Studies on urban extreme precipitation and its influencing factors are significant for prevention and reduction of meteorological disasters; however, few studies focus on hourly extreme precipitation (HEP) events due to the limitation of observation. By using the summer hourly precipitation data in Beijing from 1980 to 2020, the spatial distribution and temporal variation of HEP as well as its local environmental influencing factors are investigated. It is found that both summer precipitation amount and frequency of HEP are affected by topography, with high values in windward slope area. The summer precipitation amount of HEP is 160–200 mm, accounting for 42–47% of the annual summer precipitation amount, while the frequency proportion of HEP is only 5.5–6.0%. Although the summer precipitation amount and frequency in Beijing both decrease in the past 41 years, those for HEP present an opposite trend mainly due to the increasing HEP since 2003 and this is a phenomenon worthy of attention. A similar bimodal pattern in diurnal variation is found for the summer precipitation amount and frequency of HEP, with two peaks in 19–22 LT and 01–05 LT, respectively, indicating that HEPs are more concentrated in the evening and early morning especially in urban area. Moreover, the urbanization process of Beijing is consistent with the change trend of HEP, implying that the stronger the urban heat island intensity (UHI), the higher the probability of HEP. Furthermore, the convergence lines of terrain are also conducive to local heavy rainfall, and lower tropospheric stability (LTS) and convective available potential energy (CAPE) as well as aerosols may also affect HEP in Beijing.

Based on the prediction results of RmAPs-CAV1.0, a short time heavy rainfall in western Xinjiang was simulated and analyzed. By comparing the forecast results with the actual precipitation, it is found that RMAPS-CAV1.0 can accurately reflect the important forecast parameters of the short-time heavy precipitation system, such as precipitation level and starting time. Contrast observation and prediction results found that the terrain has forced effect to the surface wind field, the precipitation process in Yili valley special opening of the west mountain terrain, the center is located in Kazakhstan in central Asia low vortex cause lower for the westerly Yili river valley, the center is located in the Tarim Basin of central Asia for Piandongfeng middle low vortex makes the Yili river valley, cause the Yili river valley and the lower levels of vertical wind shear reinforcement; In the Yili River Valley, the combination of topography and the central Asian vortex circulation in Kazakhstan forms a low-altitude convergence line. The low-level westerly winds transport water vapor into the valley, and the northerly airflow in the valley and the northern margin of the Taklamakan Desert converge near the windward slope, and the enhanced upward movement results in the uplift of the water vapor accumulated in the valley. The result analysis of RMAPS-CAV1.0 shows that the warm and cold airflow and water vapor under the action of topography have important contributions to the generation of this precipitation.

In order to avoid the double punitive disadvantage of traditional test methods, based on the advantages of wavelet in multi-resolution analysisthe precipitation forecast products of RMAPS (Rapid-refresh Multi-scale Analysis and Prediction System) system are tested objectively. In this study, Haar is used as basis function. The test results show that the hourly precipitation forecast error of the 3km discriminant ratio of the forecast system is less than 15% for the precipitation process with the horizontal scale exceeding 94km. As the wavelet analysis scale decreases, the precipitation error gradually increases.

Precipitation forecasting at fine spatiotemporal resolution is essential for refined meteorological services, yet high-resolution data remain costly to obtain and operational model outputs (e.g., the Fuxi model at 0.25°) are too coarse to resolve small- and medium-scale precipitation structures. To address this gap, this paper proposes a precipitation forecast downscaling method based on a Correlation Diffusion Network (Diffusion-U-Net) that converts low-resolution Fuxi precipitation forecasts into high-resolution fields. The method combines the probabilistic generation capability of diffusion models with the multi-scale feature extraction advantages of U-Net, and introduces a correlation module with a channel-spatial dual attention mechanism to capture the spatiotemporal correlation features of precipitation fields. Evaluated on 2020–2023 Fuxi forecasts over North China, the proposed method outperforms traditional interpolation and deep learning baselines: it achieves a structural similarity index (SSIM) of 0.81 and a heavy-precipitation critical success index (CSI) of 0.72, and reduces the heavy-precipitation intensity bias by 6% relative to vanilla U-Net. The method restores small- and medium-scale precipitation details more accurately and provides technical support for refined meteorological forecasting.

Based on NCEP FNL 1°*1° analysis data and Xinjiang meteorological bureau area numerical operation prediction, a local precipitation event in southern Xinjiang on June 26 was analyzed, and the results showed that the water vapor involved in this process originated from the northwestern Gangetic Plain and was transported along the Indus River plain and over the northwestern end of the mountains. Driven by the Iranian low-pressure trough, water vapor at 500 hPa was transported across the northwestern end of the Gangdise Mountains, past Georgoli Peak, and northwest over the Kunlun Mountains to arrive in southern Xinjiang.