Comparing and Combining Surface Soil Moisture Products from AMSR2

Abstract:

Soil moisture is an important variable in hydrological systems as its part of the water cycle in the atmosphere, the land surface and subsurface. Microwave remote sensing is a viable tool to monitor global soil moisture conditions at regular time intervals. The Advanced Microwave Scanning Radiometer 2 (AMSR2) is a sensor onboard the Global Change Observation Mission 1 – Water that was launched in May 2012. Multiple soil moisture products from AMSR2 observations exist; these were compared and combined with special emphasis to the global scale. The first product is retrieved by the Japan Aerospace Exploration Agency (JAXA) algorithm, the other uses the Land Parameter Retrieval Model (LPRM). These two products were compared against each other and evaluated against COSMOS data over the United States, Australia, Europe and Africa. The temporal correlations highlight differences in the representation of the seasonal cycle of soil moisture. It is hypothesized that four factors, physical surface temperatures, surface roughness, vegetation and ground soil wetness conditions, affect the quality of soil moisture retrievals. The complementary between the products led to the opportunity to combine them into a superior one that benefits from the strengths of both algorithms.

These soil moisture algorithms share the same background in the radiative transfer model, but each algorithm applies different approaches to reflect various external conditions. As a result, the performance of the products is complementary in many locations in terms of bias, RMSE and, most importantly temporal correlation coefficients. Here, we present a methodology that combines the two AMSR2 based soil moisture products into a single product, which improves the overall performance by leveraging the strengths of the individual products. The new product is combined by applying an optimal weighting factor, calculated based on variance and correlation coefficients against a reference dataset. The complementary behaviour of individual soil moisture retrieval algorithms, in combination with leveraging their strengths in the combination approach, may be of interest for National Aeronautics Space Administration’ (NASA) recently launched Soil Moisture Active and Passive (SMAP) mission dedicated to soil moisture monitoring.