Resampling of microwave radiometer observations for spatial resolution matching

Multiple frequency radiometers using a common antenna produce footprints of differing sizes due to diffraction. This effect hampers physical retrieval algorithms that use observations of more than one frequency. In order to alleviate this problem for the AMSR instruments (PM-1 and ADEOS-II platforms), actual observations will be linearly combined to produce a data set of effective observations at a series of common spatial resolutions. The Backus-Gilbert method of optimal interpolation will be used to compute the necessary weighting coefficients using pre-launch measured antenna patterns prior to application of the weights when AMSR data become available. As a precursor to use with the AMSR instrument, the technique has been applied to TMI observations, where the difference between the actual and resampled data has been quantified.