The changes of low-level wind field play an important role in the formation of sudden precipitation, which can change the flow field structure in the lower atmosphere, thereby affect the stability and vertical movement of the lower atmosphere and promote the development of convective clouds. Based on wind profile radar data at Chang’an Station, observation data, the fifth generation atmospheric reanalysis data released by the European Center for Medium Range Weather Forecasting, and Doppler radar data, this study analyzed the evolution characteristics of the low-level wind field during three typical sudden precipitation events under the control of the subtropical high at the northern foothills of the Qinling Mountains in midsummer. These events occurred on August 6, 2023, from 11:00 to 12:00 (referred to as “Process I”), July 13, 2023, from 00:00 to 01:00 (referred to as “Process II”), and August 3, 2022, from 18:00 to 19:00 (referred to as “Process III”). The results show that all three events occurred under the circulation background of the subtropical high combined with the intrusion of cold air at low level, exhibiting strong suddenness. For Process I and Process II, the intrusion of cold air at low level was characterized by westerly winds, while for Process III, it was characterized by easterly winds. Before the precipitation, the atmosphere over the Chang’an region was in a significantly unstable state, with weak vertical wind shear in the middle troposphere, which was the main reason for the highly localized nature of these three precipitation events. In midsummer, the multi-year average low-level wind speed at Chang’an Station generally exhibited a single-peak pattern, the wind speed initially increased with height and then decreased. The average wind speed below an altitude of 1 000 meters did not exceed 3.14 m·s?1, and the hourly wind speed shows distinct diurnal variation characteristics. The low-level wind direction displayed a counterclockwise rotation with increasing height, shifting gradually from southwesterly to southeasterly winds.The 4-6 h before the occurrence of three sudden precipitation processes, there was a cold air intrusion process in the low-level over Chang’an, and the wind speed was significantly bigger than the multi-year average. With the continuous invasion of low-level cold air, the 2 m temperature decreased rapidly, the air pressure rose, convection was triggered, and heavy precipitation occurred. The continuous intrusion of low-level cold air could generate strong mesoscale frontogenesis in the lower atmosphere, providing energy and triggering conditions for sudden precipitation. On the other hand, due to the obstruction of the local terrain at the northern foot of the Qinling Mountains and the Guanzhong Basin, the low-level cold air was forced to rise, promoting an increase in precipitation.

In recent years, the artificial intelligence has made a breakthrough in image identification. In order to find out the practical value of artificial intelligence models in radar echo nowcasting in Wuhan City, the radar echo and precipitation observation data in Wuhan from 2015 to 2020 are used to train four deep learning models (PredRNN++, MIM, CrevNet and PhyDNet), then these trained models and radar echo observation data in flood season of 2021 are used to do nowcasting of radar echo. And on this basis, the precipitation processes are selected by using precipitation intensity and area indexes in Wuhan, and the performance of four deep learning models and optical flow method in radar echo nowcasting are tested and evaluated in Wuhan in flood season of 2021 in terms of mean square error (MSE), structural similarity index measurement (SSIM), probability of detection (POD), false alarm rate (FAR) and critical success index (CSI). The results are as follows: (1) On the whole, MSE of MIM model is the smallest, while its POD is the highest, and SSIM of MIM and PredRNN++ models are the highest. FAR of four deep learning models is lower than that of optical flow method, and it is the lowest for PhyDNet model. Except for CrevNet model, CSI of other three deep learning models is higher than that of optical flow method, and it is the highest for MIM model. (2) CSI of optical flow method is the highest during 0-12 minutes of forecast, while that of MIM model is the highest from 18 to 120 minutes, which shows the advantage of deep learning model for long prediction time. (3) With the increase of echo intensity, POD and CSI of four deep learning models and optical flow method decrease rapidly, while the variation characteristics of FAR of optical flow method and deep learning models are different. (4) For the regional precipitation processes, the prediction ability of deep learning models firstly reduces and then enhances significantly with the increase of precipitation intensity, while the optical flow method is insensitive to the change of precipitation intensity, so the increments of CSI of deep learning models are the highest under the strong precipitation processes compared with optical flow method. For the local convective precipitation processes with general intensity, the prediction ability of all models and optical flow method significantly reduces. (5) The analysis results of a rainstorm case show that deep learning models not only have prediction ability to the change of echo intensity to a certain extent, but also have better prediction ability to echo movement than optical flow method, so they have a good operational prospect.