The U.S. Geological Survey recently made all historical and current Landsat Thematic Mapper (TM) imagery freely available for download. The ready and free availability of this imagery thus opens the door to a variety of possible projects not previously undertaken by rural managers due to the former expense of purchase. For example, little is known about fluctuations of the open water portions of wetlands on Washington state’s semiarid Waterville Plateau on the northwest margin of the larger Columbia Plateau (Figure 1). We used Landsat Thematic Mapper (TM) satellite imagery of 52 wetland ponds and their associated watersheds to identify seasonal, annual, and interannual pond areas on the eastern Waterville Plateau over a 22 year period from 1986 through 2007. Specifically, we: 1) selected imagery; 2) selected pond study areas; 3) acquired imagery; 4) processed imagery; 5) classified images and mapped pond areas; 6) mapped pond watersheds; 7) acquired climate data; 8) determined geomorphic surface types; 9) determined land use types; and 10) correlated over time pond changes to climate, geomorphic surface, and land use patterns. We chose to use Landsat Thematic Mapper (TM) data because historical and current scenes became freely available beginning in late 2008, and because of its high spectral resolution, repeat imagery (16 day return interval) characteristics. However, the relatively coarse spatial resolution of TM imagery (30 m pixel) as compared to the high spatial resolution of aerial photography (1-3 m) prevented us from analyzed stream and very small pond fluctuations. Also, the large size of the datasets required us to develop automated techniques for the retrieval and analysis of the images.
Maximum open water areas of individual wetland ponds range up to nearly 4.1 hectares. Minimums range to 0 ha–i.e., ponds disappear, often leaving a salt encrusted, high albedo surface. Wetland pond areas fluctuate markedly over a particular year, generally reached their maxima in March and April at the end of the wettest months. Minima were more dispersed throughout the year. Slightly more wetland pond minima occurred in April (12%), May (14%), June (16%), July (15%), and August (16%). The slight increase in wetland pond minima occurrences in the late spring and summer months generally coincides with the typical warmer and drier months of the year. Sixteen of the wetland pond watersheds were located on end moraine, 15 on ground moraine, and 21 on scabland surfaces. When stratified by geomorphic surface type, cumulative annual end moraine and ground moraine maximum wetland pond area patterns appear quite similar while scabland pond fluctuations differ markedly. Spearman rank correlation analysis of geomorphically stratified cumulative maximum wetland pond areas supports the relationship between end and ground moraine ponds. Mean annual water year precipitation and snowfall patterns show some similarities to pond fluctuations, especially those on scabland surfaces. This suggests that the thinly till mantled scablands may respond more closely to precipitation than do the till covered end and ground moraine surfaces. However, statistical analysis does not support this hypothesis. Further, land use patterns do not visually or statistically follow pond area patterns on any of the geomorphic surface types. These results suggest that other variables may play a role in Waterville Plateau pond area fluctuations including groundwater input, basin size, and substrate. This study is significant for its scientific results in a Pacific Northwest analog setting to the Northern Great Plains Prairie Potholes region, and for its development of new geospatial techniques centered on now readily and freely available satellite imagery.