Information – based uncertainty decomposition in dual
channel microwave remote sensing of soil moisture
Abstract. NASA's Soil Moisture Active-Passive (SMAP) mission characterizes global spatiotemporal patterns in surface soil moisture using dual L-band microwave retrievals of horizontal, TBh, and vertical, TBv, polarized microwave brightness temperatures through a modeled relationship between vegetation opacity and surface scattering albedo (i.e. tau-omega model). Although this model has been validated against in situ soil moisture measurements across sparse validations sites, there is lack of systematic characterization of where and why SMAP estimates deviate from the in situ observations. Here, soil moisture observations from the US Climate Reference Network are used within a mutual information framework to decompose the overall retrieval uncertainty from SMAPs Modified Dual Channel Algorithm (MDCA) into random uncertainty derived from raw data itself and model uncertainty derived from the model’s inherent structure. The results shown that, on average, 12 % of the uncertainty in SMAP soil moisture estimates is caused by the loss of information in the MDCA model itself while the remainder is induced by inadequacy of TBh and TBv observations. We find the fraction of algorithm induced uncertainty is negatively correlated (pearson r of −0.48) with correlations between in-situ observations and MDCA estimates. A decomposition of mutual information between TBh, TBv and MDCA soil moisture shows that on average 55 % of the mutual information is redundantly shared by TBh and TBv, while the information provided uniquely from both TBh and TBv is 15 %. The fraction of information redundantly provided by TBh and TBv was found to be tightly correlated (pearson r = −0.7) to how well the MDCA output correlated to in situ observations. Thus, MDCA overall quality improves as TBh and TBv provide more redundant information for the MDCA. This suggests the informational redundancy between these remotely sensed observations can be used as independent metric to assess the overall quality of algorithms using these data streams. This study provides a baseline approach that can also be applied to evaluate other remote sensing models and understand informational loss as satellite retrievals are translated to end user products.