JoVE - Journal of Visualized Experiments
JoVE Visualize
Contact Us

A longitudinal analysis of soil salinity changes using remotely sensed imageries

  • 0Department of Soil Sciences, Gorgan University of Agricultural Sciences and Natural Resources, Gorgan, Iran. soraya.bandak@gmail.com.
Scientific Reports +

|

May 6, 2024

|

May 6, 2024

View abstract on PubMed

Summary

This summary is machine-generated.

Remote sensing effectively monitors soil salinity using indices from Landsat 8 and Sentinel 2 data. Decision Tree models accurately predict salinity, aiding land management and conservation in affected regions.

Area Of Science

  • Environmental Science
  • Agricultural Science
  • Remote Sensing

Background

  • Soil salinization severely impacts agricultural productivity, causing desertification and land degradation.
  • Large-scale field and laboratory studies for soil salinity are labor-intensive and costly.
  • Remote sensing offers a viable alternative for large-scale soil salinity assessment.

Purpose Of The Study

  • To evaluate the effectiveness of various soil salinity indices derived from remote sensing data.
  • To compare the performance of machine learning models for predicting soil electrical conductivity.
  • To identify key topographic and moisture-related indices for soil salinity mapping.

Main Methods

  • Utilized Landsat 8 (L8) and Sentinel 2 (S2) satellite imagery.
  • Employed machine learning algorithms: Random Forest (RF), eXtreme Gradient Boosting (XGBoost), Decision Tree (DT), and Support Vector Machine (SVR).
  • Correlated derived salinity indices with 280 ground-truth soil electrical conductivity (EC) samples across 24,000 hectares.

Main Results

  • The Decision Tree (DT) model demonstrated superior performance, achieving R² values of 0.85 (L8) and 0.86 (S2).
  • Key influential salinity indices identified include Multi-resolution Valley Bottom Flatness (MrVBF), moisture index, Topographic Wetness Index (TWI), and Topographic Position Indicator (TPI).
  • Time series analysis revealed significant reductions in soil salinity and sodium levels in areas with implemented drainage systems.

Conclusions

  • Remote sensing, coupled with machine learning, provides an accurate and efficient method for assessing soil salinity.
  • Topographic and moisture indices are crucial for understanding and predicting soil salinity patterns.
  • Effective land management strategies, such as drainage systems, can successfully mitigate soil salinization.

Keywords: