Retrieval of Surface Soil Water Content Using Remote Sensing in Incorporation with Phenological Traits of Crops

Abstract

【Objective】 Soil water controls crop growth and many ecological functions of terrestrial ecosystems but is difficult to measure at large scales due to soil heterogeneity. The development in air-born technologies such as drones and satellites over the past decades provides a new avenue to rapidly monitor soil water. The objective of this paper is to propose how phenological traits of crops can be used to help improve accuracy of the retrieval of soil water content using remote sensing imageries. 【Method】 The method was based on Sentinel-1 satellite imageries. We first calculated the in-come angles and the total backscattering coefficient; we then obtained the vertical transmit and vertical receive (VV) polarization, and extracted the normalized difference vegetation index (NDVI) of the crops. Using the in-come angles, backscattering coefficient and NDVI, we calculated the backscattering coefficient by removing the influence of vegetation using the water cloud model. Based on phenological traits of the crops, the relationship between the backscattering coefficient and the relative volumetric soil water content was established for crops at different growth stages. The inversion model was validated against experimental data measured from fields rotated with winter wheat and summer maize. 【Result】 Dividing growth season of the winter wheat into three stages: sowing-tillering, overwintering, turning-green and mature, while taking growth season of the maize as a single stage, the correlation coefficient between measured and estimated soil water contents at the four stages above was 0.40, 0.80, 0.91 and 0.79 respectively. Applying the method with the traits of the crops incorporated to two fields in Guzhen and Lixin, we calculated soil water distribution maps in them. Verification of the model against experimental data showed that its correlation coefficient was 0.73 for entire growth season of the winter wheat and 0.82 for growing season of the summer maize. 【Conclusion】 The regression model with phenological traits of crops at different growth stages incorporated can improve accuracy of the soil water content estimated using remote sensing imageries.

Keywords

• soil moisture content
• remote sensing inversion
• backscattering coefficient
• phenological traits