Using reanalysis precipitation data for developing intensity-duration-frequency curves in a poorly gauged city

Abstract

Study region: The Msimbazi River catchment traversing Dar es Salaam city in east-central Tanzania. Study focus: The lack of high-resolution rainfall data in cities in developing countries hinders the development of suitable intensity–duration–frequency (IDF) curves for designing and evaluating hydraulic structures. Thus, this study investigated the potential of relatively high-resolution reanalysis precipitation data for developing IDF curves in the poorly gauged Msimbazi River catchment in Dar es Salaam city. The time series of hourly ERA5-Land reanalysis precipitation data were used to generate the annual maximum series at four selected points in the river catchment. These were subsequently bias-corrected using parameters derived from limited observed rainfall data. The bias-corrected annual maximum series of reanalysis precipitation (AMSRP) data were fitted with the best probability distribution functions, which were then used to estimate quantiles of IDF curves for various annual exceedance probabilities. New hydrological insights: The findings revealed that uncorrected AMSRP data would have underestimated rainfall intensities in this region by a factor of two to six. Different areas in the Msimbazi River catchment have different rainfall distributions and IDF curves. Thus, no single set of IDF curves can represent rainfall intensities in the entire river catchment. Due to the potential of relatively high-resolution reanalysis precipitation data, designs and evaluations of hydraulic structures in poorly gauged cities are encouraged to use bias-corrected and location-specific IDF curves.

Keywords

• Annual maximum series
• Bias correction
• Bootstrap mean
• ERA5-Land reanalysis
• Msimbazi River catchment