Mean Field Bias Correction to Radar QPE as Input to Flood Modeling for Malaysian River Basins

Abstract

The occurrence of unprecedented flood events has increased in Malaysia recently. To mitigate the impact of the disaster, the National Flood Forecasting and Warning Centre has endeavored to improve the warning system to produce more accurate and reliable early warning to the public. The paper describes the use of radar composites from the radar network in Peninsular Malaysia to produce quantitative precipitation estimates (QPE) as input to the flood model. The processing of the raw radar data and the conversion of rain rate are described. The comparison between radar QPE and gauge rainfall shows that radar QPE underestimates the gauge rainfall, and the results are better at the western parts of Peninsular Malaysia compared to the eastern parts of Peninsular Malaysia. The comparison between Marshall Palmer (MP) and Rosenfeld (RF) conversion equations shows that there is not much difference in performance between the two equations. Both underestimate the rainfall, although RF estimates higher radar QPE for high rainfall intensity. The underestimated radar QPE is improved by calibration process via the Mean Field Bias (MFB) correction technique. The study introduced zoning into smaller regions for the MFB factors derivation. Results indicated that the radar QPE is much improved after the calibration process. Simulation of flood event in December 2021 for the case study of Langat River basin indicates the improvement of correlation coefficient from 0.67 to 0.99 after the calibration process via MFB for smaller zones.