INFLUENCE OF SNOW COVER RECESSION ON AN ALPINE ECOLOGICAL SYSTEM *) Markus Keller and Klaus Seidel Institut fuer Kommunikationstechnik der ETHZ CH 8092 Zuerich, Switzerland ABSTRACT In a worldwide UNESCO-program "Man and Biosphere (MAB)" the mutual effects of economical activities, land-use and natural resources are under investigation. We are concentrating on high Alpine regions. In the test site MAB-Davos virtually all available data from Climate, Forestry, Game Biology, Landscape History, Soil Science, Vegetation and Human Activities have been assembled into a Geographical Information System (GIS). For a period in 1981 a set of snow cover distribution maps has been derived using Landsat-MSS imagery. The quantitative comparison of the snow cover recession with certain GIS-layers stimulates a better understanding of natural processes and interactions. 1. INTRODUCTION In high Alpine regions snow is a dominant factor for man, fauna and vegetation. Snow cover distribution and evolution through the year determine the environmental conditions. Detailed knowledge about the development of snow cover distribution throughout the seasons helps to understand certain processes within an ecological system. Switzerland contributes to the worldwide UNESCO-program "Man and Biosphere (MAB)". In a National Research Program interest is concentrated on high Alpine regions. The mutual effects of economical activities, land-use and natural resources are under investigation. The aim is better understanding and control of those processes and interactions which are crucial ecologically and recreationally for long-term life conditions. In the testsite MAB-Davos, in which we are involved, virtually all available information has been collected in a multidisciplinary venture. Data from Climatology, Forestry, Game Biology, Landscape History, Soil Science, Vegetation and Human Activities have been assembled into a Geographical Information System (GIS). Using digital image processing techniques a large variety of coverages (features or characteristics) have been merged into a multivariate data set. *) Presented at the Eighteenth International Symposium on Remote Sensing of Environment, Paris, France, October 1-5, 1984. 1.
This study deals with the derivation of a snow cover recession map and its combinations with certain layers of the GIS. Snow cover distribution can be extracted from Landsat imagery for quantitative analyses. 2. CONTENTS OF THE GEOGRAPHICAL INFORMATION SYSTEM In the test site MAB-Davos a geographic information system (GIS) has been created. GIS's contain geo-coded information, which can be used together with programs for manipulation and display purposes to create graphic and other representations of spatial phenomena. The test site MAB-Davos is a high Alpine region located in Graubuenden/ Switzerland (N 46 50', E 9 50'). It covers an area of about 100 km 1 in an elevation zone between 1600-3200 m a.s. The population is 12000 inhabitants. The area enjoys wide popularity as a health resort and recreational area in summer and winter. It is touristically accessible for up to 23000 visitors at peak times. In the GIS a variety of relevant features from very different disciplines has been collected (Figure 1 ). Computer techniques have been used to measure quantitatively the mutual interactions between the different layers. HUMAN RCTIVITIES BIOLOGY GROUND PROPERTIES HISTORY or LRNO use LRNO Oi.WE"RSJ.!IP 3. SNOW COVER RECESSION MAP Snow cover protects wildlife and vegeta t i on during winter time. The amount of snow represents a considerable resource for hydro-electric power plante which is widely used during spring and summer time. Snow cover distribution and evolution throughout the year determine the condition of the environment. Monitoring the snow cover allows us to pre CLIMRTE DIGITRL TERRRIN MODEL Figure 1 : Information layers forming the GIS MAB-Davos dict this condition in advance. Utilizing the earth resources and technology satellites, remote sensing contributes to this problem area. Snow cover distribution can be extracted from Landsat imagery for a quantitative analysis [1 ]. In the test site MAB-Davos we selected all available satellite images for the period from April to October 1981. We processed them digitally and analyzed them in order to get "snow cover distribution maps" at the different recording times [2,3]. In a preprocessing step the digital data have been geometrically corrected corresponding to the topographic map projection used in Switzerland. The parameters for the transformation had to be determined by ground control points. For the Landsat-MSS scenes used affine transformations were found to be of sufficient precision for the mapping process. Simultaneously to the geometric correction process, a resampling of the data to a raster size of 50 x 50 m 1 has been introduced. This resolution corresponds to the sample size used for the other GIS layers. 2.
Figure 2: Snow cover distribution derived from Landsat MSS data 3.
The multispectral data set of each scene has subsequently been classified with respect to snow cover. We used supervised classification algorithms [4,5]. The quality of the training samples has extensively been controlled by ground truth from both aerial and terrestrial photographs. The analysis process of each of the different satellite images leads to thematic maps (Figure 2) showing the spatial distributions of the following categories: - snow covered areas - transition zones (pixels with about 50% snow cover) - snow free areas (aper regions) Since these maps are geometrically in register, they could be superimposed to form a multitemporal data set. Based on these multitemporal data, we derived a "snow cover recession map" representing the time and spatial variations of the snow cover within the test site. For each picture element the total number of days it appeared permanently snow covered was determined (Figure 3). The count started in the beginning of the year when all pixels were snow covered. Figure 3: Snow cover recession map Since the recording dates of the satellite scenes used are taken into account, we name this map the "absolute snow cover recession map of 1981". As one can easily see, the test site can be subdivided into zones of equal snow cover duration. 4. COMPARISON OF SNOW COVER WITH CERTAIN GIS-LAYERS As mentioned before snow cover is an important factor in high Alpine regions. The recession map is now used for a quantitative comparison with selected layers of the GIS. Snow cover recession is mainly affected by relief, irradiation and temperature. This explains the visible differences of the snow cover distribution on north-exposed compared to south-exposed slopes. 4.1 Soil Science The soil type coverage map relates to the recession map. This relationship can be seen by comparing a map representing the dominant soil types with the snow cover recession map. We argue that the soil forming factors are heavily influenced by the snow recession behaviour. We realize, for example, that the shape and areal extent of the following aper*) regions and soil types coincide (Table I): 4.
