Comparison of Longwave Diurnal Models Applied to Simulations of the Earth Radiation Budget Experiment

Abstract

Simulations of the Earth Radiation Budget Experiment with several satellite sampling schemes have been used to compare three different approaches to modeling longwave diurnal behavior observed over certain kinds of land regions. November 1978 data from the GOES satellite have been used to produce a reference set of radiation parameters over the regions of interest. The monthly average longwave radiant exitance has been estimated first with linear interpolation between satellite measurements, then with a method that replaces linear interpolations across day-night boundaries with piecewise constant extrapolations to the boundaries, and finally with a trigonometric model which replaces some of the linear interpolations that go through daytime measurements over land. This third model consists of constant extrapolation of nighttime measurements to sunrise or sunset, with a half-sine curve fitted through existing daytime measurements and constrained at sunrise and sunset to an average of the surrounding nighttime measurements. It applies only when the daytime and surrounding nighttime measurements meet certain restrictive criteria, including tests that tend to limit the trigonometric model to cloud-free regions. For all satellite sampling strategies considered, the trigonometric model gave the best overall monthly estimate of longwave radiant exitance. For non-land regions, the linear interpolation model generally gave better results than the piecewise constant model.