Kansas Academy of Science

Applications of Landsat imagery in the Great Plains

James S. Aber1, Everett E. Spellman2, Michael P. Webster1, and Laura L. Rand3

  1. Earth Science Department, Emporia State University, Emporia, KS 66801 (aberjame@esumail.emporia.edu).
  2. Kansas Dept. Health and Environment, Forbes Field, Topeka, KS 66620.
  3. Dept. of Geosciences, Fort Hays State University, Hays, KS 67601.

This article is published in the Transactions of the Kansas
Academy of Science, vol. 100, no. 1/2, p. 47-60 (1997).

Table of Contents
Introduction Conclusions
Devils Lake, North Dakota Acknowledgements
Flint Hills, Kansas References

ABSTRACT

Landsat MSS digital data and imagery are utilized to demonstrate applications in geology and hydrology for two regions of the Great Plains--Devils Lake, North Dakota and the Flint Hills, Kansas. Devils Lake occupies several, internally connected glacial basins that have no outlet at present. Consequently lake surface elevation, water volume, salinity, and biomass fluctuate significantly with climatic changes. During the period 1973-1988, the lake rose 3 m; Landsat images document an increase in surface area of more than 50%. Glacial landforms include large ice-shoved hills, ice-scooped depressions, spillway channels, and tunnel valleys. These features are interpreted as the results of glacier thrusting and melt-water erosion in connection with two converging ice lobes during a late phase of glacier advance.

In the Flint Hills of east-central Kansas, stream valleys follow distinct lineaments oriented NW-SE, NNE-SSW, and NE-SW. These lineaments correspond to crustal fractures (joints and buried faults). Based on Landsat imagery, we have identified another significant lineament trend at NNW-SSE (about 350°-170°). We propose to name this trend the Verdigris lineament, and we suggest it represents an important fracture system in east-central Kansas. The Verdigris lineament relates to bedrock joint trends in the Flint Hills region.

INTRODUCTION

Remotely sensed satellite observations from space have fundamentally changed the way in which scientists study the Earth. The Landsat satellite system was initiated in 1966 as Earth Resources Observation Satellites (EROS). The first satellite was launched in 1972, and Landsat observations of the Earth's surface have been continuous ever since. Landsats 1 through 5 carried a multispectral scanner (MSS); Landsats 4 and 5 also carried a thematic mapper (TM). Both are whisk-broom (cross-track) scanners that detect visible and infrared solar energy reflected from the Earth's surface--see Landsat program.

Archive MSS datasets are now available at low cost from the EROS Data Center of the U.S. Geological Survey. These datasets span more than 20 years of observation and represent a tremendous resource for all manner of earth science applications. Powerful computers and sophisticated GIS software are available also at modest cost for image processing and analysis of satellite data. Thus, it is now possible to conduct state-of-the-art research with Landsat digital data on a relatively small budget.

Landsat MSS digital data and imagery are utilized here to demonstrate applications in geology and hydrology for two regions of the Great Plains--Devils Lake vicinity in northeastern North Dakota and the Flint Hills of east-central Kansas. Standard techniques utilized for image processing included haze correction, stretching, false-color composites, band ratios, classification, and time-series analysis. IDRISI software was employed for image processing and analysis.

Return to Table of Contents

DEVILS LAKE, NORTH DAKOTA
Background

Devils Lake, a large natural lake located in northeastern North Dakota, has attracted much scientific interest since the early ninteenth century--see
location map. The landscape of northeastern North Dakota was heavily modified by glacier thrusting (Bluemle 1966), of which Devils Lake is a prime example--see lake overview. The lake occupies several connected depressions that were formed by glacier pushing of Cretaceous bedrock (shale) and glacial sediment. Sullys Hill, Crow Hill, and other hills immediately to the south are built of material thrust out of the Devils Lake depressions during late Wisconsin glacier advances. The ice-shoved hills are composed of a jumbled, brecciated mixture of deformed shale and glacial sediment. Relief from the bottom of central Devils Lake to the top of Sullys Hill exceeds of 200 m. The ice-scooped depressions and ice-shoved hills are among the largest and best-developed glacial landforms of this type--hill-hole pairs--in the United States (Bluemle and Clayton 1984).

A major aquifer, known as the Spiritwood aquifer, crosses Devils Lake vicinity from east-southeast to west-northwest. This aquifer is developed partly in bedrock and partly in loose sediment filling the preglacial Cannonball valley. The major depressions of Devils Lake are located above this aquifer. High ground-water pressure in the Spiritwood aquifer likely facilitated major glaciotectonic thrusting (Bluemle and Clayton 1984; Hobbs and Bluemle 1987).

Devils Lake is the terminal point of a large enclosed drainage basin that has no surface outlet at present. As a result of its terminal position, the water body of Devils Lake has undergone substantial changes in its elevation, surface area, salinity, and biomass. During the past 4000 years, Devils Lake has fluctuated several times between completely drying up and overflowing into either the Sheyenne River and/or Stump Lake basin to the east (Bluemle 1996). These changes resulted entirely from long-term climatic events.

