 4.2 0 0 3.7 0 15

3/top/100 47.5 40.9 0 0.0 0.3 0.2 0 11.0 0 13

3/lee/100 68.4 2.8 0 0.0 5.9 0.2 6.2 16.4 0.0 14

9/delta/90 23.4 73.6 0.8 0.5 0 0.1 0 1.5 0 15

12/lee/90 63.4 9.6 0 0.0 2.7 0 12.0 11.7 0 15

16/lee/80 45.7 35.6 0 0.2 0.1 0.0 0 18.4 0 15

63/ww/50 51.9 38.0 0 0.0 0.8 0 7.4 1.8 0 11

69/lee/100 57.8 31.4 0 0.2 0.8 0.5 0 9.2 0 15

70/lee/60 70.8 24.2 0 0.2 1.1 2.5 0 0.9 0.2 10

71/lee/160 55.8 34.5 0 0.0 2.7 0 0 6.9 0 10

75/lee/140 51.2 22.9 0.1 0.2 3.9 0.1 0 21.6 0 15

were separately analysed from the Dune 5 Sites 52 and 62, are described in Fig. 27. Site (top section, sampling depth 100 cm). The 52/windward/150 cm is from the original results are given in Fig. 26. dune material on the SW-rim of the basin, while Site 62/top/ 20 cm is sampled 4.4.2. Ijjävri dune field from an accumulation hummock re-de- In the Ijjävri dune field, a total of 14 sites posited on the deflation basin (photo- were studied. Mineralogical composition graph, Fig. 21). of the samples taken from them is given In addition, the chemical composition in Tables 20 and 21. of the garnet grains from both dune field, Material for samples 60/ 4 / 100 and 61/ were studied. For this purpose, four thin 4 / 100 originates from inter dune sections sections were prepared for microprobe dug into the lowest part of a deflation ba- analysis, two from each dune field. Sam- sin. Their stratigraphy reveal a series of ples 3 lee 100 cm and 8 lee 35 cm were se- coarse-grained deflation surfaces ( Chapter lected from the Muddusjävri dune field 4 . 5 ), thus the sediment is deflated delta and samples 65 lee 50 cm and 68 lee 85 cm sediment. Site 63 is located on the SW-side from the Ijjävri dune field. A total of 117 of the dune field and is originally a delta garnets were analysed. In Table 22 the mi- terrain only slightly reshaped and overlain croprobe analyses and structural formulae by aeolian material. of 4 garnet grains from two sites in Mud- The mineralogical composition of the dusjävri and two sites from Ijjävri dune dune sediment at “big deflation basin”, field are presented. 61 Table 18. Mineralogical composition (%) of the dune sites at Muddusjävri. Groups are described in the text. Dunes coded as: number/part/sampling depth (cm), where part 1 = windward side and part 3 = lee side section.

Dune Quartz Feldspars Amp.+Px. Mica HM1 HM2 Ox. RF (talc)

1/1/100 22 33 33 4 3 0 3 2 0

2/1/100 15 34 23 11 5 1 1 10 0

3/1/100 24 31 31 2 6 0 1 5 0

4/1/100 25 32 25 1 10 1 0 6 0

5/1/100 22 34 31 4 5 0 2 2 0

6/1/80 21 29 31 6 4 1 4 4 0

7/3/100 25 39 22 4 4 1 2 3 0

8/3/100 19 38 24 4 9 2 0 4 0

12/3/90 18 48 17 4 6 2 1 4 0

14/3/90 10 41 30 3 6 1 1 6 2

15/3/10021 31 28 7 8 0 3 2 0

16/3/80 17 48 17 3 9 0 0 5 1

17/3/80 16 47 20 9 4 0 1 3 0

20/3/11010 48 21 7 9 2 0 2 1

21/3/10013 66 1 2 0 0 0 18 0

The garnets originating from the area blage with sillimanite and quartz is con- of Granulite Complex proper are charac- trolled by the end member reaction 3 terised by low Ca-content, thus they can anorthite ↔ 1 grossularite + 2 sillimanite be distinguished from garnets of other or- + 1 quartz (Hörmann et al., 1980). Com- igin. The reason for the low Ca-content at parison with the sites by Hörmann et al . the garnets of the Granulite Complex can ( 1980) in Fig. 28 confirms that most of the be explained by Ca distribution between garnet grains in the Ijjävri dune area orig- 62 garnet and plagioclase, which in assem- inate from the Granulite Complex, with a Table 19. Mineralogical composition (%) of the parent sites at Muddusjävri. Groups are described in the text. Sites coded as: number/part/sampling depth (cm), where part 4 = river sand, part 5 = beach sand and part 7 = inter dune section.

Site Quartz Feldspars Amp.+Px. Mica HM1 HM2 Ox. RF (talc)

9/5/50 5 33 31 24 0 1 1 5 0

10/5/60 22 33 20 9 10 1 1 4 0

13/7/80 23 34 19 9 7 1 2 4 1

18/4/50 9 30 35 7 6 0 0 13 0

Figure 23. Muddusjävri dune field, Quartz content (%) / Feldspars content (%) against transportation distance. Dunes 1 , 2 and 3 plotted with multiple analyses. R2 = 0 . 17 (n=18, p=0 . 08) (in statistical analysis Dunes 1 , 2 and 3 are included with average values for each).

transportation distance of circa 40 km. 4.4.3. Mineralogical comparison of Garnet has been found to be durable dur- the dune fields studied ing glacial transportation in Lapland be- The mineralogical composition of the fore. The result is concordant with the ob- dune sediments is constant and unique at servations of garnet abundance in the Sokli both fields and largely reflects the miner- esker at distances of 30 to 40 km from its alogy of the parent sediments. The main source area (Perttunen and Vartiainen, minerals of the parent bedrock are sum- 1992). marized in Table 23. In fact, the main min- 63 erals are roughly the same in both fields. and the delta formations originating from The average composition of the dune fields it, rather than local sediments. is, however, different (Fig. 29). In the Granulite Complex area, one of 4.5. Lithostratigraphy of the main minerals is cordierite. It is high- the dunes studied ly probable, that cordierite does occur in 4.5.1. Muddusjävri dune field the sand samples from Muddusjävri. How- Before presenting the lithostratigraphy of ever, on the grains of the sand units it was the dunes, a brief summary of the results not possible to differentiate cordierite from of the whole sedimentary succession at feldspars. No inclusions of e.g. zircon were Muddusjävri is introduced.As discussed in observed on feldspar-like sand grains to Chapter 2 the fault and the sediment suc- confirm the presence of cordierite, either. cession at Muddusjävri dune field was in- Quartz and feldspar content at both terpreted from seismic profiles ( Chapter 2 ). fields is roughly the same. The amounts of The esker is not visible in the Muddusjävri amphiboles and pyroxenes (AP), biotite area, between the lakes Paatari and Mud- and muscovite (M) and rock fragments dusjävri, but buried beneath younger del- (RF) are clearly larger at Muddusjävri than ta and dune sediments. During the inva- at Ijjävri.All these features can be ex- sion of the Inari ice-lake into the area, the plained by smaller scale transportation nearby edge of the ice sheet was feeding process and more compact dune field pat- finer sediments to the lake forming a del- tern at Muddusjävri than at Ijjävri. The ta in the narrow bay between the Innih- most striking difference between the fields vaara and Ristivaara hills. The delta has is, however, the amount of garnet. been thought to have formed in two sepa- The pre-eminent feature for the sedi- rate accumulation stages (Tanner, 1915). ments at Ijjävri is the high content of gar- This assumption was based on the distinc- net. This can be observed in every sample, tive boundary observed under the shallow where the proportion of group Hm1 (gar- water on aerial photographs (see Fig. 14, net and epidote) varies between 16– 36% top right corner). The boundary was be- (mean 24%). The overall majority of the lieved to be there because of a steep step Hm1 minerals were, indeed, garnet. This to a lower terrace. In order to test this con- is surprising, considering that the granite cept the lake bottom was explored by scu- gneiss complex is characterised by the ab- ba diving. The bottom was seen to slope sence of garnet (Hörmann et al., 1980), gently without any trace of a steep cliff. while at Muddusjävri, on the area of gar- The boundary that was visible from the air net-cordierite-gneisses, the proportion of was noticed to be in fact a sharp bounda- garnet is smaller. In the light of the state- ry of bottom vegetation, which is lacking ment that garnet is extremely rare in the in the shallow water probably because of rocks of the granite gneiss complex (Hör- the wave and ice erosion in the bowl- mann et al., 1980), it is probable that the shaped bay.Additionally the continuous high content of garnet at the Ijjävri dune sediment input to the lake from the erod- sediments is inherited from the esker ma- ing dune faces and delta sediments ( Fig . terial. In that case, the garnet-bearing sed- 30) might be the reason for the lack of veg- iment could have been glacifluvially trans- etation. In the spring time, floods are fre- ported from the area of the granulite com- quent in Tulvalahti (the name literally plex, circa 40 km. Consequently, this sug- means “flood bay” in Finnish). For in- gests that the main source of the sediment stance, in the spring of 1993, the water level 64 at the Ijjävri dune field is the esker chain rose to the basement of the Airamo House 10%

Figure 24. Mineralogical composition of Dunes 1 and 2 at Muddusjävri dune field at the sampling depths of 100 cm and 60 cm. (Q = quartz, Fsp = plagioclase and potassium-feldspar,AP = amphiboles and pyroxenes, M = biotite and muscovite, Hm 1 = garnet and epidote, Hm 2 = zircon, apatite and titanite, Ox = magnetite and ilmenite and RF = rock fragments).

and the beach cliff eroded 0 . 5 m traced on the seismic profiles at 20 m be- (M. Helminen, pers. comm.). The deepest low the ground surface at the site of Dune part of the Lake Muddusjävri had many 3 (Fig . 3 ), whereas the delta/dune contact rounded boulders and stones on the sandy is at 8 . 85 m below the ground surface. Circa bottom. This supports Tanner’s ( 1915) as- 600 m northeast, at the Tulvalahti beach, sumption that the esker continues beneath the delta sediments can be found in the Lake Muddusjävri. eroding slope beneath the dune sediments The delta sediments overlie the esker at 1 . 4 m from the ground surface ( Fig. 31). formation. The esker/delta contact can be The palaeocurrent measurements of the 65 Dune 3

45 40 35 30 25 windward % top 20 leeside 15 10 5 0 QFsp AP MHm1 Hm2OxRF

Figure 25. Mineralogical composition of Dune 3 at Muddusjävri dune field at the sampling depth of 100 cm. For abbreviations, see the caption of Figure 24.

50 Q Q 40 FdFdss 30 APAP % 20 HMHM1 1 10 OxOx 0 RF 1233,74 RF GraiGrainn size size f fractionsractio ns(phi) (phi)

Figure 26.Variations in mineralogical composition of Dune 5 (top, 100 cm) at grain size fractions 1 . 0 , 2 . 0 , 3 . 0 and 3 . 74 phi. For abbreviations, see the caption of Figure 24.

delta sediment at Tulvalahti gave a result- dune sediment thickness and the degree of ant direction of 230 ˚ → 50˚.The contact be- dune morphologic development are in- tween the delta and the dune sediments is creasing from the SW end to the NE end erosional and it has been traced at several of the dune field. The dune accumulation locations. It is apparent, that the better the rate is apparently governed by the trans- dunes are developed morphologically, the portation distance according to the pre- 66 thicker the dune sediments are. Both the vailing wind direction of the main form- Table 20. Mineralogical composition (%) of the dune sites at Ijjävri. Groups are described in the text. Dunes coded as: number/part/sampling depth (cm), where part 1 = windward side, part 2 = top and part 3 = lee side section.

Dune Quartz Feldspars Amp.+Px. Mica HM1 HM2 Ox. RF (talc)

52/1/150 17 49 12 0 20 0 2 0 0

54/1/90 11 50 16 5 16 0 1 1 0

55/2/70 14 52 14 0 19 0 1 0 0

62/1/20 27 40 10 1 17 0 1 4 0

65/3/100 17 27 15 1 36 0 3 1 0

66/3/75 23 36 17 3 19 0 0 2 0

67/3/45 25 37 10 3 24 0 0 1 0

68/3/145 26 37 10 2 24 0 0 0 1

69/3/100 19 48 8 0 23 0 0 2 0

70/3/170 18 34 8 2 36 0 1 0 1

71/3/160 18 46 10 1 24 0 0 1 0

72/3/150 18 37 9 2 31 1 0 2 0

Table 21. Mineralogical composition (%) of the parent sites at Ijjävri. Groups are described in the text. Sites coded as: number/part/sampling depth (cm), where part 4 = deflation basin, delta sediments.

Site Quartz Feldspars Amp.+Px. Mica HM1 HM2 Ox. RF (talc)

60/4/100 21 46 14 5 11 0 2 0 1

61/4/100 15 40 19 2 21 0 3 0 0

63/4/80 14 40 17 1 3 1 0 23 1 67 50 45 40 35 30 % 25 20 15 10 5 0 QFsp AP MHm1 Hm2OxRF Original dune Secondary deposition

Figure 27. Mineralogical composition of an original dune at Ijjävri dune field (sampling depth 150 cm) compared with the composition of a secondary accumulation hummock (sampling depth 20 cm). For abbreviations, see the caption of Figure 24.

Table 22. Examples of the microprobe analyses and structural formulae of garnets at Muddusjävri and Ijjävri.

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

Figure 28. End-members of garnet grains from Ijjävri and Muddusjävri (upper part, this study) and from bedrock (lower part, according to Hörmann et al., 1980). Determination for the upper part is based on 12 oxygens and 8 cations, all iron as FeO, end-members after Deer et al., 1992. ing phase (SW-NE), hence the dune axes field ( Fig . 3 ). 13 of them were positioned at are parallel to the resultant sand transport a windward side of a dune, 13 at a crest (top) direction (Chapter 5 ). and 14 at a lee side of a dune. In addition A total amount of 44 sections were stud- to the dune sections four sites on the ied from 21 localities at Muddusjävri dune boundaries to the adjoining formations 69 Table 23. Main minerals of the bedrock units at Muddusjävri and Ijjävri.

"9#L H+>)1-1(+O A/+> IL+/ .+/#9L-: K#%#9#/O#

IARRA:,MB9+<9L/A-+># O1.4-#DP G*%#-R:4L9Q 8AL9>EQ 4-L(+1O-L:# 5"/ @&*'F6Q I#9+-M+/#/Q (L9/#>*O19R+#9+># (/#+::#: (L9/#>Q N+1>+># L/R O19R+#9+># 30;=7 ?29.L// "2 9*/= 30$F7 C,,MB9+<9L/+># (/#+:: O1.4-#DP JAL9>EQ 4-L(+1O-L:#Q G*%#-R:4L9Q N+1>+>#Q I#9+-M+/#/Q (/#+::1:# (9L/+>#: .A:O1B+>#Q )19/N-#/R# L/R #4+R1># 30;=7

45 40 35 30 25 % 20 15 10 5 0 QFsp AP MHm1 Hm2OxRF Muddusjävri Ijjävri

Figure 29. The average mineralogical composition of the dunes at Muddusjävri (n = 29 ) and Ijjävri (n = 14 ) dune field. For abbreviations, see the caption of Figure 24.

were studied. These included fluvial sedi- lowed by a concluding model of the lithos- ments on the riverbanks, esker section and tratigraphic units and their properties at delta sediments. Especially the dune/delta the Muddusjävri dune field at the end of contact between the dunes and the under- this Chapter (Fig . 38). lying delta sediments at the eroded beach Dune 4 exhibits the most common at Lake Muddusjävri were studied in detail. types of stratification in the Muddusjävri Dune lithostratigraphy is described here dune field, as depicted in Fig . 32. The stra- with examples from Dunes 4 (at three sec- ta show high and low angle foreset lami- tions), 12, 13 (interdune section) and 14 (at nae with unimodal azimuth. Dune 4 is sit- three section) (see Fig . 3 for location). uated at the northern end of the dune field, Bearing of the dip directions of the fore- along the road leading to the Hill Saukko- set strata were rather uniform (varying vaara (exact location indicated in Fig . 32 from 45˚to 55˚) indicating the wind direc- according to the Finnish grid system, see tion of the first dune forming period. The also Fig . 3 ). The road cuts the parabolic 70 detailed description of these sites is fol- dune in the middle forming a west-east orientated cross-section parallel to the cial features described here in detail. At 110 dune forming wind direction and the dune and 80 cm there are exceptionally thick sets axis. The height of the dune from the lee of dark, fine sand, 3 and 2 cm, respective- side to the top is 4 metres. Total lateral ex- ly, dipping at 25˚. The tabular foresets un- tent of the cross-section is 24 m. The ex- derlain and above the dark foreset at 80 cm act locations of the windward, top and lee are unusually thick, 10 cm each, consist- sections in the dune cross section are ing of a set of thin laminae of light sand. shown as insert in Fig. 32. There also is a 3 cm thick strata of fine dark Windward section is 1 . 2 m in height. sand at 70– 67 cm (sample 4 ww/70). Dip of The contact with the underlying sediment the dark strata is 28˚. At 67– 56 cm there is was not reached. The lowest part of the a 9 cm thick tabular foreset with thin lam- section consists of bulk sediment defined inae consisting of light, coarser (than the as intermediate to high angle tabular fore- bulk sediment) sand (dip 25˚). Contacts set lamination with thin, 0 . 5 to 1 cm thick between the foresets are sharp. laminae of dark sand alternating with 1 – 3 The strata are overlain by dune-foreset cm thick laminae of light coloured sand. cross-bedding at 55 to 40 cm, dipping at in- This foreset strata is closely packed, undis- termediate angles 10˚ to 25˚, indicating ad- turbed and planar, dipping at 25˚ and is vance of the dune. The bounding surfaces constant from 120 to 55 cm, with some spe- between the foresets are curved and the

Figure 30. The beach of Lake Muddusjävri at Tulvalahti. 71 deformation. Gently downward folding layers at 22– 25 and 30– 33 cm indicate a frost fold, which has deformed the strata. Denivation processes have caused the strata to become dimpled in this upper part of the section. Strata at 20– 6 cm are poorly stratified, loose sand with lots of roots and organic debris. The uppermost 6 – 0 cm is composed of poorly developed soil B horizon. Atthe 3 -D cut into the section the layers throughout the section are found to be continuous and almost horizontal, with slight 1 – 2 ˚ dipping towards the section. The lithostratigraphic units, indicated above with italics , are typical of all the dunes studied at Muddusjävri. In the following, more sections are described from the area. However, the description of the strata is less detailed for the elements already mentioned. The height of the top section at the dune crest is 1 . 35 m (Fig . 32). In the lowermost part of the section the bulk sediment can be described as tabular foreset strata with rather steep dipping at high-angles 30– 32˚, and with tabular bounding surfaces. This bulk sediment (from 135 to 90 cm) is inter-bedded with some noticeable foresets described in more detail. At 135 cm there is a 2 cm thick layer of coarser (than the bulk sediment) light sand, followed by steep foresets and three more thin layers of coarser light sand at 112– 110 cm, 103 – 101 ▼ cm and 99– 97 cm. The middle part of the section, at 90 to 40 cm, is defined by dune foreset cross-bedding strata, dipping at intermediate angles of 25 to 15˚. Bounding surfaces are wedge-shaped. At 83– 78 cm there is a distinctive lamina of very coarse light sand (sample 4 /top/ 85cm). Quartz Figure 31. Dune/delta contact at the Tulvalah- content of the lamina is markedly higher ti beach (indicated with an arrow). and the roundness of the grains poor. From 60 cm the strata becomes increasing- cross-bed sets are tabular or wedge-shaped. ly dimpled, though maintaining primary There is a change in the bedding at 35 to structure of foreset cross-bed sets with in- 20 cm as the foreset laminaewith disturbed termediate to low-angle dipping; 20˚ at 60 72 strata shows post- and syn-depositional cm, 10˚ at 50 cm and 5 ˚ at 35 cm. Towards Figure 32. Stratigraphy of Dune 4 on windward, top and lee sections.

the surface, at 35– 10 cm, the sediment is There is a 5 cm thick layer of light, coarser poorly stratified loosely packed dry sand, (than the bulk sediment) sand at 70– 65 cm disturbed by roots. The soil at 10– 0 cm is and another ( 2 cm) at circa 50 cm. Low- well developed with a distinguishable A 2 angle cross-bedded foreset strata are ob- horizon, circa 1 cm in thickness and a B served at 60 to 40 cm. Dimpled surfaces horizon (4 cm). At 5 cm, under a A 0 hori- indicating signs of denivation processes are zon there is a distinctive horizon of char- visible at the uppermost part of the sec- coal pieces. The pieces are rather big com- tion, at 40– 25 cm. The primary strata in- pared to other dune sites, circa 1 cm in di- dicate continuation to the foreset cross- ameter (sample 4 /top/5 cm for radiocar- bedding. From 25 cm to the surface, there bon dating). The 3 -D cut into the section is a well developed soil horizon with a soil indicates a gentle dip ( 3 – 4 ˚) of the layers B horizon at 20– 10 cm, overlain by distinc- towards the section with tabular bedding. tive A2 horizon (3 – 4 cm). On the surface Section 4 lee (Fig . 32), exhibits the same of this soil horizon there is a layer of char- basic units as the sections from the top and coal at 5 cm. The 3 -D cut into the section windward side, although the dipping of the is almost horizontal, with slight dip of 1 ˚ strata is steeper. The base of the section at into the section. 120 to 60 cm is dominated by a high-an- Another example from the Muddusjävri gle tabular foreset laminae, dipping at 30˚. dune field comes from Dune 12, which is 73 Figure 33. Lake Kotkajärvi. Note the overs- teeping of the dune leeside in the foreground and the extensive dune on the other side of the lake.

located NW from Lake Kotkajärvi, in the central area of the dune field. Lake Kotka- järvi area is characterized by the tallest dunes of Muddusjävri field ( Fig . 33), probably due to an accumulation along the lake shore, where the advance of the dunes ceased. The section 12 was excavated on lee side of the dune and it is 1 . 8 m in height, for the co-ordinates (the Finnish grid system), see Fig. 34 and for the location see Fig. 3 .

Figure 34. Stratigraphy of Dune 12, lee section. The base of the section, i.e. contact to the underlying sediments, was not reached. The lowermost part of the section consists of the strata defined as tabular high-angle foreset laminae, dipping at 25˚. At 144 cm there is a distinctive erosional contact with very coarse sand grains on the surface ( Fig . 35) indicating a deflation surface. Presence of this contact implies a strong active phase and dune advance. Above the contact, the strata are subhorizontal, very low-angle foreset laminae, dipping at 0 – 3 ˚. There is a layer of coarse, light sand at 131– 128 cm. The strata continues with planar, almost horizontal sets at 128 – 120 cm. At 120– 95 cm the foreset lamina becomes increasingly dimpled and the strata is cross-bedded, dipping at low-angles of 2 – 5 ˚. The whole upper part of the sediment section from 95 cm to the surface consists of massive unstratified sand. There are several dimpled charcoal horizons on this unit, at Figure 35. Lower part of Dune 12 on lee side. 70cm, 68 cm, 60 cm (discontinuous), 56 Note the erosional contact with very coarse cm, 50 cm (discontinuous) and at 1 cm. grains on the surface in the middle of the These buried charcoal horizons display figure. deformation structures, indicating denivation. The soil development associated with these surfaces is very poor, consisting of thin fragmental soil horizons, which are inated sand, with dimpled strata indicat- also dimpled. The dating of the charcoal ing post- and syn-depositional denivation horizons and the sand samples above the forms. In the upper part of the section the erosional contact and beneath the lowest strata is disturbed by a moderately well charcoal are discussed in Chapter 4 . 6 . developed soil B horizon at 22– 5 cm, and Site 13 at the Muddusjävri dune field the sediment becomes unstratified. At 5 cm represents an interdune section. It was ex- there is a charcoal layer overlain by a 3 cm cavated at the lowest topographic position thick soil A 0 horizon. between the Dunes 2 and 5 , on the central Dune 14 is situated at the southern cen- area of the net formed by parabolic dunes. tral part of the dune field, near Tupasuvan- Again, the exact location of the site is in- to. Fig. 37 depicts the lithostratigraphy of dicated in Fig. 36 according to the Finnish Dune 14 on lee, top and windward sections. grid system. Here, the western arm of the parabola is The lowermost part of the section studied and the sections are in reverse or- shows tabular dune foreset strata dipping der (explained in insert in Fig . 37). The coat intermediate angle of 22˚. At 34 cm there ordinates of the site are also indicated. The is an erosional contact, with a thin layer dune sections were circa 100 cm in height of coarser sand on the surface. The sequen- and the contact to the underlying sedi- tial set at 34– 22 cm consists of planar lam- ments was not reached. 75 Figure 36. Interdune section at the Muddusjävri dune field.

The sections are described in reverse dark sand interstratified with 2 – 3 cm thick order, from the lee side to the windward sets of light sand laminae. Packing of these side. The lowermost part of the lee side cross-bedded strata is moderately close. section exhibits a charcoal horizon at 100 This part is followed by the uppermost cm, indicating intense sand transport dur- sediment unit. The strata consist of tabu- ing an aeolian reactivation phase. The fore- lar foreset laminae dipping at low-angles. set laminae at the lowermost part ( 100– 50 The strata show rather constant lamination cm) are actually typical for the middle part at 45 to 15 cm, where it gradually disap- of a section at Muddusjävri, displaying pears due to soil B development. There is dune foreset cross-bedding strata, dipping a charcoal horizon at 49 cm, followed by at low-angle to intermediate angles of 10 another at 45 cm. The upper one at 45 cm to 20˚. The bulk sediment is also very sim- was radiocarbon dated. There are two ilar to those with similar setting, consist- more charcoal horizons at 23 cm (radio- 76 ing of thin, 0 . 5 to 1 cm thick laminae of carbon dated) and at 12 cm. Soil B is rath- Figure 37. Lithostratigraphy of Dune 14 at the Muddusjävri dune field.

er well developed at 15– 0 cm. There is a shows indication of denivation processes thin 1 – 1 . 5 cm thick A 2 horizon under the as the strata are dimpled. At 36 cm the lam- charcoal layer at 12 cm. ination deteriorates, and the sediment is The top section demonstrates continu- massive from 36 to 0 cm. The uppermost ous strata of tabular foreset laminae, with part of the section is characterised by sand high-angle dipping of 30˚. The foresets are with loose packing, organic debris and closely packed, however showing at 80 to roots. Buried charcoal horizons are ob- 60 cm and 30 cm post- and syn-deposition- served at 29 cm, where the horizon is frag- al deformation structures. These are prob- mental and broken and at 6 and 4 cm, ably related to the denivation processes. where the charcoal layers are continuous. The lowermost part of the windward The concluding model of the dune section comprises of foreset strata dipping lithostratigraphic units at the Muddusjävri at intermediate angle of 20˚. This foreset field is presented in Fig. 38. The dunes have strata are closely packed, undisturbed and an erosional contact with the underlying planar and are constant from 100 to 55 cm, delta sediments. Dune lithofacies model consisting of bulk sediment similar to consists of three lithostratigraphic units. Dune 4 , with no exceptional features. The properties and key elements of each There is a sharp bounding surface at 55 cm, unit are postulated in the following way: where the strata exhibit low-angle to sub- UNIT 1 : horizontal foreset laminae, with dipping at •Positioned at the lowermost part of the 0 – 1 ˚. These strata are not considered to be section, ranging usually from 135 cm to topset laminae, because those mainly con- circa 60 cm. The strata show interme- sist of climbing ripple strata, which was diate to high angle tabular foreset lam- not observed. The strata continue from 55 ination indicating a combination of to 45 cm, where the primary structure grainflow and grainfall deposition. 77 •The bulk sediment consists of tabular with the upper unit is also often ero- foreset strata with dipping at high-an- sional. gles of 32– 30˚ to intermediate angles of UNIT 2 : 20– 25˚, and with tabular bounding sur- •Positioned in the middle part of the faces. The bulk sediment consists of section, usually at 90 cm to 40 cm thin, 0 . 5 to 1 cm thick laminae of dark •The bulk sediment is defined by dune sand alternating with 1 – 3 cm thick lam- foreset cross-bedding strata, dipping at inae of light coloured sand. Often this intermediate to low-angles of 25 to 10˚, bulk sediment also has thicker foreset indicating advance of the dune. The laminae inter-bedded. Deformation bulk sediment consists of thin, 0 . 5 to 1 structures can occur. No observations cm thick laminae of dark sand inter- of charcoal horizons were made in this stratified with 2 – 3 cm thick sets of light unit. sand laminae. Dimpled strata and de- •Packing of the unit is close, at high-an- formation structures are common. One gle foresets displaying tilted rhombic charcoal horizon was detected within packing arrangement with minimum this unit. sediment porosity ( 26%), which pro- •Bounding surfaces are wedge-shaped or duces the slope of 30˚. curved between the forests and the •Lower contact with the underlain delta cross-bed sets are tabular or wedge- sediment is erosional, and the contact shaped.

UNIT 3

UNIT 2

UNIT 1

78 Figure 38. The model of the lithostratigraphic units at the Muddusjävri dune field. •Packing of these cross-bedded strata is moderately close. UNIT 3 : •The uppermost sediment unit usually ranges at circa 40 cm– 0 cm. •The strata consists of tabular foreset laminae dipping at low-angles of 0 – 5 ˚. The bulk sediment is often poorly stratified, due to disturbance of soil development, roots and denivation structures. In the uppermost part the strata is often dimpled. Most of the charcoal horizons indicating forest fires are within this unit. Towards the surface, at 40– 0 cm, the sediment is poorly stratified •Packing of the sediment is loose. Part- ly this is due to roots and the soil forming process, partly due to the denivation process.

