Pattern evaluation of operational radar reflectivity–rainfall relationship for Quitandinha River flooding events: Petrópolis, Rio de Janeiro (Brazil)

Abstract

Real-time rainfall distribution is required for decision-making based on the support of warning systems, especially in situations of imminent floodings and landslides. In this work, six relationships on radar reflectivity–rainfall relationships were calibrated and evaluated for 33 rainfall events associated with Quitandinha River flooding with occurred between 2013 and 2016. These events were chosen with the purpose of finding new relationships that could characterize with better accuracy the properties of storms and precipitation associated with Quitandinha floods. The procedure of applying variations of 5%, 10%, and 20% in reflectivity band-pass filter was used to minimize the different types of uncertainties associated with the measurement of radar reflectivity and rain gauges. Multiple linear regression analysis approach showed that radar reflectivity–rainfall relationships which were calibrated at 20% presented the lowest dimensionless coefficient of variability (CV), systematic error (BIAS), mean absolute deviation MAD), and root-mean-square error (RMSE). The calibrated expression between Marshall–Palmer and Nexrad showed the best performance among the six relationships examined.