Temperature Profile Retrieval from AMSU-A with the Aid of a priori Surface Type Information

The higher spatial resolution of the Advanced Microwave Sounding Unit (AMSU) over previous microwave sounders means a greater sensitivity to variations in surface emissivity, especially over land. Our simulation studies have shown that the presence of water bodies, topography, and transitions between land-use types within the footprint have a significant impact on measured brightness temperatures. Not accounting for these variations can degrade retrieved profile accuracy near the surface and up to the 500 mb level. A priori information on surface emissivity therefore is required for successful retrievals, especially under cloudy conditions. In this paper we examine the utility of model-based emissivity maps for retrieval of temperature profiles from AMSU-A data. These maps are derived by assigning a 50-GHz emissivity value, determined from a radiative transfer model for vegetated soils, to each of the 13 land-use classes in the MM5 terrain database. The study area chosen is the central USA which contains a variety of land cover types and sizes of water bodies. Coincident GOES VIS/IR data provide cloud cover and surface temperature information to the optimal estimation retrieval system. Retrievals with and without this a priori information are compared to temperature fields from a numerical weather prediction model and to radiosonde observations.