4.5.2. Ijjävri dune field A total amount of 31 sections were studied from 26 localities in Ijjävri dune field. Three of them were positioned at a windward side of a dune, three at a crest (top) Figure 39. Dune section 51 at the Ijjävri dune and 24 at a lee side of a dune. In addition field exhibiting a re-accumulated sand unit to the dune sections, one site on a dried overlaying a former dune surface with a up lake bottom was excavated (Site 64). charcoal horizon. Unlike at Muddusjävri, the contact with the underlying sediments was reached frequently, when the dunes were studied. Dis- tribution of the studied sites is presented into the adjoining deflation basin was ex- in Fig . 6 . cavated, and together these sections exhibit A brief look into the dune lithostratig- the succession from the base sediments raphy of the Ijjävri dune field, gives an into dune strata. In addition, section 62 impression of two sedimentary units: the describes lithostratigraphy of an accumu- closely packed stratified “old” dune strata lation hill. At the end of this Chapter , a overlain by re-accumulated loose, unstrat- model of the lithostratigraphic units at ified sand ( Fig . 39). However, more pro- Ijjävri dune field is presented. found research into the dune lithostratig- Dune 72 exhibits some of the most raphy of the field demonstrates that the common elements of stratification in the pattern is far more complicated. Ijjävri dune field. Section 72 is located at Lithostratigraphy of the Ijjävri dune the central area of the dune field (exact field is described with example sections: location indicated in Fig . 40 according to dune sites 72, 70, 67, 68 and 74 (see Fig . 6 the Finnish grid system). The contact with for locations). At Dune 74, also a section the underlying sediments was not reached. 79 The lowermost part of the strata con- of denivation processes. There is a well sists of bulk sediment defined as interme- developed soil B horizon at 85 to 75 cm

diate to high angle tabular foreset lamina- overlain by a 3 – 4 cm thick soil A 2 horizon. tion with thin, 0 . 5 to 1 cm thick laminae On the surface of this soil there is a char- of fine dark sand (consisting mainly of coal horizon at 72 cm (AMS-dated). mica and heavy minerals) alternating with The strata are overlain by low-angle to 3 – 8 cm thick laminae of light coloured subhorizontal foreset lamination at 72 to sand (consisting mainly of quartz and feld- 0 cm, dipping at angles of 7 ˚ to 0 ˚. The spars). This foreset strata are closely bounding surfaces between the foresets are packed and planar, dipping at 25˚ and is tabular. Throughout this upper part of the constant from 170 to 72 cm. Contacts be- section, the strata are poorly stratified, tween the foresets are sharp. The laminae loose and dimpled, indicating post-and are slightly dimpling, indicating influence syn-depositional deformation. Soil forma-

80 Figure 40. Lithostratigraphy of Dune 72 at the Ijjävri dune field. tion on the uppermost part of the section ing of bulk sediment of thin, 0 . 5 to 1 cm is weak. thick laminae of dark sand alternating with Section 70 is located in the same exten- 3 – 8 cm thick laminae of light coloured sive deflation basin described earlier on the sand. This foreset strata is closely packed eastern side, near the turning point indi- at 180 to 85 cm and the cross-bed sets are cated with an arrow in Fig. 20, (exact loca- wedge-shaped, dipping at 20˚ to 34˚. tion indicated in Fig. 41 according to the Bounding surfaces between the foresets are Finnish grid system). The contact with the sharp. The strata shows a distinctive ero- underlying sediments was not reached. sional contact at 112 cm, with very coarse The lowermost part of the sediment grains on the surface. This hiatus indicates unit at 180 to 61 cm consists of bulk sedi- strong deflation process. The strata are ment defined as high to intermediate an- dipping at a very high angle ( 34˚) on the gle cross-bed foreset lamination, consist- erosional surface and the foresets overlay-

Figure 41. Stratigraphy of Dune 70 at the Ijjävri dune field. 81 ing it. At 85 cm the strata becomes less ping at indermediate angle of 20˚. There stratified, and the packing of the sand is is a charcoal horizon at 25 cm, underlain

looser. The strata indicate deformation at with a thin 0 . 5 – 1 cm thick soil A 2 horizon, 75 cm, with scattered charcoal pieces and and another charcoal at 20 cm. The upper disturbed foreset. The presence of charcoal charcoal is dipping at 15˚. Between the pieces and gradually less stratified strata charcoals the sediment is stratified. The indicates a former ground surface. uppermost part from 20– 0 cm consists of A sharp, erosional contact to the sub- unstratified loose sediment with roots. Soil horizontal, planebed lamination is shown at the ground surface is poorly developed. at 61 cm. This strata continues from 61 cm Sites 67 and 68 are located at the north- to 53 cm, and indicates tractional deposi- eastern end of the dune field ( Fig. 42) along tion. The strata consists of 1 . 5 cm thick sets the same dune ridge (co-ordinates accord- of light coarse sand, inter-bedded with 0 . 5 ing to the Finnish grid system are given in cm thick sets of medium sized light sand. Fig . 43). The dune is actually situated on Contacts between the layers are erosional, the eastern side of the esker chain, and was with a coarse lamina of sand grains, indi- studied in two sections 67 and 68 (Fig s 18 cating a deflation surface. The sub-hori- and 6 , top right corner). The lithostratig- zontal strata continues from 53 to 25 cm raphy of these sites is very similar, the only with cross-bedded foreset laminae dipping difference being the presence of coarse in- at low- to intermediate angles of 0 ˚–20˚. terlayers in the lower foreset strata of Dune The strata consists of thin dark laminae 67 and inconsistent number of charcoal inter-bedded with thicker sets of light sand horizons in the strata of the sites. laminae. The sub-horizontal strata are The contact to the underlying sedi- overlain with foreset strata showing dip- ments was not detected. The lowermost

82 Figure 42.A view from Site 67 facing North. part of section 67 at 120 to 30 cm is domi- (radiocarbon dated) at 30 cm underlain by nated by intermediate angle cross-bedded a soil A 2 horizon (2 cm thick). The change foreset strata dipping at angles from 20˚ to in packing is due to deformation of the 10˚. The bounding surfaces are tabular and strata and soil formation process. The up- the strata has indications of post- or syn- per part of the section at 30- to 0 cm con- depositional deformation. These are dem- sists of unstratified, loosely packed sand, onstrated by gentle folds at 110 to 100 cm with roots and pieces of charcoal. This part and at 80 to 75 cm. The bulk sediment is indicates aeolian re-deposition. closely packed strata consisting of thin Site 68 is situated circa 1 km S-SE along foresets ( 0 . 5 – 1 cm) of dark sand inter-bed- the same dune ridge. Here also, the con- ded with 3 – 5 cm thick foresets of light sand tact to the underlying sediments was not bearing thin laminae. The strata accom- detected. The lowermost part at 170 to 95 modates unusually coarse layers of light cm of the section consists of cross-bedded sand at 109– 105 cm, 97– 95 cm and 94 –92 foreset strata dipping at intermediate an- cm. The strata becomes poorly stratified gles of 25˚ to 15˚. The bulk sediment is sim- at 59 to 33 cm, with looser packing and ilar to the bulk sediment of Site 67. The dimpled strata dipping at intermediate strata indicates deformation structures at angle of 20 ˚. There is a charcoal horizon 130 to 95 cm, with dimpled layers.At 95 cm

Figure 43. Lithostratigraphy of the Dunes 67 and 68 at the Ijjävri dune field. 83 there is a charcoal horizon underlain by a veloped consisting of a soil B horizon (8

well developed soil B horizon at 97 to 112 cm) and a 2 cm thick A 2 horizon.

cm and a soil A 2 horizon at 95 to 97 cm. Dune 74 is located in the area of para- The upper unit consists of loosely packed, bolic dunes (co-ordinates according to the unstratified sand at 95 to 40 cm. There is Finnish grid system are given in Fig. 44). another charcoal horizon at 76 cm. This The lee side of the blowout was studied layer was confirmed to be separate from ( 74) and an additional section was exca- the lower one by excavating the section vated at the bottom of the blowout. To- wider on both sides. However, the radio- gether these represent a succession from carbon age ranges are overlapping ( Chap- the base sediments into the dune forma- ter 4 . 6 ). The third charcoal horizon at 69 tion. cm was noticed to be intact in places and The lowermost part of the base section scattered in another (see discussion 5 . 5 .). has a contact with a unit of stony gravel at All three charcoal horizons were radiocar- 140 cm. Fine sand above the gravel was bon dated. The uppermost part at 40 to 10 sampled for TL/OSL -dating. The lower- cm shows low-angle foreset strata dipping most unit at 140 to 72 cm consists of un- at 5 ˚. Soil horizon at 10– 0 cm is well de- dulating low-angle (–5 ˚) laminae of fine

84 Figure 44. Lithostratigraphy of the Dune 74 and the base 74 at the Ijjävri dune field. sediments. The strata show low-angle lam- field. These are usually situated in defla- ination of fine sand, coarse sand and silt. tion basins, near the sides and behind At 115 cm, there is an undulating 0 . 3 cm “wind shields”, where the transportation thick hardened rust-brown layer of silt, capacity of the wind suddenly ceases. Such indicating ground water level.At 72– 65 cm a hummock, Site 62 was excavated at the there is an erosional contact to the upper extensive deflation basin, described earli- unit, which is consisting of low-angle to er. Fig. 45 describes the stratigraphy of the sub-horizontal planebed lamination, indi- hummock and the contact with the under- cating tractional deposition and a deflation lain dune foreset strata. The original dune process. The sets of laminae include thin foreset strata are dipping at intermediate layers of dark sand, inter-bedded with 1 – angle of 25˚. The strata are closely packed. 1 . 5 cm thick laminae of coarse light sand The upper part of the original dune strata and 3 – 5 cm thick laminae of medium sized shows a moderately well developed soil B light sand. The horizontal, 2 cm thick lay- horizon. The strata of the accumulated ers of coarser light sand at 60 and 40 cm hummock are poorly stratified with loose show indication of a deflation process, packing. There is no indication of advance, with a surface layer of coarse grains. From but the sand has rather accumulated in 40 to 0 cm the strata consist of dune fore- situ. The presence of the preserved soil B set cross-bed laminae, with dipping lami- horizon underneath the hummock indi- nae at low-angles of 15˚ to 0 ˚. There is a cates that the accumulation of the hum- deflation surface at 0 cm. mock is connected to the ongoing defla- The actual dune section 74 shows at the tion process and that there was a long pe- lowermost part at 180 to 80 cm low-angle riod of ceased reactivation before the lat- cross-bedded strata similar to the upper- est phase ( Chapter 6 ). Estimations of the most strata at the base section. The bulk duration of the soil forming processes in of the sediment is characterized with thin Lapland were discussed in Chapter 2 . laminae of dark sand ( 0 . 3 – 0 . 5 cm) alter- The concluding model of the dune nating with thicker, 3 – 5 cm thick laminae lithostratigraphic units at the Ijjävri field of light medium sized sand. The upper part is presented in Fig. 46. Dune sediments of the unit at 102 to 80 cm is post-deposi- have an erosional contact with the under- tionally altered due to soil formation. A lain sediments. The dune lithofacies model soil B horizon at 102 – 88 cm is well devel- consists of three lithostratigraphic units. oped and overlain by an unusually thick The properties and key elements of each

A 2 horizon at 88 to 81 cm. There is a char- unit are described as follows: coal horizon at 80 cm. The soil formation UNIT 1 : is overlain by another. This upper soil unit •In the sections described here, only and accommodates two charcoal horizons at 75 Site 62 at 40 to 60 cm exhibit Unit 1 . and 60 cm. The uppermost charcoal layer The lowermost parts of the Sites 70 (at is underlain by a thin A 2 horizon at 61 to 180 to 165 cm) and 68 (at 150 to 170 cm) 64 cm and a soil B horizon at 64 to 79 cm. most probably belong to Unit 1 (the sec- All three charcoal horizons were radiocar- tion should have been deeper for con- bon dated. The uppermost sediment unit firmation). Atsection 74 base, Unit 1 has at 59 to 0 cm consists of re-accumulated been eroded. In addition, Unit 1 sedi- dune sand characterized by loose packing, ments were recognised at Sites 52, 54, 63, being almost unstratified. 65, and 66. Accumulation of small hummocks is a •The unit is positioned at the lowermost common phenomenon at the Ijjävri dune part of the section, ranging from 180 cm 85 Figure 45. Site 62 at the Ijjävri dune field with an accumulation hummock.

to circa 70 cm. The strata show inter- sediment porosity ( 26%), which pro- mediate to high angle tabular foreset duces the slope of 30˚. lamination indicating a combination of •Lower contact with the underlain delta grainflow and grainfall deposition. sediment is erosional, and the contact •The dipping of the strata varies from with the upper unit can be erosional site to site at high- to intermediate an- (with deflation surfaces) or gradation- gles from 34˚ to 25˚, but is constant. The al (with an intact soil horizon). bounding surfaces are tabular. The bulk UNIT 2 : sediment consists of thin, 0 . 5 to 1 cm •Unit 2 can be observed at Dune 72 at thick laminae of dark sand (consisting 170– 72 cm, 160 to 20 cm at Site 70, at mainly of mica and heavy minerals) al- 120 to 33 cm at Site 67 , at 150 to 95 cm ternating with 3 – 8 cm thick laminae of at Site 68, at 70 to 0 cm at Site 74 base, light coloured sand (consisting mainly and at 180 to 60 cm at Site 74 lee. of quartz and feldspars). Interlayers of •Positioned in the middle part of the coarse light sand are usual. Deforma- section, ranging at 150 cm to 20 cm. tion structures, indicating the influence However, due to a hiatus between Unit of denivation processes can occur. Ob- 1 and Unit 2 , the latter can be posi- servations of soil formation and a char- tioned right on top of the base sedi- coal horizon on the contact with the ments. upper unit came from Site 62. •The strata are defined as dune foreset •Packing of the unit is close, at high-an- cross-bedded lamination, dipping at gle foresets displaying a tilted rhombic high to intermediate angles of 34˚ to 10˚, 86 packing arrangement with minimum indicating advance of the dune. The UNIT 3

UNIT 2

UNIT 1

BASE

Figure 46. The model of the lithostratigraphic units at Ijjävri dune field.

strata indicates a combination of grain- a charcoal horizon underlain by a soil flow and grainfall deposition. formation on the top of this unit. Up- •The bulk sediment is closely packed per contact can be erosional (often with strata consisting of thin foresets ( 0 . 5 – 1 deflation surfaces) or gradational (of- cm) of dark sand inter-bedded with 3 – ten with an intact soil horizon). 5 cm thick foresets of light sand bear- UNIT 3 : ing thin laminae. The strata often ac- •Positioned at the uppermost part of the commodates coarse layers of light sand. section, ranging from 90 to 0 cm. •Occurrence of charcoal horizons indi- •The unit can be defined as tabular fore- cating forest fires are usual within this set laminae dipping at low-angles of 0 – unit. 5 ˚. •A characteristic of this unit is, that the •The bulk sediment is re-accumulated strata becomes poorly stratified, with dune sand characterized by loose pack- looser packing and dimpled strata to- ing, being almost unstratified due to the wards the upper contact. Often there is disturbance of soil development, roots 87 and denivation structures. In the upper- 4.6.1. Radiocarbon dates most part the strata is often dimpled. A total amount of 54 conventional radio- •No observations of charcoal horizons carbon dates and 4 AMS dates were ob- indicating forest fires were detected in tained, from samples consisting of charcoal this unit. pieces collected from the charcoal horizons •Packing of the sediment is loose. Part- ( Table 24). From the Muddusjävri dune ly this is due to roots and the soil form- field 21 samples were collected and dated ing process, partly due to the deniva- with the conventional radiocarbon meth- tion process and the nature of the re- od (eight of them were analysed at the Ge- accumulation process. ological Survey of Finland, during 1989– 1990 and 13 at the University of Helsinki 4.6. Dating of the aeolian during 1991– 1994). From Ijjävri four sam- events ples were dated using AMS technique and Discussion of the aeolian reactivation his- 28 samples were dated using conventional tory in Lapland has been going on for radiocarbon method (two of them at the about 30 years, since the pioneering study Geological Survey of Finland and 26 at the by Seppälä ( 1971). On several occasions, it University of Helsinki). has been stressed that the aeolian history The laboratory codes ( Table 24, first still cannot be considered as thoroughly column) indicate the place where the sam- understood ( e.g. Seppälä, 1995; Käyhkö, ple was analysed; Hel (conventional radi- 1997 ; Käyhkö et al., 1999). One of the pri- ocarbon) and Hela (AMS) are samples an- mary objectives of the study ( Chapter 1 ) alysed at the University of Helsinki (with was to tackle the problem from a new per- AMS measuring at the University of Upp- spective. sala) while Su stands for Geological Sur- The aim was to determine how precise vey of Finland (conventional radiocarbon). information can actually be obtained Dune number and sampling depth in cen- about the reactivation phenomenon on the timetres are indicated in the following col- aeolian landforms using radiocarbon and umns. Note that the samples Su- 1807 and TL/OSL dating methods, if the whole suc- Su- 1869 are dated from the same charcoal cession of stabilization/activation phases horizon. There are three samples from was systematically revealed and studied. Dunes 6 , 15 and 16 with the comment With respect to the radiocarbon dates, the “double” at the sampling depth. These were concept was to date samples of an uniform series of two charcoal horizons that were material (charcoal) and to obtain an ade- accidentally combined in the laboratory. quate number of dates that would enable The relevance of these dates is discussed reliable interpretation of the aeolian his- in Chapter 4 . 6 . 3 . Date and ± columns in- tory. TL/OSL dates would be obtained dicate the result in radiocarbon years be- from selected samples, in order to clarify fore year 1950 and uncertainty which is and support the constructed radiocarbon presented with 1 sigma. Delta 13C values based chronostratigraphy. indicate that the dates are corrected for iso- Furthermore, accurate and reliable dat- topic fractionation to correspond to the ing of the lithostratigraphy and the sand δ 13C value of –25 ‰. units is essential for the concluding Calibration of the radiocarbon dates is palaeoenvironmental model ( Chapter 6 ) presented in Table 24 with two procedures. and thus provides a framework for the The first one is done with CALIB REV 4 . 4 whole study. (Stuiver and Reimer, 1993) using a sample 88 age span of 50 yrs. The calibration curve Table 24. Radiocarbon dating results of the charcoal samples from the Muddusjävri and the Ijjävri dune fields. See text for detailed explanation of the two calibration procedures used here.

Lab-no Dune Sampling Date ± δ 13C cal. yrs BP cal. yrs BP cal. yrs BP Relative area depth 14C yrs ‰, PDB (span 50 age single age under (cm) BP yrs) range(s) (2 range probability sigma) (2 sigma) distribution % Su-1867 33 1000 60 -27.60 890 1040-770 1010-790 0.933 (2) Su-1868 3131580 70 -28.40 1470 1620-1320 1610-1330 0.984 Su-1807 3192350 170 -27.00 2370 2770-1980 2770-1990 0.998 Su-1869 3192160 80 -28.40 2150 2330-1970 2340-1990 0.985 Su-1870 3294540 160 -27.90 5210 5590-4840 5490-4840 0.933 Su-1871 3344760 90 -28.10 5480 5650-5310 5650-5310 1 Su-1809 45 1320 50 -26.80 1230 1320-1170 1310-1170 0.956 Su-1810 53 1170 50 -29.20 1080 1240-960 1180-970 0.963 (2) Hel-3437 6double 680 80 -27.70 640 740-530 740-530 1 Hel-34388 6870 80 -26.50 800 930-680 930-670 1 Hel-3012121 510 100 -26.90 520 670-320 660-420 0.879 (2) Hel-3019 12 56 6560 80 -27.00 7450 7590-7310 7590-7310 1 Hel-301812606920 130 -27.10 7770 7990-7510 7980-7560 0.984 (3) Hel-3179 12 68 7770 170 -24.20 8640 9020-8200 9010-8280 0.974 (2) Hel-3017 12 70 7620 100 -27.00 8390 8590-8200 8590-8280 0.883 Hel-3439 14 12 820 90 -26.90 790 940-640 930-650 0.989 Hel-3011 14 23 1660 90 -24.20 1560 1790-1350 1740-1350 0.975 Hel-3010 14 45 2320 90 -25.00 2380 2710-2130 2540-2120 0.848 (3) Hel-3009 14 100 3360 130 -24.90 3640 3940-3330 3930-3340 0.994 Hel-3440 15 double 800 70 -26.80 780 910-650 800-650 0.821 Hel-3441 16 double 860 90 -27.00 800 940-670 930-660 0.998 Hel-3442 20 30 2360 120 -26.90 2430 2740-2130 2740-2150 0.991 Hel-3411 52 95 1150 90 -26.40 1100 1270-930 1270-930 1 Hel-34125397700 80 -28.00 650 760-530 770-540 0.995 Hel-3413 53 100 1990 90 -27.00 1940 2160-1720 2150-1710 0.995 Su-1924 53 110 4110 80 -28.60 4630 4830-4430 4830-4500 0.896 Su-1925 53 125 6120 100 -28.30 6990 7240-6750 7250-6780 0.978 Hel-3414 54 65 1470 90 -27.20 1390 1550-1200 1550-1260 0.973 (2) Hel-3415 54 50 1900 80 -26.00 1820 201-1620 2000-1690 0.947 Hel-3416 54 70 3510 90 -26.60 3780 4070-3560 3990-3560 0.972 (2) Hel-3417 54 80 4670 140 -26.70 5310 5660-4900 5590-5280 0.898 (2) Hel-3418 55 60 670 90 -27.20 640 750-520 740-520 0.991 Hela-23 63 20 3820 70 -26.30 4230 4410-4000 4420-4070 0.941 (2) Hel-3419 65 25 940 90 -27.00 840 1030-680 990-680 0.986 (2) Hel-3420 65 35 3700 90 -26.10 4080 4350- 4300-3830 0.973 3760(2) Hel-3421 66 25 880 80 -26.80 810 930-680 930-670 1 ( 2 )

89 Lab-no Dune Sampling Date ± δ 13C cal. yrs BP cal. yrs BP cal. yrs BP Relative area depth 14C yrs ‰, PDB (span 50 age single age under (cm) BP yrs) range(s) (2 range probability sigma) (2 sigma) distribution % Hel-3422 67 33 1150 90 -27.00 1100 1270-930 1270-930 1 Hel-3502 68 69 850 110 -26.70 800 970-570 970-650 0.98 (2) Hela-24 68 76 1740 75 -25.60 1680 1850-1500 1830-1510 0.978 Hel-3503 68 85 1910 120 -27.10 1840 2130-1560 2120-1550 0.998 Hel-3429 69 48 2890 110 -26.40 3040 3320-2790 3270-2780 0.966 (2) Hel-3430 69 82 6040 120 -26.70 6930 7230-6630 7210-6640 0.991 Hel-3431 69 108 6790 100 -27.50 7630 7810-7460 7790-7460 0.981 Hela-25 69 128 7650 80 -26.50 8460 8610-8230 8600-8330 0.979 (2) Hel-3016 71 118 1090 120 -26.10 1020 1260-770 1260-790 0.987 Hel-3015 71 128 2190 90 -25.70 2160 2350-1960 2350-1990 0.985 Hel-3178 71 141 3090 150 -26.80 3250 3620-2880 3590-2920 0.971 Hel-3014 71 150 5610 120 -24.90 6430 6690-6170 6670-6170 0.988 Hel-3013 71 155 5400 120 -26.10 6170 6410-5930 6360-5930 0.957 Hela-26 72 72 4380 100 -24.30 5050 5310-4710 5310-4810 0.977 (2) Hel-3504 73 35 1990 100 -27.00 1940 2300-1710 2160-1710 0.972 (2) Hel-3432 74 50 1620 90 -26.10 1520 1720-1330 1710-1330 0.997 Hel-3433 74 60 2880 80 -25.90 3020 3230-2800 3210-2840 0.949 Hel-3434 74 80 3710 130 -26.00 4060 4410-3710 4410-3800 0.945 Hel-3435 75 100 1030 80 -26.70 930 1130-750 1090-760 0.968 (2) Hel-3436 75 120 4970 100 -26.80 5740 5930-5490 5920-5580 0.973 (2)

selection was IntCal98. 14C. Version CAL- smoothing of the calibration curve. Cali- IB REV 4 . 4 uses a probability method pro- bration curve selection was the same, ducing calibrated age ranges (Stuiver et al., IntCal98. 14C. The age range produced with 1998). In this study the probability 2 sig- the 2 sigma probability method that has ma enclosing 94. 5 % area was used. The the largest relative area under the proba- ranges are presented in the output file of bility distribution is presented in the col- the version 4 . 4 with the relative areas un- umn “ cal. yrs BP single age range ( 2 sig- der the probability distribution. In Table ma)” and the area under the distribution 24, a single estimate was calculated as the is given in the last column. weighted median age estimate and present- Age estimates were plotted for both ed at column “cal. yrs BP (span 50 yrs)”. fields (Fig . 47) with the maximum age The age range(s) are presented in column ranges (first procedure, see above). Sugges- “cal. yrs BP age range(s) ( 2 sigma)”. The tions for the aeolian activation periods are smoothing of 50 yrs produced seldom marked with shaded columns. In those cas- more than one age range. However, the es, where the suggested active periods are number of age ranges other than 1 are in- synchronized on the dune fields, the darker dicated in parentheses, and the presented shading indicates the overlapping time pe- age range comprises all ranges. The second riod. These suggested periods of aeolian procedure was done with CALIB REV 4 . 3 activity are discussed in detail with refer- 90 (Stuiver and Reimer, 1993) without ence to other proxy data in Chapter 5 . 6 . Figure 47.Age estimates with age ranges (2 sigma) for the radiocarbon dates of Muddusjävri (upper part) and Ijjävri (lower part) dune fields. Suggested aeolian activation phases are presented with shaded areas (see discussion 5 . 6 ).

The probability distributions ( 2 sigma) 4.6.2. Thermoluminescence (TL) for each radiocarbon age estimate were and Optically Stimulated Luminescence plotted separately for both dune fields and (OSL) dates sum probability distribution plots were Fifteen samples were collected for TL/OSL produced using the CALIB REV 4 . 4 (Stu- dating. The number of these samples was iver and Reimer, 1993 ) tools option. The limited, hence the dating strategy was two- separate plots for the age estimate distri- fold: first to establish the dates for the butions for both fields are presented in boundaries of the lithofacies units and sec- Appendix I. The sum probability curves for ond to establish dates for the reactivation both dune fields are presented in Fig . 48. processes and an independent control for 91 Figure 48. Sum probability curves for the radiocarbon age estimates at both dune fields. For the probability distributions of each radiocarbon age estimate, see Appendix I.

the radiocarbon estimates. The results are Dating of these materials was not possible presented in Table 25. Samples K 10, K 11 and as indicated from the dating result of the K 15 were taken from the underlying gla- similar sample K5 (see discussion 5 . 6 ). ciofluvial material at the Ijjävri dune field.

92 Table 25. TL/OSL dating results of the sand samples from Muddusjävri and Ijjävri dune fields.

Sample Site depth Dose ra te TL OSL TL/OSL ± comment (cm) (mGy/a) (max) (max) K1 12 72 2,31 8 300 6 600 6 400 800 K2 12 140 2,40 11 900 10 700 10 700 1200 K3 10 70 2,50 11 500 8 900 8 700 1100 K4 75 100-110 2,65 6 000 6 600 6 600 700

K5 74 140 2,96 57 000 19 700 16 700 5000 “old” glaciofluvial K6 74 70-80 2,61 8 000 7 700 7 700 1000 K7 74 60-70 2,72 14 500 10 600 - - K8 66 120 2,75 12 100 10 400 10 200 1200 K9 66 80 2,61 10 200 9 000 8 700 1200 K10 76 102-95 - - - - - “old” glaciofluvial K11 76 52-47 - - - - - “old” glaciofluvial K12 73 40 2,91 1 900 880 1 000 200 K13 70 60 2,83 320 160 300 100 K14 14 50 2,64 9 700 8 400 8 400 1000 K15 63 70 - - - - - “old” glaciofluvial

93 5. Discussion

5.1. Morphological features c) presence of vegetation ( e.g. Galon 1959, The study areas are characterised by nets Flint 1971). Consequently, parabolic dunes of parabolic dunes (in Muddusjävri dune are known to occur in three different en- field) and blowout systems. In order to vironments (Cooke et al., 1993 and the ref- understand the morphology, it is necessary erences therein): to elucidate what morphological controls a. In cold climates, both modern and an- are effecting the initial dune formation. cient The main reason for the accumulation b. On temperate and wet-tropical coasts of sand is the presence of a topographic c. On the edges of warm deserts as both depression (Brookfield, 1983 ). A valley like active and ancient features depression also directs the local wind sys- Other factors besides vegetation, contrib- tem. Once formed, the dunes themselves uting to the formation of parabolic dunes can trap more sand (Bagnold, 1941 ), when are high moisture content, limited sand the conditions for the dune building re- supply and high wind speed. The impor- main suitable. Thus, a very important fac- tance of high moisture content has been tor in the periglacial conditions of north- emphasized by David ( 1977 ). Cooke et al., ern Lapland was the availability of sorted 1993 suggests that the series of concentric, fine material. The source area is often sit- net forming parabolic dunes testify to al- uated next to the dune field and the aeo- ternating periods of activity and stability. lian transportation distance is not very Dune form succession similar to the long. pattern presented by Kádár, 1938; Lands- An ancient vegetated dune field is es- berg, 1956 and Galon, 1959 has been inter- pecially interesting in terms of its mor- preted in early Holocene parabolic dune phology. The ancient wind system is cap- formations in Finland (Lindroos, 1972). tured in the shapes of the dunes and can According to this model, the dunes evolve be interpreted. Even the reactivation of a from (A) a single transverse dune ridge to dune field can be recognized in favorable (B) a parabolic-shaped dune. The parabol- places since the reformation of the dunes ic dune becomes increasingly deeper (C) has not been very effective. The shape of until the arms separate and the dune pa- the dunes is typically parabolic, where the rabola has become two parallel dune ridges arms of the dune open towards the wind. (D) ( Fig. 49). The formation of parabolic dunes often is Käyhkö ( 1997 ) draws attention to par- connected to a) cold climate conditions, abolic dune form being one argument for 94 b) relatively high ground water table and short lived initial dune phase by Seppälä Figure 49. Model of the transition of the dune forms, compared to the model presented by

Lindroos (1972 ). Dotted line in phase C2 indicates a joining dune front.

