Abstract: In this study LAPS is used to assimilate data observed during flood seasons from 2008 to 2010 to establish mesoscale analysisfields, respectively for the following regions: south China and Central China. Several test schemes are designed to evaluate the role of variousobservation data in LAPS. Using radiosonde observations as objective criterion, LAPS analysis data are compared with FNL reanalysis data,and root mean square errors（RMSE）are calculated to quantitatively assess the quality of LAPS mesoscale analysis fields when assimilatingall the observations. The results are shown as follows. The RMSEs with assimilating radar, radiosonde and surface observations in LAPS arethe smallest. By assimilating radar and radiosonde observations in LAPS, it has a positive impact on improving wind and temperature analysis,and the RMSEs of relative humidity at 500 hPa decrease. Temperature and relative humidity in the low levels are also improved by assimilatingsurface observations in LAPS, although it has no impact in the middle and high levels. By comparing the RMSEs of each element over thestudy period, the analyses of LAPS temperature, wind direction and wind speed are better improved than that of FNL. Height RMSE is lessthan 10 m in the middle and low levels, and 20 m in the high levels. Temperature RMSE is about 1°C. Relative humidity RMSE is less than20% in the middle and low levels. Wind direction RMSE is less than 20° above 850 hPa. Wind speed RMSE is less than 2 m·s-1.