Predicting root zone soil moisture using observations at 2121 sites across China.

Area of Science:

• Hydrology
• Soil Science
• Remote Sensing

Background:

• Root zone soil moisture (RZSM) is crucial for understanding hydrological processes, plant-atmosphere interactions, and agricultural productivity.
• Accurate RZSM estimation is vital for climate research and water resource management.

Purpose of the Study:

• To estimate RZSM across China using an optimized exponential filter (EF) method combined with a random forest (RF) regionalization approach.
• To validate the EF method's performance across multiple soil layers using extensive in situ observations.

Main Methods:

• Optimized a one-parameter (T) exponential filter at four soil layers (10-50 cm) using in situ data from 2121 sites across China.
• Developed RF classifiers for each soil layer, integrating calibrated T values with 14 soil, climate, and vegetation parameters.
• Regionalized calibrated T values using RF classifiers to generate spatial T maps for each soil layer.

Main Results:

• The EF method demonstrated good performance in predicting RZSM, with median Nash-Sutcliffe efficiency (NSE) values ranging from 0.73 (10-20 cm) to 0.27 (40-50 cm).
• The parameter T exhibited distinct spatial patterns influenced by climate regimes across China.
• Spatial maps of T were generated for each soil layer, providing valuable data for RZSM estimation.

Conclusions:

• The combined EF and RF approach offers an improved method for estimating RZSM across large regions.
• The study highlights the effectiveness of integrating in situ data with advanced modeling techniques for soil moisture assessment.
• The developed methodology has potential applicability for RZSM estimation in other global regions.