中文摘要：

太阳能光伏发电已成为仅次于水电和风能的第三大可再生能源，光伏发电受云量时空变化的影响较大，因此准确模拟云天太阳辐射的时空变化对电网安全运行至关重要。围绕如何减小中尺度气象模式的云初始场误差，进而改进云天的太阳辐射模拟这一关键科学问题，本文通过研究基于卫星资料同化的LAPS（Local Analysis Prediction System）多时间层三维云分析同化方法，改进三维云结构，并将LAPS模式输出结果作为WRF（Weather Researcha nd Forecasting）模式的初始场，模拟了2008年1月及夏季（6～8月）北京地区的总云量和总辐射的时空分布，重点分析了多云和有降水天气过程总辐射的模拟改进效果及其原因。结果表明，同化前后的总云量模拟值与观测值的时间变化趋势基本一致，但大部分时次总云量的模拟值低于观测值；大部分多云及降水时段同化后总云量模拟值较接近于实测值。1月晴天、多云天以及夏季晴天同化前后总辐射模拟值与实测值的时间变化趋势较一致，但同化前后两者的相关性差异不明显；晴天条件下同化前后总辐射模拟值均低于实测值，1月多云条件下多数时段同化后总辐射模拟误差减小不明显，与总云量的改进效果不显著有关。夏季多云、有降水及6月典型降水三种天气条件下同化前后总辐射模拟值与观测值的相关性稍差，同化后两者的相关性较同化前有所改进，尤其是6月典型降水过程改进效果较明显；同化前总辐射模拟误差较大，而同化后误差显著减小，尤其是6月典型降水过程同化后均方根误差和平均相对误差较同化前分别减小了102．6Wm-2和355．9％，最大相对误差减小更显著；同化后总辐射模拟误差小于同化前的比例高达75％，即大部分时刻同化后模拟误差小于同化前。多云和有降水天气过程总辐射模拟效果的显著改进与总

英文摘要：

Photovoltaic power is influenced by the temporal and spatial variation of cloud amounts. Therefore, to ensure safe operation of power grids on cloudy days, accuracy in simulating and forecasting temporal and spatial variations of solar radiation is critical. To reduce initial field errors in the mesoscale meteorological model and to improve the simulation accuracy of solar radiation on cloudy days, the three-dimensional cloud analysis assimilation method in the Local Analysis and Prediction System （LAPS is adopted in this study. The results are used to improve cloud simulation and are used as the initial field of the Weather Research and Forecasting （WRF model. The temporal and spatial distribution characteristics of the total cloud amount and global radiation in the Beijing area in January, June, July, and August and during the typical precipitation processes in June 2008 are simulated with the LAPS-WRF model system. This study focuses on the simulation results of global radiation with and without Fengyun satellite data assimilation and describes the reasons for the improvements on cloudy days and during the precipitation processes. The results showed that the temporal variation of simulated and observed values of total cloud amounts with and without satellite data assimilation were consistent. Without assimilation, the simulated values were significantly lower than observations in most cases. After assimilation, the simulated values of total cloud amounts were closer to observations. In addition, the correlation coefficients between simulation and observation values of global radiation before and after assimilation were higher and the differences of correlation coefficients with and without satellite data assimilation were smaller on clear and cloudy days in January and on clear days in summer. The simulation values of global radiation before and after assimilation were all lower than the measured values on sunny days. After assimilation, the error reduction of global radiation was not noticeable on cloudy