Retrieval of surface spectral emissivity in the polar regions: an optimal-estimation approach

Abstract:

The role of surface spectral emissivity, especially, the surface emissivity in the far-infrared (far-IR), in the high-latitude climate has been recognized in recent modeling studies. There have been few far-IR observations, which significantly limited our capability of characterize such far-IR surface emissivity for the need of climate modeling and polar surface energy budget study. This dilemma will be changed in the coming decade with the PREFIRE (Polar Radiant Energy in the Far-InfraRed Experiment) mission selected by NASA and the FORUM (Far-infrared-Outgoing-Radiation Understanding and Monitoring) mission selected by ESA. Both missions aim at retrieving surface spectral emissivity for high-latitude surfaces.

In preparation for the PREFIRE and FORUM missions, here we study the feasibility of using optimal estimation framework to retrieve surface spectral emissivity, surface skin temperature, and total column water vapor from synthetic clear-sky PREFIRE and FORUM observations. The synthetic radiances are calculated by feeding the ECMWF ERA5 reanalysis profiles to the PCRTM, a principle component-based radiative transfer model.

First, an information content analysis is carried out to quantify the degrees of freedom (DoF) of each measurement. For the PREFIRE measurements in January, the DoF related to surface spectral emissivity in the Arctic is 5.18, of which 2.18 are contributed by the far-IR measurements. For comparison, the DoF for surface spectral emissivity in July drops to 3.37, with only 0.12 coming from the far-IR measurements. With the same set-up for the FORUM measurements, the DoF for far-IR surface emissivity in the Arctic is 9.62 in January and 1.91 in July. The obvious seasonal difference is due to the significant change of total column water vapor in the Arctic from winter to summer. Further using the PREFIRE as a case study, we show how to select the channels for retrieval based on the information content analysis and assess the retrieval uncertainty with current knowledge of the PREFIRE noise equivalent spectral radiances.