Research progresses in parameterization schemes for estimating downward shortwave and longwave radiations are reviewed by use of satellite data and ground-based routine observations in recent years. Based on a comprehensive analysis of the relevant literatures, the theoretical backgrounds of downward shortwave and longwave radiation parameterization schemes are described. According to the clear-sky and cloudy-sky conditions, the radiation parameterization schemes are summarized and their advantages and disadvantages in the practical application are reviewed. Under clear-sky conditions, the main parameterization schemes of downward shortwave radiation are statistical methods and radiative transfer models; the longwave radiation parameterization schemes are mainly single-parameter and dual-parameter methodologies. Under cloudy-sky conditions, radiation parameterization schemes are mainly limited to the determination of cloudiness. The limitations of estimating downward shortwave and longwave radiation on the surface are the selections of appropriate parameterization scheme that need considering land cover types and the attainment of meteorological observations and satellite remote sensing data. In the future, the relatively promising estimation approach is pointed out by combination of polar-orbiting and geostationary satellite remote sensing data.

Based on the parameterization of downward shortwave radiation, upward shortwave radiation, downward longwave radiation and upward longwave radiation above the surface, a scheme is proposed to estimate instantaneous net radiation over arid and semiarid areas for clear sky days using only remote sensing observations. The scheme attempts to eliminate the need for ground meteorology data as the estimates of land surface radiation balance, by using various land and atmospheric data products available from moderate resolution imaging spectroradiometer(MODIS). The preliminary results, using 89 cases over the degraded grassland and 85 cases over the farmland during July to October in 2003, July to November in 2004 and June to October in 2005, show good agreement with ground-based observations, with the root mean square error of 47. 5 and 49. 2 W/m² respectively, which have better accuracies than the other similar studies.