Land Surface Thermal Infrared Emissivity Database Development and Modeling for Operational Weather Prediction and Climate Models

Shunlin Liang & John Townshend

Department of Geography, University of Maryland, College Park, MD 20742

Abstract

Land surface thermal infrared (IR) emissivity is a critical variable in characterizing the lower atmospheric boundary condition for successful assimilation of satellite thermal-IR observations into the numerical weather prediction (NWP) and climate models. It has been treated very approximately by various operational models. However, further improvements are urgently needed.

Our overall objective is to develop forward land surface thermal-IR radiative transfer emissivity models for NWP and climate studies. To achieve this goal, we propose to execute the following three tasks: 1). develop a high-resolution emissivity database from multiple satellite sensors (e.g., MODIS, ASTER) with multiple algorithms using a data fusion approach; 2). establish the empirical relations between emissivity and various land surface biogeophysical variables; and 3). assess, calibrate and improve existing radiative transfer emissivity models. There exist different emissivity radiative transfer models for different media including those in the JCSDA Community Radiative Transfer Model. However, they are usually calibrated by laboratory measurements or data at local scales. In many cases, they may not be valid at the landscape scale. They must be systematically evaluated and calibrated by a more accurate emissivity database, and some of the physics in those models need to be re-parameterized by the empirical relations so that they can predict the actual emissivity more realistically. The existing databases developed from laboratory measurements or combined with satellite retrievals have coarse spatial and temporal resolutions and inadequate accuracy that cannot be used to link emissivity with biogeophysical variables and to calibrate physically-based emissivity models.