( 1971). Käyhkö ( 1997 ) states that this argu- Most exposed patches on dune environment does not gain any support from the ment re-vegetate naturally in a short pe- dune literature ( c.f. Pye and Tsoar, 1990; riod of time, only a minority becoming Cooke et al ., 1993; Lancaster, 1995; Living- blowouts (Jungnerius and Van der Meulen, stone and Warren, 1996), which does not 1989). On the other hand, many blowouts consider parabolic dunes as temporary, in Lapland are several centuries old (Sep- short-lived forms that would evolve to oth- pälä, 1995). In blowouts wind is increasing- er dune types in the longer term. In fact, ly funneled through a deepened de-vege- Seppälä ( 1971) rather suggests that since tated area. Wind speeds reach a maximum there are no parallel dune ridges, the dunes in the base of the hollow, and produce have not been traveling far and thus are double vertical-axis vortiles, sweeping sand suggesting a relatively short initial dune from the base towards the sides (Hails and forming phase. The model for the devel- Bennett, 1980). There is first a positive opment of parabolic dune forms accord- feedback process: erosion increases as the ing to Pye (1982 ) is presented in Fig. 50. blowout deepens, but then negative feed- This model corresponds well to the origi- back as the blowout expands, funneling nal model presented by Lindroos ( 1972). becomes less pronounced and an equilib- The only difference is that the initial form- rium shape is reached. If erosion is severe ing of the phase one transverse dune is a dune may form on the downwind side connected to a blowout system. of the blowout and advance over nearby 95 Figure 50. The development of a parabolic dune (modified from Pye, 1982 ).

vegetation, enlarging the area of bare sand pecially its directional variability, has fre- down wind, and perhaps eventually ad- quently been regarded as the major factor vancing and expanding to become a para- determining dune type. Wind strength may bolic dune (Cooke et al., 1993). also be a factor (Ahlbrandt and Fryberger, There are several controls of dune mor- 1980) and vegetation is locally important phology. Lancaster ( 1989) has listed six (Hack, 1941 ; Ash and Wasson, 1983 ). As main factors that influence the type, align- dunes are the product of an ongoing proc- ment, size and spacing of dunes. The na- ess of sediment accumulation, the effects ture of dune sands, especially their grain of time should also be considered size and sorting characteristics was regard- (Mainguet and Callot, 1978; Walker and ed by Wilson ( 1972, 1973) as a major influ- Middleton, 1981; Lancaster 1983 ). ence on the size and spacing of dunes. The On the basis of this study, the nature of role of sand supply has been considered to the wind regime and strength, together be determinant of dune type (Mabbutt, with the sand supply provided through 1977;Wasson and Hyde, 1983 ), although deglaciation processes and the role of the satisfactory definitions of this parameter vegetation are the most important factors are difficult to achieve (Rubin, 1984 ). In contributing to the morphology of the 96 deserts the nature of the wind regime, es- dunes in the northern Finnish Lapland. Both dune fields have been reactivated repeatedly during circa 10 700 years, and the morphology reflects the whole history of the wind system affecting the fields. However, the first initial dune forming phase has contributed most to their morphology.At Muddusjävri the original dune building wind has been blowing from the SW and the reworked orientation is W-E, as demonstrated by bending of a dune parabola ( Fig. 51). Active aeolian processes are absent at the Muddusjävri dune field, due to dense vegetation and lack of forest fires during the last century. The tree rings of one storm-fallen pine in the area were calculated and they indicated well over 200 years of record (M. Helminen, pers. Figure 51. Bending of the nose of a dune comm.). indicating changed wind direction during the The nature of the parabolic dune sys- reactivation process. Example from the tem suggests that in the periglacial envi- Muddusjävri dune field. ronment of northern Finnish Lapland during the initial dune forming phase, dunes have originated from both transverse shore dunes and blowouts at the barren sand sur- total sample weight left on these sieves was face. rather minor. The mica flakes are very thin, thus their mass is unimportant in compar- 5.2. Grain size distribution ison with other minerals. In fact, Dune 1 , Dry sieving was selected as a method for with most abundant mica flake content, grain size analysis. However, several stud- consists of rather fine material ( M z = ies have shown that particle shape can have Φ 2 . 82). Most mica flakes were trapped at a significant influence on sieve-size data Φ 0 . 00 sieve, hence resulting in a 99. 97 (Komar and Cui, 1984 ; Kennedy et al ., passing weight per cent. 1985 ), and difficulties may be encountered The mean size of the dune samples var- when for instance samples contain a mix- ies between Φ 1 . 30 and Φ 3 . 00 at the Mud- ture of quartz and platy shell fragments dusjävri dune field corresponding to val- (Carter, 1982 ). Pondering the nature of the ues of medium sand to fine sand (e.g. Fried- study material, the particle shape was not man and Sanders, 1978). At Ijjävri the considered as a relevant threat. There was, range is wider, varying between Φ 1 . 50 and however an element at the sediments of Φ 3 . 40, e.g. from medium sand to very fine Muddusjävri dunes that came close to sand. However, average mean size of the causing a problem: abundant flat flakes of material is slightly finer at Ijjävri (Φ 2 . 47) mica (mainly biotite and muscovite). Dur- than at Muddusjävri (Φ 2 . 13). The finest ing the analyses, it was observed that the material of the studied dunes at Muddus- mica flakes were accumulating to the two jävri is to be found in Dune 1 (Φ 2 . 82), at uppermost sieves: Φ – 0 . 45 and Φ 0 . 00, that the eastern side of the dune field. The es- were otherwise empty. On the other hand, timated transportation distance at Dune 1 it was also noticed that the portion of the is 6250 m, the longest of the studied dunes. 97 Consequently, mean size values for some ( Φ 1 . 7 ), whereas samples below and above of the less shifted dunes ( 19 and 21) at the this unit were much finer ( Φ 2 . 5 and western side of the field are the coarsest Φ 2 . 8 ). The sample was taken from a 5 cm (with mean sizes of Φ 1 . 70 and Φ 1 . 84, re- thick layer of coarse sand. Another site spectively). The plotted average mean grain with the minimum value for main grain size from each dune against transportation size at Ijjävri was Dune 70, sample 70/lee- distance is presented in Fig. 52. Transpor- side/ 60cm ( Φ 1 . 5 ). Again this sample is tation distance was measured as the short- from a unit of whitish/yellowish, coarse est distance between a dune and a base line, sand, above an erosional contact to the which was drawn to SW-side to the dune strata below. The sample was primarily field. There are some dunes like 17 and 16, taken for SEM- analysis.

which do not fit well to the model. Dune Average value for sorting, σ 1 , is identi- 17 is located near Lake Paatari at the south cal on both fields, 0 . 65. The value is indi- end of the dune field. Probably the mate- cating moderately well sorted material. The rial has not traveled far in aeolian trans- values vary between 1 . 00 and 0 . 40 (mod- portation, but may have inherited features erately sorted to wellsorted ) at Muddusjävri from beach sand. The other location, Dune and between 1 . 10 and 0 . 40 (poorly sorted 16 is situated at the border of the dune field to well sorted ) at Ijjävri. The values of 1 . 00– as well. It locates west from the village of 1 . 10 are indeed rather poorly sorted. The Tirro, at the other side of the River Vasko, two most poorly sorted samples at Mud- at the western limit of the Muddusjävri dusjävri dune field derive from Dunes 16/ field. top/50 cm (with mean grain size of Φ 1 . 3 ) The finest material observed at Ijjävri and 2 /top/ 100 cm (mean grain size Φ 1 . 7 ). field is from Dunes 63 (Φ 2 . 8 – 3 . 1 ) and 66 Typically during the transportation proc- ( Φ 2 . 5 – 3 . 4 ). The coarsest dune material at ess sorting becomes better and mean grain Ijjävri was found at Dune 67/leeside/95cm size smaller, e.g. towards the surface in a

Figure 52.Average mean grain size (phi) against transportation distance on Muddusjävri dune field. R2 = 0 . 52 (n = 16, p = 0 . 002 ). Trendline and statistical analysis were done ignoring Dunes 16 98 and 17(labeled with numbers, see text. If Dunes 16 and 17 are included R2 = 0 . 21 (n = 18 , p = 0. 06). dune section and from windward side to both fields. The slight positive trend on the leeside, although the variation within Muddusjärvri suggests a little more pro- a dune is not large ( Fig . 53). At both dune nounced tail of fine material compared fields it was noticed that the average pa- with log-normal distribution (symmetri- rameter values grouped according to the cal). Also the average value for kurtosis, K G sampling sections, did not differ very much evokes very similar results on both dune from each other ( Tables 8 and 11). This can fields; 1 . 05 at Muddusjävri and 1 . 09 at be observed both on an individual dune Ijjävri. This value implicates mesokurtic ( Fig . 53) and, in fact, on every dune, while frequency distribution, indicating Gaus- comparing the values from windward-side sian profile of the grain size distribution. to leeside. The transition of the grain size The original idea of transforming log- parameters becomes more apparent when arithmically both grain size and grain fre- comparing sites from different areas of the quency scales by Bagnold ( 1937) has be- dune fields. At Muddusjävri, where it is come later popular among statisticians. possible to use the concept of transporta- Log-hyperbolic parameters were rediscov- tion distance, this can be seen at Figs 52 and ered in late 1970 s and have received quite 54. The best sorted material at Muddus- a lot attention ( e.g. Barndorff-Nielsen, jävri originates from Dune 1 (with longest 1977 ; Bagnold and Barndorff-Nielsen, 1980 ; transportation distance). The average val- Barndorff-Nielsen et al ., 1982 ; Hartmann, ues for inclusive graphic deviation for each 1988). The point of the method is to gain dune at Muddusjävri field are plotted more information about the tails of a grain against transportation distance in Fig. 54. size distribution. However, in this study,

The values from the Sites 21, 17 and 19 with both Sk1 and K G grain size parameter val- the shortest transportation distance ( 500 ues indicate that the grain size distribu- m) are not as poorly sorted as might be tions are indeed very symmetrical. Hence expected. They all, with Dune 16 locate the information obtained from the tails near the western border of the dune field. remains very limited and unsubstantial. Estimation of the transportation distance For this reason, log-hyperbolic parameters is not reliable there, because the area next are not used. to the dune field in west is peat-covered. Many attempts have been made to dif- Nevertheless, a descending trend of inclu- ferentiate between sediments from differ- sive standard deviation values indicating ent environments using bivariate scatter- better sorting can be observed relative to gram plots ( e.g. Friedman, 1961 ; 1967 ; transportation distance. At Ijjävri dune Besler, 1983 ; Khalaf, 1989) and statistical field the poorest sorting was observed at analysis of graphical grain size parameters.

Dunes 67/windward/20 cm ( σ 1 = 1 . 00), 70/ Plots of mean size against sorting and windwardw/ 60 cm ( σ 1 = 1 . 10), 76/top/40 cm skewness have generally proved most use-

( σ 1 = 1 . 10) and interdunes 61/ 30cm ( σ 1 = ful (Pye and Tsoar, 1990). Besler ( 1983 ) was

1 . 00) and 74/ 130 cm ( σ 1 = 1 . 00). Site 76 at able to differentiate between sedimentary Ijjävri is actually a shore formation, hav- environments at Sahara, Namib and Kala- ing only 30 cm of reworked aeolian mate- hari deserts. The usefulness of her concept rial on top, which explains the poor sort- “response diagram”, bivariate plot of mean ing and coarse grain size ( Φ 0 . 5 ). grain size against sorting, has later received

Inclusive graphic skewness, Sk1 values criticism by Vincent ( 1985 ) and Thomas from the dune fields average out + 0 . 10 at ( 1986; 1987). As there is considerable over- Muddusjävri and – 0 . 03 at Ijjävri dune lap between grain size parameters from field, indicating symmetrical skewness on different environments, the use of statis- 99 Figure 53. Mean grain size, Mz , Standard deviation, σ 1 , Skewness, Sk1 , and Kurtosis, KG on windward, top and leeside sections on standard sampling depths 100– 110, 70 and 30 cm at Dune 20, Muddusjävri.

Figure 54.Average mean sorting ( σ 1 ) against transportation distance on Muddusjävri dune field. R 2 = 0 . 43 (n=15, p=0 . 008). Trendline and statistical analysis were made ignoring outlier points at 100 the trasportation distance of 500 m (indicated with a frame, see text for discussion). tical analyses like factor analysis (Klovan, Bivariate plots of main grain size and 1966) and multiple discriminant analysis skewness, as well as mean grain size and (Moiola et al ., 1974) have been able to pro- sorting are being presented in Figs 55 and duce better discrimination (Pye and Tsoar, 56. 1990). On the dune fields of this study, the The response diagram (modified Fried- relationship between the source material, man-diagram; Besler, 1983 ) does not fit dune sediments and adjoining deposits has very well to the data. The M z / σ 1 values of been studied by several methods. Al- the samples do not form distinguishable though, the grain texture analysis en- populations that would differentiate aeo- hanced with SEM technique ( Chapter 4 . 4 ) lian environments and thus reactivation. In should be most useful in this context, the Fig. 55A the aeolian mobility population is bivariate plots of grain size parameters of- mainly formed by the samples of the Dune fer an important insight into the matter. 1 that has the longest transportation dis-

A Muddusjävri dune field

0,6 k 0,4 , S

ss 0,2 dune

ne 0 delta/interdune ew -0,2 Sk -0,4 01234 Mean grain size (phi)

B Ijjävri dune field

0,6 1

k 0,4 0,2 , S dune

ss 0 delta

ne -0,2 -0,4 interdune ew -0,6 Sk -0,8 00,5 11,5 22,5 33,5 4 Mean grain size (phi)

Figure 55. Bivariate plot of Mean grain size, Mz, against skewness Sk1 on Muddusjävri dune field A, and on Ijjävri B. Diamond shapes dune samples, squares and triangles for other sediments. 101 Figure 56. Bivariate plot of Mean grain size, Mz, against Standard deviation, σ 1 , on Muddusjävri dune field A, and on Ijjävri B. Diamond shapes indicate dune samples in both graphs, squares indicate delta and interdune sediments (A), or delta sediments (B) and triangles stand for inter- 102 dune sediments (B). tance from the base line, which is sufficient Ijjävri (Fig . 58). The comparison of the Figs to explain the values for both mean grain 57 and 58 demonstrates the fact that the size and sorting of the material. The other grain size of the material at the Ijjärvi dune two values at M z Φ 2 . 8 / σ 1 0 . 9 are from field is finer than that at Muddusjävri. Dune 7 , which is located at the edge of Lake Muddusjävri, at a physical boundary 5.3. Grain shape of the dune field. Accumulation of dune The surface texture of quartz grains reflects fronts at the lakeside can explain moder- the effects of aeolian and sub-aqueous ately sorted material combined with small transportation especially well in grain size mean grain size rather than origin from of 0 . 71 mm (Cailleux, 1952). In visual com- fluvial river sand. However, most of the M z / parison of the surface textures at other

σ 1 values from the Muddusjävri dune field grain sizes; Φ 1 . 00 (0 . 50 mm) and Φ 0 . 00 fall in the aeolian stability field, which ( 1 . 00 mm), the concept was accepted and seems reasonable. The response diagram the grain size of Φ 0 . 49 (0 . 71 mm) was se- does not work very well with the data from lected for both binocular and SEM-analy- Ijjävri either. Even though most of the val- ses. Roundness of the quartz grains de- ues fall at the aeolian stability field, the creases clearly as the grains get finer, due large number of values at “the river sands” to sorting of the grains into different trans- -category is not satisfactory. Those values portation loads. Transportation along the derive from both dune and deflation ba- surface by rolling produces effective abra- sin sections. The “river sand” population sion to larger grains, while the finer ones consists of Sites 61, 65, 66, 67, 68 and 74, remain less rounded in the suspension and while only 61, 66 and 74 embody fluvial in the saltation processes (Bagnold, 1954). sediments. Dunes 67 and 68 are located at From the very beginning of the analyses, the northern end of the dune field. Sites it became obvious that the geological set- 63, 52 and 54 locate at the NW-side of the ting of the study areas provided a suitable field and their M z / σ 1 values indicate aeo- case for grain shape studies. The quartz lian mobility. The material from Site 63 is grains had inherited glacial, sub-aqueous delta sediment, and should therefore indi- and aeolian elements, which were clearly cate rather river sand quality than aeolian distinguishable. mobility. Dunes 52 and 54 are indeed re- Binocular microscopy analysis was ap- activated and the category aeolian mobil- plied to 225 samples, and a total of 22 500 ity fits well. quartz grains were visually categorized. The log-probability curve has been The results presented in Table 16 indicate widely used as a tool for tracking the orig- that at Muddusjävri most of the quartz inal grain populations produced by the grains were classified into groups C, sub- sorting processes in entrainment and angular (average 51. 4 %) and B, angular transport. The populations can be distin- (average 24. 4 %) and whereas at Ijjävri guished as separate segments of the curve most of the grains fell in classes C, sub- ( e.g. Visher, 1969 ; Middleton, 1976). How- angular (average 44. 0 %) and D, sub- ever, later the sedimentary significance of rounded (average 34. 9 %). It can be seen in these segments has been much debated Fig. 59 that the distribution of the materi- (Walton et al ., 1980; Christiansen, 1984). al is very Gaussian at Muddusjävri, where- Examples of grain size frequency histo- as at Ijjävri the skewness of the distribu- grams and cumulative weight percentage tion of the material is slightly negative. curves for some samples are being present- Nearly 80% of the entire material falls in ed from Muddusjävri (Fig . 57) and from classes C and D at Ijjävri. Material in Ijjävri 103 Figure 57. Grain size frequency histograms and cumulative weight percentage curves for the samples 3 lee 100 cm and 15 lee 60 cm from the Muddusjävri dune field.

is better rounded than at Muddusjävri in ie et al ., 1987). The exceptions are cases general. It can also be seen at Table 16 that where the sands have been recycled from class C has maximum percentages on both older sedimentary cycles, as in Sossusvlei, fields and that minimum values are rath- Namib, where the reddish dune sand orig- er high. Sub-angular (class C) grains are inates from the underlying red sandstone predominate in both dune fields. Com- (Kotilainen, 1989). Most of the dune sand pared to other dune studies, from cold and grains in the Simpson Desert of Australia hot environments around the world, the are sub-angular to angular, with no notice- result is cogent. able rounding being accomplished in the Although, the impression was given in present dune environment (Folk, 1978). many early papers that desert dune sands Goudie and Watson ( 1981) examined sand are typically near-spherical and well grains in 108 dune sand samples from all rounded ( e.g. Sorby, 1877 ), more recent over the world and concluded that well studies have indicated that most quartz rounded grains are relatively rare (about dune sand grains are actually not well 8 % of the grains examined). Their results 104 rounded (Goudie and Watson, 1981; Goud- indicate that in dune sands from the Thar Figure 58. Grain size frequency histograms and cumulative weight percentage curves for the samples 65 windward 100 cm and 74 lee 145 cm from the Ijjävri dune field.

A % 30 B 20

0 ABCDEF Roundness classes

Figure 59. Average proportions (%) of quartz grains in roundness classes A (very angular) to F (well rounded). A Muddusjävri and B Ijjävri dune fields. Classification according to Powers (1953) and Shepard ( 1963 ). 105 and California the predominant shape was sub-angular. Lindroos ( 1972) inferred that in post-glacial dunes of North Karelia, Fin- land, the roundness values varied between 0 . 20 and 0 . 36 indicating very angular to angular form and the sphericity values varied between 0 . 65 and 0 . 70 indicating moderate sphericity (scale after Krumbein and Sloss, 1963 ). Seppälä ( 1971) used graniformameter to determine roundness of the dune sand grains from Kaamasjoki-Kiel- lajoki area, Inari, Finland. He came to the conclusion that dune material did not manifest aeolian abrasion at any significant extent; in fact, the roundness of dune material did not differ from the roundness of glaciofluvial sand or even till. Table 13 presents roundness analyses from parent materials adjoining the Mud- dusjävri dune field. Material from Site 9 , at sampling depths of 80 cm and 45 cm (delta at the shore of Lake Muddusjävri, Figure 60.Roundness indexes based on at the northern limit of the dune field) is average proportions of each roundness class clearly less rounded than average dune from Muddusjävri dune field and their material from Muddusjävri. However, sand geometric mean values. Transportation samples from Site 22 (esker) and Site 24 distance from the base line in meters. (esker) demonstrate roundness values very similar to dune samples. Also delta samples 23 (depth 100 cm) and 9 / 120 cm indi- units, is not well rounded (Goudie and cate congruent roundness values as aver- Watson, 1981) is interesting. Quartz grains age dune material. Apparently the differ- of the first sedimentary cycle resist aeolian ence in rounding degree is not dependent abrasion very well. The role of short trans- on the transportation agent, but rather the portation distance and the influence of par- grain size. tial vegetation cover will be reconsidered There are many aspects contributing to in the synthesis of all data (Chapter 6 ). apparently poor roundness of the dune In order to be able to relate roundness sand. Folk ( 1978) deduced that poor values to transportation distance at Mud- roundness of the material in the Simpson dusjävri dune field, a roundness index for Desert,Australia, was due to partial fixing the samples was calculated using the geo- by vegetation and thus indicates that the metric means for each roundness class and grains have not been blown a great dis- their average proportions: (0 . 14 A% + tance from their fluvial source sediments. 0 . 21 B% + 0 . 30 C% + 0 . 41 D% + 0 . 59 E% This conclusion could be drawn from the + 0 . 84 F%)/ 100 . Roundness index for the roundness data of the dune material on dunes at Muddusjävri are presented in Fig. both of the dune fields studied in this the- 60.Variation of the index is minimal and sis. However, the fact that dune material, plotting against the transportation dis- 106 unless recycled from older sedimentary tance does not indicate correlation. Figure 61. Upturned silica plates on quartz grain surface. Sample 3 /top/100 cm grain 2 . Magnifica- tion of the original grain surface is 465 x.

The other parameter studied with bin- Germany. Another surface texture feature ocular microscopy was the amount of that can be associated with aeolian envi- matt/ shiny quartz grains. Tables 12 and 14 ronment is dish-shaped concavities. On signal that the proportions of matt and the contrary, conchoidal fracture, flat shiny particles vary widely within a dune, cleavage face and cleavage-planes are in- as indicated by rather high values of stand- dicative of glacial environment. Sub-aque- ard deviation. The surface texture may re- ous and glacial/sub-aqueous environments veal more when studied with SEM. are characterized with conchoidal fracture, Upturned silica planes are mainly asso- mechanical V-forms, straight or slightly ciated with aeolian environment, but are curved grooves and flat cleavage face also indicative of source material or loess. (Krinsley and Doornkamp, 1973 ), Grains with upturned silica planes occur The proportion of upturned silica plates for instance on recent dune sands in Aus- that are indicative of aeolian transporta- tralia and Libya and on Upper Pleistocene tion, is the most abundant surface texture dune sand from Roslyn NY., USA. Weath- class in most of the samples ( Table 17; Fig . ered granites in Scotland and Ireland set 61). The proportion of the class varies be- an example of source material with up- tween 85. 7 % and 45. 7 % for the dune sam- turned silica plates, whereas loess particles ples. The sample 9 /delta/90 cm from Mud- with similar features have been found from dusjävri signals a more modest proportion 107 for upturned silica plates, 23. 4 %. These smaller ( 2 . 8 % to 40. 9 %). Conchoidal frac- plates appear as more or less parallel ridges tures develop typically in quartz particles ranging in width from 0 . 5 to 10 µ m and in the glacial grinding process and colli- have been interpreted to result from break- sions during transportation. In dune sam- age of quartz along cleavage planes in the ples where the percentage of imprint class- quartz lattice (Pye and Tsoar, 1990). Exper- es indicating subaqueous/glacial transpor- imental evidence shows that the spacing tation remain as high as 40%, the aeolian and size of upturned plates might be reworking of the material has not been boardly related to wind energy (Krinsley very effective. and Wellendorf, 1980). Apart from aeolian The series of samples from Dune 3 , environments, the presence of upturned sampling depth 100 cm from windward, silica plates has been connected to weath- top and lee side sections appears to be ered granite, a source material of many rather confusing ( Table 17). The average aeolian sediments. The appearance of the values for top section 3 /top/ 100 samples surface is the same. The presence of up- are different compared to the two other turned silica plate -surfaces in delta mate- samples. The relation between aeolian/ gla- rial indicates that the feature is partly in- cial imprints is not logical. There is no suc- herited from the parent material and does cession from lee side to the windward side. not imply aeolian phenomena alone. If the individual results of 3 /top/100 sam- Dish-shaped concavities are also linked ple are perused, it can be noticed that the to aeolian transportation. Unlike other sur- variation on surface textures between the face features, these are indicative only for grains is immense. For instance the per- aeolian abrasion. Krinsley and Doornkamp centage for upturned silica plates varies ( 1973) mention examples of quartz grains between 98. 6 % to 3 . 5 %. The number of with dish-shaped concavities from Libya studied grains was only 13 from the sam- and Antarctica. These concavities are be- ple 3 /top/10, which seems rather an insuf- lieved to develop as a result of direct im- ficient number of grains. pacts rather than glancing blows during Considering the transportation distance saltation (Pye and Tsoar, 1990). This indi- and sedimentation process at the Muddus- cates quite high wind velocities. The pro- jävri dune field, the SEM-analysis results portions of dish-shaped concavities varies are logical. Dune 3 is most advanced, Dune from 5 . 9 % to 0 %, the latter in delta sedi- 12 in the middle of the field and Dune 16 ment. The presence of these features dem- at the western limit of the field. On the onstrates the effect of wind action on the other hand in Ijjävri, the samples indicate quartz grains. In fact, dish shaped concav- quite good maturation of the quartz par- ities are more frequently found on perigla- ticles that is increase in the amount of up- cial aeolian grains rather coastal sand (Pye turned silica plates, approaching the sur- and Tsoar, 1990). face, as the comparison of the samples 69, Sub-aqueous and glaciofluvial/ glacial 70 and 71 shows. transportation leaves imprints classified in In conclusion, the SEM analysis of categories 2 , 3, 4, 7 and 8 . Conchoidal frac- quartz grains proved to be a precise and tures ( Fig. 62) and cleavage planes are the useful method. The disadvantage is that it most abundant features of this kind in the is very laborious and time consuming. The quartz grain surface textures of the stud- number of studied grains has to be at least ied samples. The amount of these is high- 15 to 20 in environments with versatile sur- est in the delta sample 9 / 90 cm ( 73. 6 %), face textures. 108 while in dune samples the proportion is Figure 62. Conchoidal fracture on quartz grain surface. Sample 3 /lee side/100 cm, grain 2 . Magnification of the original grain surface is 465 x.

5.4. Grain mineralogy the transportation process sorts grains of The minerals present on both fields are the same mass into a certain transporta- typical for dune sediments (Pye and Tsoar, tion load, and subsequently to the same 1990), and the only deviant mineral sediment layer. When transported, small present is talc. However, talc occurs only particles composed of high density mate- at percentages of 1 – 2 % and was found at rial may display the same behaviour (hy- 9 out of 50 thin sections. The hardness of draulic equivalence) as much larger parti- talc is very low, 1 on Mohs’ scale. Presence cles composed of low-density material of talc grains indicates a very short-lived (Pye and Tsoar, 1990). Hardness is indica- grinding process during sedimentation tive of the intensity of the grinding proc- and is associated with talc-bearing rock ess, as the less hard minerals tend to abrade fragments in these samples. Some proper- first, and the maturity of an aeolian sedi- ties of the main minerals present in the ment is often demonstrated by a higher study sediments are presented in Table 26. content of hard minerals, like quartz. The Especially important are density and hard- hardness of garnet is roughly the same as ness. The aeolian transportation process is that of quartz, but the density is much highly selective, and sorting of the mate- greater ( Table 26). rial is effective. Grain size is strongly con- A series of four thin sections of sieved nected to mineralogy, i.e. grain density, as material was studied from the dune sam- 109 Table 26. Properties of the minerals present on the studied dune sediments.

