DANIELK. ZIGICH KENNETHE. KOLM Department of Geology and Geological Engineering South Dakota School of Mines and Technology Rapid City, SD 57701 Evaluating the Effectiveness of Landsat Data as a Tool for Locating 1 Buried Pre-Glacial Valleys in Eastern South Dakota Winter, spring, and early summer 1:500,000-scale Landsat mosaics provide the most useful information for locating pre-glacial buried valleys. INTRODUCTION The regional study area, consisting of the URIED PRE-GLACIAL alluvial valleys, which Coteau du Missouri and James River Lowland B yield significant quantities of groundwater in provinces (lat. 43' -46"N, long. 97" -lOOOW), is eastern South Dakota, have traditionally been dis- located in eastern South Dakota, east of the Mis- covered by expensive, time-consuming drilling souri River (Figure 1).Local study sites, including methods. The objective of this study was to de- Jerauld, Aurora, and Walworth Counties, were velop a rapid and cost-effective method using chosen due to the availability of drill data sub- Landsat data to delineate buried pre-glacial val- sequently acquired there (Figure 1). ABSTRACT:A rapid, low-cost procedure for delineating pre-glacial buried valley aquifers in eastern South Dakota from Landsat imagery was developed and evaluated. Curvilinearllinear, anomalous tonal pattern, and drainage overlays were prepared using winter and spring, Landsat band 5, band 7, and false-color infrared mosaics, and were analyzed to define subtle characteristics related to known buried valleys. Drainage characteristics, curvilinearllinear patterns, and anomalous sinuous tonal patterns, which do not correlate with the trends or characteristics of glacial features, were mapped as indicators of buried valleys. Subsequent drilling confimned the effectiveness of this procedure. Due to till thicknesses and glacial dynamics, delineation of buried valley aquifers is most accurate near the Missouri River. The best guides to buried valleys are present day streams which reoccupy pre-glacial valleys, or anoma- lous chains of lakes, ponds, and depressions, and associated soils and vegetation which overlie unexhumed buried valleys. leys and associated aquifers in eastern South PREVIOUSWORK Dakota. The primary goals are to determine which Flint (1955) summarized the Pleistocene glacial imagery best delineates buried valleys, to develop geology and geomorphology that characterizes the a method of manual spatial and spectral interpre- surface ofeastern South Dakota (Figure 2). He also tation for predicting the locations of buried val- discusses the subsurface geology as it pertains to leys, and to evaluate these techniques by com- the Coteau du Missouri and James River Lowland paring valley locations predicted by Landsat provinces (Figure 3). Ultimately, Flint (1955) analysis with those subsequently located by the comments on the geologic evolution of the Mis- South Dakota State Survey's drilling program. souri River drainage (Figure 4). Tipton (1975) PHOTOGRAMMETRICENGINEERING AND REMOTE SENSING, 0099-111218214812-1891$02.25/0 Vol. 48, No. 12, December 1982, pp. 1891-1901. @ 1982 American Society of Photogrammetry PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING, 1982 buried aquifers have been located in South Da- kota (Hyland and Andrawis, 1977) using low alti- tude thermal imagery and in Indiana using Land- sat imagery (Peterson et al., 1975). Lucas and Taranik (1977) demonstrated that springtime Landsat scenes of the midwestern United States were superior for the regional interpretation of glacial features. The general procedure employed in this study was to determine the surface characteristics ex- hibited by known buried valleys in the field, to delineate these characteristics on Landsat imag- ery, and to extrapolate these results to other areas where buried valleys have not been discovered. Generally, the field characteristics of buried val- leys are subtle at best (e.g., soil changes) or non- existent (Figure 5).Only large scale changes in the character of the land, such as topographic breaks FIG.1. Location of the study area in eastern South in the Coteau or drainage pattern anomalies, can Dakota. Note that Walworth County is located entirely be used to delineate preglacial valleys. Therefore, within the Coteau du Missouri of the Great Plains Prov- Landsat mosaics, which provide a regional and ince, whereas Jerauld and Aurora Counties are partially seasonal view of the study area, are the most suit- within the James Basin of the Central Lowland Province. able tools for this study. Preliminary Landsat investigations were con- summarizes the more recent geologic and geohy- ducted at the scales of 1:1,000,000 and 1:500,000. drologic work pertaining to the region that has County maps showing areas of known pre-glacial