Table I: Aper regions and dominant soil types number of days l to become aper l 189-242 153-188 corresponding dominant soil type Regosols Podzols *) aper = snow-free The established theory considers the climate as a soil forming factor [6]. With the above example we suggest an approach for a more detailed quantitative investigation with respect to soil forming factors. 4.2 Vegetation Vegetation is heavily influenced by snow coverage. The areal extent of each vegetation type in the different aper zones has been computed. The different vegetation types show different dynamic behaviour within the snow recession period. The vegetation types build groups with similar distributions over the aper regions. Better time resolution of the recession map could lead to a better separation of the vegetation types. 4.3 Game Biology Wildlife is dependent on the snow cover recession. Hibernating species are influenced the most. As an example we extracted from the GIS the distribution map of marmots (Marmota marmota L.) within the test site and compared it with the snow recession map. Finally we com~uted the index of occurence Ni for the marmots in the different aper regions \Table II). Table II: number of days to become aper index Ni Index of occurences Ni of marmots (Marmota marmota L.) <97 0.2 97-152 0.85 153-188 1.46 189-242 1.04 >242 0.24 Ni = 1 says: marmots occur as often as can be expected based on the areal extent of the region with respect to the test site. Table II confirms quantitatively the experience that marmots prefer to live in regions with an hibernation period of about 220-250 days. 4.4 DTM Based Snow Melt Runoff Calculations Existing water runoff models [7] use as input parameters the areal extent of snow cover in different elevation zones. For the calculation of the absolute or relative areas of snow coverages, we combined the DTM with the individual snow distribution maps (Figure 2). Table III gives as an example these values for a reduced test site area of about 43 km 2. Table III: Absolute and relative snow coverages in different elevation zones elevation 7-APR-81 l-jun-81 7-JUL-81 30-AUG-81 zone traneition snow transition snow tranaition snow transition snow [m a. J zone covered zone covered zone covered rone covered " " 1600-2100 110.8 ha 761. 5 ha 82.0 ha 14.3 ha 5.0 ha 1. 5 ha 13.2 84.3 ' 9.1 1.6' 0.6 ' 0.2 ' 0 0 2100-2600 42.8 ha 2420.ll ha 511. 8 ha ' 1656.0 ha 331. 8 ha 681. 5 ha 18.3 ha 12.ll ha 1.2 98.2 20.8 ' 67.2 13.5 27.7' e.5' 2600-3200 1.8 ha ' 9457.5 ha ' 88. 3 ha 858. 5 ha ' 78.0 ha ' 663.3 ha 67.3 " 7 ha ' 103.8 ha 0.2 ' 99.8 ' 9.3 ' 90.6 ' 8.2 67.2 ' 8.1 ' 11.0 ' 5.
5. CONCLUSIONS Snow Cover Distribution as derived from Landsat-MSS data is significant to the GIS in that it aids in: - recognition of relationships between biological and pedological properties and snow cover recession behaviour - derivation of life conditions - prediction of snow melt runoff The proposed method can be useful in fundamental research with respect to the environment and also in monitoring of renewable resources on a regular basis. 6. ACKNOWLEDGMENTS We wish to thank the Swiss National Research Foundation who funded this project. 7. REFERENCES [1] RANGO, A. and MARTINEC, J. Application of a Snowmelt-Runoff Model Using Landsat Data Nordic Hydrology, 10, 225-238 (1979) [2] LICHTENEGGER, J., SEIDEL, K., HAEFNER, H. and KELLER, M. Snow Surface Measurements from Digital Landsat-MSS Data, Nordic Hydrology, 12, 275-288 (1981) [3] SEIDEL, K., ADE, F. and LICHTENEGGER, J, Augmenting Landsat MSS Data with Topographie Information for Enhanced Registration and Classification, IEEE Trans. Geosc. & RS, GE-21, 252-258 (1983) [4] [5] [6] [7] DUDA, R.O. and HART, P.E. Pattern Classification and Scene Analysis, Wiley Interscience, New York, 1973 SCHUERMANN, J. Polynomklassifikatoren fuer die Zeichenerkennung, Oldenburg Verlag, Muenchen, 1977 SCHEFFER, F., und SCHACHTSCHABEL, P. Lehrbuch der Bodenkunde Ferdinand Enke Verlag, Stuttgart, 11. Auflage (1983) MARTINEC, J., RANGO, A. and MAJOR E. The Snowmelt-Runoff Model (SRM) User's Manual NASA Reference Publication 1100 (1983) 6.