Mineral Composition Density Hardness (kg/m 3 ) quartz SiO 2 2650 7

albite NaAlSi 3 O 8 2600-2630 6-6.5

labradorite (Ca,Na)Al(Al,Si) Si2 O 8 2690-2720 6-6.5

anorthite CaAl 2 Si2 O 8 2740-2760 6-6.5

orthoclase, KAlSi3 O 8 2550-2630 6-6.5 microcline pyroxenes (Ca,Mg,Fe) 2 (Si,Al) 2 O 6 3200-3580 5-6.5

hornblende NaCa2 (Mg,Fe,Al)5 (Si,Al) 8 O 22(OH)2 3000-3470 5-6

garnet (Fe,Al,Mg,Mn,Ca)5 (SiO4 ) 3 3560-4320 7-7.5

epidote Ca2 (Al,Fe) 3 (SiO4 ) 3 (OH) 3350-3500 6-7

zircon ZrSiO 4 4600-4700 7.5

titanite; (rutile, TiO2 3900-4250 5.5-6.5 anatase)

apatite Ca 5 (PO 4 ) 3 (F,Cl,OH) 3100-3250 5

muscovite KAl2 (Si 3 Al)O10(OH,F)2 2770-2880 2.5

+2 +3 biotite K(Mg,Fe ) 3 (Al,Fe )Si 3 O 10(OH,F)2 2900-3300 2.5-3

+2 +3 magnetite Fe Fe2 O 4 4850-5200 5.5-6.5

ilmenite FeTiO3 4600-4720 5-6

ple 5 /top/100 cm. The variation in the min- ess. The proportion of quartz was most eralogical composition of different grain substantial in the grain size fractions of size fractions reflects precisely the grain Φ 1 . 0 (0 . 5 mm) and Φ 2 . 0 (0 . 25 mm). Feld- size/grain density relationship. The grain spars were most abundant in the fraction size fractions are governed by mineralogy, of Φ 1 . 0 (0 . 5 mm), (42% of total miner- i.e. grain density ( Fig. 26), indicating in- als), while amphiboles and pyroxenes were 110 tense control of the aeolian sorting proc- most common in Φ 3 . 74 (0 . 075 mm) fraction ( 40%). Number of rock fragments de- The comparison between “parent ma- creased drastically from 28% to 7 % as the terials” and dune sediments is not very in- grain size fraction was getting finer, where- formative. If the nature of the adjoining as the number of heavy minerals increased sediments (samples 10, 13 and 18 from from 0 % to 8 %, respectively. Muddusjävri and 60, 61 from Ijjävri) is The grouping of the minerals (Q, Fds, considered, it becomes apparent that these AP, M, HM 1 , HM2 , Ox, RF, T) was com- sediments are not representative of the piled in order to simplify the point count- source materials. Site 10/ 5 / 60 is taken from ing routine without reducing the level of the sediment unit overlaying the erosion- information. The relationship between al contact with the delta material. It is quartz and feldspar content is a significant probable that the material is aeolian in or- indicator for the intensity of the aeolian igin, as the transition to the surface aeo- transportation process ( Fig. 23). There are, lian sediment units is continuous. Site 13/ however, some aspects that have to be con- 7 / 80 represents an interdune section. The sidered, while interpreting these results. sediment at the depth of 80 cm is proba- Plagioclase was often partly altered to mus- bly aeolian in origin, only the top 34 cm covite, epidote and sericite, in which case of the strata is re-deposited interdune ma- the grain was counted as feldspar. On some terial with an erosional contact to old dune occasions, it was difficult to distinguish material underneath. Site 18/ 4 / 50 is sam- between the groups of rock fragments and pled from the bank of the River Vaskojoki. feldspars, when the grain was highly al- The interpretation of the stratigraphy of tered. Also, discrimination of polycrystal- the Site 18 suggests that the sample at the line quartz grains and rock fragments con- depth of 50 cm is aeolian reworked mate- taining feldspars and quartz was quite dif- rial. At Ijjävri, the Sites 60 and 61 are both ficult. Nevertheless, the relationship be- interdune sections. Their material is deflat- tween the transportation distance and Q/ ed former delta material. Proper delta ma- Fds ratio seems relevant ( Fig. 23), suggest- terial is presented with samples 9 / 5 / 50 ing that the counting routine was fairly from Muddusjävri and 63/ 4 / 80 from competent. However, the large variation in Ijjävri, which are both thin sections of a set of values derived from a single dune, fixed delta stratigraphy. The effect of grain indicates some controversy, especially, as size is apparent at sample 9 / 5 / 50 (compare the differentiation of the mineralogy on to Fig . 26), because the grain size of the various parts of a dune, or at different sam- cross-bedded delta layers is very fine. The pling depths, seems to be insignificant high amount of mica is apparent, as the ( Figs 24 and 25). The proportion of quartz fine dark inter layers of the cross-bedding does not increase from windward side to are formed by parallel mica flakes ( Fig. 63). the lee side and from the bottom sediments to the surface, as might be expected. The 5.5.Dune lithostratigraphy transportation mechanism is not effective The lithofacies analysis of the dune fields enough to be apparent on such a small concluded that it is possible to identify a scale. However, over the whole dune field, similar succession of dune sediment units the variation in mineralogical composition in both fields. The depositional succession of the dunes is indicative of the transpor- has certain key elements shared by both tation distance. This transition towards dune areas, evidently showing the aeolian more mature sediment can be detected on history of post-glacial dune fields in north- both dune fields (Tables 18, 20), and even ern Lapland. However, there are features on the “big deflation basin” ( Fig. 27). in which these two dune fields do differ. 111 Figure 63. Thin section of a delta structure sample .

Those differences are mainly due to the Kee ( 1966) detected dominance of high- role of vegetation in the aeolian process. angle foreset in the lower part of a para- It has been demonstrated that the in- bolic dune from New Mexico while thin, ternal structures of vegetated coastal dunes low-angle to horizontal cross-sets were showed low-angle dips with an unimodal found in the upper part of the dune, es- azimuth that could be correlated directly pecially on the windward side. Pye and with the prevailing wind (Goldsmith, 1973; Tsoar ( 1990) suggest that this reflects the 1985 ). Although the first dune forming fact that the upper levels of a dune are af- phase has produced high- to intermediate fected more by cutting and filling events angle foreset strata, the unimodal azimuth in response to changing wind conditions. can be observed in both dune fields in the The wind system in northern Lapland is Unit 1 lithofacies. Development of the known to have changed to a more wester- wind system and evidence from the dune ly direction towards the end of Boreal fields will be discussed in Chapter 6 . chron (e.g. Aartolahti, 1976 ). Consequent- In both lithofacies models tabular pla- ly, the cross-bedded strata at Unit 2 are nar foresets are more common in the lower more complicated, although this may also parts of dunes (Unit 1 ) while thinner, reflect other signals. wedge-planar cross-bed sets become more Actually, the dominance of low-angle common towards the top (Unit 2 ). This cross-beds on coastal dunes has been con- 112 pattern has been observed elsewhere. Mc- sidered to reflect sand accumulation on a well vegetated surface (Goldsmith, 1973; of deflation surfaces. Bigarella ( 1975) found Yaloon, 1975). The presence of vegetation several erosional surfaces buried beneath has undoubtedly affected the dune stratig- high-angle cross-strata especially in the raphy on both dune fields especially when crestal area of a parabolic dune in Brazil. the most intense dune forming phase was He also noted that in the nose the strata over. The influence of vegetation can be dipped at high angles ( 29˚–34˚) while in the seen also as post-depositional deformation arms strata were found to be bidirection- caused by roots. This phenomenon is ap- al normal to the dune axis. Also studies of parent on both dune fields and on the parabolic dunes at Cape Flattery, North proximity of both former and present Queensland, Australia by Pye (1980) indi- ground surface (Units 2 and 3 ). cated that the deposition on the nose area The unstratified or poorly stratified of a parabolic dune is different from the strata can also imply a niveo-aeolian proc- arms. Accumulation of grainflow and ess during deposition. This may be the key grainfall deposits was found to occur only factor at the Ijjävri dune field. The defla- on outer slopes of the arms near the nose, tion process is observed to be most effec- whereas further upwind these slopes were tive in early winter before the continuous well vegetated with a well developed soil snow cover develops (Dijkmans, 1990). A horizon. The inner slopes of the arms Such deposition typically produces loose all along the dune were deflational or cov- unstratified strata detected in Ijjävri (Unit ered with a thin veneer of sand in trans- 3 ). On the other hand, the denivation proc- port parallel to the dune axis. Buried soil ess produces dimpled strata. Since the ad- A horizons were exposed along the inner vance rate of the dunes is low, the preser- slopes of the arms, especially close to the vation potential of the denivation features dune nose where the blowout troughs were is low, because of deflation (Dijkmans, being widened. Considering the position 1990). Koster and Dijkmans ( 1988) con- of a dune section in relation to the dune cluded that fossil denivation structures will morphology it can be interpreted that at occur preferably in the slip face strata in Ijjävri the dunes with a nose section are 55, the form of various local disturbances of 67, 72 and 73. Lithostratigraphy of these the aeolian stratification. Fossil denivation dune sections do not signal any special fea- forms can be observed on every section of tures linked to their position. However, it dunes in the Muddusjävri dune field, al- has to be considered that the dunes were though the quantity of such deformation not studied in toto and that the dune field increases from windward to lee side and itself is not a good example of a parabolic from the base of the section towards the dune system. At the Muddusjävri dune surface ( Fig. 32). This is suggesting that the field the dunes form a parabolic dune net, re-activation process has eroded the stra- but there the erosional deflation layers are ta from the windward side and that the ef- rare and comparison can not be made. fect of post-depositional deformation is Over-steepening of the strata can be greater that syn-depositional. observed on both dune fields. This is typ- In the Ijjävri dune field one character- ical in places where the advance of dunes istic feature is the presence of several de- is stopped and the accumulation produc- flation surfaces in the foreset strata (Units es over-steepening of the lee side. The con- 1 and 2 , and their contacts). Site 70 even tinuous presence of vegetation and the exhibits a sharp, erosional contact followed binding effect of the roots contribute to by 8 cm thick set of sub-horizontal, pla- this phenomenon. This can be observed nar laminated sand, indicating a succession along the western side of Lake Kotkajärvi 113 at the Muddusjävri dune field as well as at 5.6. Dating of the aeolian the circular dune ridge (Sites 67 and 68) events and at Ijjävri. chronostratigraphy At the Muddusjävri dune field there has Over the last few years luminescence dat- been controversy regarding the grain size ing has been improved considerably, both and grain property results of the Dunes 17, in the methods for the estimation of the 21 and 19. The estimated transportation equivalent dose (with the development of distance of all these dunes is only 500 m. the single-aliquot regenerative-dose, SAR Dune 17 is situated at the southern end of protocol, Murray and Wintle, 2000) and in the dune field, near Sotkaniemi, Dune 21 the measurement facilities (Bøtter-Jensen at the western side of the River Vaskojoki and Murray, 1999). With these new tech- and interdune section 19, between the Riv- niques OSL dating of quartz grains has er Vaskojoki and Lake Kotkajärvi. The successfully been applied to samples from grain size analyses of the sand samples a wide range of depositional environments from these sites indicate that the materi- and for a wide range of ages (Murray and als are not as poorly sorted as might be Olley, 2002 ). These environments include expected. The dune lithostratigraphy of fluvial (Stokes et al ., 2001 ) and even gla- these sites suggest that the transportation cial (Rhodes, 2000) environments. Parks distance has indeed been longer than the and Rendell ( 1992) found three deposition- estimation. The strata, especially at Site 21, al phases of aeolian silt during the Late indicates well developed stratigraphy. It Pleistocene namely 10– 25 ka BP, 50– 125 ka seems that at Site 21, the transportation BP and > 170 ka BP using TL dating of distance spans over the present paludified loess deposits in Southeast England. On area on the western side. Sedimentary data the other hand, Ballarini et al . ( 2003 ) have at Sites 17 and 19, suggests aeolian trans- shown that OSL dating of quartz gives ac- portation along the shores of the River curate ages at decadal time scale for coastal Vaskojoki and Lake Paatari. dune sand that formed over the last 300 On the basis of the average thickness of yrs as documented on historical maps and the re-accumulated sand (Unit 3 ) it can be other sources. Also Li et al. (2002 ) have concluded that re-activation of the dunes obtained very young optical age estimates has been more a significant phenomenon as the youngest age on the uppermost sand at Ijjävri than at Muddusjävri. This could unit in the northeastern deserts of China be related to different vegetation history, yielded an age of only 40 yrs. but is also governed by the distribution However, the dating results given in this and age range of the charcoal horizons in- study are based on feldspar for both TL dicating forest fires and re-activation and OSL method. This was a common pro- ( Chapter 4 . 6 ). The observation, that at the cedure in 1993, when the dating was per- Site 68 the uppermost charcoal was intact formed. The combination of the methods in places and scattered at other places, is is thought to give the most accurate age important. It suggests that some of the ( Table 25), because the bleaching process scattered pieces of charcoal in the upper tends to be incomplete leaving a significant part of other sections can actually repre- residual charge which gives more interfer- sent an independent, yet deformed char- ence on the TL signal than on the OSL sig- coal layer rather than re-deposited charcoal nal. The combination age is based on the material from elsewhere. age difference between the two methods which enables the estimation of the actu- 114 al residual charge and results in a most accurate age assessment. The closer the dates average age for the former forest, which with the two methods, TL and OSL are, the typically in Lapland has trees of ages from better the bleaching process has been leav- zero to 300 years, it can be estimated that ing the sample with an insignificant resid- an uncertainty of not more than ± 150 yrs ual charge. Samples K 5 , K 10, K 11 and K 15 (M. Eronen, pers. comm., 2004 ) has to be were taken from the underlying glacioflu- added to the calibrated radiocarbon esti- vial material at Ijjävri. The only “dated” mate in order to convert the age to calen- glaciofluvial sample K 5 from Site 74 (depth dar years. It can be concluded that taking 140 cm) resulted in highly diverged age es- into account the calibration age range ( 2 timates with TL and OSL methods ( 57 000 sigma) of 8590– 8200 yrs for the radio- / 19 700 ), thus indicating incomplete carbon date and the extra ± 150 yrs for the bleaching. Consequently, these samples can forest age, the combination age estimate only be interpreted as being deposited dur- 5600– 7200 yrs for TL/OSL date seems to ing deglaciation of the Late Weichselian ice be more reasonable, although leaving the sheet, as the nature of the samples does not results inconsistent. allow for more precise dating with TL/OSL Comparison of the age estimates can methods. AtSite 74 (Fig . 44) two other TL/ also be done at two sites from Ijjävri, with OSL dated samples were collected at radiocarbon date / TL-OSL date proximi- depths 70– 80 cm and 60– 70 cm represent- ty. At Site 75, the sand sample underneath ing water lain sediment and aeolian sedi- a 5740 cal. yrs BP ( 2 sigma range 5930– ment, respectively. The sediment deposit- 5490 yrs) old charcoal horizon was dated ed in water resulted in an age estimate TL with TL 6000(max)/ OSL 6600 (max), 14 500 (max)/ OSL 10 600 (max) and the and a combination age estimate given at aeolian sand TL 8000 (max)/ OSL 7700 6600 ± 700 yrs. Here the age estimates (max). The greater age difference in water seem reasonable and reliable. However, at lain sediment indicates again problems Site 73 the results are again more confus- with zeroing the signal, whereas the age ing. There the age estimate underneath a estimates for the aeolian material are in 1940 cal. yrs BP old charcoal horizon (2 better agreement. sigma range 2300– 1710 yrs) gives TL 1900 In many cases the use of the combina- (max)/OSL 880 (max) and the combination ages seems to be rather unfortunate. tion age of 1000 ± 200 yrs. The TL dates are known to give too old The radiocarbon date/TL-OSL date ages (see above), yet these ages would of- comparison results from Sites 12 and 73 are ten fit perfectly to the overall geological indeed confusing. Considering the sedi- and geochronological context; the sedimentation and stabilisation processes, the mentological evidence and the radiocar- TL/OSL date from an underlying sand unit bon age estimates. This is the case at the should actually compare to maximum age Site 12 (Fig. 34), where the age estimate of a former burnt down forest, which underneath a 8400 cal. yrs BP old char- makes the age difference even bigger in coal horizon gives TL 8300 (max)/OSL these cases. There has been only one ear- 6600 (max) and the combination age of lier attempt to combine luminescence dat- 6400 ± 800 yrs. However, when comparing with radiocarbon estimates in relation ing radiocarbon ages to TL/OSL dates, the with aeolian history of the northernmost influence of the original age of the former Lapland. Käyhkö ( 1997 ) combined infra- forest into the radiocarbon ages, has to be red stimulated luminescence (IRSL) dates included. Given that the age estimate de- with conventional radiocarbon dates in an rived from a charcoal layer represents an attempt to determine the aeolian history 115 of the whole subarctic Fennoscandia. He of correlating the soil age estimates derived collected some dating samples over a large from different organic fractions (NaOH- area comprising the whole northernmost soluble humic acid or NaOH-insoluble res- Finland, 81 000 km 2 and based his inter- idue) should be made with caution and pretation of the aeolian history in addition may cause problems (Martin and Johnson, to pre-existing dating results from the lit- 1995). On the contrary, charcoal is inert erature on 7 radiocarbon age estimates and and many of the problems related to bulk 14 IRSL dates. He dated some sequences dating of soil organic matter can be ig- with both techniques, often dating a char- nored when using pieces of charcoal for coal horizon with radiocarbon method radiocarbon dating. For instance, Wintle and the overlying sand unit with IRSL ( 1990) in her review of TL dating of loess technique. Combination of the techniques deposits remarks that TL dates for inters- was not very successful, but a discrepancy tadial soils in Europe are in agreement with between the dating results obtained by the the very few radiocarbon dates on reliable radiocarbon and luminescence (IRSL) material, such as charcoal. The fact that methods was reported as the time span charcoal is inert, provides however, anoth- between the dates was thousands of years er source of uncertainty. Charcoal can be (Käyhkö, 1997). Possible explanations, for recycled from one unconsolidated forma- example the 2000 yrs time gap between tion into a new deposit and dating of char- the two results at Melajärvi site included coal should therefore not be regarded as too old radiocarbon age due to long ap- reliable until evidence is found to support parent mean residence time (AMRT) in soil its contemporaneous coexistence (Watch- carbon, a too young IRSL date due to in- man and Twidale, 2002 ). The number of complete bleaching and the possibility that charcoal horizons varies within a single the dating results are correct and the time dune site especially at Ijjävri indicating gap is caused by slow rim deposition along patchy deflation and thus erosion of the the side of the blowout. In the light of the charcoal horizons. Also, it has been noticed above discussed cases of Sites 12 and 73, that a charcoal horizon can be intact in one where the underlying sand unit yields place at a dune and scattered in another younger age estimate than the capping section only one meter further ( Chapter charcoal horizon, the possibility that the 5 . 5 ). The inconsistency of TL/OSL dates both dating results are correct, must be and radiocarbon age estimates is probably overruled. The reliability of TL/OSL dates due to mixing of old charcoal into a char- has been discussed above, with the conclu- coal layer. The radiocarbon age estimates sion that the combination age is the best should therefore be interpreted as maxi- available estimate. Consequently, the reli- mum ages for a former forest, in particu- ability of the radiocarbon dates has to be lar the age estimates of the uppermost considered. Käyhkö ( 1997 ) suggests that charcoal horizons in a series of charcoal AMTR effect results with too old radiocar- horizons. bon ages in his study. However, usually the The dating results pinpoint some key AMTR factor has been linked to organic issues on the history of the dune fields, soils and problems with their radiocarbon which are discussed here in relation with dating ( e.g. Geyh et al., 1983 ) as the soil the suggested periods of aeolian activity organic matter is always of mixed age and described in Fig . 47. the decomposition and humification processes do not necessarily cease completely 116 after the burial. Furthermore, the attempts Beginning of the aeolian history, The end of the primordial dune building circa 10 800 cal. yrs BP phase, circa 8 700 cal. yrs BP The bottommost dune material above the To define the end of the primordial dune glaciofluvial base sediments was TL/OSL building phase on the aeolian history of dated 10 700 ± 1200 yrs old at the Mud- the dune fields, the age estimates of the dusjävri (sample K 2 , Dune 12) and 10 200 oldest charcoal horizons were crucial. The ± 1200 yrs old at Ijjävri (sample K 8 , Dune oldest radiocarbon age estimates at Mud- 66). At Muddusjävri the sample K 2 actu- dusjävri were 7770 ± 170 yrs BP ( 8640 ally represented a deflation surface within cal. yrs BP, range 9020– 8200) and 7620 the oldest aeolian unit. The TL/OSL age ± 100 yrs BP ( 8390 cal. yrs BP, range 8580– estimate 10 700 (range 11 900 – 9500) yrs 8200) dating the same charcoal layer BP is in good agreement with the estima- (sampling depths 68 and 70 cm, respective- tion of the deglaciation at Muddusjävri ly). The oldest obtained radiocarbon age region ( circa 11 000 cal. yrs BP; e.g. Saarn- estimate from the Ijjävri dune field was isto, 1973; Seppälä, 1980; Johansson and 7650 ± 80 yrs BP ( 8460 cal. yrs BP, range Kujansuu, in press) and suggests that the 8610– 8230). It has to be recalled that the aeolian activation started immeaditely af- problems with “old charcoal” mixing with ter the land emerged from the water and the charcoal layers (discussed above) has desiccated. On the other hand, sample K 3 no impact on these oldest charcoal hori- from the lake shore bank at the northern zons. Therefore these age estimates can be end of the Muddusjävri dune field was col- considered to be the most reliable radio- lected above the erosional delta/dune con- carbon dates. The first stabilization phase tact and resulted in TL/OSL date 8700 ± on both dune fields appears to have taken 1100 yrs, which is considerably younger place about 2200 years after the beginning than the date from the deflation surface at of the aeolian history and the first dune the Dune 12. This suggests that the contact building phase on both fields occurs from has indeed been erosional and that the hi- circa 10 900 – 8700 years before present. atus must affect also the bottommost dune Käyhkö ( 1997 ) discussed the distribu- sediments. At the Ijjävri dune field, sam- tion of the available radiocarbon dates pri- ple K 9 was collected from dune material or to his study remarking that the date above the fluvial sediments. The age esti- obtained by Seppälä ( 1971 ) 7160 ± 200 BP mate 10 200 ± 1200 yrs is in good agree- (Hel-31) ( circa 8000 cal. yrs BP) was the ment with the regional deglaciation histo- first to be published and is by far the old- ry, although deglaciation took place at est one also. He concludes that the age ob- Ijjävri earlier than at Muddusjävri, and one tained by Seppälä was exceptionally old would have expected the oldest dune age and did not indicate any stabilization estimates from these areas to be in reverse phase in a larger regional scale. The oldest order. Nevertheless, the age estimate from charcoal horizons dated in this study yield- Ijjävri bottommost dune sediments sug- ed even older ages, supporting the idea of gests also the rapid beginning of the aeo- a rather early end of the primordial dune lian phase after the deglaciation, although building phase at least at the Muddusjävri it is probable that erosion has effected the and Ijjävri areas. initial dune sediments at the dune section 66 and that there are even older dune sediments in the area that remain undated.

117 First reactivation phase, ping a sand unit bearing marks of soil for- R1 8 200–7 800 cal. yrs BP mation and pieces of charcoal ( Fig. 27). It The first reactivation phase appears to have is not rare on both dune fields, Muddus- taken place on both dune fields almost si- jävri and Ijjävri, that a charcoal horizon is multaneously at circa 8200– 7800 years scattered and not intact. In such cases ex- before present lasting about 400 yrs. The tra sections near the first one often reveal start of this suggested period of aeolian an intact charcoal horizon on the strati- activity coincides with the 8 . 2 ka cooling graphic position of the scattered pieces at event reported on North Atlantic marine the first section. The pieces of charcoal in records (Bond et al., 1997 ) and on Green- the underlying sand unit strongly suggest land ice cores GRIP and GISP2 (Dansgaard that there was an older charcoal horizon et al ., 1993; Grootes et al ., 1993; O’Brien et and most importantly, that the charcoal al., 1995). The influence of this 8 . 2 ka event horizon Seppälä was dating was not the has also been detected on terrestrial oldest one, but was bearing “old charcoal” records in Central Europe ( e.g. Tinner and element, as all the charcoal horizons do Lotter, 2001 ) and in lake sediments in Swe- excluding the very first one. Particularly in den and Finland (Korhola et al ., 2002 ). A this case, as the scattered pieces of char- connection between the 8 . 2 ka cooling coal indicate erosion and re-deposition of event and the aeolian reactivity periods on the former charcoal horizon. Consequent- the Muddusjävri and Ijjävri dune fields has ly, the date obtained by Seppälä ( 1971) is been suggested earlier (Kotilainen and Ko- not a strong argument against the climat- tilainen, 2001 ; Kotilainen, 2002 ). Reactiva- ic significance of the first reactivation tion of the dune fields might be related to phase detected on the Muddusjävri and reorganization of the atmospheric and sur- Ijjävri dune fields. face ocean circulation over Greenland and North Atlantic. It is possible that this “ 8 . 2 Second reactivation phase, R2 : ka cooling event” resulted from surface wa- 7 400–5 700 cal. yrs BP at Muddusjävri, ter freshening and thus decrease in the 7 500–7 200 cal. yrs BP at Ijjävri North Atlantic Current northward trans- The second reactivation phase started at port of surface water causing cooling of the the Muddusjävri dune field at circa 7400 high latitudes. This mechanism might de- cal. years before present and in Ijjävri cir- crease precipitation over the subarctic Fen- ca one hundred years earlier, at 7500 cal. noscandia and thus trigger the onset of the yrs BP. However, the data suggests that at aeolian reactivation in the region. the Muddusjävri dune field the aeolian On the other hand, the above men- phenomena has been continuing rather tioned age estimate 7160 ± 200 (Hel-31) longer, possibly even to 5700 cal. years BP, obtained by Seppälä ( 1971 ) of a charcoal while at Ijjävri the phase has initially last- horizon from the Kiellajoki dune field, ed only a few hundred years to 7200 cal. some 40 km north from the Muddusjävri years BP. During the time period 7200 to dune field does not support any large scale 5800 cal. yrs BP there were two periods interpretation of the climatic significance of reduced stability probability at 6500 to of this reactivation phase. Calibrated age 6700 cal. yrs BP and at 5900 to 6000 cal. for the sample is with 2 sigma range 7635– yrs BP at the Ijjävri dune field, but no con- 8357 cal. yrs BP. However, on Fig s 27 and tinuously active phase as the one detected 28 (Seppälä, 1971, p. 37 and 40) it can be at Muddusjävri. On the other hand, it is seen, that the charcoal horizon Seppälä was possible that the long period at Muddus- 118 sampling for radiocarbon dating was cap- jävri is artificial due to missing charcoal horizons because of the insufficient Muddusjävri at circa 4800– 5700 cal. yrs number of dunes studied and a strong de- BP. On the other hand, the data Seppä flation processes. In an attempt to consid- ( 1996) presented from northernmost Fin- er this, a review of the concurrent climate land, with the most southern sampling site, proxy data is necessary. Lake Rautuselkä only 20 km northeast The climate proxy data over this peri- from Ijjävri, dates the decline of the tree- od, or actually over a more extensive time lines and a major environmental change slice circa 7000– 4000 cal. yrs BP, indicate towards a moister climate at 6000– 5500 development of a well-documented dry cal. yrs BP.According to the pollen record climatic period (Seppä et al ., 2002 ). For the climate in northern Lapland remained example, lake-level reconstructions from colder and moister after the hypsithermal small lake basins in the Finnish tree line (Seppä, 1996 ). In conclusion, the biostrati- region indicate that lake-levels during this graphic climate proxy data suggests that period were circa 3 – 4 m lower than at the long active period in the Muddusjävri present (Hyvärinen and Alhonen, 1994; dune field is possible. The more compli- Korhola and Rautio, 2002 ). Pine arrived at cated pattern at the Ijjävri dune field could the Lake Rautuselkä area at 8080 cal. yrs be result of the pine advance from the BP ( 20 km northeast from Ijjävri), which north and the more stable vegetation there marks a change in climate towards drier than on the tree line region at Muddus- conditions after the moist and warm lat- jävri. Pine arrived to the Lake Inari basin ter part ( circa 10 000– 8000 cal. yrs BP) of from the north at around 8500 – 8000 cal. the early-Holocene (Seppä, 1996). Accord- yr BP. The pine migration further south ing to Eronen et al . ( 2002 ) the evidence of and southwest was delayed and dia- the megafossils and pollen data are mutu- chronous in nature reaching western Finn- ally supportive in indicating that the time ish Lapland only 6800 cal. yrs BP (Seppä, of maximum extent of pine in Finnish La- 1996 ). The Ijjävri area has been covered pland was between 6800 and 4500 cal. yrs with pine trees at circa 6000 cal. yrs BP BP (see Chapter 2 . 4 ). The mid-Holocene (Kultti et al., in press; see Chapter 2 . 3 ). represents the hypsithermal period in the far north of Fennoscandia with mean July Third reactivation phase, R3: temperatures 1 . 2 ˚C to 1 . 5 ˚C higher that at 4 800–3 900 cal. yrs BP at Muddusjävri, present at around 8000– 5500 cal. yrs BP 4 600–4 300 cal. yrs BP at Ijjävri (Seppä, 1996). A dry and warm climate The third reactivation phase started at the with a low ground water table is favoura- Muddusjävri dune field at circa 4800 cal. ble to the suggested long active aeolian years before present and at Ijjävri circa two period at the Muddusjävri dune field. hundred years later, at 4600 cal. yrs BP. However, the pollen influx values from Again, at the Muddusjävri dune field the three lake sediment cores in the north Fen- aeolian phenomena has lasted longer, con- noscandian tree line area suggest that the tinuing to 3900 cal. years BP, while at maximum extent of dry, oligotrophic Ijjävri the phase has lasted only three hun- heaths and consequently the driest period dred years to 4300 cal. years BP. Even this in northwestern Finland dates to 5000– short period at Ijjävri is presented only 5900 cal. yrs BP (Seppä et al., 2002 ). Re- with slightly reduced stability probability grettably, this period is concurrent with a ( Fig . 48). It can be argued that the third phase of forest fires, indicated by charcoal reactivation phase occurs only in the Mud- horizons bearing evidence of forests grow- dusjävri dune field. Examination of the ing and stabilizing the dune area in the probability distribution sum curves of the 119 calibrated radiocarbon age estimates on fields at circa 1200 cal. yrs BP, which might Fig. 48 shows that the patterns of reacti- indicate a short reactivation on the Mud- vation processes are not concurrent at the dusjävri and Ijjävri dune fields. dune fields for the last 4500 years before present. The activation processes can not Fifth and the latest reactivation phase, therefore be linked to global or not even R 5: 500 cal. yrs. BP onwards at Ijjävri regional climate signals but are local in The latest reactivation phase can only be nature. detected at Ijjävri dune field, where it still Major change in climate towards is an ongoing phenomena. The period of moister and colder conditions occurred in the aeolian reactivation phase coincided northernmost Finland at circa 6000– 5500 with the well documented LIA event at cir- cal. yrs BP as discussed above and in Chap- ca 1550– 1850 AD ( 400– 100 yrs BP) ( e.g. ter 2 . 4 . ( e.g. Seppä, 1996; Korhola et al., Bradley and Jones, 1993). The connection 2002 ). Retreat of the forest was followed between dune reactivation and LIA event by an advance in tundra vegetation, while in Finland has been suggested before (Aar- the late-Holocene tundra was distinctly tolahti, 1976; 1980; Käyhkö, 1997 ). The different from the early-Holocene chiefly dune area at Ijjävri is very susceptible to mid-Arctic, sparsely vegetated tundra. The erosion due to its fragile vegetation cover. pollen assemblage indicates increase in During the last 500 years human impact moisture, and gradual spread of the peat on the area has increased. Reindeer hus- lands.After this major change, the pollen bandry and off road vechiles (ORV) are record shows no further climate signal for both contributing to the recent reactiva- the remaining 5000 years of late- tion phenomenon (Kotilainen, 1991). The Holocene. significance of the latest activation phase together with the human impact to the re- Fourth reactivation phase R 4: activation history will be discussed in the 3 300–2 700 cal. yrs BP at Muddusjävri, concluding Chapter 6 . 2 800–2 400 cal. yrs BP at Ijjävri Looking at the radiocarbon age distri- The earlier, towards the end of phase R 3 butions, it is impossible to note, that the detected diachronous activation pattern two dune fields represent different activa- continued. The activation phases overlap tion phase nowadays. The youngest radi- only briefly at circa 2700– 2800 cal. years ocarbon charcoal horizon at Muddusjävri BP ( Fig. 47). However, the circa 2 . 8 ka cool- dune field gives a radiocarbon age estimate ing event has been detected in many proxy of 320 – 670 cal. yrs BP ( 2 sigma), at the records and the possible climate signal here sampling depth of 1 cm, indicating that no can not be totally ignored ( e.g. Denton and reactivation followed the forest fire. There Karlén, 1973, alpine glacier expansions; van is a tall, dense pine forest at Muddusjävri Geel et al., 1996, Netherlands cooling event; today and no forest fires have been record- Bond et al ., 1997 ; 2001, North Atlantic drift ed in the area during the last century. If ice indices; Tiljander et al ., 2003 , lake sed- the age estimate for an average full grown iment properties in central Finland). forest ( 0 – 300 yrs) in the study regions is Vliet-Lanoë et al. (1993)suggested an correct, we would have a 150 yrs BP radio- aeolian reactivation period concurrent carbon age estimate from the charcoal, if with the Medieval Warm Period, MWP cir- the forest were to burn down now. The ca 1000– 700 yrs BP and at Hietatievat, charcoal would be mixed with the older Enontekiö. The probability distribution charcoal in places and the interpretation 120 curve (Fig. 48) has a decline on both dune that these two charcoal layers 1 cm apart represent different forests would be diffi- achronous charcoal horizons at the dune cult to make even if the age estimates field were caused by very local forest fires, would result in 150 and 510 yrs BP, respec- which did not trigger any large scale reac- tively. The forest fire at the sampling depth tivation. For this reason, the combination of 1 cm did not cause any activation phase ages Hel-3437 , Hel-3440 and Hel-3441 were at the dune field. It is probable that the di- included to the data.