According to established records, the maximum historical lake level was attained in the early 1800s, when it exceeded 1440 feet (439 m) elevation (Bluemle 1995), toward the end of the Little Ice Age. Between 1826 and 1830, the lake probably overflowed into Stump Lake, and it possibly may have overflowed to the Sheyenne River during part of this time. After the 1860s, lake level declined until 1940, when it fell to its lowest historical elevation of 1400 feet (427 m), at which time the lake was only 1 m deep--see Devils Lake record. The lake subsequently rose and exceeded 1425 feet (434½ m) during most of the 1980s (Fig. 3). Following a slight decline in the late 1980s and early 1990s, the lake is again rising; the water surface surpassed 1437 feet (438 m) elevation during the summer of 1996. Rising water threatens to flood the city of Devils Lake, Camp Grafton, roads, fields, sewage treatment plants, and other human facilities--see Ziebach Pass. The city has received funding to construct a dike that would provide protection up to 1440 feet, but if the current trend continues the lake may exceed this elevation during 1997--see Devils Lake hydrograph.

Both the prehistorical and historical fluctuations of Devils Lake level correlate with general climatic variations (Callendar 1968; Wiche 1986). Another factor that influences Devils Lake is storage of surface water runoff in the upstream chain of lakes. This chain of lakes also acts as a trap for sediment, so that little sediment is washed into Devils Lake. In addition, Devils Lake may loose or gain water from the underlying Spiritwood aquifer. Changes in water level of Devils Lake are, thus, caused by climatic variations, as modified by the surface and subsurface characteristics of the drainage basin. Recent modifications of drainage, agricultural practices, and other human factors have caused only slight effects on water levels in Devils Lake (Bluemle 1995); however, runoff from land north of the lake has washed in substantial salts and agricultural wastes.

Interpretation of glacial geomorphology

An interpretation of glacial geomorphology can be given based primarily on visual examination of Landsat MSS images, along with color-infrared air photographs and conventional topographic maps, in combination with ground observations. Sullys Hill forms a focal point for the glacial landforms in the subscene--see winter Landsat image. Two series of ice-shoved ridges and associated source depressions (lake basins) converge at Sullys Hill. One series extends southwest to Crow Hill and on to the west. The other trends southeast, past Devils Heart Butte, to near Horseshoe Lake. These two series are associated respectively with the North Viking and Cooperstown ice margins. Ice-shoved hills are discontinuous along these two series, and the sizes of hill-hole pairs diminish away from Sullys Hill--see location map.

These two trends of ice-shoved hills may be interpreted as features formed by thrusting at the margins of two converging ice lobes (Aber et al. 1993). One lobe advanced locally from the north or northwest in the West Bay/Crow Hill vicinity, while the other moved locally from the northeast in the East Bay/Devils Heart Butte region, with Main Bay/Sullys Hill at the junction of the two. The two ice lobes probably advanced together, or at least at about the same time, to the North Viking and Cooperstown ice margins. However, some parts of the Crow-Sullys Hill complex may have formed in association with the Heimdal or Pekin ice margins.

A well-defined melt-water drainage system can be seen south of the ice-shoved hills. Four drainage routes are: Big Coulee, Black Slough, the channel that begins at Twin Lakes, and the Long Lake channel. These drainage channels, particularly Big Coulee, are quite distinctive on Landsat images. All four channels carried melt water away from the glacier margin at the ice-pushed hills into the Sheyenne valley a short distance to the south. To the north, Sixmile Bay may have been eroded by a subglacial melt-water stream flowing toward the Devils Lake basin. Creel Bay may have a similar origin. These bays/valleys are here interpreted as tunnel valleys formed by melt-water erosion under the ice.

Good preservation of this assemblage of glacial landforms suggests that subsequent glaciation either did not advance over this vicinity or did little to modify the hill-hole pairs and associated landforms. The association of large ice-shoved hills, melt-water spillways, deposits of hydrodynamic blowouts--see Devils Heart Butte, and tunnel valleys indicates that a substantial volume of water was released during creation of these landforms. This water was derived from both the glacier and the Spiritwood aquifer (Bluemle 1993).

Interpretation of lake hydrology

The results of surface-area calculations for Devils Lake demonstrate that significant changes have taken place during the past two decades (Table 1). In 1973, Devils Lake had a minimum extent of 120-125 km²--see 1973 image. From 1973 to 1983, the lake rose in elevation and spread in area. Lake level/area then stabilized with minor fluctuations from 1983 to 1988; surface area during this phase varied in the range of 180 to 200 km²--see 1988 image. Between 1973 and 1988, the lake rose 3 m and increased its surface area by >50%--see lake change. Following 1988, lake level began to fall in response to drought conditions. By 1991 the lake had declined to 434 m (1423 feet) elevation, and surface area was only 166 km². The changes in lake surface area, as determined from Landsat images, follow the same general trends as lake elevation fluctuations. It is apparent from these data that small variations in lake elevation result in large changes of surface area. Additional data of this kind could be used to derive an empirical relationship between lake surface area and elevation.

