The Use of DEM and Satellite Data for Regional Scale Soil Databases
New, quantitative methods and data sources for characterizing small scale soil resources have been demonstrated. AVHRR and coarse spatial resolution DEM were designed for mapping large areas of the world quickly and cost effectively. The method combines digital elevation data, “ground truth” information, including the soil taxonomic class for measured soil locations, and a time series of satellite images to form a digital soil database. The results show that using ancillary information such as AVHRR data and DEM derivatives from the national to continental level surveys is among the most promising tools for geographers and soil surveyors. The AVHRR data is often used for land cover studies but its usefulness in soil studies has not yet been proven. This study is a representative example of the usefulness of AVHRR data in characterizing the soil-forming environment and delineating soil patterns, particularly when integrated with other data for describing the soil landscape, such as the DEM, slope, curvature and PDD. The predictive power of AVHRR and similar low spatial resolution satellite data sources could be further improved with the development of soil sensitive filters. Mention should be made of the potential improvement of the products derived from these data sources with the use of better quality data provided by satellites that have been launched recently. Neither the AVHRR nor the DEM-derivatives show high correlation with the soil classes, but both represent a great portion of the environmental variability. In general, the more uncorrelated information is extracted from DEM and AVHRR, the better explanation of the spatial soil variability is achieved with an integrated use of them. The images of AVHRR time series show a relatively low correlation, thus each of the new dates adds much potential information on the soils. The studies also highlighted the great help of surface vegetation in soil remote sensing, as indicated by the high R² value of Band 1 and NDVI. The importance of the short-term weather history of the study area was also demonstrated. Terrain information and terrain variables were primarily developed for large scale local studies. Small scale mapping of large regions presents different issues, like over-generalization and over-smoothing of the soil information. The terrain features with smaller extents are dissolved into a larger neighborhood. As a smoother terrain map is created, a lot of detail is lost and less variability is observable. Many of the terrain attributes are useless with this approach. Elevation, slope, relief intensity, potential drainage density and the curvature variables are the most informative digital variables for characterizing the soil-landscape in small scale inventories. The resulting soil databases will have all the advantages of quantitatively derived databases, including consistency, homogeneity, and reduced data generalization and edge-matching problems. Although the results from the above procedures are believed to be accurate enough to serve as a basis for global and regional studies, they should be checked and further revised by local and regional experts to ensure quality. Research should continue on improving the procedures, augmenting the pedon data with new field sampling, and incorporating new image and DEM data sources. One of the most important results of these studies is the demonstration of the usefulness of these data sources for small scale soil mapping and the overall validity and representatitivity of the AVHRR-terrain/soil correlation within the temperate region of the world. Further studies will need to be performed to test the use of AVHRR and terrain data for other climate zones of the World, where potential problems, like continuous cloud cover, may occur.