Developing hourly rainfall maps from open meteorological data sources to analyze the spatial extent and temporal evolution of flash flood events

Abstract

Research on flash floods and related geological hazards has often focused on spatial hazard assessment. However, studies on the temporal evolution of such disasters remain limited, primarily due to the lack of hourly rainfall data. Conventional rain gauge stations fail to capture the full spatial extent of disaster impacts and are typically unevenly distributed, especially in mountainous regions. This paper proposes a novel approach using the R programming language to retrieve, process, and calibrate open-access meteorological datasets in order to generate historical hourly rainfall maps. The proposed method is tested on three extreme rainfall events in Northern mountainous Vietnam: (1) the flash flood on August 1, 2025 in Điện Biên and Sơn La; (2) the flash flood on May 18, 2025 in Tuyên Quang and Bắc Kạn; and (3) the flash flood on August 5, 2023 in Yên Bái and Lai Châu. Results indicate that both the intensity and spatial distribution of hourly rainfall change rapidly. Peak hourly rainfall intensities often exceed 60 mm, and the locations experiencing such rainfall levels are typically those suffering the most severe damage. Notably, the spatial extent of flash flood impacts generally overlaps with areas where 24-hour accumulated rainfall exceeds 200 mm. The workflow for generating hourly rainfall evolution maps provides a reliable basis for delineating disaster-affected areas in alignment with actual damage observations. It also serves as a foundation for developing data-driven tools to establish quantitative thresholds for flash flood early warning systems