JoVE - Journal of Visualized Experiments
JoVE Visualize
Contact Us

Optimal inversion model for cultivated land soil salinity based on UAV hyperspectral data

  • 0Department of Geography and Spatial Information Techniques, Ningbo University, Ningbo 315211, Zhejiang, China.
Ying Yong Sheng Tai Xue Bao = the Journal of Applied Ecology +

|

February 10, 2025

|

February 10, 2025

View abstract on PubMed

Summary

This summary is machine-generated.

Unmanned aerial vehicle (UAV) hyperspectral technology accurately maps soil salinity in cultivated lands. Combining spectral data transformations with Random Forest regression provides a reliable method for monitoring and managing soil salinization.

Area Of Science

  • Agricultural Science
  • Remote Sensing
  • Environmental Monitoring

Background

  • Soil salinization severely impacts agricultural productivity and food security.
  • Accurate soil salinity data is crucial for effective land management and remediation strategies.
  • Hyperspectral remote sensing offers a promising non-invasive approach for large-scale soil salinity assessment.

Purpose Of The Study

  • To develop and validate an optimal model for accurately inverting soil salinity in cultivated lands using UAV hyperspectral data.
  • To compare the performance of different machine learning models (SVR, BPNN, RFR) for soil salinity inversion.
  • To evaluate the effectiveness of various spectral transformation techniques and feature band selection strategies.

Main Methods

  • Acquisition of hyperspectral remote sensing data using Unmanned Aerial Vehicles (UAVs).
  • Application of spectral transformation techniques (e.g., first-order differential) to hyperspectral data.
  • Feature band subset selection tailored to different land use statuses.
  • Comparison of Support Vector Machine (SVR), Back Propagation Neural Network (BPNN), and Random Forest regression (RFR) models.
  • Validation of model accuracy using statistical metrics (R², RMSE, RPD).

Main Results

  • The optimal soil salinity inversion model combined first-order differential spectral transformation data with Random Forest regression (RFR).
  • This model achieved a high coefficient of determination (R²) of 0.885.
  • Separately extracting feature bands for different land use statuses significantly improved overall model accuracy (RMSE = 0.413, RPD = 4.208).

Conclusions

  • UAV-based hyperspectral remote sensing, coupled with optimized spectral transformations and machine learning (RFR), provides a highly accurate method for regional soil salinity inversion.
  • This approach offers valuable scientific support for the prevention and control of soil salinization in cultivated lands.
  • The findings serve as a reference for high-precision soil salinity monitoring using advanced remote sensing technologies.

Keywords: