Estimation of Surface Net Radiation from Solar Shortwave Radiation Measurements

This study develops a method to estimate surface daytime net radiation (R_n) from solar shortwave radiation measurements with help of the conventional meteorological observations and vegetation indices from satellite. Data collected at twenty-two sites in the U.S. and two sites on the Tibetan Plateau (China) from 2000 to 2006 are used to develop and validate the method. The land cover types of the sites vary among desert, semi-desert, croplands, grasslands and forest. The elevations of the sites range from 98 m to 4700 m. The results show that our method estimates R_n accurately. The bias varies from -7.8 W m^-2 to 9.7 W m^-2 (plusmn3% in relative value) for different sites, while the root mean square error ranges from 12.8 W m^-2 to 21 W m^-2 (from +5% to +9% in relative value) for different sites with an average of 16.9 W m^-2 (+6% relative) for all sites. For all sites the correlation coefficient is about 0.99. The correlation coefficient between the measured and predicted annual abnormal (year-average subtract multi-year average) in daytime R_n is as high as 0.91, indicating that the method can be used to accurately estimate long-term variation in R_n.