IWHR

Main Participants: LIU Ronghua, WU Zebin, TIAN Jiyang, KAN Guangyuan, SUN Chaoxing, DOU Yanhong, LIU Qi, ZHANG Xiaolei, ZHAI Xiaoyan, LIU Xiao

1.1 Background

The forecasting and early warning of torrential rain and flash flood serve as the primary technical means for the prevention of flash flood disasters, and also represent an international technical challenge. At present, China has basically achieved two methods for flash flood disaster prevention: meteorological risk early warning and real-time monitoring early warning. The meteorological risk early warning offers a long forecast lead time but suffers from low accuracy, while the real-time monitoring early warning boasts higher accuracy but lacks sufficient lead time. In order to achieve the "effective unification of lead time and accuracy of flash flood early warning" and scientifically guide the evacuation and risk avoidance of populations in mountainous regions, the research team has been engaged in the research on heavy rainfall monitoring, forecasting, and imminent early warning of flash flood since 2018, and has achieved notable results. The related findings have been put into practical use on the national flash flood disaster monitoring, forecasting and early warning platform, as well as provincial platforms in Fujian, Shaanxi and other provinces. These achievements provide robust technical support for the implementation of the national flash flood disaster prevention projects, and play a vital role in enabling the Ministry of Water Resources to grasp the risk situation of flash flood disasters across the country during the flood season, and to supervise and guide local efforts in flash flood disaster prevention.

1.2 Contents

· Developing a multi-source rainfall fusion method incorporating satellite, radar and ground station data, with consideration of the localized characteristics of rainfall.

· Developing a nowcasting technology that is suitable for different regions across the country, different radar frequency bands, and different types of rainfall.

· Developing a soil moisture simulation technology based on the water-heat balance theory and coupled equations.

· Developing a dynamic rainfall warning indicator analysis method and establishing a technical system of flash flood disaster nowcasting and early warning.

1.3 Achievements

· Having proposed a multi-source rainfall fusion approach that integrates "global products to supplement monitoring blind spots, rainfall radar to capture rainband trends, and ground stations to correct rainfall errors", developed a multi-source rainfall fusion method considering localized rainfall characteristics, and created "space-air-ground" three-dimensional rainfall monitoring products that cover mountainous regions across the country.

· Having proposed a rainfall spectrum decomposition method and developed a nowcasting technology for predicting rainfall in mountainous regions across the country within the next two hours, based on spectrum decomposition techniques and deep learning, thereby improving the accuracy of rainfall nowcasting by 20%.

· Having established a set of bidirectional coupled equations for the water and heat balance processes, proposed a quantification scheme for key physical quantities, and developed a dual-time-step finite volume spatial-temporal discretization method, improving the simulation accuracy by 10%.

· Having proposed a forward trial calculation method for dynamic rainfall warning indicators that considers soil moisture changes and spatial-temporal heterogeneity of rainfall, and established a sample database of dynamic early warning indicators for villages and a surrogate model based on random forests, increasing the forecast and early warning hit rate to over 75%.

1.4 Application

The achievements have been applied by the Ministry of Water Resources, 29 provinces and the Xinjiang Production and Construction Corps, and 2,076 county-level organizations, benefiting more than 300 million people. From 2021 to 2024, the national level issued 550 nowcasting and early warnings, while local authorities issued 5,775 forecasts and early warnings, and sent over 4.203 million early warning text messages. As a result, 370,000 people were evacuated, significantly enhancing the level of flash flood disaster early warning, strongly supporting the decision-making and deployment for flash flood disaster prevention efforts, and yielding substantial benefits in disaster prevention and mitigation.

Figure 1 National High-resolution Three-dimensional Rainfall Monitoring and Fusion Technology

Figure 2 Rainfall Nowcasting Technology

Figure 3 Soil Moisture Field Construction Technology

Figure 4 Dynamic Early Warning Indicator Analysis Technology

Key Technologies and Application of Multi-source Rainfall Monitoring, Forecasting, and Imminent Early Warning of Flash Flood