been completed by the South Dakota State and bedrock surfaces (delineating pre-glacial valleys) United States Geological Surveys. were reduced to scales of 1:1,000,000 and Previous investigations show that Landsat may 1:500,000 and combined to form one regional base be used as a tool for locating shallow groundwater map for each scale. The locations of known buried aquifers in South Dakota (Moore and Meyers, valleys were drafted directly from the base maps 1972a, 197213; Myers and Moore, 1972; Rahn and to acetate overlays. These overlays were then Moore, 1977; Heinemann et al., 1972). More deeply placed upon each of the Landsat mosaics in order FIG.2. Map of eastern South Dakota showing generalized Pleis- tocene glacial geology (after South Dakota State Geological Survey Education Series Map #2, 1971). EVALUATING LANDSAT DATA FOR LOCATING PRE-GLACIAL VALLEYS l;.'.-l I0 0 50 IIL~s VERTICLL IXLGGCRLIION LPPROXI~LIELI1311MES FIG.3. East-west geologic cross-section of eastern South Dakota (after Darton, 1909; Flint, 1955) showing almost imperceptible dips of bedrock surfaces in the region. to correlate areas of known buried valleys with Winter scenes were chosen for interpreting drain- spatial and spectral patterns and characteristics age patterns and density, and topographic features observed on Landsat imagery. because the low solar angle and light snow cover Five mosaics of Landsat scenes, each with 10 enhance the landscape. Shadows that are created percent or less cloud cover, provide a regional by low sun angles and sun azimuths outline and, view needed to study subtle surficial characteris- therefore, enhance topography. A light, total snow tics of buried valleys that extend for several miles cover eliminates most of the tonal and color varia- (Table 1). A winter mosaic was constructed to tions due to cover type. Therefore, the surface ap- study drainage and topographic characteristics as- pears to have a more uniform brightness, which is sociated with buried valleys (Table 1).Spring and less distracting in drainage analyses. summer mosaics were used to study the spectral Spring (April, May, early June) images have characteristics of buried valleys associated with been determined as optimal for glacial studies in vegetation, soil type and soil moisture. the midwest (Lucas and Taranik, 1977). During Landsat false-color infrared composites, and this season, such factors as minimal crop cover, Band 5 and Band 7 scenes obtained from different limited land-use or farm practices, maximum open seasons and years (wet versus dry) were compared water (lakes and swamps, for example) and tonal using the methods of Lucas and Taranik (1977). contrast due to soil types and soil moisture, and FIG.4. Diagram of the State of South Dakota depicting the generalized evolution of the eastern South Dakota drainages. (A) Pre-glacial drain- age: Note the east-west river trends; (B) Late Wisconsin glacial advance (arrow)and formation of Missouri River (MR): Missouri River and dotted line indicate approximate ice border; (C) Present day drainage: Note the lobate or curvilinear river trends east of the Missouri River due to strong glacial influence; (D) East-west trending, pre-glacial buried valleys (dotted) underlying glacial drift. PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING, 1982 large vegetative vigor differences greatly enhance the utility of these images for analysis (Figures 6a, b, and c). Surficial characteristics of buried valleys, in- cluding chains of lakes, closed depressions, and anomalous drainage patterns, are commonly best exhibited on springtime images (such as illus- trated in ~igurgs6ar b, and c). In addition, spring a snow meltprovides low lying, poorly drained areas characteristic of some buried valleys with abundant water. As a result, there are distinctive reflectance differences between these areas and well-drained areas associated with the sloping FIG. 5. Oblique aerial view of James River Lowland lands of valley sides and moraines. These wet near Mitchell, South Dakota. This area is characterized conditions prevent the farmer from planting in low by subtle depressions containing water, Tonka soils, lying areas during early spring, resulting in easily and grassland vegetation (after Johnson et al., 1974). recognizable dark patterns on imagery (Figure 6a). Early summer Landsat scenes may provide ous pattern on 1978 springtime imagery (Figure better buried valley patterns than springtime im- 6a). The pattern cannot be seen on a 1974 spring ages at some localities. Both low lying depressions image (Figure 6b). The dark pattern is caused by a and adjacent unplanted uplands may appear uni- chain of small ponds and depressions. In 1974, formly dark gray on springtime imagery. On early South Dakota was experiencing