Assimilation of multiresolution radiation products into a downwelling surface radiation model: 2. Posterior ensemble implementation

Abstract

Satellite-derived radiation products are used to condition a high-resolution estimate (space and time) of total downwelling radiation within a data assimilation (DA) framework. The prior ensemble (part 1 of this study) is conditioned using existing, multiresolution radiation products in an ensemble-based DA scheme to yield a high-resolution posterior ensemble. The resulting posterior ensemble implicitly captures the complex spatiotemporal uncertainties inherent in satellite-derived radiative flux estimation. In addition, application of the DA framework reduces these uncertainties while simultaneously downscaling coarse-scale spatial information content from the existing, multiresolution radiation products. Results suggest that the magnitude of the gain matrix is greatest when longwave (LW) estimates assimilate LW measurements and shortwave (SW) estimates assimilate SW measurements. Correlations in space are less pronounced between broadband fluxes of different type. The greatest reduction in uncertainty was achieved via simultaneous assimilation of North American Land Data Assimilation-LW and Pinker-SW products. Comparison of the conditioned posterior ensemble against a network of independent, ground-based radiometer observations in the Southern Great Plains of the United States during the 14 month study period showed a maximum reduction of 8 W m -2 (∼32%) in LW uncertainty and 2 W m -2 (∼6%) in SW uncertainty. The posterior (conditioned) ensemble could eventually be used as forcing in distributed terrestrial hydrologic model applications or land data assimilation systems with the goal of better understanding and characterizing the variability and uncertainty of the hydrologic response at the land surface. © Copyright 2010 by the American Geophysical Union.