High spatiotemporal resolution vegetation index time series can facilitate enhanced remote sensing monitoring of soil salinization

Background and aims

Timely and accurate knowledge of the spatiotemporal variation characteristics of soil salinization is paramount. The vegetation index (VI) time series holds significant promise in soil salinization monitoring, yet studies on using high spatiotemporal resolution remain limited. This study aimed to evaluate the effectiveness of high spatiotemporal resolution VI time series for soil salinization monitoring.

Methods

First, an optimized Gap Filling and Savitzky–Golay filtering (GF-SG) method was proposed to reconstruct high-quality Landsat NDVI time-series data. Second, three inversion models (Models A, B, and C) were established to assess the performance of the high spatiotemporal resolution VI time series in soil salinization monitoring. Model A was developed using single-temporal Landsat images, while Models B and C were developed by incorporating MODIS and high-quality Landsat NDVI time-series data, respectively. Finally, we achieved the inversion and spatiotemporal variations monitoring of soil salinization in the Yellow River Delta (YRD) based on the optimal model.

Results

The Model C demonstrated the highest prediction accuracy (R² = 0.84, RMSE = 0.64 mS cm⁻¹). The optimal model predictions show that soil salinization in the YRD gradually decreases from coastal to inland areas, with an overall improving trend from 2004 to 2022.

Conclusion

The high spatiotemporal resolution VI time series significantly improves the predictive and generalization capabilities of the model and can be effectively used for spatiotemporal dynamic monitoring of soil salinization.