干旱区生态环境知识资源中心

A multi-spectral outgoing longwave radiation (OLR) estimation technique is being applied to GOES sounder data to study the diurnal cycle of OLR. Currently, all satellite-based OLR estimation techniques require the use of models of the diurnal cycle in order to estimate time-averaged OLR from any scene. In this study we use the first half of July 1998 hourly GOES sounder radiance observations with the Ellingson et al. (1994) multi-spectral OLR estimation technique to determine the instantaneous, daily- and monthly-averaged OLR from 100 x 100 km scenes over North America and the adjacent Gulf of Mexico. The instantaneous data are compared with collocated data from the TRMM/CERES to establish the instantaneous accuracy of the multi-spectral estimation techniques for homogeneous scenes. The time-averaged data are used to estimate the accuracy of the ERBE/CERES and HIRS diurnal cycle models as applied to daily- and monthly averaged products at different spatial resolutions.

Our analysis shows:

RMS errors depend on sampling time and interval;

RMS errors of daily-averaged OLR for V x V regions may be kept <4 Wm(-2) with a 3 hour sampling interval; and

The clear-sky OLR diurnal cycle for an underlying desert surface resembles a sine curve.