L-band Radiosonde Data Analysis And Assimilation Experiment In Meso-scale Model

The generation switch from59-701sounding system to L-band electronic soundingsystem was completed in2011in China. L-band upper-air sounding system can provide densesecond/minute level sounding data. Although weather forecast and numerical weatherprediction both can benefit from the application of high-vertical-resolution sounding data,weather analysis and assimilation business based on L-band high-resolution vertical soundinginformation have not been carried out yet in China. The main reason is the lack ofunderstanding of second/minute level sounding data, and the method research and techniquedevelopment for effective usage of it are particularly insufficient. Research work aboutL-band high-vertical-resolution sounding data assimilation is very few. This research is amore thorough work focused on L-band second/minute level sounding data analysis and itsassimilation application into meso-scale operation model. By using L-band sounding dataacross China during3years, surface rainfall data and NCEP global analysis data, qualitycontrol of L-band second/minute level sounding data is explored, the sounding data basicfeatures are analyzed, application of second level data in weather analysis is initiallyresearched, and numerical experiments and comparative study are carried out using regionalmeso-scale numerical weather prediction system GRAPES-Meso(CMA) and BJ-RUC(NorthChina Regional Meteorological Center) targeted on some basic problems aboutsecond/minute level sounding data assimilation., The sparsification method of second leveldata and the method for introducing second level data into assimilation system are explored.Also the effects of L-band vertical intensive observation and balloon drift assimilation onmodel results are researched. The main conclusions are as follows:(1)Singular value exist in second/minute level sounding data and basic quality control isnecessary. There is obvious pulse for the second level wind from present operation algorithm.Based on60s time window, using improved second level sounding wind calculation methodproposed by Wang Mian and Li Wei, wind pulse influence can be reduced while keeping theauthenticity of rapid change of wind direction and speed with height. Then present operationsecond level sounding wind can be modified to some extent, and the modification amplitudeis small in the lower level, but more evident above150hPa or in winter. In summer，the sounding balloons show obvious latitudinal drift in the north of35°N, and the longitudedeviation increases quickly above500hPa. On the average, the difference is larger than0.5°above100hPa, while in the south of35°N latitudinal drift and longitudinal drift are bothweak on average. Therefore, ballon drift should not be ignored for high resolution meso-scalemodel.(2)Second level sounding data is more accurate than conventional sounding data indetermining elements such as Lifting Condensation Level, Free Convection Level,equilibrium level and the height of maximum pseudo-equivalent potential temperature.Accordingly, calculated Convective Available Potential Energy(CAPE) and ConvectiveInhibition(CIN) are more precise. CAPE calculated from second level data is generally largerthan that calculated from conventional sounding data, but their distribution in changing withtime are basically in agreement. For wind and temperature vertical distribution in theboundary layer, the overall situation reflected by conventional and second level soundingdata are consistent, but second level sounding can capture detailed information about verticaldistribution of boundary layer meteorological elements (such as structure change of thermalinversion layer and weak wind) that conventional sounding can’t capture, and thisinformation is valuable for hazy weather change monitoring and short term weather forecast.(3) Sparsification of second level sounding data before assimilation is necessary forcurrent vertical resolution condition of operation model. The sparse scheme which matchesmodel levels is better than those sparses schemes which are uniformly-spaced, and highresolution sounding data can get more reasonable assimilation application through it. L-bandminute level sounding, as natural sparsification of second level sounding, can be used directlyfor higher vertical resolution meso-scale model system (eg: with BJ-RUC system), but for alower vertical resolution system, it is less effective than sparse scheme that matches modellevels.(4)Large-scale situation fields and general characteristics of precipitation derived fromsecond/minute level sounding assimilation by means of profile or point-by-point method aresimilar to those derived from conventional sounding assimilation, but there is still adistinguishable impact on the entire forecast area. Point-by-point assimilation ofseconds/minute sounding indicates that the differences of wind field assimilation caused bythe balloon horizontal drift is more significant at200hPa altitude and the jet stream positionof northern China. Moreover, differences between second/minute sounding assimilation andconventional one mainly occur in areas where the differences of vertical levels betweenminute sounding and conventional sounding are large. The extra information provided bysecond/minute sounding can partially affect the subsequent precipitation forecast.(5)Batch tests shows that Point-by-point assimilation of minute level observations has the best analysis capability, but in subsequent conventional meteorological elementsforecasting the advantage of initial fields gradually weakens, but there is still improvementfor the upper troposphere wind forecasting. For torrential rain, whichever method is used toassimilate minute level sounding, there is a positive effect.(6)After proper sparsification of second level sounding data, current nationalGRAPES-Meso operation systems and operation model system BJ-RUC in North RegionalCenter, both can reasonably absorb the L-band high-resolution sounding information.Although precipitation forecasts of two systems has different degrees of closeness to theobservation, after assimilating second/minute level sounding both systems have a positiveimprovement on details of their own precipitation forecast and intensity of precipitationcenter, which indicates that second/minute level sounding assimilation can have a positiveimpact on precipitation forecast, and is system-independent which means the conclusionhas acertain universality.