Ensemble rainfall-runoff forecasting based on different physical parameterization schemes for small and medium catchments
- Article
- Figures
- Metrics
- Preview PDF
- Reference
- Related
- Cited by
- Materials
Abstract:
The ensemble method can efficiently reduce the uncertainty of rainfall forecast.The ensemble rainfall forecast is established through random disturbance of the initial field.Different numerical weather prediction models are used to form the forecasting ensemble.The ensemble rainfall forecast is build based on a combination of different physical parameterization schemes,which is regularly used for rainfall forecasting under unknown weather conditions.The selection of the physical parameterization scheme has a significant impact on forecasting results.A single physical parameterization scheme is difficult to adapt to different rainfall processes,which brings great uncertainty in the forecast.Ensemble forecast based on physical parameterization scheme can effectively reduce the uncertainty of rainfall forecast,which can provide reliable rainfall information for flood forecast. Thirty-six physical parameterization schemes based on the WRF model are used to establish the ensemble rainfall forecast.The relative error (ER),critical success index (ICS),and the root mean square error (ERMS) are used to comprehensively evaluate the rainfall forecast.Meixi distributed hydrological model is constructed based on China flash flood hydrological model (CNFF-HM).The peak flood discharge error,peak present time error,and Nash efficiency coefficient are used to evaluate the flood forecast.The coupled meteorological and hydrological system is formed by the WRF model and Meixi distributed hydrological model.The research also uses a statistical model that is developed based on the heteroscedastic extended Logistic algorithm to postprocess the rainfall ensemble forecast results. For rainfall storms caused by Saola typhoon,the ERs based on 36 schemes are between 0.88% and 21.00%.In spatial dimension,the ICSs are between 0.736 8 and 0.758 2,and the ERMSs are between 0.133 1 and 0.221 6.In the time dimension,the ICSs are both 0.687 5 and the ERMSs are between 0.592 4 and 0.760 0,respectively.The error of peak flow discharge based on coupled meteorological and hydrological systems is 11.3%.With rainfall forecasting postprocess,the error of peak flow discharge is 3.97%.Likewise,for rainfall storms caused by Hagibis typhoon,the ERs based on 36 schemes are between 24.32% and 68.51%.In spatial dimension,the ICSs are between 0.347 0 and 0.487 9,and the ERMSs are between 0.521 6 and 0.845 1.In the time dimension,the ICSs are between 0.329 2 and 0.435 6,and the ERMSs are between 1.300 1 and 1.634 9,respectively.The error of peak flow discharge based on coupled [JP2]meteorological and hydrological systems is -86.89%.With rainfallforecasting post-process,the error of peak flow discharge is -48.95%. Forecasted rainfall in spatial dimension performs better than that in time dimension with different physical parameterizations schemes.The ensemble rainfall forecast is appropriately used for flood forecast with the coupled meteorological and hydrological system,which can efficiently reduce the forecasting uncertainty.Reasonable postprocessing methods should be used to process the numerical rainfall forecast.For the rainfall with even spatiotemporal distribution,flood forecast with the coupled meteorological and hydrological system has certain advantages compared to flood forecast based on observed rainfall.For the rainfall with uneven spatiotemporal distribution,the forecast still has room for improvement.
Keywords:
Project Supported:
Clc Number:
Mobile website
® 2024 All Rights Reserved
冀公网安备13010502002656号
冀公网安备13010502002656号