How can statistics analysis for high-resolution rainfall fields help future urban flood forecasting for multiple purposes in poor-gauged areas?

Código

I-EBHE0148

Autores

Mateo Hernández Sánchez, Marcos Roberto Benso, Luis Miguel Castillo Rápalo, Eduardo Mário Mendiondo

Tema

WG 1.03: Urban Water - Urbanization phenomenon & adequate water management

Resumo

There has been an increase in the frequency and intensity of rainfall extreme events, increasing hazards in urban areas such as flash floods and landslides. Disaster risk management requires flood modeling to assess the potential impacts of floods and to promote strategies for reducing social, economic, and environmental risks. Furthermore, high temporal resolution rainfall datasets are crucial for addressing current and future flood risks under increasing environmental and climate changes. Despite the lack of data availability that affects least developed countries (LDCs), proposing reproducible and reliable protocols for data quality control is still a bottleneck for practitioners and researchers. This paper presents a research project to address the importance of data reliability for urban hydrology and presents an automatic quality control procedure for sub-hourly rainfall data. A case study was conducted in the Aricanduva river basin of 100 km² of drainage area, located in the São Paulo city, Brazil. With a combination of high density of impervious areas and medium-to-high slope portions, the downstream zone faces flash floods problems. This watershed is monitored with 5 rainfall gauges within and 14 in the vicinity with a distance less than 5 km from the watershed. The rainfall gauges have, on average, 5 years of information within a registration period of 9 years (2012?2021) provided by the rain gauge network of the Flood Warning System of São Paulo-SAISP. The quality control procedure presented in this study is structured by two steps: (i) for each independent gauge station, a 1-hour filtering of spurious rainfall peaks, with long periods of missing data, clogging and long rainless periods, were used as a data analysis to assess possible equipment malfunctions affecting the overall data accuracy, giving inaccurate measurements of rainfall patterns and difficulties in the identification of trend; (ii) analysis of the consistency with neighbor gauge stations through double-mass analysis and time series analysis to identify patterns, trends, and anomalies in the study area. After the application of the proposed procedure, it is expected to identify the station's periods with faults and assess the performance of the filters applied in the procedure. The proposed study can contribute for future research, due to it could be applied in typical poorly gauge urban conditions, since this type of data status is commonly found in underdeveloped and developing countries. It identifies and provides longer periods of rainfall records to be employed in the calibration of hydrological and hydraulic couplets models. This, in turn, facilitates more accurate analyses of extreme events, improves strategies for flash floods risk reduction, and supports better assessments of runoff water quality . Furthermore, the refined data and methodologies enable the development of Digital Twin models that could integrate Early Warning Systems (EWS), advanced flood control strategies based on rainfall forecasting information, and innovate runoff water reuse systems. These integrated approaches can significantly advance urban water management and disaster risk reduction in challenging data environments.