Abstract:
Land data assimilation systems merge estimates from land models with observations of the land surface state based on their respective uncertainties and aim to produce estimates that are superior to the model estimates and observations alone. Poorly specified model and observation error parameters (such as error standard deviations), however, negatively affect the quality of the assimilation products. For very poor input error parameters, the assimilation estimates may even be worse than model estimates without data assimilation. Adaptive data assimilation approaches dynamically estimate the input error parameters along with the state estimates by continually adjusting the error parameters in response to internal filter diagnostics. The observation error standard deviation is therefore a by-product of the adaptive assimilation procedure. In this presentation, we use an adaptive data assimilation approach developed at the NASA Global Modeling and Assimilation Office to derive error estimates for the NASA surface soil moisture product from the Advanced Microwave Scanning Radiometer for the Earth Observing System (AMSR-E). Such error estimates are not supplied with the AMSR-E data and are therefore a genuine contribution to the AMSR-E product.