Submitted to: Meeting Abstract
Publication Type: Proceedings
Publication Acceptance Date: 6/17/2004
Publication Date: 10/28/2004
Citation: Crow, W.T., Bindlish, R. 2004. The impact of incorrect model error assumption on the assimilation of remotely sensed soil moisture into hydrologic model. In: Proceedings of the Second Model and Data Assimilation (CAHMDA), October 25-27, 2004, Princeton, New Jersey. p. 145-148. Interpretive Summary:
Technical Abstract: Recent advances in the development of sequential land data assimilation techniques have demonstrated that remote sensing observations of surface soil moisture can improve the dynamic representation of root-zone soil moisture and streamflow in hydrologic models. However, much of the available evidence is based on identical twin experiments using synthetically generated, and artificially perturbed, measurements. These experiments, while extremely useful diagnostic tools for evaluating filter efficiency, typically simplify or avoid a number of key complexities facing operational efforts to assimilate spaceborne observations. One typical assumption in synthetic experiment is that the statistical nature of model errors is perfectly known. In reality, error in hydrologic model predictions comes from a wide variety of sources and manifests itself within multiple model state variables. Consequently, error information required by sequential data assimilation filter is almost never available in operational settings. The flexibility of approaches like the Ensemble Kalman filter with regards to model error is frequently cited to support its use in hydrologic data assimilation. However, such flexibility cannot be properly exploited if the structure and source of model errors cannot be constrained in some way. Generally, little is known about the impact of poorly specified model error on the efficiency of assimilating remote observations into hydrologic models. This analysis describes a set of synthetic experiments where by model errors used to generate the Ensemble Kalman filter are statistically different from errors used to originally perturb the model.