LAND SURFACE EMISSIVITY ESTIMATION AT 89 AND 150 GHz FROM AMSU-B MEASUREMENTS

Most of the works on estimation of land surface emissivity from satellite radiometric measurements are performed at the frequencies d 100 GHz in the past two decades [1-3]. In the frequency range of AMSU-B (89-183 GHz) the land surface emissivity problems draw attention from scientific community recently mainly because of difficulty associated with precipitation retrievals over land. For liquid precipitation, unlike the ocean areas, the lack of radiometric contrast between land surface and lower atmosphere in the frequency range of 10-37 GHz presents a serious problem for reliable rain rate retrievals [4]. As a consequence, rain rate retrievals over land have to rely on surrogate scattering signals at high frequencies, which originate mainly from storm-associated frozen hydrometeors aloft above the freezing level. For solid precipitation (snowfalls), scattering signals from AMSU-B measurements have been used to retrieve snowfall rate in recent years [5-6]. In both cases, a priori knowledge of surface emissive properties is essential to accurate and reliable precipitation retrievals over land. In this paper, we focus on estimation of land surface emissivity at 89 and 150 GHz from AMSU-B measurements on January 21, 2007, around 1130 UTC, over an 400 km by 400 km area centered around the C3VP (Canadian CloudSat Calipso Validation Project, 44.23qN, 79.87qW) site under a clear-sky condition. The Goddard WRF (Weather Research and Forecast) modeled atmospheric temperature, humidity and cloud profiles, as well as surface temperatures [7] are used to aid this estimation; these data sets are modeled in hourly intervals and at a spatial resolution of 1 km. Two methods are explored: (1) estimation based on AMSU-B 89 and 150 GHz measurements and WRF-modeled parameters alone, and (2) estimation based on AMSU-B measurements, WRF-modeled parameters, and additional matching with AMSU-B 183 GHz measurements for a more realistic atmospheric condition at the time of the AMSU-B pass. In both methods,