121 6. Palaeoenvironmental model of the dune fields based on the data presented in this study

he data obtained in this study suggests dating) presented in this study has been re- T that aeolian activity can be linked with arranged according to the Unit 1 – 3 , from regional, even global climate at least in the where each sample has been taken.All the early to the mid-Holocene. During the dune site descriptions were studied and the Holocene all combinations of temperature boundaries between the Units defined. and precipitation have occurred in north- Samples representing each Unit are com- ern Finnish Lapland ( e.g. Seppä et al., 2002 ; bined in order to define special nature of Kultti, 2004 ; see Chapter 5 . 6 ). The combi- each Unit and the palaeoenvironment dur- nation of cold and dry indicating less rain, ing their deposition. Morphology will be lower ground water table and more sensi- discussed in context with the initial formative vegetation is most favourable to aeo- tion of dune fields (Unit 1 ), reworking of lian activity. Combination of warm and the material (Unit 2 ) and present day proc- dry climate leads to enhanced evaporation, esses (Unit 3 ). The chapter will be conclude thus lower ground water table and sensi- with a discussion of the pattern, mecha- tive vegetation and was very much in fa- nism and significance of the reactivation vour for a prolonged aeolian activity after phenomena with reference to human im- a forest fire. On the other hand, combina- pact and climate. tions of moist and warm or moist and cold climate can not be seen as supportive for 6.1. Muddusjävri dune field aeolian activity. Increased moisture result- The concluding model of the dune lithos- ed in thicker snow cover and raised ground tratigraphic units at Muddusjävri field was water table. The data also suggests, that cli- presented in Chapter 4 . 5 . 1 (Fig. 38). For the mate conditions alone can not explain re- palaeoenvironment model all the samples activation processes on the dune fields in were classified according to their strati- the late Holocene. In the following a graphic position. The majority of the sam- palaeoenvironmental model for the post- ples were positioned in the Unit 2 . Regard- glacial development of the two dune fields ing the sand samples (grain size and bin- is presented in connection with their sed- ocular grain shape analyses), 34 samples imentary units and the sediment proper- represented Unit 1 , 80 Unit 2 and only 9 ties. samples Unit 3 at the Muddusjävri dune The model will be based on the lito- field ( Tables 27 and 28). Samples that were facies units described from both dune taken across a contact of the units were fields. All the data (lithostratigraphical, excluded from the data. Variation between 122 mineralogical, grain shape, grain size and the average values retrieved from the Units 1 – 3 are discussed in the beginning of the by seven radiocarbon age estimates and description (Unit 1 ). one TL/OSL date, and Unit 3 by twelve ra- Two of the SEM samples from the Mud- diocarbon age estimates. The dates will be dusjävri dune field represent Unit 1 ( 3 /ww/ identified in discussion of each Unit. 100 cm and 3 /top/ 100cm) and three were The sand samples were taken at three positioned in the Unit 2 (3 /lee/100cm, 12/ uniform depths in the Muddusjävri dune lee/90cm and 16/lee/80 cm). Table 17 de- field (with some additional samples of spe- scribes the average percentages of each cial interest, see Chapter 3 . 1 . 3 ). Therefore imprint class interpreted on the quartz the number of samples associated with grain surfaces of these samples. The min- each lithostratigraphic Unit 1 – 3 is indica- eralogical composition of 15 dune samples tive of their sediment thickness. Thus, it from Muddusjävri were studied. Eight of indicates the significance and duration of them originate from Unit 1 and seven from each sedimentation cycle. Unit 2 (80 sam- Unit 2 (Table 29). Dating samples are dis- ples) is the major unit at the dune field, tributed in the following way: Unit 1 was while Unit 3 (9 samples) is not well devel- represented by three TL/OSL dates, Unit 2 oped.

Table 27. Average grain size parameters for the samples associated with

Units 1–3 at the Muddusjävri dune field. Mean grain size is indicated by Mz

( Φ ), Inclusive Graphic Standard Deviation for sorting, σ 1 , Inclusive Graphic

Kurtosis, KG Inclusive Graphic Skewness, Sk 1 and the number of samples in each unit by n.

Grain size parameters Lithostratigraphic unit M z σ 1 K G S k 1 n ( Φ ) UNIT 1 2.08 0.65 1.05 0.09 34

UNIT 2 2.14 0.65 1.04 0.10 82

UNIT 3 2.16 0.71 1.07 0.12 9

Table 28. Average values of the binocular microscopy results for grain shape parameters associated with Units 1–3 at the Muddusjävri dune field. Percentages of the grains classified to roundness classes A–F and roundness index according to them are indicated with percentages of shiny and matt particles and the number of samples.

Lithostratigraphic unit A % B % C % D % E % F % Roundness shiny matt n Index % % UNIT 1 2 27 54 19 1 0 0.30 51 49 34

UNIT 2 1 24 50 23 2 0 0.31 41 59 80

UNIT 3 5 23 50 20 2 0 0.30 29 71 9 123 Table 29. Average values for Mineralogical composition (%) of the samples in Unit 1 and 2 at the Muddusjävri dune field.

Quartz Feldspars Amp.+Px. Mica HM1 HM2 Ox. RF (talc) n

Unit 1 22 33 32 4 6 1 2 3 0 8

Unit 2 14 47 18 6 6 1 0 7 0 7

Unit 1 (Muddusjävri dune field) the SW. General development of the wind The Unit 1 represents the initial dune system in Finland is characterised by NW building period in early Holocene. It is winds during the Preboreal and Boreal positioned at the lowermost part of the times (Aartolahti, 1976). Also Seppälä section and the lithostratigraphy is char- ( 1971) concluded that the dunes in Kaa- acterized with intermediate to high angle masjoki-Kiellajoki area were formed by NE tabular foreset lamination indicating a winds. However, the thickness of the dune combination of grainflow and grainfall sediments at the Muddusjävri dune field deposition. The bulk sediment is close are increasing towards the northeastern packed and consists of thin, 0 . 5 to 1 cm end of the dune field, hence the dune axes thick laminae of dark sand alternating with are parallel to the resultant sand transport 1 – 3 cm thick laminae of light coloured direction.The lithostratigraphy of the Unit sand. Often this bulk sediment also has 1 deposit confirms this suggestion of the thicker foreset laminae inter-bedded. De- initial dune forming wind to be SW-NE formation structures, like dimpled strata, orientated. are common implying that niveo-aeolian Grain size distribution of the dune sand processes operated syn- and post-deposi- in the Unit 1 shows modest variation com- tionally. The fact that these denivation paring to the other Units ( Table 27). A structures have remained intact only in the vague trend of sediment fining towards the lee sides of the dunes indicates that the surface can be seen. However, the mean advance rate of the dunes has been high. grain size and sorting are correlating with There were no observations of charcoal the transportation distance at the Muddus- horizons in this unit. These lithostrati- jävri dune field ( Fig. 52 and Fig. 54), and graphic features indicate that the material the finest and best sorted material ( Φ 2 . 82;

was deposited with unimodal azimuth, σ 1 0 . 40) can be found in the Dune 1 with high wind velocities and sparse vegetation. longest distance downwind. It can be con- However, the dominance of tabular pla- cluded, that the grain size variation be- nar foresets in the Unit 1 can be a result of tween the Units can not be expected, while a low relief at the time of the deposition all the Units consist of material with var- rather than unimodal wind system. How- ying maturity in the aeolian transportation ever, at the Muddusjävri dune field, the process. wind system is governed by the local to- However, the average values for the 34 pography (see Chapters 2 . 5 and 4 . 1 . 1 ). The samples from Unit 1 establish a standard morphology of the parabolic dune net sug- for the early Holocene dune material at the 124 gests an initial dune forming wind from Muddusjävri dune field. The mean grain size is Φ 2 . 08 (circa 0 . 250 mm), which is SEM evidence from the quartz grains at quite close to the average value for the Unit 1 (samples 3 /ww/100 and 3 /top/100 , whole dune field (Φ 2 . 13). The value indi- Table 17) does not indicate any common cates fine sand according to Friedman and features. Both samples were studied with Sanders ( 1978). The inclusive graphic a quite satisfactory number of grains ( 15 standard deviation, σ 1 (0 . 65) indicates and 13, respectively). In Unit 1 , the inter- moderately well sorted material. The av- bedded layers of coarse sand are common, erage Inclusive graphic skewness, Sk 1 val- and in fact, the sample 3 /top/100 represents ue of 0 . 09 shows that the grain size distri- such a layer. However, the strong evidence bution of the material in Unit 1 is symmet- of aeolian abrasion detected in the sam- rical. Also the K G (inclusive graphic kurto- ple 3 /ww/100 indicates high wind veloci- sis) value of 1 . 05 indicates mesokurtic ties and transportation rates during the (Gaussian) distribution of the grain size. deposition of Unit 1 . For the palaeoenvironmental modelling, Mineralogical composition of Unit 1 these values can contribute the following: dunes at the Muddusjävri dune field ap- The material of Unit 1 is not remarkably pears to be clearly different from Unit 2 less finer or less sorted than the material samples. The quartz content of Unit 1 sam- at the subsequent Units. Evidently, the ae- ples varies between 19– 25%, with an aver- olian process at the early Holocene has age of 22%. The samples of Unit 2 contain been effective. less quartz (average 14%, range 10– 18%), Binocular grain shape studies of the which is surprising. The average quartz material from Unit 1 show that the distri- content of the Muddusjävri dunes is 18%. bution of the sand grains at the roundness However, the amount of amphiboles and classes A-F demonstrate no variation com- pyroxenes in Unit 1 (32%) is higher than pared to the other Units ( Table 28). Again, the average value for the whole dune field it has to be stated, that such result was not ( 26%) and Unit 2 (18%), indicating an ef- expected as the sand samples in each Unit fective grinding process during the depo- are originating from every part of the dune sition of Unit 1 . field (see above). However, it is interesting There were three TL/OSL dates associ- that the proportions of shiny/matt parti- ated with Unit 1 . Sample K 2 from Dune 12/ cles are indicative for the sediment depth 140 cm resulted in a combination age gradient. The proportion of shiny particles 10 700 ± 1200 yrs and sample K 14 Dune decreases towards the surface (from Unit 14/ 50cm 8400 ± 1000 yrs.Also the sam- 1 to Unit 3 ), while the proportion of the ple K 3 Dune 10/ 70 cm belongs to Unit 1 matt particles in increasing, which suggests ( 8700 ± 1100 yrs). The sample from the increasing maturity of quartz grains to- Dune 10 at the NE end of the dune field wards the surface. indicates the onset of aeolian phenomena, The material in the Unit 1 is character- but is likely to have a hiatus underneath. ised by class C quartz grains ( 54%), which The samples 12/ 140 cm and 14/ 50 cm indi- indicates sub-angular rounding.An aver- cate strong deflation and re-accumulation age of 27% was classified into the class B during Unit 1 . The onset of the aeolian (angular). The values correspond very well phenomena at the Muddusjävri dune field to the average values of the whole Mud- is controlled by the age estimate of the de- dusjävri dune field ( 51% and 24%, respec- glaciation on the area, which took place at tively). The proportion of shiny quartz circa 10 800 yrs ago (Kujansuu et al., 1998; particles ( 51%) dominates the matt ones Johansson and Kujansuu, in press). These ( 49%) in Unit 1 slightly. dates establish the time period of Unit 1 125 formation at circa 10 800 to 8 400 ± 1 000 One charcoal horizon was detected with- yrs before present. in this unit. These lithostratigraphic features indicate that the material was depos- Summary of Unit 1 ited with changing wind directions, mod- (Muddusjävri dune field): erate wind velocities and with the presence •The initial dune building period in the of vegetation. early Holocene at circa 10 800 to 8400± Hence, the dominance of the cross-bed 1000 yrs before present. sets with wedge-shaped or curved bound- •The initial dune forming wind has been ing surfaces in Unit 2 can be a result of a SW-NE orientated. higher (compared to Unit 1 ) relief at the •The lithostratigraphy is characterised time of the deposition responding to a by tabular foreset strata. changing wind system. In the Muddusjävri •The dune material is defined as fine, dune field the wind system is governed by moderately well sorted sand, mainly the local topography, as discussed above in sub-angular, dominance of shiny quartz context with Unit 1 . However, some mor- particles ( 51%), with surface texture in- phological features interpreted at the Mud- dication of aeolian transport. Main dusjävri dune field, like a nose formation mineralogical component is amphib- ( Chapter 4 . 1 ), are indicative of a western oles and pyroxenes (32%). shift of the wind direction during the dep- •The data suggests high wind velocities osition of Unit 2 . It is probable, that the and transportation rates with effective higher relief together with the shifting aeolian processes (sorting and grind- wind system (causing more filling and cut- ing) during the deposition of Unit 1 . ting) contributed both to the complex cross-bed strata of Unit 2 . Unit 2 (Muddusjävri dune field) Grain size distribution of the dune sand Unit 2 is the most significant deposit at the in Unit 2 is indicated by 82 sand samples. Muddusjävri dune field. It is most complex Unit 2 sand is characterised by a mean size and thickest accommodating most of the of Φ 2 . 14, which corresponds with the av- analysed sand samples. It is positioned in erage value for the whole dune field the middle part of the dune section and ( Φ 2 . 13) and indicates fine sand. The inclu-

the lithostratigraphy is characterized with sive graphic standard deviation, σ 1 (0 . 65) dune foreset cross-bedding strata, dipping indicates moderately well sorted material.

at intermediate to low-angles of 25˚ to 10˚, The average inclusive graphic skewness, Sk 1 indicating advance of the dune. The bulk value of 0 . 10 shows that the grain size dis- sediment consists of thin, 0 . 5 to 1 cm thick tribution of the material in Unit 2 is sym-

laminae of dark sand interstratified with metrical. Also the K G (inclusive graphic 2 – 3 cm thick sets of light sand laminae. kurtosis) value of 1 . 04 indicates mesokur- Packing of these cross-bedded strata is tic (Gaussian) distribution of the grain moderately close. Dimpled strata and de- size. For the palaeoenvironmental model- formation structures are common imply- ling, these values can contribute the fol- ing the presence of niveo-aeolian process- lowing: The material of Unit 2 is slightly es and also plant roots during and after the finer than in Unit 1 , but not remarkably deposition. However, these deformation finer or better sorted than the material at structures occur at every side of a dune, Unit 1 . The aeolian process during the dep- not only at the lee side (as was the case in osition of Unit 2 has been effective, pro- Unit 1 ) indicating that the advance rate of ducing typical grain size parameter values 126 the dunes has not necessarily been high. for inland dunes. Binocular grain shape studies of the has to be taken into account. Also the dif- material from Unit 2 indicate similar dis- ficulties in differentiating polycrystalline tribution of the material into the round- quartz from rock fragments of quartz and ness glasses as in Units 1 and 3 (Table 28). feldspars has to be considered as an source The results for Unit 2 are based on 80 sam- of error (class RF in Table 29, see also ples. The proportions of shiny/matt par- Chapter 5 . 4 ). The amount of amphiboles ticles are indicative for the sediment depth and pyroxenes in Unit 2 (18%) is lower gradient, as discussed above, and the pro- than the average for the whole dune field portions of shiny/matt particles in Unit 2 ( 26%) and Unit 1 (32%), indicating an ef- are 41% and 59%, respectively. The mate- fective grinding process during the depo- rial in Unit 2 is characterised by class C sition of Unit 2 . quartz grains ( 50%), which indicates sub- There are seven radiocarbon age esti- angular rounding. The rounding index of mates and one TL/OSL date associated 0 . 31 indicates a vaguely better rounding with Unit 2 . The laboratory codes for the compared with other Units (roundness in- radiocarbon dates are: Su-1870 , Su- 1871 , dex on both 0 . 30). An average of 24% was Hel-3019, Hel-3018, Hel-3179, Hel-3017, and classified into the class B (angular), while Hel-3009 (see Table 24). They indicate that 23% was classified into class D, indicating the deposition of Unit 2 took place be- sub-rounded grains. However, the values tween 8640 cal. yrs BP and 2380 cal. yrs correspond very well to the average values BP. The oldest charcoal dates have been of the whole Muddusjävri dune field (class associated with Unit 2 , and the upper limit C 51% and class B 24%). for Unit 2 is indicated by the oldest char- SEM analyses from the quartz grains at coal horizon associated with Unit 3 . The Unit 2 (samples 3 /lee/100 , 12/lee/ 90, and 16/ only TL/OSL date representing Unit 2 is lee/80, Table 17) indicate some common sample K 1 of Dune 12/ 72cm. The sample features. The samples from the dunes 3 and was dated 6400 ± 800 yrs (see discussion 12 are showing very similar surface tex- in Chapter 5 . 6 ). tures. Sample 16 is not in deep contrast with them either. All samples were stud- Summary of Unit 2 ied with a quite satisfactory number of (Muddusjävri dune field): grains ( 14, 15 and 15, respectively). Unit 2 •The deposition of Unit 2 took place be- quartz grain surface textures are character- tween 8640 cal. yrs BP and 2380 cal. yrs ised by the dominance of textures indica- BP. tive for aeolian transportation. This result •The wind direction during deposition indicates an effective aeolian process and has shifted from the SW-NE to the W- aeolian maturation of the quartz particles E orientation. during the deposition of Unit 2 . •The lithostratigraphy is characterised The differences between Unit 1 and Unit by cross-bedded strata. 2 mineralogical composition were dis- •The dune material is defined as fine cussed above (see Unit 1 ). The samples of (slightly finer than in Unit 1 ), moder- Unit 2 contain less quartz (average 14%, ately well sorted sand, mainly sub-an- range 10– 18%) than Unit 1 . The amount of gular (with slightly better roundness quartz has been widely accepted as a meas- than Unit 1 ), dominance of matt quartz ure of dune sand maturity. However, the particles ( 59%), with strong surface tex- differences are small and the above dis- ture dominance of features indicating cussed problem of each unit accommodat- aeolian transport. The main mineralog- ing material with versatile aeolian history, ical component is feldspars ( 47%). 127 •The data suggests moderate wind veloc- mal distribution. However, the K G (inclu- ities and transportation rates with ef- sive graphic kurtosis) value of 1 . 07 indi- fective aeolian processes (sorting and cates mesokurtic (Gaussian) distribution grinding) and aeolian maturation of of the grain size. For the palaeoenviron- quartz particles during the deposition mental modelling, these values can con- of Unit 2 . tribute the following: The material of Unit 3 is slightly finer than in Units 1 or 2 , but Unit 3 (Muddusjävri dune field) not remarkably finer and actually, less well Unit 3 is the most limited deposit at the sorted than in Units 1 and 2 . However, the Muddusjävri dune field. It is the thinnest aeolian process during the deposition of unit, however it accommodates most of the Unit 3 has been effective, producing typi- radiocarbon dated charcoal horizons. It is cal grain size parameter values for inland positioned in the uppermost part of the dunes. dune section and the lithostratigraphy is Binocular grain shape studies of the characterized with tabular foreset laminae material from Unit 3 are indicating quite dipping at low-angles of 0 ˚–5 ˚. The bulk similar distribution of the material into the sediment is loosely packed, often poorly roundness glasses as in Units 1 and 2 (Ta- stratified, due to the disturbance of soil ble 28). The results for Unit 3 are based development, roots and denivation struc- only on 9 samples The proportions of tures. In the uppermost part the strata is shiny/matt particles are indicative for the often dimpled. Most of the charcoal hori- sediment depth gradient, as discussed zons indicating forest fires are within this above, and the proportions of shiny/matt unit. Towards the surface, at 40– 0 cm, the particles in Unit 3 are 29%/71%. The ma- sediment is poorly stratified. Sediment terial in Unit 3 is characterised by class C thickness and deformation structures that quartz grains ( 50%), which indicates sub- occur at every side of a dune, are indicat- angular rounding.An average of 23% were ing that the advance rate of the dunes has classified into the class B (angular), while not been high. It is more likely that the re- 20% were classified into class D, indicat- activation of the dunes has been patchy, ing sub-rounded grains. These values do and that the large areas were not activated not differ much from the average values of simultaneously. the whole Muddusjävri dune field (class C Grain size distribution of the dune sand 51% and class B 24%). in Unit 3 is indicated only by 9 sand sam- Mineralogical composition of Unit 3 ples. The Unit 3 sand is characterised by a dunes at the Muddusjävri dune field was mean size of Φ 2 . 16, which is slightly finer not studied. None of the thin sections se- than the average value for the whole dune lected for the analysis were positioned in field ( Φ 2 . 13) and the Units 1 and 2 .The val- Unit 3 . ue indicates fine sand. The inclusive graph- There are twelve radiocarbon age estimates associated with the Unit . The lab- ic standard deviation, σ 1 (0 . 71) indicates 3 moderately sorted material. The average oratory codes for the radiocarbon dates are: Su- , Su- , Su- , Su- , Su- inclusive graphic skewness, Sk 1 value of 1867 1868 1807 1869 0 . 12 shows that the grain size distribution 1809, Su- 1810, Hel-3437, Hel-3438, Hel- of the material in the Unit 3 is positively 3012, Hel-3439, Hel-3011, Hel-3010, Hel- skewed. The skewness is not very clear 3440 , Hel-3441, and Hel-3442 (see Table (positively skewed + 1 . 0 –+0 . 3 ), but indi- 24). They indicate that the deposition of cates a presence of a more pronounced tail Unit 3 has taken place since 2370 cal. yrs 128 of fine material compared with a log nor- BP. The lower limit of Unit 3 is defined with the date of the oldest charcoal hori- ing the same procedure, all the samples zon associated with Unit 3 , and the upper were classified according to their strati- limit for Unit 3 is the present day situation. graphic position for the palaeoenvironment model.Again, the majority of the Summary of Unit 3 samples were positioned in Unit 2 . Regard- (Muddusjävri dune field): ing the sand samples (grain size and bin- •The deposition of Unit 3 has been tak- ocular grain shape analyses), 10 samples ing place since 2 380 cal. yrs BP. represented Unit 1 , 53 samples Unit 2 and •The prevailing wind direction during 21 samples Unit 3 at the Ijjävri dune field deposition has been from the west to ( Tables 30 and 31). Samples that were tak- the east. en across a contact of the units were again •The lithostratigraphy is characterised excluded from the data. Variation between by unstratified, deformed strata. the average values retrieved from the Units •The dune material is defined as fine 1 – 3 are discussed in the beginning of the (slightly finer than in Units 1 and 2 ), description (Unit 1 ). moderately sorted sand (less sorted that One of the SEM samples from the in Units 1 and 2 ), mainly sub-angular Ijjävri dune field represent Unit 1 (63/ww/ (with slightly better roundness than 50cm), three were positioned in Unit 2 (69/ Unit 1 ), dominance of matt quartz par- lee/100cm, 71/lee/160 cm and 75/lee/140 cm) ticles ( 71%). and one in Unit 3 (70/lee/60cm). Table 17 •The data suggests moderate to weak describes the average percentages of each wind velocities and transportation rates imprint class interpreted on the quartz with effective aeolian processes (sorting grain surfaces of these samples. The min- and grinding) and aeolian maturation eralogical composition of twelve dune of quartz particles during the deposi- samples from Ijjävri were studied. Five of tion of Unit 3 . them originate from Unit 1 , six from Unit 2 and one from Unit 3 (Table 32). Dating 6.2. Ijjävri dune field samples are distributed in the following The concluding model of the dune lithos- way: Unit 1 was represented by two TL/OSL tratigraphic units at Ijjävri field was pre- dates, Unit 2 by 34 radiocarbon age esti- sented in Chapter 4 . 5 . 2 (Fig. 46). Follow- mates and four TL/OSL dates, and Unit 3

Table 30. Average grain size parameters for the samples associated with Units 1–3 at the Ijjävri dune field. Mean grain size is indicated by Mz ( Φ ), Inclusive Graphic Standard Deviation for sorting by σ 1 ,

Inclusive Graphic Kurtosis, KG Inclusive Graphic Skewness by Sk1 and the number of samples in each unit by n.

129 Table 31. Average values of the binocular microscopy results for grain shape parameters associated with Units 1–3 at the Ijjävri dune field. Percentages of the grains classified to roundness classes A–F and roundness index according to them (see Chapter 4.3.1) are indicated with percentages of shiny and matt particles and the number of samples.

Table 32 . Average values for Mineralogical composition (%) of the samples in Unit, 1 , 2 and 3 at the Ijjävri dune field.

