Soil Inventory From Digital Analysis of Satellite Scanner and Topographic Data

Abstract

A soil inventory of Chariton County, Missouri was prepared using computer‐aided analysis of Landsat Multispectral scanner (MSS) data to determine how spectral soil maps produced by digital analysis of satellite MSS data can accelerate and improve the quality of the National Cooperative Soil Survey Program.Landsat data collected on 15 Apr. 1974 were spatially registered at a scale of 1:24,000 and overlaid with ancillary data in the form of digitized township, watershed, and physiographic boundaries. The county was classified into 30 spectrally separable classes using computer‐implemented pattern recognition techniques. Approximately 64% of the county was identified as bare soil and was classified into 14 spectrally separable soil classes. The remaining 36% was classified as forest, pasture, and close‐grown crops.The digitized boundary data allowed the computer to manipulate, compile, extract, and present spectral class information within topographic or political units. The physiographic boundaries were used to delineate three landscape positions of interest: (i) bottomlands, (ii) gently sloping uplands, and (iii) moderately steep uplands. These delineations allowed the computer to differentiate between spectrally similar soils which occur on distinctively different landscape positions. Spectral classes of soil were correlated with individual soil series and taxonomic classifications for both the bottomland and uplands units.While a spectral classification of soils alone cannot distinguish between widely different soils exhibiting similar spectral responses, it can aid in identifying meaningful divisions of soils. By combining digitized ancillary data with MSS data, a more detailed delineation of soils can be provided as compared to information derived solely from MSS data. Further refinement of this technique will continue to increase the usefulness of satellite MSS data as an aid in soil survey.

Ground radiometry and airborne multispectral survey of bare soils

“…The reasons for this are many, but undoubtably include the relatively poor spatial resolution of LANDSAT MSS images, and the inability of present sensors to provide information on subsurface features. However, there is evidence from other countries that multispectral data from sensors aboard aircraft and satellites can provide a useful data source for soil survey, and complement traditional data collection methods (see, for example, Weismiller et al 1977, Mathews et al 1973, Kristof and Zachary 1974 The test areas used in the present study are shown in figure I and include representatives of many of the major soil groups of the British Isles. The present report is restricted largely to data collected from areas C (the 'Winchester test area') and D (the 'Petersfield test area') during the Natural Environment Research Council (NERC) MSS-83 Airborne Thematic Mapper (ATM) campaign.…”

A Geographical Analysis of Image Interpretation Methods

“…[I91 and by Weismiller et a/. [17], both of whom combined multispectral remote sensing data with non-image terrain information to prepare an inventory of soil resources.…”

Likely locust infestation areas in western New South Wales, Australia, located by satellite

“…However, Weismiller et al (1977), Gerberman & Neher (1979), Kornblau & Cipra (1983) and Dwivedi (1985) have shown that soils can be classified effectively according to their spectral characteristics.…”