Deconvolution of passive microwave brightness temperatures
Data assimilation of all-sky microwave brightness temperatures remains difficult. One issue that confounds the cross-channel comparisons necessary for deriving observation operator statistics is the different footprint sizes of different channels. For each channel, the underlying signal is convolved over a different antenna pattern. In order to assimilate such observations, this convolution must be taken into account, yet this complex observation operator poses problems for many data assimilation systems. Using a 3DVar-style approach, a methodology for deconvolving brightness temperatures is presented using an observation operator that simulates the antenna pattern of a conical scanning satellite such as GPM/GMI or SSMIS. A method to iteratively improve the background error covariance used in the 3DVar deconvolution is also introduced. These deconvolved observations are also error quantified, and therefore this pre-processing step makes the all-sky satellite microwave observations much easier to assimilate.
Prof. Jeff Steward University of California, Los Angeles