Table 1. Devils Lake Surface Area (Spellman 1993).
Date Km² Miles²
05/14/73 122.6 47.3
07/07/73 120.9 46.7
07/16/83 180.8 69.8
05/31/84 197.2 76.1
07/18/84 181.4 70.0
09/28/84 185.3 71.5
10/01/85 182.4 70.4
05/29/86 194.4 75.1
06/19/88 184.9 71.4
09/23/88 186.5 72.0
10/25/88 188.4 72.7
08/28/90 171.8 66.3
08/31/91 166.4 64.3

Suspended sediment in surface water of the lake is indicated by higher reflectance of visible light--see autumn image. In general, Main Bay and East Bay are deeper and cleaner compared to West Bay, which often displays muddy water (Fig. 7). The images in which West Bay appears to have suspended sediment correlate with episodes of relatively high winds. This bay is quite shallow (1-2 m deep), so wind-driven waves are able to stir up lake-bottom sediment. Matson and Berg (1981) noted a similar situation in their study of the Great Salt Lake, Utah. This relationship between wind, suspended sediment, and lake depth could be used for region-wide classification of deep and shallow lakes. This knowledge of lake depth could be applied, in turn, to improve regional estimates of lake storage capacity.

Algal blooms have an appearance similar to land vegetation on Landsat MSS images. The algae is dark in visible bands and bright in short-infrared bands. The blooms often take place in summer or autumn, when lake water temperature is relatively warm--see algal mat. Large algal blooms are depicted on certain images in East/Black Tiger Bays and central Devils Lake/Creel Bay. Extensive algal mats are apparently more common in the deep, clear waters of central Devils Lake and East Bay, as opposed to the shallow muddy water of West Bay. This difference may reflect the ability of sunlight to penetrate deeper in clean water and support more photosynthesis. However, algal blooms can take place in any portion of Devils Lake.

Return to Table of Contents

FLINT HILLS, KANSAS

The Flint Hills are a bedrock terrain that stands above lower plains to the east and west--see
physiographic map. Prior to the Miocene, major river systems flowed from west to east across what are now the Flint Hills and the landscape possessed less topographic relief than today (Aber et al. 1995). Since the Miocene, stream downcutting has led to massive erosion across all eastern Kansas. As the regions to the east and west were lowered, drainage diversions took place, and the Flint Hills emerged as a resistant bedrock massif--see Florence Limestone.

It is clear that bedrock structural features have exerted a strong influence onstream erosion patterns in east-central Kansas--see river map. Major lineaments are oriented toward the northwest (Fall/Neosho) and north-northeast (Walnut). These lineaments are defined by straight, continuous stream valleys, discontinuous valley segments, or en echelon stream valleys. The major lineaments correspond to well-known surficial joint sets, which in turn give the appearance of being related to deep-seated faults (Berendsen and Blair 1991).

Valley patterns are especially conspicuous on Landsat images of the Flint Hills region, because of high topographic relief and strong contrast in landuse between valleys (row crops) and uplands (pasture). Landsat images reveal linear stream valleys that meet at angular junctions--see autumn image. Valleys in upper portions of the South Fork Cottonwood River, North Branch Verdigris River, and headwaters of the Fall River are parallel, trending about 350°. In fact, the Fall River headwater valleys are in direct alignment, across the drainage divide, with the North Branch Verdigris and South Fork Cottonwood River valleys. We propose to call this NNW-SSE trend (350°-170°) the Verdigris lineament, and we suggest that it represents an important fracture pattern in bedrock of east-central Kansas-see joint set 4.

The upper North Branch Verdigris River takes an abrupt turn toward the east, changing its direction by 100°. This feature is one of the most prominent drainage anomalies in Kansas. During the Neogene, the ancestral Verdigris River presumably flowed across this region from west to east (Aber et al. 1995). Subsequent stream capture by the Cottonwood drainage diverted the upper ancestral Verdigris, leaving the unusual barbed drainage pattern in the North Branch Verdigris River.

Return to Table of Contents

CONCLUSIONS

Archive Landsat MSS digital datasets and processed images have proven valuable for a variety of uses in the Great Plains. Hydrologic applications include documenting changes in lake surface area, turbidity, and algal growth. Landsat MSS images provide synoptic views of glacial geomorphology, drainage patterns, and bedrock structures that facilitate interpretation of significant geological features. We conclude that archive Landsat MSS datasets are a resource with great potential for research applications at modest cost.

ACKNOWLEDGEMENTS

Primary support for this research was given by a grant from NASA (NA-G8215). Additional support was provided by the North Dakota Geological Survey, North Dakota State Department of Health and Consolidated Laboratories, Kansas Geological Survey, Kansas Space Grant Consortium, and Emporia State University. J.P. Bluemle and P. Berendsen reviewed the article and offered many helpful suggestions. All Landsat MSS digital datasets were obtained from the EROS Data Center.

Return to Table of Contents

REFERENCES

Return to Table of Contents

Return to Geospatial Analysis Symposium

Comments to aberjame@esumail.emporia.edu