loading page

Soil salinity estimation based on machine learning using the GF-3 radar and Landsat-8 data in the Keriya Oasis, Southern Xinjiang, China
  • +2
  • Sentian Xiao,
  • Ilyas Nurmemet,
  • Nuerbiye Muhetaer,
  • Jing Zhao,
  • Adilai Abulaiti
Sentian Xiao
Xinjiang University College of Resource and Environment Sciences
Author Profile
Ilyas Nurmemet
Xinjiang University College of Resource and Environment Sciences

Corresponding Author:[email protected]

Author Profile
Nuerbiye Muhetaer
Xinjiang University College of Resource and Environment Sciences
Author Profile
Jing Zhao
Xinjiang University College of Resource and Environment Sciences
Author Profile
Adilai Abulaiti
Xinjiang University College of Resource and Environment Sciences
Author Profile

Abstract

Soil salinization has been an important environmental problem globally, particularly in oasis areas in arid zones. The advantages of using multi-source data, combining radar and optical remote sensing data, and applying machine learning-based algorithms to these data could be beneficial for addressing the soil salinization problem. The current research on salinity estimation still needs to be deepened. To overcome this shortcoming, this study combines the environmental covariates extracted from the Gaofen-3 (GF-3) radar data, Landsat-8 multispectral data, and digital elevation model (DEM) data to explore the advantages of radar remote sensing in detecting soil salinity. The soil salinity distribution degree in the Keriya Oasis is mapped using a machine-learning-based method, and the advantages of different sensor images in predicting soil salinity are evaluated. Three soil salinity inversion models are constructed using measured electrical conductivity (EC) data, the random forest (RF), gradient boosting tree (GDBT), and extreme gradient boosting (XGBoost) models. Also, five classes of optimal environmental covariates are used. The results show that the best accuracy corresponding to an R 2 of 0.87, a root mean square error (RMSE) of 6.02, and a relative percent deviation (RPD) of 2.77 is achieved by the RF model on the GF-3+Landsat-8 data. Therefore, using multi-source data can fully exploit the advantages of both radar and optical data and has been demonstrated to be a more effective method for mapping soil salinity in the study area. In the importance analysis of independent variables, the salinity index (SI), normalized difference vegetation index (NDVI), and DEM contributed the most to the prediction of soil salinity. In this study, the radar polarization decomposition characteristics are incorporated into the inversion of soil salinity modeling as an environmental covariate, providing an innovative and efficient method for soil salinity estimation in arid areas.