Determination of flood-inducing rainfall and runoff for highly urbanized area based on high-resolution radar-gauge composite rainfall data and flooded area GIS data

Abstract

This study derived the Flood-Inducing-Rainfall (FIR) and Flood-Inducing-Runoff (FIRO) for urban flood warning. For this, we derived a relationship between the watershed time of concentration and accumulated rainfall depth for 261 flood events at 239 watersheds of Seoul with areas between 0.5 and 6.8 km2 during the years 2010 and 2011, based on 10-minute 1β × 1 km radar-gauge composite rainfall field. This relationship was further categorized by discrete ranges of the proportion of the flooded area in the watershed (FP) and coefficient of variation (CV) of the rainfall time series. Then, minimum possible rainfall intensity calculated for each of the classified depth-duration relationship was determined as FIR for the specified range of FP and CV. A similar methodology was applied to derive FIRO, which used runoff depths that were estimated using the NRCS Curve Number method. We found that the FIR and FIRO varied between 37–62 and 10–41 mm/h, respectively. FIR and FIRO increase with an increase in FP, suggesting that greater rainfall causes larger flooded area. As rainfall CV increases, FIR and FIRO decrease, suggesting that the temporally concentrated rainfall requires less rainfall to cause the flood. A flood warning system that checks the accumulated rainfall at each of the 239 Seoul watersheds every 10 min was developed to verify the FIR and FIRO against the 21 flood events that occurred between 2012 and 2015. When flooding was tolerated on 5 percent of the watershed area, the ratios of hit, miss, and false alarm of the warning system based on the accumulated rainfall was 44.2, 9.5, and 55.8%, respectively. The ratios of hit, miss, and false alarm of the warning system based on the runoff was 66.7, 4.7, and 33.3% respectively. The flood warning system based on gauge-only or radar-only data gives approximately 3 times as much false warnings compared with the system based on radar-gauge composite data. Also, the warning system that does not consider rainfall temporal variability gave approximately twice as much false warnings as the system that considered temporal variability.