Abstract:
Using the 3-hour densely-observed radiosonde and the 1-hour automatic weather station data obtained from Baolian, Chaoyang and Daxing station in Beijing from August 28, 2016 to September 2, 2016, based on the WRF model and the WRFDA three-dimensional variational assimilation system, this paper conducted assimilation experiments on only assimilating densely-observed radiosonde (S-DA), only assimilating automatic weather station data (A-DA), and simultaneously assimilating the above two types of data (M-DA), to study the improvement effect of three data assimilation schemes on the numerical simulation of the boundary layer in the Beijing area. The results show that: (1) In the vertical direction, densely-observed radiosonde plays a core role in improving simulation results, which can reduce the root mean square errors of temperature, humidity, and wind within the boundary layer range by 65%, 61%, and 22%, respectively. The automatic weather station data also contributes to the simulation results in the vertical direction, but the improvement is small and the impact range is low. The results of the M-DA test are similar to those of the S-DA test. (2) In the horizontal direction, the improvement effect of automatic weather station data is mainly reflected in a wide range of impact, with strong improvement efforts in densely-observed radiosonde, but the impact range is small. The advantages of combining two types of data in the M-DA experiment can make the simulation results closer to the observation results. (3) In terms of assimilation timeliness, the assimilation test has a relatively long effect on improving the thermal state within the boundary layer, and a relatively short effect on improving the humidity state and dynamic structure. Among them, the M-DA test can extend its effect on improving the thermal state up to 6 hours in the forecast, and its effect on improving the humidity and dynamic structure up to 3 hours in the forecast. In summary, the simultaneous assimilation of densely-observed radiosonde and automatic weather station data is more effective than assimilating either data alone. The two types of data can complement each other"s shortcomings after assimilation, which can greatly improve the initial field of the model and thus improve the accuracy of boundary layer simulation results to a certain extent.