Clear-Sky Longwave Downward Radiation Estimation by Integrating MODIS Data and Ground-Based Measurements

Surface energy budget has great influence on earth surface biogeophysical and biogeochemical processes. As one of the components of surface energy budget, the longwave downward radiation (LWDR) is essential to many land applications. In this research, a new LWDR estimation scheme under clear sky by integrating MODIS thermal bands radiance and in situ measurements through the multivariate adaptive regression splines (MARS) has been proposed. The MODIS products and corresponding multi-network in situ LWDR observations under clear skies in 2004 were extracted for model training and cross-validation. The in situ LWDR observations from seven SURFRAD sites in 2008 were used to test the new method. In training phase, the cross-validation RMSE, bias, and R-square of the new method are, respectively, 24.66 W/m(2), -0.446 W/m(2), and 0.88 in the daytime and 23.16 W/m(2), -0.272 W/m(2), and 0.92 in the nighttime. The test results at the seven SURFRAD sites show that the proposed method has good performance, with RMSE, bias, and R-square of 17.15 W/m(2), 0.028 W/m(2), and 0.91 in the daytime and 14.08 W/m(2), -3.609 W/m(2), and 0.94 in the nighttime, respectively. More importantly, the error analysis reveals that the predicted residual errors of the proposed method have a weak correlation with related meteorological parameters at global scale, which reflects the stability and robustness of the new method.