Solar radiation data are important parameters of crop model,hydrological model and climate change model,however,the distribution of solar radiation sites is scarce and uneven throughout the country, and it is difficult to obtain spatial continuous solar radiation by using only rare radiation site data interpolation or extrapolation.Therefore,the lack of solar radiation data restricts the development of the relevant model,and the neural network on the solar radiation has a good predictability,many Artificial Neural Network ensemble models were developed to estimate solar radiation using routinely measured meteorolological variables,but it did not consider cloud,aerosol,and precipitable water vapor influence on solar radiation.In this paper,we used cloud,aerosols,atmospheric precipitable water vapor from MODIS atmosphere remote sensing products and conventional meteorological data including air pressure, temperature,sunshine duration and latitude and elevation,based on the LM-BP neural network model to simulate the 90 conventional weather stations in Eastern China from 2001 to 2014.The results show that the model has a good fit of 0.95,and the root mean square error is controlled within 2MJ·m~(-2).The average deviation error is between-1MJ·m~(-2) and 1MJ·m~(-2).Finally,using the simulated values of the model and the measured values of 13 radiation sites,the spatial distribution of the annual solar radiation in the East China region from 2001 to 2014 is obtained by spatial interpolation and the spatial variation trend is analyzed.