Quartz Feldspars Amp.+Px. Mica HM1 HM2 Ox. RF (talc) n

Unit 1 19 37 11 2 29 0 1 1 0 5

Unit 2 18 45 13 2 21 0 0 0 0 6

Unit 3 27 40 10 1 17 0 1 0 0 1

by one TL/OSL date. The dates will be packed and consists of thin, 0 . 5 to 1 cm identified in discussion of each Unit. thick laminae of dark sand alternating with The number of sand samples associat- 3 – 8 cm thick laminae of light coloured ed with each Unit 1 – 3 is indicative of the sand. Inter layers of coarse sand are usual. sediment thickness and thus, duration of There are deformation structures, like each sedimentation cycle. Unit 3 is well dimpled strata implying the presence of developed at the Ijjävri dune field, al- niveo-aeolian processes syn- and post- though it represents a shorter time slice depositionally. The fact, that these deniva- compared to Unit 3 at Muddusjävri. tion structures have remained intact only at the lee sides of the dunes indicates that Unit 1 (Ijjävri dune field) the advance rate of the dunes has been Unit 1 represents the initial dune building high. There was one observation of a char- period in early Holocene also at Ijjävri. It coal horizon associated with this unit (Site is positioned at the lowermost part of the 62), but usually the contact to the upper section and the lithostratigraphy is char- unit is erosional. These lithostratigraphic acterized with intermediate to high angle features indicate that the material was tabular foreset lamination indicating a probably deposited with unimodal azi- combination of grainflow and grainfall muth, high wind velocities and sparse veg- 130 deposition. The bulk sediment is close etation. At the Ijjävri dune field the local relief cates moderately well sorted material. The does not control the local wind system. average Inclusive graphic skewness, Sk 1 val- The tabular planar foresets in Unit 1 indi- ue of 0 . 02 shows that the grain size distri- cate prevailing wind from the WNW dur- bution of the material in Unit 1 is symmet- ing the deposition. Again, the unimodal rical. Also the K G (inclusive graphic kurto- azimuth can be a result of a low relief at sis) value of 1 . 06 indicates mesokurtic the time of the deposition rather than an (Gaussian) distribution of the grain size. unimodal wind system. The dunes of the For the palaeoenvironmental modelling, Ijjävri field are isolated by till areas and do these values can contribute the following: not form a continuous dune field ( Chap- The material of Unit 1 is slightly finer and ter 4 . 1 . 2 ). The sediment thickness can not better sorted than the material in the sub- therefore be indicative of transportation sequent Units. The dune material from distance, but is governed by the availabili- Unit 1 (Ijjävri) represents the finest and ty of the source material. The dunes asso- best sorted dune sand of the study. Evi- ciated with the delta (southern end of the dently, the aeolian process during the ini- dune field) or with the esker (northern tial dune forming phase in the early part of the field) have the thickest depos- Holocene has been effective. its. The sites, were Unit 1 has been detect- Binocular grain shape studies of the ed (Dunes 63, 66, 65 and 62) are all locat- material from Units 1 – 3 show that the dis- ed near the deltaic source area in the tribution of the sand grains at the round- southern part of the dune field. The sec- ness classes A-F demonstrate no variation tions at the Sites 66 and 62 are excavated ( Table 31). However, it is interesting that into the deflation basins. the proportions of shiny/matt particles are Grain size distribution of the dune sand giving some indication for the sediment in Unit 1 shows some variation comparing depth gradient within Units 1 and 2 , where- to the other Units ( Table 30). In contrary as Unit 3 stands out. Again, this is proba- to the fining trend at the Muddusjävri bly a result of various sediments sources field, a trend of sediment coarsening to- during the whole post-glacial development wards the surface can be observed. How- of the dune field. In addition to the sort- ever, the variation in mean grain size and ed sediments, the till areas within the dune sorting are not great. Evidently, the aeo- field have been contributing to the sedi- lian sorting and grinding processes have ment supply. not been able to produce trends of better The material in Unit 1 is characterised sorting or fining in Ijjävri as easily as in by class C quartz grains ( 47%), which in- Muddusjävri due to the uneven distribu- dicates sub-angular rounding.An average tion of sorted source material at the area. of 30% were classified into the class D (sub- The till areas within the dune field have rounded) and an average of 17% were clas- been acting as sediment source as well. sified into the class B (angular). The val- The average values of the 10 samples ues correspond quite well to the average from Unit 1 establish a standard for the values of the whole Ijjävri dune field ( 44%, early Holocene dune material at the Ijjävri 35% and 15%, respectively). However, the dune field. The mean grain size is Φ 2 . 56, proportion of shiny quartz particles ( 59%) which is finer than the average value for dominates the matt ones ( 41%) in Unit 1 , the whole dune field ( Φ 2 . 47). The value indicating the large amount of quartz indicates fine sand according to Friedman grains with glacial/sub-aquatic features. and Sanders ( 1978 ). The inclusive graphic SEM evidence from Unit 1 relies on the standard deviation, σ 1 value of 0 . 58 indi- analysis of the quartz grains from the sam- 131 ple 63/ww/50cm ( Table 17). Regrettably, the Summary of Unit 1 (Ijjävri dune field): sample was studied with an unsatisfacto- •The initial dune building period in the ry number of grains, 11. However, there is early Holocene beginning at circa 10900 strong evidence of aeolian abrasion (class yrs before present. 1 / 52%) indicating high wind velocities dur- •The initial dune forming wind has been ing the deposition of Unit 1 . WNW-ESE orientated. Mineralogical composition of Unit 1 •The lithostratigraphy is characterised dunes at the Ijjävri dune field was studied by intermediate to high angle tabular with 5 thin sections. The quartz content of foreset strata. Unit 1 samples varies between 17– 26%, •The dune material is defined as fine, with an average of 19%. The amount of moderately well sorted sand, mainly quartz in Unit 2 is the same ( 18%), but in- sub-angular, dominance of shiny creases in Unit 3 (27%). The average quartz quartz particles ( 59%), with clear in- content of the Ijjävri dunes is the same as dication of aeolian transport in the sur- in Unit 1 , 19%. The amount of amphiboles face texture. and pyroxenes in Unit 1 (11%) shows no •The main mineralogical component is significant difference compared to Units 2 amphiboles and pyroxenes ( 37%). The and 3 or the average value for the whole amount of heavy minerals (mainly gar- dune field ( 12%). The most interesting fea- net) is high, 26% giving an indication ture regarding the mineralogical compo- of saturation of the higher density min- sition of the Ijjävri samples is the amount erals to Unit 1 . This in turn, can be an of garnet (category HM1 ). The amount of indication of higher wind velocities HM1 minerals in Unit 1 is 29%, higher than during the deposition of Unit 1 . in Units 2 (21%) and 3 (17%), giving an in- •The data suggests high wind velocities dication of saturation of the higher den- and transportation rates with effective sity minerals in Unit 1 . This in turn, can aeolian processes (sorting and grind- be indication of higher wind velocities ing) during the deposition of Unit 1 . during the deposition of Unit 1 . There were two TL/OSL dates associat- Unit 2 (Ijjävri dune field) ed with Unit 1 . Sample K 8 from Dune 66/ Unit 2 is the most significant deposit also 120 cm resulted in a combination age at the Ijjävri dune field. It is most complex 10 200 ± 1200 yrs. Sample K 9 from the Site and thickest accommodating most of the 74/ 70cm gave inconsistent TL and OSL sand samples analysed ( 53) and radiocar- dates ( 14 500 and 10 600 yrs, respectively) bon dated charcoal horizons ( 34). It is po- leaving the sample without a combination sitioned in the middle part of the dune sec- age (due to incomplete zeroing of the sig- tion and the strata is defined as dune fore- nal). Sample 66/ 120cm is taken from the set cross-bedded lamination, dipping at lower part of Unit 1 , indicating early post- high to intermediate angles of 24˚ to 10˚, glacial mobilization of the dunes. The on- indicating the advance of the dune. The set of the aeolian phenomena at the Ijjävri bulk sediment is closely packed, consisting dune field is controlled by the age estimate of thin, 0 . 5 to 1 cm thick laminae of dark of the deglaciation on the area, which took sand interstratified with 3 – 5 cm thick fore- place at circa 10 900 yrs ago (Kujansuu et sets of light sand bearing thin laminae. al., 1998; Johansson and Kujansuu, in Dimpled strata, deformation structures press). These dates can establish only the and looser packing are common towards onset of Unit 1 formation at circa 10 900 the upper contact. Deformation structures 132 yrs before present. are implying presence of niveo-aeolian processes and plant roots during and af- Binocular grain shape studies of the ter the deposition. However, these defor- material from Unit 2 indicate similar dis- mation structures occur at every side of a tribution of the material into the round- dune, not only at the lee side (as was the ness glasses as in Units 1 and 3 (Table 31). case in Unit 1 ) indicating that the advance The results for Unit 2 are based on 49 sam- rate of the dunes has not necessarily been ples. The proportions of shiny/matt par- high. Often there is a charcoal horizon ticles ( 37%/63%, respectively) indicate ae- underlain by a soil formation on the top olian maturation of the quartz particles of this unit. These lithostratigraphic fea- compared to Unit 1 . The roundness of the tures indicate that the material was depos- material in Unit 2 is characterised by class ited with changing wind directions, mod- C (sub-angular) quartz grains ( 44%), and erate wind velocities and with the presence class D (sub-rounded) grains ( 37%). The of vegetation. rounding index of 0 . 34 indicates a vague- In the similar way as at the Muddus- ly better rounding compared with other jävri, the dominance of the cross-bed sets Units. As indicated by the rounding index, with wedge-shaped or curved bounding the rounding of the quartz grains in every surfaces in Unit 2 can be a result of a high- Unit in Ijjävri dune field is better than in er (compared to Unit 1 ) relief at the time the Muddusjävri dune field. An average of of the deposition responding to a chang- 12% were classified into the class B (angu- ing wind system. It is probable, that the lar), while 5 % were classified into class E, higher relief together with a shifting wind indicating rounded grains. However, the system (causing more filling and cutting) values correspond very well to the average contributed both to the complex cross-bed values of the whole Ijjävri dune field (class strata of Unit 2 . C 44% and class D 35%). Grain size distribution of the dune sand SEM analyses from the quartz grains in in Unit 2 is indicated by 53 sand samples. the Unit 2 (samples 69/lee/100, 71/lee/160, Unit 2 sand is characterised by a mean size and 75/lee/140 , Table 17) indicate some of Φ 2 . 42, which corresponds with the av- common features. Especially the samples erage value for the whole dune field from the Dunes 69 and 71 are showing very ( Φ 2 . 47) and indicates fine sand. The in- similar surface textures. Sample 75 is not clusive graphic standard deviation, σ 1 val- in deep contrast with them either. Com- ue of 0 . 71 indicates moderately sorted ma- pared to the sample 63/ww/50 (Unit 1 ) they terial. The average inclusive graphic skew- indicate increase in the textures indicating ness, Sk 1 value of –0 . 05 shows that the aeolian abrasion. Unit 2 quartz grain sur- grain size distribution of the material in face textures are characterised by the dom-

Unit 2 is symmetrical. Also the K G (inclu- inance of textures indicative of aeolian sive graphic kurtosis) value of 1 . 07 indi- transportation. This result indicates effec- cates mesokurtic (Gaussian) distribution tive aeolian process and aeolian matura- of the grain size. For the palaeoenviron- tion of the quartz particles during the dep- mental modelling, these values can con- osition of Unit 2 . tribute the following: The material of the Samples of Unit 2 contain the same Unit 2 is slightly coarser and less sorted amount of quartz (average 18%, range 11– than in the Unit 1 , but not remarkably. The 25%, n = 6 ) as Unit 1 . Mineral composi- aeolian process during the deposition of tion of Unit 2 is characterised by increased the Unit 2 has been effective, producing amount of feldspars and decrease in gar- typical grain size parameter values for in- net (HM1 ) content. The stable quartz con- land dunes. tent together with the decrease in heavy 133 mineral content implies weaker wind con- fective aeolian processes (sorting and ditions compared to Unit 1 . grinding) and aeolian maturation of There are 34 radiocarbon age estimates quartz particles during the deposition and four TL/OSL dates associated with of Unit 2 . Unit 2 . Sample K 6 from the Site 74/ 80cm was dated 7700 ± 1000 yrs old, and is tak- Unit 3 (Ijjävri dune field) en from the contact with the delta sedi- Unit 3 is the uppermost deposit at the ments, however, the section continues with Ijjävri dune field. The volume of Unit 3 is Unit 2 sediments, indicating a hiatus be- quite significant. The lithostratigraphy is tween the delta sediments and Unit 2 . The characterized with tabular foreset laminae other TL/OSL dates are: K46600 ± 700 yrs, dipping at low-angles of 0 ˚–5 ˚. The bulk K 98700 ± 1200 yrs and K 12 1000 ± 200 sediment is re-accumulated dune sand yrs. They indicate that the transition from characterised by loose packing and often Unit 1 to Unit 2 sediments occurred at cir- poor stratification. Deformation structures ca 8700 ± 1200 yrs before present. caused by roots, soil formation and deni- The laboratory codes for the radiocar- vation processes are common. No charcoal bon dates are indicated in the Table 24, all horizons were detected within this unit. the samples (numbers 52– 75) from the Grain size distribution of the dune sand Ijjävri dune field represent Unit 2 . They in Unit 3 is indicated by 21 sand samples. indicate that the deposition of Unit 2 took Unit 3 sand is characterised by a mean size place between 8460 cal. yrs BP and 640 of Φ 2 . 35, which is slightly coarser than the cal. yrs BP (minus the average age of a for- average value for the whole dune field est, 150 yrs). The oldest charcoal dates have ( Φ 2 . 47) and Units 1 and 2 . However, the been associated with Unit 2 , and the up- value indicates fine sand. The inclusive per limit for Unit is indicated by the 2 graphic standard deviation, σ 1 value of youngest charcoal horizon associated with 0 . 60 indicates moderately well sorted ma- Unit 2 , which actually indicates the stabi- terial. The average inclusive graphic skew- lization phase before the latest activation ness, Sk 1 value of 0 . 00 shows that the grain phase, Unit 3 . size distribution of the material in the Unit

3 is symmetrical. K G (inclusive graphic kur- Summary of Unit 2 (Ijjävri dune field): tosis) value of 1 . 03 indicates mesokurtic •The deposition of Unit 2 took place be- (Gaussian) distribution of the grain size. tween circa 8500 yrs BP and 700 yrs BP. For the palaeoenvironmental modelling, •The wind direction during deposition these values can contribute the following: has shifted from the WNW-ESE to The material of Unit 3 is slightly coarser more western wind direction. than in Units 1 or 2 , but not remarkably. •The lithostratigraphy is characterised However, the aeolian process during the by cross-bedded strata with high to in- deposition of Unit 3 has been effective, termediate angles. producing typical grain size parameter val- •The dune material is defined as fine ues for inland dunes. (slightly coarser than in Unit 1 ), moder- Binocular grain shape studies of the ately sorted sand, mainly sub-angular material from Unit 3 indicate quite simi- and sub-rounded quartz grains. The ae- lar distribution of the material into the olian maturation of the quartz particles roundness glasses as in Units 1 and 2 (Ta- can be observed compared to Unit 1 . ble 31). The results for Unit 3 are based on •The data suggests moderate wind veloc- 28 samples. The proportions of shiny/matt 134 ities and transportation rates with ef- particles are not corresponding to aeolian maturation of quartz particles as the val- There is one TL/OSL date associated ues indicate increase in shiny particles with the Unit 3 . Sample K 13, 300 ± 100 yrs compared to Unit 2 . The proportions of was dated from Dune 70/ 60 cm. The Unit shiny/matt particles in Unit 3 are 49%/51%, 3 represents the latest, still ongoing reac- respectively. The material in Unit 3 is char- tivation phase at the Ijjävri dune field. It acterised by class C quartz grains ( 43%), is approximated to have started circa 500 which indicates sub-angular rounding.An years ago. average of 26% were classified into the class B (angular), while 29% were classi- Summary of Unit 3 (Ijjävri dune field): fied into class D, indicating sub-rounded •The deposition of Unit 3 has been tak- grains. These values are indicating less ing place during the past 500 years. rounded quartz grains comparing to the •The prevailing wind direction during average values of the whole Ijjävri dune deposition has beenW-E orientated. field (class B 15%), and demonstrate the •The lithostratigraphy is characterised poorest roundness within Units 1 – 3 of by loose, unstratified, deformed strata. Ijjävri. However, the difference is not dis- •The dune material is defined as fine tinctive enough, to allow any palaeoenvi- (slightly coarser than in Units 1 and 2 ), ronmental conclusions to be drawn. moderately well sorted sand, mainly Unit 3 is represented by one SEM sam- sub-angular, a dominance of matt ple: 70/lee/ 60cm. The percentage of surface quartz particles ( 51%), strong indica- textures indicating aeolian abrasion is the tion of aeolian abrasion in the quartz highest of all studied samples from Ijjävri surface texture. and is in contrast with the results from the •The data suggests moderate to weak binocular analysis. The sample 70/lee/ wind velocities and transportation rates 60cm is taken from a strata indicating re- with less effective aeolian processes activation and accumulation followed by (sorting and grinding) and aeolian mat- deflation ( Fig . 41). The high content of ae- uration of quartz particles during the olian imprints in the quartz grains indi- deposition of Unit 3 . cates that the tractional deposition type produces aeolian surface textures. The 6.3. Climate or human impact sample does not represent a typical Unit 3 Evidently, both the dune fields have expe- sand. Unfortunately, there are no SEM rienced cycles of aeolian reactivation and samples from the usual unstratified loose stabilization during the Holocene. Arche- sediment of Unit 3 . ological records indicate that the areas Mineralogical composition of Unit 3 were inhabitated since the early Inari Stone dunes at the Ijjävri dune field is represent- Age ( 8000– 3000 BP; Carpelan, 1988 ). The ed by one thin section.When interpreting trapping pit systems detected on both dune the results of this single sample represent- fields indicate hunting of wild deer in the ing the whole unit, one has to be cautious. areas ( Chapters 2 . 1 . 5 and 2 . 2 . 5 ). However, The variations between the samples in Units the theory of the early use of fire in order 1 and 2 were great. However, the sample in- to force the animals in the required direc- dicates, that the quartz content has in- tion (Tolonen, 1983 ) has later been criti- creased to 27%. The content is higher than cized (P. Halinen, pers. comm., 2003 ). Oth- the maximum quartz contents in Units 1 erwise, the early human impact must have and 2 . This is giving an indication of aeo- been an insignificant contributor to the lian maturation of the sediment. Heavy reactivation phenomena, quite in contrast mineral content has decreased to 17%. to the present day situation. 135 On the other hand, the forest fire et al ., 1997 ) and lake sediments e.g. in Swe- records and studies related to the Holocene den and Finland ( e.g. Korhola et al., 2000) forest fire frequency (Chapter 2 . 3 ) indicate and Estonia (Veski et al., 2004 ). Reactiva- that the natural forest fire frequency in the tion of the dune fields might be related to pine dominated forest is different from reorganization of the atmospheric and sur- birch-pine forests. It can be concluded, that face ocean circulation over Greenland and the forest fire frequency and the maximum North Atlantic. It is possible that this “ 8 . 2 extent of forest fires is governed by the ka cooling event” resulted from surface wa- shifting of the forest zones. Both the dune ter freshening and a decrease in the North fields have been covered with pine forest Atlantic Current northward transport of for most of the post-glacial time. The re- surface water causing cooling of the high treat of the pine forest from the Ijjävri area latitudes. This mechanism might decrease had begun already circa 6000 cal. yrs BP precipitation over the sub-arctic Fennos- (Seppä, 1996), but the temperature rise candia and thus trigger the onset of the during MWP has produced pine forest on aeolian reactivation in the region. The re- the area again (Kultti et al., in press). After sults for the aeolian reactivation history MWP at circa 1000 cal. yr BP, there has ( Chapter 5 . 6 ) suggest coupled ocean–at- been no pine forest on the Ijjävri area. This mosphere forcing of the early- and mid- leads to the assumption, that natural for- Holocene aeolian history in the subarctic est frequency alone has easily caused the Finnish Lapland. The early Holocene cli- dune activating forest fires on both dune mate was most favourable for aeolian proc- fields for most of the post-glacial time. esses as discussed in Chapter 5 . 6 . The mid- Forest fires are a common and natural phe- Holocene reactivation phenomena can also nomena in coniferous forest. They do trig- be linked to climate variation. At the stud- ger dune reactivation, but the duration of ied dune fields the different pattern of re- the reactivation phase is crucial and con- activation phenomena during the second trolled by climate and/or human activity. reactivation phase is suggested to have re- This is demonstrated on Unit 3 of Mud- sulted from differences in forest migration. dusjävri dune field, where the forest fires However, the data suggests that the climate alone have not resulted in dune reactiva- can not explain the reactivation processes tion. In conclusion, human activity as a during late-Holocene for the last 4500 yrs promoter for dune reactivation prior to before present. The reactivation phenom- 1000 cal. yrs BP is improbable in the stud- ena are not concurrent of the dune fields ied dune fields. and do not indicate any global or regional The climate signal on the duration of climate signal. the active phases can be detected on both The latest reactivation phenomena at the dune fields during early- and mid- Ijjävri dune field, from circa 500 yrs BP to Holocene. The duration of the primordial the present is promoted by both climate dune building phase circa 10 800 cal. yrs and human impact. The LIA event pro- BP to circa 8700 cal. yrs BP is controlled vides the triggering factor for the begin- by climatic factors. The earliest aeolian re- ning of the phase. Käyhkö ( 1997) detected activity phase R 1 can be compared to the the latest phase of aeolian activity at 1100– rapid cooling around 8 . 2 ka observed in 1650 AD across the area he had studied. He many proxies around the North Atlantic was not able to indentify any unique de- region. It has been recorded e.g . in Green- flation triggering factor, but came to the land ice-core proxies (O’Brien et al ., 1995), conclusion that several agents – such as the 136 North Atlantic deep-sea sediments (Bond LIA event, reindeer trampling and forest fires – acted in combination. However, above. The continuous aeolian activity there was a change in the vegetation pat- from the last onset circa 500 yrs ago is sup- tern and forest fire capacity after circa 1000 ported by human activity including rein- yrs BP at the Ijjävri area, as discussed deer husbandry and off road vehicles.

137 7. Conclusions and recommendations for future research

he main conclusions discussed can be ulite Complex type garnet at the Ijjävri T summarised as follows: sediments implies a minimum sediment The morphological nature of the dune transportation of 40 km along the esker fields, a parabolic dune system, is indicat- chain. Quartz/Feldspar relation is indica- ing a cold climate, relatively high ground tive for transportation distance in the water table, high moisture content, limit- Muddusjävri dune field (n = 18 , p = 0. 08). ed sand supply and the presence of vege- The palaeoenvironmental model result- tation during the deposition. The morpho- ed in the following main conclusions: logical frame of the dune fields originates from the initial dune forming period dur- Unit 1 ing the deposition of Unit 1 and met only Unit 1 , the initial dune building period, minor modification at the deposition of took place in the early Holocene at cir- Units 2 and 3 . ca 10800 to 8700 yrs before present.

Mean grain size, Mz, and sorting, σ 1 cor- The initial dune forming wind during relate with the transportation distance at the the deposition of Unit 1 has been SW- Muddusjävri dune field. The dune materi- NE orientated at Muddusjävri and al at Ijjävri is in general finer in grain size WNW-ESE orientated at Ijjävri. Unit 1 than at Muddusjävri. Grain size distribution is characterised by high wind velocities on both dune fields indicates high symme- and transportation rates with effective try and good sorting which are indicative aeolian processes (sorting and grinding) for aeolian deposition. Most of the samples during the deposition on both dune were indicative for aeolian stability. fields. Quartz grain surface texture study by SEM proved to be a precise and successful Unit 2 method in this environment. The imprint Unit 2 represents on both dune fields classes indicative of aeolian, sub-aqueous the longest accumulation series with and glacial transportation were distin- several stabilization-reactivation cycles guishable. The roundness of the quartz and shifting wind directions. The dep- grains does not reflect anything in this en- osition of Unit 2 took place between vironment, whereas the amount of matt/ 8640 cal. yrs BP and 2380 cal. yrs BP shiny particles is more informative. at Muddusjävri and between circa 8500 Mineralogical composition of the dune cal. yrs BP and 700 cal.yrs BP at Ijjävri. fields largely reflects the mineralogy of the Dune reactivation phases R 1 (8200– 138 parent materials. High amounts of Gran- 7800 cal. yrs BP), R2 (7400– 5700 cal. yrs BP at Muddusjävri and 7500– 7200 region. Reactivation of the dune fields cal. yrs BP at Ijjävri), R3 (4800– 3900 might be related to the reorganization of cal. yrs BP at Muddusjävri and 4600– the atmospheric and surface ocean circu- 4300 cal. yrs BP at Ijjävri) and R 4 lation over Greenland and the North At- ( 3300– 2700 cal yrs BP at Muddusjävri lantic. It is possible that this “ 8 . 2 ka cool- and 2800– 2400 cal. yrs BP at Ijjävri) ing event” resulted from surface water have taken place during the deposition freshening and decrease in the North At- of Unit 2 . The wind direction during lantic Current northward transport of sur- deposition has shifted from SW-NE to face water causing cooling of the high lat- W-E orientated at Muddusjävri and itudes. This mechanism might decrease from WNW-ESE to more western at precipitation over the sub-arctic Fennos- Ijjävri during the deposition of Unit 2 . candia and thus trigger the onset of the The data suggests a dune advance with aeolian reactivation in the region. The re- moderate wind velocities and transpor- sults for the aeolian reactivation history tation rates and with effective aeolian suggest coupled ocean–atmosphere forcing processes (sorting and grinding) and of the early- and mid-Holocene aeolian aeolian maturation of quartz particles history in the subarctic Finnish Lapland. during the deposition of Unit 2 . However, the data suggests that the climate variability can not explain the reactivation Unit 3 processes during late-Holocene, for the last The deposition of Unit 3 has been tak- 4500 yrs before present. The latest reacti- en place since 2380 cal. yrs BP at Mud- vation phenomena at the Ijjävri dune field, dusjävri and since 500 yrs before from circa 500 yrs BP to the present is pro- present at Ijjävri. Dune reactivation moted by both climate and human impact. phase R 5 (since 500 yrs BP) is an on- The LIA event provides the triggering fac- going phemonena at the Ijjävri dune tor for the beginning of the phase. There field. The prevailing wind direction was a change in the vegetation pattern and during deposition has been from W-E forest fire capacity after circa 1000 yrs BP at Muddusjävri and at Ijjävri. The at the Ijjävri area. The continuous aeolian lithostratigraphy is characterised by un- activity from the last onset circa 500 yrs stratified, deformed strata. The data ago has been supported by human activi- suggests moderate to weak wind veloc- ty including reindeer husbandry and off ities and transportation rates with ef- road vehicles. fective aeolian processes (sorting and The inconsistency of TL/OSL dates and grinding) and aeolian maturation of radiocarbon age estimates is probably due quartz particles during the deposition to the mixing of old charcoal into a char- of Unit 3 . coal layer. In order to control the “old char- Human activity as a promoter for dune coal” effect several AMS dates from a sin- reactivation prior 1000 cal. yrs BP is im- gle charcoal layer should be taken and probable on the studied dune fields. On the compared with the dates of this study. other hand, the climate signal on the du- TL/OSL dating has suffered from poor ration of the active phases can be detect- sampling technique. More precise sedi- ed on both the dune fields during early- ment sampling technique with sampling and mid-Holocene. The earliest aeolian boxes rather that plastic tubes is recom- reactivity phase R 1 can be compared to the mended. rapid cooling around 8 . 2 ka observed in The nature of the sedimentation proc- many proxies around the North Atlantic ess is evidently complicated. On both dune 139 fields the contact between the bottommost er, it is difficult and uncertain to find a se- dune sediments and the delta indicates not quence with continuous, undisturbed sed- only erosion of the delta but also of the imentation. The dune fields have through- dune sediments. The stratigraphy of the out their history constantly been reworked dunes was also noticed to vary within a by wind action, leading to complicated dune as charcoal horizons together with sedimentation/deflation patterns in their the sand units have experienced patchy sedimentary structure. This means that any deflation. It is necessary to include sedi- stratigraphic work on dune fields requires mentological and stratigraphical observa- well planned field work with an adequate tions to any study on dune fields. Howev- number of study sections.

140 Acknowledgements

Professor Veli-Pekka Salonen has acted as at the Godwin laboratory, at the Univer- a supervisor of this work during 2003 – sity of Cambridge during 1993 – 1996. 2004 . His encouragement, help and guid- Most of the laboratory work was done ance have been essential in finishing this there. Several people helped with my project. Professor Emeritus Joakim Don- work, especially Mr Simon Crowhurst, ner was my supervisor during 1992– 1996, who programmed the procedure for dig- and I want to thank him for the numer- itizing SEM-images. I also wish to thank ous field trips and discussions that have Docent Philip Gibbard, Sir Nicholas influenced this work and my way of think- Shackleton, Dr. Roy Switsur, Dr.Antho- ing probably more than I realise. I also ny Carter, Dr. Deborah Hemming, Dr. wish to thank Professor Juha Karhu, Head Neil Loader and the others at the God- of Geology Department, for his positive win laboratory for scientific discussions attitude towards my PhD work. Financial and good company. support from the Academy of Finland Most of the radiocarbon- and all TL/ ( 1992– 1994; grants 1011946 and 1011922) OSL-analyses have been carried out at the and the Graduate School of Geology (1 . 10.– Dating Laboratory, University of Helsin- 31. 12. 2003 ) is greatly appreciated. ki, under the supervision of Professor The manuscript was reviewed by Pro- Högne Jungner. He kindly revised the fessor Juha-Pekka Lunkka and Dosent Pe- Chapter 4 . 6 (Dating of the aeolian events). ter Johansson. Their comments improved Some of the radiocarbon dates were ana- the thesis considerably. Mr David Wright lysed at the Geological Survey of Finland revised the English language. by Ph.Lic. Tuovi Kankainen. The Muddusjävri dune field area was Thin sections of the sand samples were selected for this study following a recom- prepared at the University of Helsinki, mendation by Professor Matti Saarnisto, Department of Geology by Mr Lars Rämö who had mapped the Quaternary depos- and Ms Helena Korkka. The supervision of its of the area in the 1960’s for the Geo- Professor Juha Karhu and Dr. Paula Ko- logical Survey of Finland. He acted also as sunen is highly appreciated with the exam- the editor of the manuscript and his efforts ination of the thin sections. Geologist Bo and help are greatly appreciated. The Johansson (GSF) kindly analysed the com- Ijjävri dune field area was selected for this position of garnet grains with the micro- study by the recommendation of Profes- probe technique. M.Sc. Akseli Torppa took sor Matti Eronen. the photograph of the structure thin sec- I was very fortunate to be able to work tion ( Fig. 45). 141 M.Sc. Techn. Hannu Koivula, Finnish dusjärvi Experimental Farm of the Univer- Geodetic Institute, provided information sity of Helsinki in Kaamanen was used as for the land uplift in the northernmost a base during several field trips. Finland. Lecturer Petri Halinen, Depart- I sincerely thank all my colleagues at the ment of Archeology, University of Helsin- Department of Geology for their encour- ki, discussed with me of the theories of agement, scientific discussions and relax- early human impact in Lapland. Dr. Pent- ing moments; especially M.Sc. Anu Kaak- ti Sepponen, Finnish Forest Research In- inen, Dr. Seija Kultti, Dr. Heikki Seppä, stitute, Rovaniemi, provided data of the M.Sc. Anu Hakala, M.Sc. Minna Väliran- vegetation cover on the dune areas. Ms ta, M.Sc. Samu Valpola and Ms Kirsi-Marja Kirsti Luoto, University of Helsinki, De- Äyräs. partment of Geography, has drawn the I wish to express my deepest gratitude original figure used as template for the Figs to my mother and to my family;Aarno, 1 and 2 . Ms Kirsti Keskisaari (GSF) draw Ada and Anton, who have always been the Figs 3 and 6 . there for me. All the dune excavations dur- Mr Matti Helminen, Tulvalahti, Inari, ing the field work were dug by Dr. Aarno provided some valuable information of the Kotilainen. From the field work in 1993 to Muddusjävri area. M.Sc. Martti Laakko- the hectic writing period in the autumn nen, Vantaa, is thanked for his computing 2003 , this work has been a factor in our program that was used for determining the lives. Sometimes it has been more in the grain size parameters. background, just nagging the back of my Tvärminne Zoological Station (Univer- head, sometimes overriding everything sity of Helsinki), Lammi Biological Station else. Besides my family, friends, relatives (University of Helsinki) and Archipelago and colleagues, numerous artists have Research Institute (University of Turku) eased the occasional dark moments and provided some peace and quiet during the given me strength. Thank you, everyone! most intensive writing periods. The Mud-

