Submitted to: Applied and Environmental Soil Science
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: February 26, 2011
Publication Date: May 1, 2011
Repository URL: http://handle.nal.usda.gov/10113/57962
Citation: Browning, D.M., Duniway, M.C. 2011. Digital soil mapping in the absence of field training data: A case study using terrain attributes and semiautomated soil signature derivation to distinguish ecological potential. Applied and Environmental Soil Science. 2011, Article ID 421904:1-12. Interpretive Summary: Soil properties are critical for understanding patterns of vegetation community composition and primary productivity as well as ecological potential in arid and semi-arid ecosystems globally. Digital mapping of soil properties at broad spatial scales is commonly hampered by a lack of supporting ancillary data to train the classification. This is especially so in spatially extensive and often remote dryland ecosystems. We present a statistically-based approach to derive spectral signatures for classifying soils without prior knowledge and then characterize mapped soil classes to facilitate interpretation of soil properties. We demonstrate that it is possible to extract ecologically meaningful information about soil properties from a remotely sensed perspective. Extensive field sampling and knowledge of ground conditions is not required a priori, but the approach using established image processing techniques does require post hoc field efforts to characterize soil class maps generated. The method provides a basis for mapping soil classes across landscapes and/or for effectively stratifying sites to make the most efficient use of resources. As such, the approach to delineate soils and focus field sampling efforts would benefit state and federal land managers (i.e., NRCS, BLM, and USFS) as well as ecologists and soil scientists.
Technical Abstract: Spatially-explicit data for soil properties governing plant water availability are needed to understand mechanisms influencing plant species distributions and predict plant responses to changing climate. This is especially important for arid and semi-arid regions. Spatial data representing surrogates for soil forming factors are becoming widely available (e.g. spectral and terrain layers). However, field-based training data remain a limiting factor, particularly across remote and extensive drylands. We present a method to map soils with Landsat ETM+ imagery and high resolution (5 m) terrain (IFSAR) data based on statistical properties of the input data layers that do not rely on field training data. We then characterize soil classes mapped using this semi-automated technique. The method distinguished spectrally distinct soil classes that differed in subsurface rather than surface properties. Field evaluations of the soil classification in conjunction with analysis of long-term vegetation dynamics indicate the approach was successful in mapping areas with similar soil properties and ecological potential.