IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing · 2015 · 47 citations · 38 references
Natural SciencesEarth SciencesSoil SciencesRemote SensingSoil ChemistrySoil ScienceErosionRemote Sensing Image AnalysisSoil ErosionSoil Horizon IdentificationSoil HorizonsSoil TextureAirborne Hyperspectral SensingErosion MappingHyperspectral AnalysisLaboratory Soil CharacterizationSoil Erosion MappingSoil Erosion ProcessesSoil PropertiesSupport Vector Machine Classification
The determination of surface soil properties is an important application of remotely sensed hyperspectral imagery. Moreover, different soil properties can be associated with erosion processes, with significant implications for land management and agricultural uses. This study integrates hyperspectral data supported by morphological and physico-chemical ground data to identify and map soil properties that can be used to assess soil erosion and accumulation. These properties characterize different soil horizons that emerge at the surface as a consequence of the intensity of the erosion processes, or the result of accumulation conditions. This study includes: 1) field and laboratory characterization of the main soil types in the study area; 2) identification and definition of indicators of soil erosion and accumulation stages (SEAS); 3) compilation of the site-specific MEDiterranean Soil Erosion Stages (MEDSES) spectral library of soil surface characteristics using field spectroscopy; 4) using hyperspectral airborne data to determine a set of endmembers for different SEAS and introducing these into the support vector machine (SVM) classifier to obtain their spatial distribution; and 5) evaluation of the accuracy of the classification applying a field validation protocol. The study region is located within an agricultural region in Central Spain, representative of Mediterranean agricultural uses dominated by a gently sloping relief, and characterized by soils with contrasting horizons. Results show that the proposed method is successful in mapping different SEAS that indicate preservation, partial loss, or complete loss of fertile soils, as well as down-slope accumulation of different soil materials.
38
R: A Language and Environment for Statistical Computing
R Core Team · 2000 · 353K citations · Full text
Natural Sciences, Computer Science, Statistical Computing +10
Numerical recipes in C: the art of scientific computing
Choice Reviews Online · 1993 · 18.1K citations · Full text