Predicting soil chemical characteristics in the arid region of central Iran using remote sensing and machine learning models
Contact us if these videos are not relevant.
Contact us if these videos are not relevant.
View abstract on PubMed
Summary
This summary is machine-generated.Digital Soil Mapping (DSM) effectively mapped soil salinity in arid Iran using remote sensing. An ensemble model integrating Landsat 8 data and machine learning provided accurate predictions for calcium, carbonates, and sulfates.
Area Of Science
- Digital Soil Mapping (DSM)
- Remote Sensing
- Soil Science
- Machine Learning Applications
Background
- Arid regions like Gavkhouni sub-basin face challenges in sustainable land management due to salinity and water scarcity.
- Accurate soil property data is crucial for effective land management but often scarce in data-limited environments.
- Digital Soil Mapping (DSM) offers advanced techniques to bridge soil data gaps.
Purpose Of The Study
- To assess the performance of six machine-learning models for mapping four salinity-related soil properties (Ca, CaCO3, CaSO4, SO4) in an arid region of Iran.
- To identify key environmental predictors influencing soil salinity using remote sensing data and digital elevation models.
- To evaluate an integrated approach for mapping multiple soil attributes simultaneously.
Main Methods
- Employed 34 environmental covariates derived from Landsat 8 imagery and a digital elevation model.
- Collected 96 surface soil samples (0-20 cm depth) for analysis of Ca, CaCO3, CaSO4, and SO4.
- Assessed six machine-learning models (RF, CART, SVR, GAM, GLM, ensemble) using tenfold cross-validation, with predictor selection via VIF and Boruta algorithm.
Main Results
- The ensemble model demonstrated superior performance, achieving high R2 values: 0.89 for Ca, 0.84 for CaCO3, 0.79 for SO4, and 0.73 for CaSO4.
- Key predictors varied by soil attribute: Elevation and TVDI for Ca; TCB and TCW for CaCO3; Band 5 (B5) and TCB for CaSO4; TCB, B5, and Band 7 for SO4.
- The study successfully mapped multiple salinity-related soil properties in an integrated manner.
Conclusions
- Remote sensing-based DSM, particularly using Landsat data, shows significant potential for enhancing soil monitoring in data-scarce, arid environments.
- The developed models and identified predictors can improve soil assessment and support sustainable land management in resource-limited regions.
- The availability of free satellite data provides valuable opportunities for advancing soil science and management globally.
Contact us if these videos are not relevant.
Contact us if these videos are not relevant.
