A nowcasting procedure based on downscaling and merging radar retrievals at different scales and resolutions
crossref(2023)
摘要
Convective precipitation represents one of the most critical issues in urban areas and its occurrence may have dangerous effects on the local economy, environment and population. Due to the large temporal and spatial variability, numerical weather prediction models frequently fail to predict such precipitation events and early warning systems may be ineffective. In this scenario, radar-based nowcasting models built on extrapolation of the most recent observed precipitation fields may represent powerful procedures for providing accurate short-term forecasts. In this study we proposed a nowcasting approach for merging relevant information from two radar products with different resolutions and scales: (i) high-resolution (HR) observations retrieved by an X-band weather radar in a small study area (the metropolitan area of Cagliari, located in Sardinia, Italy), and (ii) low-resolution (LR) mosaic data provided by the Italian Civil Protection national radar network (the whole Italian country). To this aim, we developed a downscaling-based procedure for combining the corresponding spectral data by generating an artificial power spectrum that captures the spectral information of the large (small) frequencies from the high (low) resolution data spectrum. Specifically, the proposed methodology is applied through the following steps: (1) acquisition of the latest HR and LR radar frames; (2) application of the downscaling and merging procedure to combine the large-scale precipitation patterns detected in LR data with the statistical properties of the smaller scales emerged in HR observation; and (3) run of nowcasting models to produce forecasts for different lead times. The effectiveness of the presented approach is then assessed by comparing the corresponding forecast performance evaluated for the study area through several metrics with that obtained by applying the nowcasting models only to HR data.
更多查看译文
AI 理解论文
溯源树
样例
![](https://originalfileserver.aminer.cn/sys/aminer/pubs/mrt_preview.jpeg)
生成溯源树,研究论文发展脉络
Chat Paper
正在生成论文摘要