chlorophylls, algae, and floating oil slicks. The Lake Ontario study exemplifies a potential use of Landsat in providing a small-scale

assessment of pollution, internal lake dynamics, and water quality that could not be acquired routinely from any other single, existing source.

Ground water analysis

Ground water accounts for more than 90 percent of the water available for man's use. Anomalies in vegetation, lithology, soils, soil moisture,

and their patterns of distribution have been used to infer the presence of ground water from Landsat data.

Moore and Deutsch (1975) described present and potential uses of Landsat to locate new aquifers, to study aquifer recharge and discharge, to estimate ground-water pumpage for irrigation, to predict the location and type of aquifer management problems, and to locate and monitor strip mines which commonly are sources of acid mine drainage. Deutsch (1974)

described the use of Landsat data to delineate glacial and alluvial aquifers in the Columbia River basin and the crystalline rock aquifers of New York State. Taiti (1975) demonstrated the use of Landsat for predicting ground-water occurrence in the semi-arid lava plains between Mt. Kenya and the Ewaso-Nygeria River of Kenya. Moore (1977) was able to infer ground-water occurrence from analysis of landforms, drainage, and vegetation patterns on Landsat data. He also has developed procedures and keys for the detection of aquifers on Landsat imagery (Moore, 1976). These techniques are now being employed by the EROS Program in a new series of "Water Resources Remote-Sensing Workshops." The first of these was held in December 1977.

Land cover analysis

The Geography Program of the Geological Survey devised efficient computer techniques to produce a land cover map of the Washington, D.C., metropolitan area compiled from October and April 1973 Landsat data.

The map, at a scale of 1:1,000,000, is overprinted with place names and landmarks as well as census tracts on the Universal Transverse Mercator

rectangular coordinate system.

Computer analysis of Landsat image tapes has also enabled land use mappers to work at scales as large as 1:25,000. The methods developed on high-speed computer processing systems such as the ILLIAC-IV show promise of providing major inputs to the 1980 census which will relate detailed land-cover classes with population statistics acquired by standard interview methods. By superimposing census tract boundaries on land-cover classes, the relationship of population to land-cover class can be established and used periodically to monitor areas of great population change.

In 1977, NASA and the GS started a 3-year cooperative Application Systems Verification and Transfer project to test the feasibility of the operational use of Landsat digital data for change detection and updating land use and land cover maps being produced by the GS Geography Program.

25

Land capability mapping

Mapping and describing the capability of the land (rather than

solely its present use) provides information for decisions on potential, beneficial, and detrimental uses of the land. A method of land capability mapping, the so-called "land systems approach," was developed in

Australia, using aerial photographs and field methods, and was later adopted in South Africa and the Soviet Union.

Present research by the EROS Program uses the Australian philosophy but is based on analysis of Landsat images. This method permits the grouping of land classes based on combinations of geological, soil, and vegetational characteristics. Landsat data permit a uniform and repeatable mapping of land systems and an objective comparison of maps

for different areas.

A test of the feasibility of mapping integrated terrain units was conducted in part of southwestern Queensland, Australia, in cooperation with the Queensland Department of Primary Industries (Robinove, 1977). The primary purpose was to test the use of digitally classified terrain units for grazing land management. A recently published map of "land systems," (made by aerial photointerpretation and ground surveys) integrated terrain units composed of vegetation, soil, and geomorphic features,

and was used as a basis for comparison with digitally classified Landsat multispectral images. The land systems, in turn, each have a specific

grazing capacity for cattle (expressed in animals per km2) which is

determined by sampling and measurement of the biomass and its nutrient quality.

Landsat images, in computer-compatible tape form, were digitally

classified into distinct spectral classes to determine their correspondence

to land systems.

Inasmuch as some spectral classes did correspond to

known land systems, mapping of the systems could have preceded field work and acted as a guide to field sampling and detailed land system description,

as well as giving a measure of the location, area, and extent of each

system.

A similar experiment, now in progress in the northeastern Mojave Desert in California, has shown that the technique, using Landsat data, is applicable to that region. Land capability mapping is applicable to information needs of Federal land management agencies such as the

Bureau of Land Management, National Park Service, and Forest Service.