Soil Landscape Estimation and Evaluation Program (SLEEP)

The Soil-Landscape Estimation and Evaluation Program (SLEEP) is a software tool for use in the ArcGIS-environment developed to predict soil attributes and provide inputs to the Soil and Water Assessment Tool (SWAT). The latter is used to simulate stream flow, crop yield, sediment transport and nutrient transport through input of terrain, climatic, land use/cover and soil data. Whereas the first three datasets have increasingly become available, with increasing accuracy, through satellite remote sensing, obtaining high-resolution soil data is still a challenge. Currently, the majority of soil data are available as soil maps, in polygon format, which do not allow extraction of layers of individual soil attributes, thus ignoring the spatial variability of soils within these polygons. SLEEP aims to address this problem by using measured soil properties (viz. soil depth and percentage content of clay, silt, sand, stone and organic matter) at different locations in a watershed along with the geographical co-ordinates of the easurement locations, to produce the spatially distributed soil properties for the whole watershed in the form of raster data. The essential inputs of SLEEP are a high-resolution digital elevation model (e.g. SRTM) and field observations, or, in their absence, legacy soil data/maps. Additional layers, such as satellite images and auxiliary data, improve the prediction accuracy. The model contains a series of steps (menus) to facilitate iterative analysis in order to derive many terrain attributes for characterization of individual pixels as well as the characteristics of the contributing area. The model then subdivides the entire watershed into sub-watersheds and classifies these into groups. For each group a linear regression model is used to predict soil attributes from the terrain attributes and auxiliary data.  The final step is to merge the predicted soil attributes for the entire watershed or study area.

An application of the tool demonstrated acceptable accuracy and better spatial distribution of soil attributes compared with two spatial interpolation techniques, inverse distance weighing and kriging. The robustness of the SLEEP approach is indicated by a low sensitivity of SWAT prediction to the number of field observations using SLEEP-generated soil attribute values.