DETERMINING AND APPLYING THE RESPONSE OF EPHEMERAL LAKES TO EXTREME FLOOD EVENTS USING EO Dr Robert G. Bryant and Michael P. Rainey SCEOS and Department of Geography, University of Sheffield, Winter Street, Sheffield S10 2TN, UK, Phone Number: +44 (0)114 222 7966, Fax Number: +44 (0) 114 222 7912 Email address: [email protected] Abstract It is known that closed lake-volumes fluctuate in response to changes in evaporation and precipitation rates within their catchment basin. Measurement of lake-volume changes are, therefore, not only important for hydrological and economic purposes, but can also provide a climate record. By integrating the precipitation over the catchment basins, they can provide a fuller picture of precipitation changes than the more localised in situ measurements. Playas are common features of closed, arid basins, and are the most sensitive of closed systems to regional climate changes. However, as playa systems operate under a high-evaporation/low-rainfall regime (often > 20:1), hydrological changes are not generally expressed in terms of simple volumetric changes within a lake. Instead, playas record changes in terms of variations in the timing, magnitude, frequency and residence time of specific ephemeral flooding events; generally a much more complex response, that is less well understood Using a time-series of NOAA-AVHRR data (1979-present), and associated climate data for playas in southern Tunisia, appropriate methods are presented for the accurate monitoring and detection of ephemeral flooding within specific catchments. Results have led to a greater understanding of playa response to of these often-extreme events, and further analyses of specific large floods have shown that values of P-E (or the hydrologically effective precipitation, PE) can be estimated for specific dryland catchments by inverting a simple water-balance model allied to an EO-derived curve of basin volume against area. These observations may make an important contribution to the wider monitoring and understanding of precipitation/evaporation changes within drylands using closed-lake systems. 10th Australasian Remote Sensing and Photogrammetry1 Conference 86 Introduction General Circulation Models (GCMs) predict significant 21st climate changes that are likely to increase drought frequencies and intensities in subtropical and temperate drylands (e.g. Hulme, 1992). Irrespective of the nature of forcing mechanism, empirical data suggest that hydrological and ecological system responses to these changes are likely to be rapid and un-buffered. Most perennial lakes within drylands respond in a simple and well-understood fashion to changes in the balance of precipitation and evaporation (P-E) over the lake and catchment by changing in volume; and hence predictably in both area and level (Street- Perrott and Harrison, 1985). For closed lakes, the area of the perennial lake represents equilibrium between run-off from the catchment and the water deficit over the lake (Khofield and Harrison, 2000). Although changes in lakes can be caused by local, non- climatic factors, regionally synchronous changes are reasonably assumed to be climatically determined (Mason et al., 1994). Measurements of lake-area/volume changes are, therefore, not only important for hydrological and economic purposes, but can also provide a proxy climate record (Birkett, 2000). In more arid dryland regions, however, perennial lakes are rare. In particular, playas (arid, ephemeral lakes without surface outflow) are more common. These systems have been recognized as being the most sensitive of closed systems to regional changes in rainfall patterns, and are also commonly unaffected by contemporary human interference or management (Bryant et al., 1994ab; Bryant, 1996). However, as playa systems operate under a high-evaporation/low-rainfall regime (often > 20:1), hydrological changes resulting from changes in P-E are not generally a simple volume/area response. Instead, playas demonstrate changes in the timing, magnitude, frequency and residence time of specific ephemeral flooding events; generally a much more complex response, that is less well understood (Bryant, 1999). Nevertheless, these environments, if monitored properly, may have a crucial role in providing contemporary and past (i.e. last 20-30 years) information relating to regionally synchronous changes in P-E for specific dryland regions for which climate data are either sparse or unreliable. These data may then be helpful in both validating GCM predictions for specific drylands, and helping to understand past sedimentary records from playas basins (e.g. Sweezey et al., 1999) The aim of this research was to examine the application of a time-series of EO data (NOAA-AVHRR) to the study of climatically sensitive (e.g. Richards and Vita-Finzie, 1982; Rognon, 1987) playas in southern Tunisia and eastern Algeria (principally the Chott el Djerid). Specific objectives of the study were to: (i) monitor monthly changes in lake areas and seasonal water