The prediction of monthly rainfall is greatly beneficial for water resources management and flood control projects. Machine learning (ML) techniques, as an increasingly popular approach, have been applied in diverse climatic regions, showing their respective superiority. On top of that, the ensemble learning model that synthesizes the advantages of different ML models deserves more attention. In this study, an ensemble learning model based on stacking approach was proposed. Four prevalent ML models, namely k-nearest neighbors (KNN), extreme gradient boosting (XGB), support vector regression (SVR), and artificial neural networks (ANN) are taken as base models. To combine the outputs from the base models, the weighting algorithm is used as second-layer learner to generate predictions. Large-scale climate indices, large-scale atmospheric variables, and local meteorological variables were used as predictors. R2, RMSE and MAE, were used as evaluation metrics. The results show that the performance of base models varied among the nine stations in the Taihu Basin, while the stacking approach generally performed better than the four base models. The stacking model showed better performance in spring and winter than in summer and autumn. During wet months, the accuracy of model prediction varied more significantly. On the whole, based on performance evaluation measures, it is concluded that the proposed stacking ensemble multi-ML model can provide a flexible and reasonable prediction framework applicable to other regions.
Given the fact that researchers require more specific spatial rainfall information for storm flood calculation, hydrological risk assessment, and water budget estimates, there is a growing need to analyze the spatial heterogeneity of rainfall accurately. This paper provides insight into rainfall spatial heterogeneity in urban areas based on statistical analysis methods. An ensemble of short-duration (3-h) extreme rainfall events for four megacities in China are extracted from a high-resolution gridded rainfall dataset (resolution of 30 min in time, 0.1° × 0.1° in space). Under the heterogeneity framework using Moran’s I, LISA (Local Indicators of Spatial Association), and semi-variance, the multi-scale spatial variability of extreme rainfall is identified and assessed in Shanghai (SH), Beijing (BJ), Guangzhou (GZ), and Shenzhen (SZ). The results show that there is a pronounced spatial heterogeneity of short-duration extreme rainfall in the four cities. Heterogeneous characteristics of rainfall within location, range, and directions are closely linked to the different urban growth in four cities. The results also suggest that the spatial distribution of rainfall cannot be neglected in the design storm in urban areas. This paper constitutes a useful contribution to quantifying the degree of spatial heterogeneity and supports an improved understanding of rainfall/flood frequency analysis in megacities.
The Global Precipitation Measurement (GPM) Integrated Multi‐satellitE Retrievals for GPM products (i.e., IMERG) provide new‐generation satellite precipitation measurements. For urban contexts, however, the issues of its bias and insufficient resolutions still exist. This study aims to develop high‐precision and high‐resolution (e.g., 0.01°/1 h) data for a metropolitan region based on IMERG and gauge precipitations. The original IMERG product is evaluated using hourly in situ precipitations from 47 gauges. A spatial downscaling‐calibration (DC) technique is then developed to enhance the IMERG using the Normalized Difference Vegetation Index. The results show the limited capability of IMERG to capture sub‐daily precipitation and high‐intensity precipitation. The proposed DC method significantly improves IMERG performance, with correlation coefficient (CC) increasing from 0.07 to 0.75, and probability of detection improving from 0.34 to 0.90 at the hourly scale. In terms of spatial rainfall distribution, 86% of mean absolute error and 80% of RMSE are improved with CC increasing from 0.07 to 0.91 on average. Additionally, the calibrated downscaled product provides finer information in local areas, capturing three times more spatial variabilities of urban precipitation against the original IMERG input data. The results highlight the necessity of improving urban observations for flood risk management at fine spatiotemporal resolutions.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.