Article ID | Journal | Published Year | Pages | File Type |
---|---|---|---|---|
8894002 | Geoderma | 2018 | 16 Pages |
Abstract
Soil attributes (clay, organic matter and moisture) directly influence land surface temperature (LST). Although there are several studies using soil spectra measured by satellites, soil evaluation through LST is still scarce. The objective of this research was to define the influence of soil attributes on LST and satellite image spectra. The study area (198â¯ha) is located in São Paulo state, Brazil. Soil samples were collected in a 100â¯Ãâ¯100â¯m (0-0.2â¯m) regular grid. A Landsat 5 image, with bare soils, was acquired and LST was extracted using the inversion of Planck's function in band 6. Land surface emissivity was estimated using the Normalized Difference Vegetation Index threshold method. Reflectance values were extracted from bands 1 to 5 and 7. Linear regression (LR) models were calibrated for soil attributes prediction. Each model used a different set of covariates: (a) LST; (b) elevation; (c) spectral reflectance; and (d) all predictors. Ordinary kriging was performed and its results were compared to maps obtained from LR. There was significant correlation between soil attributes and reflectance, LST, and elevation. Models using only elevation presented poor performance for prediction of clay, sand, OM, and iron oxides; models using LST, moderate; and Vis-NIR-SWIR bands, good. The use of LST for estimating soil attributes increases the predictive performance when associated with surface reflectance, improving the validation of models. Mapping of clay, sand, OM and iron oxides using Landsat 5 products can strongly enhance agriculture management approaches.
Related Topics
Physical Sciences and Engineering
Earth and Planetary Sciences
Earth-Surface Processes
Authors
Veridiana Maria Sayão, José A.M. Demattê, Luis Gustavo Bedin, Marcos Rafael Nanni, Rodnei Rizzo,