balance, (ii) determine the detailed hydrologic response of the playa to rainfall events, and (iii) briefly examine the extent to which RS of playa flooding may be used to give important information regarding contemporary changes in regional rainfall patterns within other parts of North Africa. A summary of the principal findings of this work are presented here. More detailed results will be available shortly (Bryant and Rainey, in prep; Bryant et al., in prep) 10th Australasian Remote Sensing and Photogrammetry2 Conference 87 Study Area This closed basin forms part of the Zone of Chotts, a large lowland depression that stretches from southern Tunisia to central Algeria. The Chott el Djerid basin forms the northeastern and one of the lowest extremities of the Bas Saharan Basin. This artesian basin covers most of the Algerian and Tunisian Sahara and extends to Morocco and Libya, enclosing the whole of the Grand Erg Oriental (Roberts and Mitchell, 1987). The Chott el Djerid is an ephemeral salt playa situated in an arid-zone closed basin which has a catchment area of 10,500km2 The Chott el Djerid itself has a surface area of approximately 5360 km2, and is situated at a latitude of around 340N (Bryant et al., 1994). The climate in the Chott el Djerid basin is Pre-Saharan. A dry period occurs for 7-8 months between April and November. Average temperature recorded at Kibili in January is 9.4 0C. For the same station, the average July temperatures are 32 0C. The mean annual temperature for this region is 20.9 0C. Mean annual rainfall in Kibili is 89 mm rising to approximately 150mm in Gafsa. The region of North Africa studied is characterized by a wet-and-dry season regime; the winters being cool and generally wet (dominated by secondary depressions), and the summers dry and hot (dominated by the expanded north Atlantic anticyclone). Extreme rainfall events in this region (e.g. the winters of 1969 and 1990) have generally been associated with occasional desert depressions, which intensify as they move eastward along the Atlas Mountains. Most of the rainfall therefore occurs as inland orographic or convectional thunderstorms (Berndtsson, R., 1989). Evaporation at Kibili is at a maximum in the dry season, reaching between 2520 and 2550 mm a-1 (almost 30 times the mean annual rainfall). Alternations of wetter and drier periods throughout the Pleistocene and Holocene in this part of Tunisia have been recognised for some time (e.g. Richards and Vita-Finzi 1982; Rognon, 1987; Fontes and Gasse, 1989; Causse et al., 1989; Swezey et al., 1999). Figure 1. AVHRR (1,2,1) image from January 1990 showing the location of the main closed ephemeral basins within the "Zone of Chotts". This image was collected just after a severe rainfall event (1:20) which affect the whole region, and resulted in the flooding of the ephemeral lakes. 10th Australasian Remote Sensing and Photogrammetry3 Conference 88 Data Procurement The Institut National de la Meteorologie, Tunis provided detailed climate data for Tunisia, covering the period 1979 to 1998. Annual climate almanacs from the same institution were also used to provide decadal (10-day) precipitation and evaporation figures for selected years. Latterly, climate data for Algeria were acquired via the NOAA National Climate Data Center. AVHRR data (Appendix) covering the same period were obtained from the NERC satellite receiving station at Dundee University. Suitable cloud-free HRPT images of the ephemeral lakes were identified over the Internet and subsequently ordered from the archive in level 1B - NOAA format, with associated calibration files. In all, a total of 154 images representing monthly intervals spanning the period 15th October 1983 to 23rd March 1996 were obtained. Each image was geo-rectified (1st order), with sun-angle and radiometric corrections also being applied. Following this, the data were re-projected to a latitude/longitude co-ordinate system and spatially sub-set to the regions needed for further analyses. The image sub- sets then underwent an image-to-image geo-correction, which defined a consistent pixel- size for the data set (1.1km; Wu and Liu, 1997). Landsat MSS imagery from 1990 of Southern Tunisia was supplied to this project by NERC. Using this higher spatial resolution data, the objective was to test the accuracy of the AVHRR lake area detection technique (see Section 3.2.3). 180 600 El Ninio La Ninia 160 500 140 120 400 100 300 80 60 200 Evaporation (mm) Precipitation (mm) 40 100 20 0 0 O-83 A-84 A-84 A-85 A-85 J-86 M-86 S-86 F-87 J-87 O-87 F-88 J-88 M-89 J-89 D-89 A-90 S-90 J-91 A-91 J-92 Figure 2. Climate data for southern Tunisia. Evaporation (yellow), Precipitation (blue), ENSO events are also labeled. Results Determining Flooding Styles within the Zone of Chotts using EO. Flooding within a closed ephemeral system is principally a function of either direct precipitation (often strongly seasonal, with high inter-annual variation), or groundwater inflow (which can be a direct function of seasonal evaporation rates) (7).