142 References

Aario, L., 1960.Vegetation zones of Lapland. Andersen, B.G., 1968. Glacial geology of west- Atlas of Finland. Helsinki. 162 pp. ern Tromsø, North Norway. Norges Geolo- Aartolahti, T., 1967 . Über die Dünen von Urja- giske Undersøkelse 256, 1 – 160. la. Comptes Rendus de la Societe Ge- Arhipov, S.A. & Volkov,V.I., 1980. Paleogeo- ologique de Finlande 39, 105– 121 . graphia Zapadio-Sibirskoi ravnini v maxi- Aartolahti, T., 1973. Morphology, vegetation and mum pozdneziryanskogo oledenenia. Nov- development of Rokuanvaara, an esker and osibirsk, Nauka, 107 pp. dune complex in Finland. Fennia 127, 53 pp. Arnalds, O., Gisladottir, F.O., & Sigurjonsson, Aartolahti, T., 1976 . Lentohiekka Suomessa H., 2001. Sandy deserts of Iceland: an over- (Driftsand in Finland). Suomalainen Tie- view. Journal of Arid Environments 47 (3 ), deakatemia. Esitelmät ja pöytäkirjat 1976, 359 – 371 . 83– 95 (in Finnish). Ash, J.E. & Wasson, R.J., 1983 . Vegetation and Aartolahti, T., 1980. Periglasiaalisen morfolo- sand mobility in the Australian desert dune gian tutkimus Suomessa (The research of field. Zeitschrift für Geomorphologie, Sup- periglacial morphology in Finland). Terra 92 plement Band 45, 7 – 25. ( 2 ), 74– 87. Atlas of Finland, 1987. Folio 131 . Climate. Na- Ahlbrandt, T.S. & Fryberger, S.G., 1980. Eolian tional Board of Survey & Geographical So- deposits in the Nebraska sand hills. U.S. ciety of Finland. 32 pp. Geological Survey Professional Paper 1120- Bagnold, R.A., 1937. The size-grading of sand A. by wind. Proceedings of the Royal Society Ahlbrandt, T.S. & Fryberger, S.G., 1981. Sedi- of London, Series A 163 , 250 – 264. mentary features and significance of inter- Bagnold, R.A., 1938. Grain structure of sand dune deposits. Society of Economic Palae- dunes and its relation to their water content. ontologists and Mineralogists, Special Pub- Nature 142, 403– 404 . lication 31, 293 – 314 . Bagnold, R.A., 1941 . The Physics of Blown Sand Allen, T., 1981. Particle size measurement. Lon- and Desert Dunes. London: Methuen. 265 don: Chapman and Hall. 806 pp. pp. Al-Saleh S. & Khalaf, F.I., 1982 . Surface textures Bagnold, R.A., 1954. Experiments on a gravity- of quartz grains from various recent sedi- free dispersion of large solid spheres in a mentary environments in Kuwait. Journal of Newtonian fluid under shear. Proceedings of Sedimentary Petrology 52 (1 ), 215 – 225. the Royal Society of London, Series A 255, Alestalo, J., 1979. Land uplift and development 49– 63. of the littoral and Aeolian morphology on Bagnold, R.A. & Barndorff-Nielsen, O.E., 1980. Hailuoto, Finland. Acta Universitatis Oulu- The pattern of natural grain size distribu- ensis, Series A, no. 82, Geologica 3 , 109– 120. tions. Sedimentology 27, 199– 207 . Andersen, B.G., 1965 . Glacial chronology of Ballarini, M., Wallinga, J., Murray,A.S., van western Tromsø, North Norway. Geological Heteren, S., Oost, A.P., Bos, A.J.J. & van Eijk, Society ofAmerica Special Paper 84, 35– 54. C.W.E., 2003 . Optical dating of young coast- 143 al dunes on a decadal time scale. Quaternary system for measuring OSL from quartz sam- Science Reviews 22, 1011– 1017. ples. Nuclear Tracks and Radiation Measure- Barndorff-Nielsen, O.E., 1977 . Exponentially ment 20, 549– 553. decreasing distributions for the logarithm of Bøtter-Jensen, L. & Mejdahl, V., 1988. Assess- particle size. Proceedings of the Royal Soci- ment of beta dose-rate using a GM multi- ety of London, Series A 353 , 401– 419 . counter system. Nuclear Tracks and Radia- Barndorff-Nielsen O.E., Dalsgaard, K., Hal- tion Measurement 14 (1 / 2 ), 187– 191. green, C., Kuhlman, H., Møller, J.T. & Schou, Bøtter-Jensen, L. & Murray,A.S., 1999. Devel- G., 1982 .Variation in particle size over a opments in optically stimulated lumines- small dune. Sedimentology 29, 53– 65. cence techniques for dating and retrospec- Banerjee, D., Bøtter-Jensen, L., & Murray,A.S., tive dosimetry. Radiation Protection Dosim- 1999. Retrospective dosimetry: Preliminary etry 84, 307 – 315. use of the single aliquot regeneration (SAR) Bradley, R.S. & Jones, P.D., 1993. “Little Ice Age” protocol for the measurement of quartz summer temperature variations: their na- dose in young house bricks. Radiation Pro- ture and relevance to recent global warm- tection Dosimetry 84, 421 – 426. ing trends. The Holocene 3 , 367 – 376. Bateman, M.D., 1994. The Chronology of Cov- Brander, G., 1934. Suomen Geologinen Yleis- ersand Deposition in Britain. Unpublished kartta – General geological map of Finland. PhD-thesis, University of Sussex, England. Lehti- Sheet C 3 . Kuopio. Maalajikartan se- 215 pp. litys- Quaternary deposits. Geologinen Toi- Bergqvist, E., 1981. Svenska inlandsdyner. Över- mikunta. (Geological Survey of Finland). sikt och förslag till dynreservat (Abstract: Briffa, K.R., Bartholin, T.S., Eckstein, D., Jones, Ancient inland dunes in Sweden. Survey and P.D., Karlén, W., Swingruber, F.H. & Zetter- proposals for dune reserves). Naturvårdsver- berg, P., 1990. A 1400-year tree-ring record ket Rapport 1412. Solna, Sweden. 103 pp. of summer temperatures in Fennoscandia. Besler, H., 1983 . The response diagram: distinc- Nature 346 , 434 – 439. tion between aeolian mobility and stability Bristow, C.S., Bailey, S.D. & Lancaster, N., 2000 . of sands and aeolian residuals by grain size The sedimentary structure of linear sand parameters. Zeitschrift für Geomorphologie dunes. Nature 406, 56– 59. Neue Folke, Supplement -Band 45, 287 –301. Brookfield, M.E.,1983 . Eolian sands. In Walker, Bigarella, J.J., 1975. Lagoa dune field, State of R.G. (ed.) Facies Models ( 2 nd edition). Ge- Santa Catarina, Brazil – a model of eolian oscience Canada Reprint Series 1 , 91– 103 . and pluvial activity. Boletim Paranaense de Cailleux, A., 1952. Morphoskopische Analyse der Geociências 33, 133 – 167 . Geschiebe und Sandkörner und ihre bedeu- Bigarella, J.J. & Salamuni, I., 1961 . Early Meso- tung für die Paläoklimatologie. Geologishe zoic wind patterns as suggested by dune Rundschau 40, 11– 19. bedding in the Botucatu Sandstone of Bra- Calkin, P.E. & Rutford, R.H., 1974. The sand zil and Uruguay. Geological Society of dunes of Victoria valley,Antarctica. The America Bulletin 72, 1089– 1105. Geographical Review 64, 189– 216 . Bond, G., Kromer, B., Beer, J., Muscheler, R., Carpelan, C., 1988. Ihmisen vuosituhannet Evans, M.N., Showers, W., Hoffman, S., Lot- itäisessä Inarissa. In: S. Lammi (ed.) Poh- ti-Bond, R., Hajdas I., & Bonani, G., 2001. joinen erämaa Kessi-Vätsäri, 46– 68. O&G Persistent Solar Influence on North Atlan- Kustannus, Vaasa. tic Climate During the Holocene. Science Carroll, D., 1939. Movement of sand by wind. 294, 2130– 2136. The Geological Magazine 76, 6 – 23. Bond, G., Showers,W., Cheseby, M., Lotti, R., Carroll, S.B. & Bliss L.C., 1982 . Jack pine-lichen Almasi, P., de Menocal, P., Priore, P., Cullen, woodland on sandy soils in northern Sas- H., Hajdas I., & Bonani, G., 1997. A perva- katchewan and northeastern Alberta. Cana- sive millennial-scale cycle in North Atlan- dian Journal of Botany 60, 2270 – 2282. tic Holocene and glacial climates. Science Carter, R.W.G., 1982 . Some problems associat- 278 , 1257 – 1266. ed with the analysis and interpretation of 144 Bøtter-Jensen, L. & Duller, G.A.T., 1992. A new mixed carbonate and quartz beach sands, illustrated by examples from north-west Ire- vision of the Finnish Quaternary. Commen- land. Sedimentary Geology 33, 35– 56. tationes physico-mathematicae Societas Sci- Castel, I.Y., 1991. Late Holocene eolian drift entiarum Fennica 40, 191– 198. sands in Drenthe (The Netherlands). PhD Donner, J.J., 1995. The Quaternary history of -thesis, Department of Geography, Univer- Scandinavia. Cambridge University Press. sity of Utrecht. ISBN 90- 6809- 143 - 3 . 157 pp. 200 pp. Castel, I.I.Y., Koster, E.A. & Slotboom, R.T., Donner, J.J., Ashour, M.M., Embabi, N.S. & Sii- 1989. Morphogenetic aspects and age of Late riäinen, A., 1999. The Quaternary geology of Holocene eolian drift sands in Northwest a playa in Farafra, Western Desert of Egypt. Europe. Zeitschrift für Geomorphologie, Annales Academiae scientiarum Fennicae. Neue Folge 33, 1 – 26. Geologica – Geographica 160, 49– 112. Catt, J.A., 1977. Loess and coversands. In: Shot- Douglas, M.S.V., Smol, J.P. & Blake W. Jr., 1994. ton, F.W. (ed.), 1977, British Quaternary Marked Post-18th Century Environmental Studies, 221 – 230. Change in High-Arctic Ecosystems. Science Christiansen,C. 1984.A comparison of sediment 266, 416 – 419 . parameters from log-probability plots and Drebs, A., Norlund, A., Karlsson, P., Helminen, log-log plots of the same sediments. Dept. J. & Rissanen, P., 2002 . Finnish Meteorolog- Geol., Univ. Aarhuus, Geoskriffer 20. ical Institute, 2002. Climatological statistics Clemmesen, L.B., Murray,A.S., Bech, J-H. & in Finland 1971– 2000. Climatological statis- Clausen, A., 2001 . Large-scale Aeolian sand tics in Finland No. 2002 , 1 . movement on the west coast of Jutland, Drenova, A.N., 1994. Reconstruction of wind Denmark in late Subboreal to early Subat- activity by studying dune sands. In: Derby- lantic time – a record of climate change or shire E., (ed.) Wind Blown Sediments in the cultural impact? Geologiska föreningens i Quaternary Record. Abstracts : QRA Meet- Stockholm förhandlingar (GFF) 123, 193 – ing in Royal Holloway, London, U.K. 203 . Ehrenberg, C.G., 1847. The Sirocco dust that fell Cooke, R., Warren, A. & Goudie, A.S., 1993 . at Genoa on the May 16th 1846. Quarterly Desert Geomorphology. University College Journal of the Geological Society of London London, London. 526 pp. 3 , 25– 6 . Dansgaard, W., Johnsen, S.J., Clausen, H.B., Eronen, M., Glückert, G., Plassche, O. van de, Dahl-Jehnsen, D., Gundestrup, N.S., Ham- Plicht, J. van de & Rantala, P., 1995. Land mer, C.U., Hvidberg, C.S., Steffensen, J.P., uplift in the Olkiluoto-Pyhäjärvi area, Sveinbjörnsdottir,A.E., Jouzel, J. & Bond, G., southwestern Finland, during the last 8000 1993 . Evidence for general instability of past years. ReportYJT- 95- 17. Teollisuuden Voi- climate from a 250-kyr ice-core record. Na- ma Oy. 26 pp. ture 364 , 218 – 220. Eronen, M. & Huttunen, P., 1987. Radiocarbon- David, P.P., 1977. Sand dune occurrences in dated subfossil pines from Finnish Lapland. Canada. Dept. of Indian and NorthernAf- Geografiska Annaler 69 A (2 ), 297– 304 . fairs, National Parks Board, Repost. Ottawa. Eronen, M., Hyvärinen, H., & Zetterberg, P., 183 pp. 1999a. Holocene humidity changes in north- Deer,W.A., Howie, R.A., & Zussman, J., 1992. ern Finnish Lapland inferred from lake sed- An introduction to the rock forming min- iments and submerged Scots pines dated by erals, 2 nd edition. Longman. 696 pp. tree rings. The Holocene 9 , 569– 580. Denton, G.H. & Karlén, W., 1973. Holocene cli- Eronen, M., Lindholm, M., Saastamoinen, S., & matic variations: their pattern and possible Zetterberg, P. , 1999b. Variable Holocene cli- cause. Quaternary Research 3 , 155– 205. mate, tree line dynamics and changes in nat- Dijkmans, J.W.A., 1990. Niveo-aeolian sedimen- ural environments in northern Finnish La- tation and resulting sedimentary structures: pland. Chemosphere: Global Change Sci- Sondre Stromfjord area, Western Greenland. ence 1 , 377 – 384 . Permafrost and Periglacial Processes 1 (2 ), Eronen, M., & Zetterberg, P. , 1996. Climatic 83– 96. changes in northern Europe since Late Gla- Donner, J.J., 1971. Towards a stratigraphical di- cial times, with special reference to dendro- 145 climatological studies in northern Finnish quency distribution of beach and river Lapland. Geophysica 32, 35– 60. sands. Journal of Sedimentary Petrology 37, Eronen, M., Zetterberg, P., Briffa, K.R., Lind- 327 – 354 . holm, M., Meriläinen, J. & Timonen, M., Friedman, G.M. & Sanders, J.E.,1978. Principles 2002 . The supra-long Scots pine tree-ring of sedimentology.Wiley & Sons, New York. record for Finnish Lapland: Part 1 , chronol- 792pp. ogy construction and initial inferences. The Galehouse J.S., 1971. Point counting. In: Carv- Holocene 12 (6 ), 673 – 680. er (Editor), Procedures in Sedimentary Pe- Filion, L.,1984.A relationship between dunes, trology, New York,Wiley Interscience, 385 – fire and climate recorded in the Holocene 409 . deposits of Quebec. Nature 309, 543– 546. Galon, R.,1959. New investications of the inland Filion, L. 1987. Holocene development of par- dunes in Poland. Przeglad geograficzny 31, abolic dunes in the central St. Lawrence low- Supplement, 93– 110. land, Quebec. Quaternary Research 28, 196– van Geel, B., Buurman, J. & Waterbolk, H.T., 209. 1996.Archaeological and palaeological in- Filion, L., & Payette, S., 1989. Subarctic lichen dications of an abrupt climate change in the polygons and soil development along a col- Netherlands, and evidence for climatologi- onization gradient on eolian sands. Arctic cal teleconnections around 2650 BP. Journal and Alpine Research 21, 175 – 184. of Quaternary Science 11, 451 – 460. Filion, L., Saint-Laurent, D., Desponts, M. & Geyh, M.A., Roeschmann, G., Wijmstra, T.A. & Payette, S., 1991. The late Holocene record Middeldorp A.A., 1983 . The unreliability of of aeolian and fire activity in northern Que- 14C dates obtained from buried sandy pod- bec, Canada. The Holocene, 1 (3 ), 201– 208. zols. Radiocarbon 25, 409– 416 . Finnish Meteorological Institute, 1991. Clima- Goldsmith, V., 1973. Internal geometry and or- tological statistics in Finland 1961– 1990. igin of vedetated coastal sand dunes. Jour- Supplement to the Meteorological yearbook nal of Sedimentary Petrology 43, 1128 – 1142. of Finland, vol. 90, part 1 . Goldsmith, V., 1985 . Coastal dunes. In: Davis, Flint, R.F.,1971. Glacial and Quaternary geolo- R.A.(ed.) Coastal sedimentary environ- gy.Wiley & Sons, New York. 892 pp. ments, 2 nd edn. 303 – 378 . Springer, New Yo rk. Folk, R.L.,1966.A review of grain-size param- Goudie, A. S. & Watson, A., 1981. The shape of eters. Sedimentology 6 (2 ), 73– 93. desert sand dune grains. Journal of Arid Folk,R.L., 1978.Angularity and silica coatings Environments 4 , 185 – 190. of Simpson Desert sand grains, Northern Goudie, A. S., Warren, A., Jones, D.K.C. & Territory. Journal of Sedimentary Petrolo- Cooke, R.U., 1987. The character and possi- gy 52, 93– 101. ble origins of the aeolian sediments of the Folk, R.L. & Ward, W.C., 1957. Brazos River Bar: Wahiba Sand Sea. The Geographical Jour- A Study in the Significance of Grain Size nal, 153, 231– 256. Parameters. Journal of Sedimentary Petrol- Grootes, P.M., Stuiver, M., White, J.W.C., ogy 27 (1 ), 3 – 26. Johnsen, S. and Jouzel, J., 1993. Comparison Foster, D.R., 1983 . The history and pattern of of oxygen isotope records from the GISP2 fire in the boreal forest of southeastern Lab- and GRIP Greenland ice cores. Nature 366, rador. Canadian Journal of Botany 61, 2459– 552 – 554. 2471. Haapanen, A. & Siitonen, P., 1978. Kulojen esiin- French, H.M., 1976. The Periglacial environ- tyminen Ulvinsalon luonnonpuistossa. ment. Longman, London and New York. 309 (Summary: Forest fires in Ulvinsalo strict pp. nature reserve). Silva Fennica 12, 187– 200. Friedman, G.M., 1961 . Distinction Between Hack, J.T., 1941 . Dunes of the Western Navaho Dune, Beach, and River Sands from Their County. The Geographical Review 31( 2 ), Textural Characteristics. Journal of Sedi- 240 – 263 . mentary Petrology 31, 514 – 529. Hails, J.R. & Bennett, J., 1980.Wind and sedi- Friedman, G.M., 1967 . By name processes and ment movement in coastal dune areas. 17th 146 statistical parameters compared for size fre- coastal engineering conference, 1565 – 1575. New York: American Society for Civil Engi- Johansson, P., Sahala, L. & Virtanen, K., 2000. neers. Rantamerkit, tuulikerrostumat ja moreeni- Hartmann, D., 1988. The goodness-of-fit to ide- muodostumat geologisina luontokohteina. al Gauss and Rosin distributions: a new Summary: The most significant raised grain size parameter – discussion. Journal beaches, aeolian and morainic landforms in of Sedimentary Petrology 58, 913– 917. Finland. Geologian tutkimuskeskus, Tut- Heikkinen, O. & Tikkanen, M., 1979. Glacial kimusraportti – Geological Survey of Fin- flutings in northern Finnish Lapland. Fen- land, Report of Investication 151 . 76 pp. nia 157 (1 ), 1 – 12. Johansson, P. & Kujansuu, R. (eds.), (in press). Helama, S., Lindholm, M., Timonen, M., Mer- Pohjois-Suomen maaperä ja kvartäärigeo- iläinen, J. & Eronen, M., 2002. The supra- loginen kehitys. Maaperäkarttojen 1 : long Scots pine tree-ring record for Finnish 400 000 selitys. Summary: Quaternary de- Lapland: Part 2 , interannual to centennial posits and history northern Finland. Ex- variability in summer temperatures for 7500 plnation to the maps of Quaternary depos- years. The Holocene 12 (6 ), 681– 687. its 1 : 400 000. Geologian tutkimuskeskus, Hellemaa, P., 1998. The development of coastal Tutkimusraportti – Geological Survey of dunes and their vegetation in Finland. Fen- Finland, Report of Investication. nia 176 (1 ), 111– 221 . Jungner, H., 1979. Radiocarbon Dates I. Radio- Herlin, R., 1896. Tavastmons erosionsterasser carbon Dating Laboratory, University of och strandlinjer. Fennia 12 (7 ). Helsinki. Report no. 1 , Helsinki, 131 pp. Hirvas, H., 1991. Pleistocene stratigraphy of Jungner, H., Korjonen, K., Heikkinen, O., & Finnish Lapland. Geological Survey of Fin- Wiedemann, A.M., 2001. Luminescence and land, Bulletin 354 . 123 pp. radiocarbon dating of dune series at Cape Högbom, I., 1923.Ancient inland dunes of Kiwada, Oregon, USA. Quaternary Science northern and middle Europe. Geografiska Reviews 20, 811– 814 . Annaler 5 , 113– 243 . Jungnerius, P.D. & van der Meulen, F.,1989. The Hörmann, P.K., Raith, M., Raase, P., Acker- development of dune blowouts as measured mand, D. & Seifert, F., 1980. The granulite with erosion pins and sequential air photos. complex of Finnish Lapland: petrology and Catena 16, 369 – 376. metamorphic conditions in the Ivalojoki- Kádár, L., 1938 . Die periglazialen Binnendqnen Inarijärvi area. Geological Survey of Fin- des norddeutschen und polischen Flachen- land, Bulletin 308. 95 pp. landes. C.R. du Congrés International de Hunter, R.E.,1977. Basic types of stratification Geographie Amsterdam. Actes du congrés I, in small eolian dunes. Sedimentology 24, 167 – 183 . 361 – 387. Käyhkö, J. 1997 .Aeolian activity in subarctic Hyvärinen, H., 1973 . The deglaciation history Fennoscandia – distribution, history and of eastern Fennoscandia – recent data from modern processes. University of Reading, Finland. Boreas 2 (2 ), 85– 102 . United Kingdom. Unpublished PhD -thesis. Hyvärinen, H., 1975 .Absolute and relative pol- 314 pp. len diagrams from northernmost Fennos- Käyhkö, J.A., Worsley, P., Pye, K., & Clarke, candia. Fennia 142 , 1 – 23. M.L., 1999.A revised chronology for Aeo- Hyvärinen, H., 1976. Flandrian pollen deposi- lian activity in subarctic Fennoscandia dur- tion rates and tree-line history in northern ing the Holocene. The Holocene 9 (2 ), 195– Fennoscandia. Boreas 5 , 167 – 175. 205. Hyvärinen, H. & Alhonen, P., 1994. Holocene Kennedy, S.K., Meloy, T.P. & Gurney, T.E., 1985 . lake level changes in the Fennoscandian Sieve data- size and shape information. Jour- tree-line region, western Finnish Lapland: nal of Sedimentary Petrology 55, 356– 360. diatom and cladoceran evidence. The Khalaf, F.I., 1989. Desertification and aeolian Holocene 4 , 251 – 58. processes in the Kuwait desert. Journal of Ivanov, A.D., 1966. Eolovie peski Zapadnogo Arid Environments 16, 125– 145. Zabaikalya i Pribaikalya. Ulan-Ude Buryat- Klovan, J.E., 1966. The use of factor analysis in skoe knijnoe izdatelstvo, 230 pp. determining depositional environments 147 from grain size distributions. Journal of Sed- aeolian deposits and denivation forms, with imentary Petrology 36, 115– 125 . special reference to the Great Kobuk Sand Kocurek, G., 1981. Significance of interdune de- Dunes, northwestern Alaska. Earth Surface posits and bounding surfaces in aeolian Processes and Landforms 13, 153– 170 . dune sands. Sedimentology 28, 753 – 780 . Koster, E.A., Castel I.I.Y. & Nap, R.L., 1993 . Gen- Kocurek, G. & Dott, R,H. Jr 1981. Distinction esis and sedimentary structures of Late and uses of stratification types in the inter- Holocene Aeolian drift sands in northwest pretation of eolian sand. Journal of Sedi- Europe. In: Pye, K. (ed.) The Dynamics and mentary Petrology 51, 579 – 595. Environmental Context of Aeolian Sedi- Kolstrup, E., 1982 . Coversand and coversand mentary Systems. Geological Society Special stratigraphy in southern Denmark. Ge- Publication No. 72, 247– 267 . ografisk Tidskrift 82, 88– 91. Kotilainen, M., 1989. Dunes ´ 89 Meeting Nami- Kolstrup, E., Grun, R., Mejdahl V., Packman S. biassa.Geologi 41 (7 ), 144– 145 . Helsinki. C. & Wintle A.G., 1990. Stratigraphy and Kotilainen, M.M., 1990. Muddusjavren dyy- thermoluminescence dating of Late Glacial nikentän jääkaudenjälkeinen kehityshistoria. cover sand in Denmark. Journal of Quater- (Postglacial development of the Muddusja- nary Science 5 , 207 – 224 . vre dune field). University of Helsinki, De- Komar, P.D. & Cui, B., 1984. The analysis of partment of Geology. Unpublished M.Sc.- grain size measurements by sieving and set- thesis. 102 pp. tling tube techniques. Journal of Sedimen- Kotilainen, M.M., 1991. Aavikkopaholaisen tary Petrology 54, 603 – 614 . jäljillä – dyynikenttien uudelleenaktivoitu- Korhola, A. & Rautio, M., 2002. Cladocera and minen Pohjois-Lapissa. Summary: Some re- other brachiopod crustaceans. In: Smol, J.P. , activated dune fields in Northern Lapland, Birks, H.J.B. & Last, W.M. (eds.) Tracking Finland. Geologian tutkimuskeskus – Geo- Environmental Change Using Lake Sedi- logical Survey of Finland, Tutkimusraport- ments, 4 : Zoological Indicators, KluverAc- ti – Report of Investication 105, 105– 113 . ademic Publishers: Dordrecht, 5 – 41. Kotilainen, M.M., 2002. Subarctic aeolian reac- Korhola, A., Vasko, K., Toivonen, H.T.T. & tivity in northern Finnish Lapland: a 10 ka Olander, H., 2002. Holocene temperature record. In Lee, J.A. & Zobeck, T.M., 2002. changes in northern Fennoscandia recon- Proceedings of ICAR5 /GCTE-SEN Joint structed from chironomids using Bayesian Conference, International Centre for Arid modeling. Quaternary Science Reviews 21, and Semiarid Lands Studies, Texas Tech 1841 – 1860. University, Lubbock, Texas, USA Publication Korhola, A., Weckström, J., Holmsrtöm, L. & 02- 2 , 395– 397 . Erästö, P., 2000.A quantitative Holocene cli- Kotilainen, M.M. & Kotilainen, A.T., 2001 . Early matic record from diatoms in northern Fen- Holocene subarctic Fennoscandian aeolian noscandia. Quaternary Research 54, 284 – 294 . activity and North Atlantic climate variabil- Korsman, K., Koistinen, T., Kohonen, J., Wen- ity: possible land-sea correlation. The 25th nerström, M., Ekdahl, E., Honkamo, M., Id- Nordic Geological Winter Meeting,Abstract man, H. & Pekkala, Y., 1997 . Bedrock map Volume, p. 115. of Finland 1 : 1000 000. Geological Survey Kotivuori, H., 1986. Inventointikertomus. Mu- of Finland, Espoo, Finland. seoviraston esihistorian toimiston arkisto, Koster, E.A., 1978. De stuifzanden van de Velu- 22– 39. Unpublished report. we; een fysisch geografische studie (The eo- Krinsley, D.H. & Doornkamp J.C., 1973. Atlas lian drift sands of the Veluwe, Central Neth- of quartz sand grain textures. Cambrigde erlands; A physical geographical study): dis- University Press. 91 pp. sertation, University ofAmsterdam, Amster- Krisley, D.H. & Wellendorf, W., 1980. Wind ve- dam, 195 pp. locities determined from surface textures of Koster, E.A.,1988.Acient and modern cold-cli- sand grains. Nature 282 , 372– 373 . mate aeolian sand deposition: a review. Jour- Krumbein, W.C., 1934. Size frequency distribu- nal of Quaternary Science 3 , 69– 83. tions of sediments. Journal of Sedimentary 148 Koster, E.A. & Dijkmans, J.W.A., 1988. Niveo- Petrology 4 , 65– 77. Krumbein, W.C. & Sloss, L.L., 1963 . Stratigra- glasial and post-glacial dunes in North Kare- phy and Sedimentation, Second Edition, lia, eastern Finland. Geological Survey of W.H. Freeman and co, San Francisco, 660 Finland, Bulletin 254. 85 pp. pp. Livingstone, I. & Warren,A., 1996. Aeolian Ge- Krygowski, B.,1964 . Graniformametria mechan- omorphology: an introduction. Longman. iczna. Teoria, Zastosowanie. Zusammenfas- 256 pp. sung: Die mechanische Graniformametrie – Luho, V., 1948. Suomen kivikauden pääpiirteet. Theorie und Anwendung. Gesellschaft der Otava, 174 pp. Freunde der Wissenschaften in Poznan. Ab- Lumme, E.,1934. Die Flugsandfelder und handlungen der geographisch-geologischen Dunengebiete Finnlands nach Litteraturbe- Kommission II: 4 , 1 – 112. legen zusammen-gestellt. University of Kujansuu, R., Eriksson, B. & Grönlund, T. 1998. Turku, Department of Geography, Publica- Lake Inarijärvi, northern Finland: sedimen- tion 11. 77 pp. tation and late Quaternary evolution. Geo- Maarleveld, G.C., 1960. Glacial and periglacial logian tutkimuskeskus, Tutkimusraportti – landscape forms in the central and north- Geological Survey of Finland, Report of In- ern Netherlands. KNAG Tijdschrift 77, 298 – vestication.143 . 25 pp. 304 . Kultti, S., 2004 . Holocene changes in treelines Mabbutt, J.E., 1977. Desert Landforms. Canber- and climate from Ural Mountains to Finn- ra:Australian National University Press. 340 ish Lapland. Yliopistopaino. Helsinki, 33 pp. pp. Kultti, S., Mikkola, K., Virtanen, T., Timonen, Mahaney, W.C., 2002 .Atlas of sand grain sur- M. & Eronen, M., in press. Past changes in face textures and applications. Oxford Uni- the Scots pine treeline and climate in Finn- versity Press, New York. 237 pp. ish Lapland – evidence from pine megafos- Maikawa, E. & Kershaw, K.A., 1976. Studies on sils and lake sediments. The Holocene. lichen-dominated systems. XIX. The post- Lahtinen, A. & Mäkinen, K., 1985. Map of Qua- fire recovery sequence of black spruce-li- ternary deposits 1 : 400 000, Sheet 38 + 48, chen woodland in the Abitau Lake Region, Inari. Geological Survey of Finland, Espoo N.W.T. Canadian Journal of Botany 54, Finland. 2679– 2687. Lancaster, N., 1983 . Controls of dune morphol- Mainguet, M. & Callot, Y., 1978. L’Erg de Fachi- ogy in the Namib sand sea. In: Brookfield, Bilma, Tshad – Niger. Memoires et Docu- M.E. & Ahlbrandt, T.S. (eds.), Eolian Sedi- ment, Centre National de Recherches Scien- ments and Processes, Developments in Sed- tifiques 18. 18 pp. imentology 38. Amsterdam, Elsevier. Mäkelä, E., Sarmaja-Korjonen, K. & Hyvärinen, Lancaster, N., 1989. The Namib Sand Sea: Dune H., 1994. Holocene forest history of the forms, Processes and Sediments. Balkema. Pöyrisjärvi area north of the coniferous tree- 180 pp. line in western Finnish Lapland: a pollen Lancaster, N., 1995. Geomorphology of Desert stratigraphical study. Bulletin of the Geolog- Dunes. Routledge. London. 312 pp. ical Society of Finland 66, 81– 94. Landsberg, S.Y., 1956. The orientation of dunes Markov, K.K., 1955.Ancient continental dune in Britain and Denmark in relation to wind. of Europe. Ocherky po geographii chetver- The Geographical Journal 122, 176 – 189. tichnogo perioda. Gos. izd. geographich- Leiviskä, I., 1907 . Über die Oberflächenbildun- eskoy literaturi, M. 1955, 3 – 27. gen Mittel-Ostbottniens und ihre Entste- Marthinussen, M., 1962 . 14C -datings referring hung. Fennia 25 (2 ). 113 pp. to shore lines, transgressions, and glacial Leiviskä, I.,1929 . Über die Ose Mittel-Finnland. substages in Northern Norway. Norges Ge- Fennia 51 (4 ). 207 pp. ologiske Undersøkelse 213 , 118– 169 . Li, S.-H., Sun, J.-M. & Zhao, H., 2002 . Optical Marthinussen, M.,1974. Contributions to the dating of dune sands in the northeastern Quaternary Geology of North-easternmost deserts of China. Paleogeography, Paleocli- Norway and the CloselyAdjoining Foreign matology, Paleoecology 181, 419 – 429 . Territories. Norges Geologiske Undersøkelse Lindroos, P., 1972. On the development of late- 28, No. 315. 149 Martin, E. 1988. Relationship between dune for- Moiola, R.J., Spencer,A.B. & Weiser, D., 1974. mation and Baltic sea transgressions in Es- Differentiation of modern sand bodies by tonia. In: Winterhalter, B. (ed.) The Baltic linear discriminant analysis. Transactions – Sea. Geological Survey of Finland Special Gulf Coast Association of Geological Soci- Paper 6 , 79– 85. eties 24, 321 – 326 . Martin, C.W. & Johnson, W.C., 1995. Variation Murray,A. & Clemmesen, L.B., 2001. Lumines- in radiocarbon ages of soil organic matter cence dating of Holocene Aeolian sand fractions from Late Quaternary buried soils. movement, Northern Jutland, Denmark. Quaternary Research 43, 232 – 237. Quaternary Science Reviews (Quaternary McBride, E.F., 1971. Mathematical treatment of Geochronology) 20, 751– 754. grain distribution data. In: Carver, R.C. (ed.) Murray,A.S. & Olley, J.M., 2002 . Precision and Procedures in sedimentary petrology, 109– accuracy in the optically stimulated lumi- 127. New York,Wiley-Interscience. nescence dating of sedimentary quartz: a McIntyre,A. & Be,A., 1967 . Modern Coccol- status review. Geochronometria 21, 1 – 16. ithophoridae of the Atlantic Ocean: I. Pla- Murray, A.S. & Wintle, A.G., 2000. Lumines- coliths and Cyrtoliths, Deep Sea Research 14, cence dating of quartz using an improved 561– 597 . single-aliquot regenerative-dose protocol. McKee, E.D., 1966. Structures of dunes at White Radiation Measurements 32, 57– 73. Sands National Monument, New Mexico Näkkäläjärvi, O., 1964 . Preliminary Report on (and a comparison with structures of dunes Investication of Wild Reindeer Trapping Pits from other selected areas). Sedimentology in Finland. Studia Ethnographica Upsalien- 7 , 1 – 69. sis XXI, 228 – 233. McKee, E.D. & Tibbits, G.C., 1964 . Primary Niemelä, J., Ekman, I., & Lukashov, A. (Eds.), structures of a seif dune and associated de- 1993 . Quaternary deposits of Finland and posits in Libya. Journal of Sedimentary Pe- Northwestern part of Russian Federation trology 34, 5 – 17. and their resources. Scale 1 : 1000 000. Es- McManus, D.A., 1963 .A criticism of certain us- poo: Geological Survey of Finland, Special age of the phi notation. Journal of Sedimen- Paper 24 Part 2 (Maps). tary Petrology 35, 792– 796. Niessen, A.C.H.M., Koster, E.A. & Galloway, J.P. , Meriläinen, K., 1965 . Lehti-Sheet C 8 – 9 , Inari- 1984 . Periglacial sand dunes and eolian sand Utsjoki. Suomen geologinen yleiskartta – sheets; an annotated bibliography. U.S. Ge- General geological map of Finland. Kivila- ological Survey Open-File Report, 84– 167 . 61 jikartta – Pre-Quaternary rocks. Geological pp. Survey of Finland. Nowaczyk, B.,1976. The genesis and develop- Meriläinen, K. 1970. Taka-Lapin suurtektooni- ment of inland dunes in the western part of sista rakenteista. Geologi 22, 139 – 144. Hel- the Warsaw-Berlin pradolina in the light of sinki. examinations of the structure, granulation Meriläinen, K. 1976. The Granulite Complex and stratigraphy of the deposits which built and Adjacent Rocks in Lapland, northern them. Prace Komisji Geograficzno Geolog- Finland. Geological Survey of Finland, Bul- icznej Poznanskie Towarzystwo Przyjaciol letin 281 , Helsinki. Nauk, Tom 16, 108 pp. Middleton, G.V., 1976. Hydraulic interpretation O’Brien, S.R., Mayewski, P.A., Meeker, L.D., of sand size distributions. Journal of Geol- Meese, D.A., Twickler, M.S., & Whitlow, S.I., ogy 84, 405– 426. 1995. Complexity of Holocene Climate as Miettinen, A., 2002 . Relative sea level changes reconstructed from a Greenland ice core. in the eastern part of the Gulf of Finland Science 270, 1962 – 1964 . during the last 8000 years. Annales Oescher, H. & Riesen T., 1965 . Bern radiocar- Academiae Scientiarum Fennicae, Geologi- bon dates IV. Radiocarbon 7 , 3 . ca-Geographica 162 , 100 pp. Okko, V.,1954. Periglasiaalisesta morfologiasta Mikkola, E., 1932. On the Physiography of Late- Suomessa (Periglacial morphology in Fin- Glacial Deposits in Northern Lapland. Fen- land). (In Finnish with English summary). 150 nia 57 (1 ), 1 – 88. Terra 66( 2 ), 54– 57. Helsinki. Pachur, H.J. & Hoelzmann, P., 1991. Palaeocli- gin of the loess. Geological Magazine 9 , 293– matic implications of late Quaternary lacus- 305. trine sediments in Western Numibia. Sudan Ristaniemi, O., Eronen, M., Glückert, G. & Ran- Quaternary Research 36, 257 – 276 . tala, P., 1997. Holocene and recent shoreline Parks, D.A. & Rendell, H.M., 1992. Thermolu- changes on the rapidly uplifting coast of minescence dating and geochemistry of western Finland. Journal of Coastal Research loessic deposits in Southeast England. Jour- 13, 397 – 406. nal of Quaternary Science 7 , 99– 107 . Rubin, D.M., 1984. Factors determining desert Parrish, J.T. & Peterson, F., 1988. Wind direc- dune type. Nature 309, 91– 92. tions predicted from global circulation Ruegg, G.H.J., 1983 . Periglacial eolian evenly models and wind directions determined laminated sandy deposits in the Late Pleis- from eolian sandstones of the western Unit- tocene of northwest Europe, a facies unre- ed States – a comparison. Sedimentary Ge- corded in modern sedimentological hand- ology 56, 261 – 282 . books. In: Brookfield, M.E. & Ahlbrandt, Perttunen, M. & Vartiainen, H., 1992 . Glacioflu- T.S. (eds.) Eolian Sediments and Processes, vial transport of clasts and heavy minerals Elsevier Amsterdam, 455– 482 . from the Sokli carbonatite complex, Finn- Saaranen, V. &Mäkinen, J., 2002 .Determina- ish Lapland. Geological Survey of Finland, tion of postglacial rebound from the three Bulletin 366. precise levellings in Finland: status in 2002 . Porter, J. J., 1962 . Electron microscopy of sand Proceedings of the XIV General Meeting of surface textures, Journal of Sedimentary the Nordic Geodetic Commission (ed. Petrology 32, 124 – 135 . M. Poutanen, H. Suurmäki), 171 – 174 , ISBN Possnert, G. 1984.AMS with the Uppsala EN 951 - 711 - 247- 5 . tandem accelerator. Nuclear instruments & Saarnisto, M., 1973 . Contributions to the late- methods in physics research, Series B 5 , 159– Quaternary history of the Lutto river valley, 162 . Finnish Lapland. Commentationes physico- Possnert, G., 1990. Radiocarbon dating by the mathematicae Societas Scientiarum Fenni- Accelerator Technique. Norwegian Archae- ca 43, 11– 20. Helsinki. ological Review 23 (1 – 2 ), 30– 37. Saastamoinen, S., Kultti, S. & Eronen, M., 2000. Powers, M.C., 1953.A new roundness scale for A disturbance in the growth of pine in Finn- sedimentary particles. Journal of Sedimen- ish Lapland in the 4 th century BC. 24th Nor- tary Petrology 23, 117– 119. dic Geological Winter Meeting, Trondheim, Pye, K., 1980. Geomorphic evolution of coastal Norway, Jan. 6 .–8 . Abstracts, p. 168. sand dunes in a humid tropical environ- Sallenger, A.H., Jr 1979. Inverse grading and ment: North Queensland. Unpublished hydraulic equivalence in grain flow depos- PhD-thesis, University of Cambridge, U.K. its. Journal of Sedimentary Petrology 49, Pye, K., 1982 . Morphological development of 553– 562 . coastal dunes in a humid tropical environ- Salmi, M., 1963 . On the subfossil Pediastrum ment, Cape Bedford and Cape Flattery, Algae and Molluscs in the Late-Quaternary North Queensland. Geografiska Annaler sediments of Finnish Lapland. Archivum 64A, 212 – 227. Societatis Zoologicar Botanicae Fennicae Pye, K. & Tsoar, H., 1990 .Aeolian sand and sand “Vanamo” 18, 105– 120. dunes. Unwin Hyman. London. 396 pp. Sarmaja-Korjonen, K., 1995.The relationship Raukas, A., 1966. On the composition and strat- between Holocene fire frequencies and cli- ification of dune sands in Estonia. In: Eesti matic changes in Finland. In Heikinheimo, Geograafia Seltsi Aastaraamat 1966, Estoni- P. (ed.) International conference on past, an Academy of Sciences, Tallinn, 72– 89. (in present and future climate: proceedings of Russian, summary in English). the SILMU conference held in Helsinki, Fin- Rhodes, E.J., 2000. Observations of thermal land, 22– 25 August 1995, p. 106– 109. transfer OSL signals in glacigenic quartz. Sarnthein, M. & Diester-Haass, L., 1977. Eolian Radiation Measurements 32, 595– 602 . sand turbidites. Journal of Sedimentary Pe- Richthofen, F. von, 1882. On the mode of ori- trology 47, 868– 890. 151 Sauramo, M., 1924 . Suomen Geologinen Shuisky, Y. D., 1986. Studies of aeolian process- Yleiskartta. Lehti B 2 . Tampere. Maalajikar- es on seashores of the Soviet Union. Ge- tan selitys. Geologinen Komissio, Helsinki. ografiska Annaler 68 A, 33– 40. Schwan, J.C.G., 1988. Sedimentology of cover- Siegert, M.J., Dowdeswell, J.A., Hald, M. & Sv- sands in northwestern Europe: a study to endsen, J.-I., 2001. Modelling the Eurasian early Holocene Aeolian sand sheets in Eng- Ice Sheet through a full (Weichselian) gla- land, The Netherlands and the Federal Re- cial cycle. Global and Planetary Change 31, public of Germany. Dissertation, Free Uni- 367 – 385 . versity of Amsterdam, Amsterdam, 135 pp. Simola, M., 1979. Dyyni- ja rantasedimenttien Seppä, H.,1996 . Post-glacial dynamics of vege- raekoostumusten eroista. University of Hel- tation and tree-lines in the far north of Fen- sinki, Department of Geology and Palaeol- noscandia. Fennia 174 , 1 – 96. ogy. Unpublished Lic.-thesis no. 41. Seppä, H. & Birks, H.J.B., 2001. July mean tem- Slota, P.J., Jull, A.J., Linick, T.W. & Toolin, L.J., perature and annual precipitation trends 1987. Preparation of small samples for 14C during the Holocene in the Fennoscandian accelerator targets by catalytic reduction of tree-line area: pollen based climate recon- CO. Radiocarbon 29, 303 – 306. structions. The Holocene 11, 527– 539. Smith, H.T.U., 1949 . Pleistocene Research.A Seppä, H. & Birks, H.J.B., 2002 . Holocene cli- Review by the Committee on Interrelations mate reconstructions from the Fennoscan- of Pleistocene Research, National Research dian tree- line area based on pollen data Council. 11. Physical effects of pleistocene from Toskaljavri. Quaternary Research 57, climatic changes in nonglaciated areas: eo- 191– 199. lian phenomena, frost action and stream ter- Seppä, H. Nyman, M., Korhola, A. & Weck- racing. Bulletin of the Geological Society of ström, J., 2002 . Changes of treelines and al- America 60, 1485 – 1516. pine vegetation in relation to post-glacial Smith, H.T.U., 1964. Periglacial eolian phenom- climate dynamics in northern Fennoscandia ena in the United States. Report of the VIth based on pollen and chironomid records. International Congress on Quaternary,War- Journal of Quaternary Science 17, 287 – 301 . saw IV, 177 – 186. Seppälä, M., 1971 . Evolution of eolian relief of Sokolow, N.A., 1894. Die Dünen. Bildung, Ent- the Kaamasjoki-Kiellajoki river basin in wicklung und Innerer Bau. Berlin. 298 pp. Finnish Lapland. Fennia 104 , 1 – 88. Helsin- Sollid, J.L., Andersen S., Hamre N., Kjeldsen O., ki. Salvigsen O., Sturod S., Tveitå T. & Wilhelm- Seppälä, M., 1972. Location, morphology and sen A., 1973. Deglaciation of Finnmark, orientation of inland dunes in Northern North Norway. Norsk Geografisk Tidsskrift, Sweden. Geografiska Annaler 54A ( 2 ), 85– Bind 27, Hefte 4 . 104 . Sorby, H.C., 1877. The application of the mi- Seppälä, M., 1980. Deglaciation and glacial lake croscope to geology. Monthly Microscopi- development in the Kaamasjoki river basin, cal Journal 17, 13– 136. Finnish Lapland. Boreas 9 , 311– 319. Oslo. Sorsa, V., 1965. Pollenanalytische Untersuchun- Seppälä, M., 1981. Forest fires as activator of gen zur spätquartären Vegetations- und Kli- geomorphic processes in Kuttanen esker- maentwicklung im östlichen Nordfinnland. dune region, northernmost Finland. Fennia Annales Botanici Fennici 2 , 301– 413 . Helsin- 159 (1 ), 221 – 228 . ki. Seppälä, M., 1995. Deflation and redeposition Stetler, L. D. & Gaylord, D. R., 1996. Evaluating of sand dunes in Finnish Lapland. Quater- eolian-climatic interactions using a region- nary Science Reviews 14, 799– 809. al climate model from Hanford, Washing- Sepponen, P., 1985. The Ecological classification ton (USA). Geomorphology 17, 99– 113 . of sorted forest soils of varying genesis in Stokes, S., Bray, H.E., & Blum, M.D., 2001. Op- northern Finland. Communicationes Insti- tical resetting in large drainage basins: tests tuti Forestalis Fenniae 129 . 77 pp. of zeroing assumptions using single-aliquot Shepard, F.P., 1963 . Submarine Geology. New procedures. Quaternary Science Reviews 20, 152 York: Harper & Row. 557 pp. 879– 885 . Stuiver, M. & Reimer, P. J., 1993 . Extended 14C Tolonen, K., 1983 . The Post-glacial Fire Record. Data Base and Revised CALIB 3 . 0 14C Age In:Wein, R.W. and D.A. MacLean (eds.) The Calibration Program. Radiocarbon 35, 215 – Role of Fire in Northern Circumpolar Eco- 230. systems, 21– 44. John Wiley & Sons Ltd. Stuiver, M., Reimer, P.J., Bard, E., Beck, J.W., Tutkovskiy, P. A., 1909. Iskopaemie pustini Sev- Burr, G.S., Hughen, K.A., Kromer, B., Mc- ernogo polusharia. Zemlevedenie kn. 1 – 4 , Cormac, F.G., v.d. Plicht, J. & Spurk, M., p. 28– 64. (English summary). 1998. INTCAL98 Radiocarbon age calibra- Vincent, P.J., 1985 . Some Saudi Arabian dune tion 24, 000– 0 cal BP. Radiocarbon 40, 1041– sands: a note on the use of the response di- 1083 . agram. Zeitschrift für Geomorphologie 29, Synge, F., 1969. The raised shorelines and de- 117– 122. glaciation chronologyof Inari, Finland, and Visher, G.S.,1969 . Grain size distributions and South Varanger, Norway. Geografiska An- depositional processes. Journal of Sedimen- naler 51 A, 4 , 193 – 206. Stockholm. tary Petrology 39, 1074– 1106. Tanner, V., 1915. Studier öfver kvartärsystemet Vliet-Lanoë, B. van., Seppälä, M. & Käyhkö, J., i Fennoskandias nordliga delar III. Bulletin 1993. Dune dymanics and cryoturbation fea- de la Commission Géologique de Finlande tures controlled by Holocene water level 38, p. 1 – 815. Helsinki. change, Hietatievat, Finnish Lapland. Geol- Tanner, V., 1930. The problems of an esker II. logie en Mijnbouw 72, 211 – 224 . Bulletin de la Commission Géologique de Veski, S., Seppä, H. & Ojala, A., 2004 . Cold event Finlande 92, 33– 39. Helsinki. at 8200 yr B.P. recorded in annually lami- Terasmae, J. & Weeks, N.C., 1979. Natural fires nated lake sediments in eastern Europe. Ge- as an index of paleoclimate. Canadian Field ology 32, 681– 684 . Naturalist 93, 116– 125 . Wadell, H., 1933. Sphericity and roundness of Thomas, D.S.G., 1986. The response diagram rock particles. Journal of Geology 40, 443 – and desert sands – a note. Zeitschrift für 451 . Geomorphologie 30, 363 – 369. Wadell, H., 1935 .Volume, shape and roundness Thomas, D.S.G., 1987. Discrimination of dep- of quartz particles. Journal of Geology 43, ositional environments using sedimentary 250– 280. characteristics in the Mega-Kalahari, central Walker, R.G. & Middleton G.V., 1981. Facies southernAfrica. In: Frostick, L. & Reid, I. Models 4 : Eolian sands. In: R.G. Walker (eds) Desert sediments ancient and modern, (ed.), Facies Models. Geoscience Canada 1 . 293– 306. Oxford: Blackwell. Walton, E.K., Stephens, W.E. & Shawa, M.S., Tikkanen, M. & Heikkinen, O., 1993 . Deflation 1980. Reading segmented grain-size curves. of sand dunes in NW Finnish Lapland. Geological Magazine 117, 517 – 524 . Third International Geomorphology Con- Wasson, R.J. & Hyde R., 1983. Factors determin- ference, Hamilton. Programme with Ab- ing dune types. Nature 304 , 337– 339. stracts, p. 155. Watchman, A.L. & Twidale, C.R., 2002. Relative Tikkanen, M. & Heikkinen, O., 1995. Aeolian and “absolute” dating of land surfaces. landforms and processes in the timberline Earth-Science Reviews 58, 1 – 49. region of northern Finnish Lapland. Zeszy- Wiedemann, A.M., 1990. The coastal parabolic ty Naukowe Uniwersytetu Jagiellonskiego. dune system at Sand Lake, Tillamook Coun- MCLVI Prace Geograficzne – Zeszyt 98, 67– ty, Oregon, USA. In: Davidson-Arnott, R., 90. (ed.), Proceedings of the Symposium on Tiljander, M., Saarnisto, M., Ojala, A.E.K. & Coastal Sand Dunes 1990 . National Research Saarinen, T., 2003 . A 3000-year paleoenvi- Council, Ottawa, 341 – 358 . ronmental record from annually laminated Wilson, I.G.,1972.Aeolian Bedforms; Their de- sediment of Lake Korttajärvi, central Fin- velopment and origins. Sedimentology 19, land. Boreas 26, 566– 577 . 173 – 210 . Tinner,W. & Lotter,A., F., 2001. Central Euro- Wilson, I.G.,1973. Ergs. Sedimentary Geology pean vegetation response to arupt climate 10, 77– 106. change at 8 . 2 ka. Geology 29 (6 ), 551 – 554. Wintle, A., 1990.A review of current research 153 on TL dating of loess. Quaternary Science Evidence on climatic variability and prehis- Research 9 , 385 – 397 . toric human activities between 164 BC and Yaloon, D.H., 1975. Discussion of “ Internal ge- AD 1400 devired from subfossil Scots pines ometry and origin of vegetated coastal ( Pinus Sylvestris L.) found in a lake in Ut- dunes” Journal of Sedimentary Petrology 45, sjoki, northernmost Finland. Geological 359 . Society of Finland, Bulletin 66, 107 – 124 . Zackrisson, O., 1977. Influence of forest fires on Zetterberg, P., Eronen, M., & Briffa, K., 1995. A the North Swedish boreal forest. Oikos 29, 7500-year long pine tree-ring record from 22– 32. Finnish Lapland and its implications to Zeeberg, J.J., 1993 .Aeolian redistribution of gla- paleoclimate studies. In: Heikinheimo, P. cial silt and sand in Estonia and the Baltic (ed.) International Conference on Past, region. Rapport GEOPRO 1993. 12. Institute Present and Future Climate, Proceedings of of Geographical Research, Rijksuniversiteit the SILMU Conference held in Helsinki, Utrecht, The Netherlands. Finland, 22– 23 August 1995. Publications of Zetterberg, P., Eronen, M., & Briffa, K., 1994. the Academy of Finland 6 / 95, 151 – 154.

