Subpixel-scale variability of rainfall and its application to mitigate the beam-filling problem

Surface rain radar data from areas surrounding Japan and in the tropics are used to study the variability of rainfall over an area similar to a satellite microwave radiometric instantaneous field of view (FOV). The results of such variability are applied to radiative transfer simulations to modify the brightness temperature versus rain rate relationship. First, the surface radar data from different geographical regions are used to develop a relationship between fractional rain cover and average rain rate over microwave radiometric FOV-sized areas. This resultant relationship is also compared with spaceborne Tropical Rainfall Measuring Mission (TRMM) Precipitation Radar (PR) observations. It is found that the rainfall type (convective or stratiform) accounts for the most of variations in this relationship. The sub-FOV rainfall variability is then investigated using the same surface radar data sets. The conditional probability distribution function of precipitation within the FOV-sized area is parameterized in terms of average rain of the area. The form of the model function does not depend upon the characteristics of the two radars and their geographical locations and can be parameterized by a combination of two lognormal distributions. By analyzing collocated TRMM Microwave Imager (TMI) and PR data, it is shown that the averaged relationship between observed brightness temperatures and rain rates at 19.3 and 37 GHz shows a significant disagreement with that simulated by a plane-parallel radiative transfer model for convective rain events, although it shows a better agreement for stratiform rain events. Presumably, this disagreement is mainly caused by variability of the rain field within satellite instantaneous FOV. With the application of the fractional rain area and the probability distribution function derived from this study to the radiative transfer simulations, we are able to bring results of the radiative transfer simulations much closer to the observations.