More Precise Monitoring of Soil Moisture Content. 2024

The utilization of unmanned aerial vehicle (UAV) remote sensing allows for fast and effective soil moisture content (SMC) monitoring. Scale conversion can obtain remote sensing images at multiple resolutions and offer the possibility of improving the estimation accuracy of SMC, which is significant for precision agriculture. To explore the effect of upscaling conversion on SMC monitoring with UAV remote sensing, multispectral images and SMC data at different soil depths were acquired. The multispectral image with an original spatial resolution of 0.1 m was continuously upscaled to 6 m (with a step of 0.1 m and a total of 60 levels). The correlation coefficient method was adopted to determine the optimal scale, and spectral indices at the original and optimal scales were calculated. SMC inversion models at different depths were constructed at the original and optimal scales by multiple linear regression (MLR) and back propagation neural networks (BP), respectively. The results showed that the correlation of UAV multispectral data with SMC displayed a change in first enhancement and then fluctuating weakness with the increase of the upscaled levels. The optimal scale of UAV multispectral remote sensing for monitoring SMC at all soil depths was about 2 m. The BP models were significantly better than the MLR models. For example, in the SMC estimation at 0-20 cm, the RMSE of the BP model was reduced from 0.0110 to 0.0087 (20.9%). This study can provide theoretical and technical references for improving the accuracy of UAV multispectral remote sensing for SMC monitoring.