154 Appendix I Radiocarbon dates of the Muddusjävri dune field. Probability Distributions

Muddusjävri dune field, n 22

cal. BP 155 Appendix II Radiocarbon dates of the Ijjävri dune field.

Probability Distributions

Ijjävri dune field, n 34

156

cal. BP

ANNALES ACADEMIAE SCIENTIARUM FENNICAE GEOLOGICA-GEOGRAPHICA (Formerly Annales Academiae Scientiarum Fennicae A III, Geologica-Geographica)

121. Laaksonen, Kyösti: Seasonal variations in the influence of seas and inland waters upon mean air temperatures in Fennoscandia (1921–1950). (33 pp.) 1976 ...... 8,– 122. Ruotsalo-Aario, Ritva, and Aario, Leo: The effect of environmental factors and distance from the functional centre upon the density of rural settlement in the province of Kuopio in 1967. (52 pp.) 1977 ...... 12,– 123. Alhonen, Pentti: Penecontemporaneous deformation structures in a glaciofluvial environment: a new site from southern Finland. (12 pp.) 1978 ...... 5,– 124. Núñez, Milton G.: The Vantaa phosphate survey, a practical illustration of the method. (16 pp.) 1978 ...... 5,– 125. Ruotsalo-Aario, Ritva: The effect of distance from the commune centre on population density in the province of Kuopio in 1967. (19 pp.) 1978 ...... 5,– 126. Jauhiainen, Erkki: Geographical interpretation of photographs using equidensity film. (10 pp.) 1978 ...... 5,– 127. Kvasov, D.D.: The Late-Quaternary history of large lakes and inland seas of Eastern Europe. (71 pp.) 1978 ...... 16,– 128. Pyökäri, Mauri: Mixed sand and gravel shores in the southwestern Finnish Archipelago. (126 pp.) 1979 ...... 24,– 129. Gibbard, P.L.: Late Pleistocene stratigraphy of the area around Muhos, North Finland. (38 pp.) 1979 ...... 10,– 130.Saarnisto, Matti: Holocene emergence history and stratigraphy in the area north of the Gulf of Bothnia. (42 pp.) 1981 ...... 10,– 131.Donner, Joakim, and Jungner, Högne: Radiocarbon dating of marine shells from southeastern Australia as a means of dating relative sea-level changes. (44 pp.) 1981 ...... 10,– 132. Kivinen, E., and Pakarinen, P.: Geographical distribution of peat resources and major peatland complex types in the world. (28 pp.) 1981 ...... 7,– 133. Lehtovaara, Jyrki J.: Palaeozoic sedimentary rocks in Finland. (35 pp.) 1982 ...... 9,– 134. Aartolahti, Toive, and Eronen, Matti (editors): Studies on the Baltic shorelines and sediments indicating relative sea-level changes. (219 pp.) 1982 ...... 38,– 135. Koutaniemi, Leo, and Keränen, Reijo: Lake Oulujärvi, main holocene developmental phases and associated geomorphic events. (48 pp.) 1983 ...... 11,– 136. Donner, Joakim: The identification of Eemian interglacial and Weichselian interstadial deposits in Finland. (38 pp.) 1983 ...... 9,– 137. Löytönen, Markku, and Vähäkylä, Pekka: Interactive network analysis: an approach based on APL. (20 pp.) 1983 ...... 6,– 138. Liivrand, Elsbet: The Interglacials of Estonia. (16 pp.) 1984 ...... 5,– 139. Tolonen, Kimmo, Huttunen, Pertti, and Jungner, Högne: Regeneration of two coastal raised bogs in eastern North America. (51 pp.) 1985 ...... 24,– 140.Donner, Joakim:Weichselian indicator erratics in the Hyvinkää area, southern Finland. (20 pp.) 1986 ...... 12,– 141. Pyökäri, Mauri: Relation of grain-size distribution to sediment transport in shore zones, SW Finnish Archipelago. (34 pp.) 1986 ...... 16,– 142. Alhonen, Pentti: Cultural eutrophication of Lake Lippajärvi in southern Finland and related restoration problems. (19 pp.) 1986 ...... 10,– 143. Aartolahti, Toive: Contorted structures in Quaternary glaciofluvial deposits in southern Finland. (51 pp.) 1987 ...... 24,– 144. Tolonen, Kimmo: Natural history of raised bogs and forest vegetation in the Lammi area, southern Finland studied by stratigraphical methods. (51 pp.) 1987...... 22,– 145. Eronen, Matti (editor): Dendrochronology around the Baltic. (147 pp.) 1987 ...... 54,– 146. Malisa, Elias: Geology of the tanzanite gemstone deposits in the Lelatema area, NE Tanzania. (160pp.) 1987 ...... 64,– 147. Forsström, Lars: The northern limit of pine forest in Finland during the Weichselian interstadials. (24 pp.) 1988 ...... 12,– 148. Donner, Joakim, and Anto Raukas (editors): Problems of the Baltic sea history. (35 pp.) 1988 .... 18,– 149. Donner, Joakim: The Eemian site of Norinkylä compared with other interglacial and interstadial sites in Ostrobothnia, western Finland. (31 pp.) 1988 ...... 16,– 150.Gibbard, P., Forman, S., Salomaa, R., Alhonen, P., Jungner, H., Peglar, S., Suksi, J., and Vuorinen, Antti: Late Pleistocene stratigraphy at Harrinkangas, Kauhajoki, Western Finland. (36 pp.) 1989 .. 18,– 151.Mäkelä, Ulla: Geol ogical and geochemical environments of precambrian sulphide deposits in South-western Finland. (103 pp.) 1989 ...... 42,– 152. Palomäki, Mauri: The density of Finnish settlement system. (20 pp.) 1989 ...... 12,– 153. Delusin, Irene: The Holocene pollen stratigraphy of Lake Ladoga and the vegetational history of its surroundings. (66 pp.) 1991 ...... 30,– 154. Strömberg, Bo:A connection between the clay varve chronologies in Sweden and Finland. (32 pp.) 1990...... 16,– 155. Korhola, Atte: Paleolimnology and hydroseral development of the Kotasuo bog, Southern Finland, with special reference to the Cladocera. (40 pp.) 1990...... 20,– 156. Petäjä-Ronkainen, A., Peuraniemi, V., and Aario, R.: On podzolization in glaciofluvial material in Northern Finland. (19 pp.) 1992 ...... 10,– 157. Tikkanen, Matti, and Korhola, Atte.: Divergent successions in two adjacent rocky basins in southern Finland. (26 pp.) 1993 ...... 14,– 158. Jantunen, Tuija: A late litorina transgression in the district of Porvoo in southern Finland. (40pp.) 1995 ...... 20,– 159. Mikkonen, Kauko: Structural change of manufacturing industries in a declining economy. A Case Study of Finland, 1990–1993. (28 pp.) 1998 ...... 14,– 160.Donner, Joakim (editor): Studies of playas in the Western Desert of Egypt. (146 pp.) 1999 ...... 120,– 161. Ikonen, Liisa and Ekman Ilpo: Biostratigraphy of the Mikulino interglacial sediments in NW Russia: The Petrozavodosk site and a literature review. (88 pp.) 2001...... 20 € 162. Miettinen, Arto: Relative Sea Level Changes in the Eastern Part of the Gulf of Finland during the Last 8000 Years. (102pp.) 2002...... 25 € 163. Halla, Jaana: Origin and Paleoproterozoic reactivation of Neoarchean high-K granitoid rocks in eastern Finland. (105pp.) 2002...... 25 € 164. Donner, Joakim: Summing up: half a century of Quaternary geology. (190pp.) 2003...... 25 € 165. Mäkinen, Joni: The Development of Depositional Environments within the Interlobate Säky- länharju-Virttaankangas Glaciofluvial Complex in SW Finland. (65 pp.) 2003...... 15 €

1 € = 5.9457 FIM