Daily average net radiation (DANR) is an important variable for estimating evapotranspiration from satellite data at regional scales, and is used for atmospheric and hydrologic modeling, as well as ecosystem management. A scheme is proposed to estimate the DANR over large heterogeneous areas under clear-sky conditions using only remotely sensed data. The method was designed to overcome the dependence of DANR estimates on ground data, and to map spatially consistent and reasonably distributed DANR, by using various land and atmospheric data products retrieved from MODIS (Moderate Resolution Imaging Spectroradiometer) data. An improved sinusoidal model was used to retrieve the diurnal variations of downward shortwave radiation using a single instantaneous value from satellites. The downward shortwave component of DANR was directly obtained from this instantaneous value, and the upward shortwave component was estimated using satellite-derived albedo products. Four observations of air temperature from MOD07_L2 and MYD07_L2 data products were used to derive the downward longwave component of DANR, while the upward longwave component was estimated using the land surface temperature (LST) and the surface emissivity from MOD11_L2. Compared to in situ observations at the cropland and grassland sites located in Tongyu, northern China, the root mean square error (RMSE) of DANR estimated for both sites under clear-sky conditions was 37 W m−2 and 40 W m−2, respectively. The errors in estimation of DANR were comparable to those from previous satellite-based methods. Our estimates can be used for studying the surface radiation balance and evapotranspiration.