Evaluation of passive microwave land surface emissivities for improved precipitation retrievals over land for GPM-era algorithms—part I: comparison of inversion methodsp

Traditionally, the retrieval of precipitation over land from passive microwave sensors has been severely limited due to the highly variable land surface emissivity, thus limiting most current retrieval methods to high frequency scattering techniques which are only indirectly related to surface rainfall. Ideally, if the land surface emissivity could be retrieved with a high degree of accuracy (especially over complex surfaces such as arid and snow covered surfaces), more physical retrieval algorithms that retrieve the liquid water content of clouds and precipitation below the freezing level and the ice water path above the freezing level could be developed. This is a primary goal of the Global Precipitation Measurement (GPM) mission.

There are a wealth of emissivity products that have been developed by the scientific community, however, the majority of these have been geared for NWP model data assimilation, primarily under clear sky conditions. Additionally, these are primarily for frequencies at or below 89 GHz. Wetting of the underlying surface due to active precipitation may greatly alter the emissivity, in particular, at frequencies where the atmosphere is partially transparent to the clouds and precipitation. Under the auspices of NASA's Precipitation Measurement Missions (PMM) Science Team, a study was embarked upon that is assessing the similarities and differences of several off the shelf emissivity products, as well as other techniques being tested by the precipitation community. All of these methods derive the emissivity by removing the atmospheric effects (e.g., water vapor and clouds) and using the surface temperature. It is the purpose of this paper to describe the comparison study and present the results to date.

It should be noted that two other companion papers are also being submitted – one that also evaluates emissivity derived through forward calculations, using a Land Surface Model to drive the surface conditions, and one the presents the sensitivity of the precipitation retrievals to uncertainties in the emissivity.