Retrieval of atmospheric parameters from SSM/I using a variational method

A different approach to the extraction of useful information from SSM/I data, and one which is being adopted increasingly, is that of optimal estimation. In this method, the required physical parameters are estimated simultaneously and are the best set of parameters which explain the observed SSM/I brightness temperatures (top of atmosphere radiances). A great advantage of this variational approach is that it allows the retrieval to take place whilst keeping the derived atmospheric profile consistent with high quality a priori information in the form of a numerical weather prediction (NWP) forecast of the atmospheric state at the time of the SSM/I data. A further advantage of this technique (referred to as 1DVAR, as it is a one-dimensional variational retrieval), is that a direct byproduct of the technique is an estimate of the errors of the retrieval process. In assessing the performance of the 1DVAR retrieval scheme, it is shown that the assimilation of total precipitable water vapour will improve the accuracy of the NWP initial conditions (the analysis). In addition, since surface windspeed retrievals are tightly constrained by the background field, this parameter may be used as a diagnostic tool for the 1DVAR retrieval scheme itself. Regional biases in this parameter may be used to diagnose cross-talk between parameters in the retrieval or unmodelled elements which may be making the scheme sub-optimal