Multi-Sensor Earth Observation and Spatially Validated Machine Learning for Soil Texture Mapping in Shiradi Ghat, Western Ghats
Journal of Intelligent Decision Making and Information Science · 07.09.2026 · 3 (11s) · 368-388
DOI: 10.17762/jidmis.v3.3044 ↗Kurz gefasst
Fachartikel in der Zeitschrift Journal of Intelligent Decision Making and Information Science, online erschienen am 07.09.2026. Verfasst von Swapnalaxmi K und 1 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
Accurate soil-texture information is essential for understanding soil variability and supporting sustainable land, watershed, and ecosystem management in heterogeneous mountainous landscapes. This study developed a reference-guided multisensor Earth observation and spatially validated machine-learning framework for soil-texture mapping in Shiradi Ghat, Western Ghats. Sentinel-1 SAR, Sentinel-2 MSI, SRTM DEM, and OpenLandMap soil-property data were integrated, with all predictors harmonized to a common 10 m grid and subsequently aggregated to the 250 m reference scale. A total of 18 predictors, including spectral bands, vegetation and soil indices, radar backscatter, polarization difference, elevation, slope, and aspect transformations, were generated. Random Forest (RF), XGBoost, and Multilayer Perceptron (MLP) were compared using an 80:20 random hold-out strategy, followed by 5-fold spatial block cross-validation at 1.25 and 2.5 km scales. RF achieved the best baseline performance, with R² values of 0.2178, 0.1635, and 0.1096 for clay, sand, and silt, respectively. Enhanced RF increased these to 0.2505, 0.1929, and 0.1369. Under 2.5 km spatial validation, R² decreased to 0.1351, 0.0434, and 0.0245, respectively. Mapping across 9,790 valid cells produced mean fractions of 30.08% clay, 42.14% sand, and 27.79% silt.
Zitieren
APA
K, S., & S, V. S. (2026). Multi-Sensor Earth Observation and Spatially Validated Machine Learning for Soil Texture Mapping in Shiradi Ghat, Western Ghats. Journal of Intelligent Decision Making and Information Science, 3(11s), 368-388. https://doi.org/10.17762/jidmis.v3.3044
Vancouver
K S, S VS. Multi-Sensor Earth Observation and Spatially Validated Machine Learning for Soil Texture Mapping in Shiradi Ghat, Western Ghats. Journal of Intelligent Decision Making and Information Science. 2026;3(11s):368-388. doi:10.17762/jidmis.v3.3044
BibTeX
@article{K2026,
author = {K, Swapnalaxmi and S, Vijaya Shetty},
title = {Multi-Sensor Earth Observation and Spatially Validated Machine Learning for Soil Texture Mapping in Shiradi Ghat, Western Ghats},
journal = {Journal of Intelligent Decision Making and Information Science},
year = {2026},
volume = {3},
number = {11s},
pages = {368--388},
doi = {10.17762/jidmis.v3.3044}
}
EUVIMED