Evaluation of reanalysis soil moisture products using Cosmic Ray Neutron Sensor observations across the globe

Soil moisture influences many physical processes in hydrology, meteorology, and agriculture, such as evapotranspiration, infiltration, runoff generation, drought development, crop growth, among others. Robust and accurate soil moisture estimates are needed for drought monitoring, climatology research and hydrological model initialization. Compared to in-situ soil moisture measurements and satellite products, reanalysis soil moisture products are becoming good alternatives for analysis at the global scale due to their long temporal coverage. However, there are a great variety of reanalysis products available and choosing a suitable product that is consistent with the observed soil moisture condition is of significant interest.

In this study, we evaluate the performance of seven reanalysis products including ERA5-Land, CFSRv2, MERRA2-Land, JRA-55, GLDAS-Noah v2.1, CRA40, and GLEAM against field measurements from 109 sites with Cosmic Ray Neutron Sensors (CRNS). CRNS provide estimates of root-zone soil moisture at the field scale (~250m radius from sensor). The sites used in this study are located in the United Kingdom (51 sites), United States (40 sites), and Australia (18 sites). Metrics describing the temporal correlation (Pearson correlation coefficient, R) for the daily time series, seasonal cycle and anomaly time series, bias (mean square error, MSE) as well as root mean square difference (unbiased root mean square error, ubRMSE) are employed to quantify agreement between reanalysis products and measurements.

As an example, for the UK sites, CFSRv2 and GLEAM soil moisture products have lower errors in terms of temporal correlation and bias, while MERRA2-Land and GLDAS datasets exhibit higher errors. Reanalysis soil moisture products tend to have poorer behaviour at wet sites, with low temporal correlation and high bias. Relatively low correlation coefficient values are also found at sites with organic soils and low bulk density. This study provides guidelines for researchers about choosing the reanalysis soil moisture products.