The development of machine learning (ML) provides new means and methods for accurate climate analysis and prediction. This study focuses on summer precipitation prediction using ML algorithms. Based on BCC CSM1.1, ECMWF SEAS5, NCEP CFSv2, and JMA CPS2 model data, we conducted a multi-model ensemble (MME) prediction experiment using three tree-based ML algorithms: the decision tree (DT), random forest (RF), and adaptive boosting (AB) algorithms. On this basis, we explored the applicability of ML algorithms for ensemble prediction of seasonal precipitation in China, as well as the impact of different hyperparameters on prediction accuracy. Then, MME predictions based on optimal hyperparameters were constructed for different regions of China. The results showed that all three ML algorithms had an optimal maximum depth less than 2, which means that, based on the current amount of data, the three algorithms could only predict positive or negative precipitation anomalies, and extreme precipitation was hard to predict. The importance of each model in the ML-based MME was quantitatively evaluated. The results showed that NCEP CFSv2 and JMA CPS2 had a higher importance in MME for the eastern part of China. Finally, summer precipitation in China was predicted and tested from 2019 to 2021. According to the results, the method provided a more accurate prediction of the main rainband of summer precipitation in China. ML-based MME had a mean ACC of 0.3, an improvement of 0.09 over the weighted average MME of 0.21 for 2019–2021, exhibiting a significant improvement over the other methods. This shows that ML methods have great potential for improving short-term climate prediction.

The rainstorm intensity formula and the design of rainstorm hyetographs are important aspects in drainage design standards. Against the backdrop of climate change, in most cities in Jiangsu Province of China, significant trends of increasing intensities of heavy rainfall are apparent. The parameters of the rainstorm intensity formula are no longer applicable in the current context of significantly stronger rainstorms. To adapt to this change, we first contrasted the fitting accuracy of the Gumbel distribution, the exponential distribution, and the Pearson III distribution for the rainstorm intensity formula in Jiangsu. It was found that the Gumbel distribution has the smallest relative mean square error in most cities, proving that it provides the best estimation of rainstorm intensity formula parameters. Therefore, the rainstorm intensity formula parameters for 13 cities was revised using the Gumbel distribution based on the rainfall data from 1991 to 2020. Then, the precipitation with a 100-year return period was calculated using the revised formula. Moreover, to compensate for the lack of storm hyetographs that have been designed for Jiangsu, we designed short-duration rainstorm hyetographs for 13 cities using the Chicago hyetograph method and the Pilgrim and Cordery (PC) method. The results show that most of the short-duration rainstorms lasted between 45 and 120 min and were dominated by single-peaked patterns, with the peak position typically occurring in the first half of the rainfall cycle. The peak coefficient distribution of short-duration rainstorms shows that short-duration rainstorms in the south reached their peak rainfall intensity earlier than those in the north. On this basis, using the Chicago method and PC method, short-duration storm hyetographs were designed, which could be used in the design of drainage systems to provide support in effectively reducing urban flood threats. By comparing the hyetographs with real short-duration rainstorm patterns, it was found that the precipitation process designed using the PC method was most similar to the actual precipitation process. However, the PC method was found to be highly dependent on local precipitation data, whereas only the rain peak coefficient is required to design the Chicago rainstorm hyetograph. Therefore, we primarily recommend hyetographs designed using the PC method for Jiangsu Province’s 13 major cities, while we recommend the Chicago hyetograph for the surrounding areas of the 13 cities that have no meteorological stations or lack data.

This study harnesses ground observation data collected between 1980 and 2021 and ERA5 hourly data to thoroughly implement trend and correlation analysis techniques to explore the spatiotemporal dynamic characteristics of daily and hourly extreme precipitation in the Sichuan Basin. The investigation delineates these characteristics and probes into the potential triggers of extreme hourly rainstorms. The findings unveil the following: (1) A general increase in extreme rainfall volume, contribution rate, intensity, and dispersion, along with a decline in frequency and proportion of rainstorm areas, indicating the concentration of daily-scale severe rainstorms. The basin’s edge receives more precipitation than the bottom, exhibiting latitudinal variations. (2) The northernmost mountainous regions have less frequent, less intense rainstorms influenced by terrain, whereas the northeastern region experiences more frequent, dispersed rainstorms. (3) Extreme hourly rainstorms predominantly occur at night, with rainfall amount, intensity, and frequency declining at 21:00 compared to 19:00. (4) Summer experiences the highest risk of extreme rainstorms, with annual and monthly datasets displaying a rising trend in the frequency, dispersion, and intensity of intense hourly rainstorms. (5) Peak values of extreme hourly rainstorms are growing, with two distinct periods for their frequency: 1:00–9:00 and 10:00–24:00, with an increase in the former and a decrease in the latter. (6) Normalized difference vegetation index (NDVI) values ascend from southwest to northeast within the basin on a ten-day scale, correlating with the distribution of hourly extreme precipitation.

Soil evaporation (Es) and crop transpiration (Tc) are important components of water balance in cropping systems. Comparing the accurate calculation by crop models of Es and Tc to the measured evaporation and transpiration has significant advances to the optimal configuration of water resource and evaluation of the accuracy of crop models in estimating water consumption. To evaluate the adaptation of APSIM (Agricultural Production Systems simulator) in calculating the Es and Tc in Nanjing, APSIM model parameters, including the meteorological and soil parameters, were measured from a two-year field experiment. The results showed that: (1) The simulated evaporation was basically consistent with the measured Es, and the regulated model can effectively present the field evaporation in the whole maize growth period (R² = 0.85, D = 0.96, p < 0.001); and (2) The trend of the simulated Tc can present the actual Tc variation, but the accuracy was not as high as the evaporation (R² = 0.74, D = 0.87, p < 0.001), therefore, the simulation of water balance process by APSIM will be helpful in calculating Es and Tc in a humid area of Nanjing, and its application also could predict the production of maize fields in Nanjing.