Application Of Doppler Radar Data In The Diagnosis And Forecast Of Heavy Rainfall

Doppler radar which provides data at high spatial and temporal resolution over a large area is widely used to analyze and forecast strong convective weather. Due to the later construction of the Doppler weather radar network in our country, in terms of depth application and research of Doppler radar data, there is a big gap between our country and the advanced countries.In order to understand the application of Doppler radar data in mesoscale weather system as well as the dynamical structure of precipitation system, a large-area and more accurate retrieved result is aimed to be captured, the method of single-Doppler radar wind retrieval based on the two-step variational is applied to the dual-Doppler radar. The data come from the Dual-Doppler Radar (S-Band) in Hefei and Fuyang, Anhui province. A heavy rainfall process occurred in Anhui Province on 7-10th June in 2010 is analyzed. The results show that this continuous rainfall is associated with the wind fields which cantains a convergence in low layers and a divergence in high layers, a cyclonic wind rotation and a wind shear line caused by southeaster and east wind.Although data of Doppler radar are widely used in the retrieval wind and numerical prediction based on high spatial and temporal resolution, most studies focus on S-band radar and a few of study for the shorter wavelength (eg C-band radar) weather radar. So how about the application results with C-band weather radar in heavy rain? Based on the study of S-band radar, a heavy rainfall process occurred in Tianshui on 19 June 2013 is analyzed for the retrieval wind from Tianshui C-band radar. The results show that, the wind field can be retrieved well by the two-step variational method from the C-band radar. The south and southeast winds provide warm and wet air to the regions of heavy rainfall. The low-level jet and warm advection have significant contribution to the precipitation. The wind has a good correspondence with the radar reflectivity. In addition, the reflectivity of C-band radar is smaller than the S-band radar with the same intensity of observed precipitation. Therefore, the reflectivity of C-band radar should be corrected when quantitative precipitation is estimated.In order to apply the wind data of C-Band Doppler radar to numerical weather prediction model better, the retrieval wind field from the heavy rainfall process occurred in Tianshui on 19 June 2013 is processed into a normal form of radiosonde observation. Then the processed wind is assimilated with the WRF-3DVAR system and the forecast is implemented by WRF model. Four experiments were designed to evaluate the performance of the assimilation of radar retrieval wind. The results show that:assimilation of retrieval wind can extend the positive impact on rainfall forecast up to 12h, the improvement is particularly significant especially in the 3-9h after the assimilation and 9-12h forecast has a positive effect to some extent. However, the 0-3h accumulation precipitation forecast after the assimilation is not improvemed evidently when compared with Experiment CTL. In addition, the cycle assimilation is better than only assimilate once but not the more the better.Further more, in consideration of the underestimation of radar-derived quantitative precipitation, the precipitation is classified into stratiform precipitation, convective precipitation and warm rain by a fuzzy logic algorithm and an automated technique which uses vertical profiles of reflectivity (VPR). Finally, the reflectivity values are converted into rainfall rate using an adaptive Z-R relation based on the different rain types, and then the 1-hr rainfall is accumulated from rain rate. The data come from Hefei Doppler radar and gauge of Anhui province from June to August in 2010. In order to evaluate the performance of cloud classification, the following three experiments are constructed to have a comparion:experiment Ⅰ, CINRAD WSR-98D default Z-R relationship (Z=300I1.4) is applied on converting reflectivity to rain rate; experiment Ⅱ, according to the fuzzy logic algorithm, rainfall rates are derived from convective and stratiform corresponding Z-R relationship; experiment Ⅲ, according to different rain types, an adaptive Z-R relation is applied on convective, stratiform and warm rain. By comparing the above three experiments with precipitation from gauge observations respectively, the results show that the rain rates derived from all experiments correspond well with gauge observations. In addition, experiments Ⅱ and Ⅲ have improved the problem of underestimation significantly. Particularly, the accuracy of experiment Ⅲ is higher than the others in the aspect of all precipitation fields as well as the heavy rain.