TIGGE: Medium range multi model weather forecast ensembles in flood forecasting (a case study).

Title
TIGGE: Medium range multi model weather forecast ensembles in flood forecasting (a case study).
Technical memorandum
Date Published
01/2008
Secondary Title
ECMWF Technical Memoranda
Number
557
Author
Florian Pappenberger
J. Bartholmes
J. Thielen
E. Anghel
Publisher
ECMWF
Abstract The performance of a hydrological multi model flood forecast system with inputs provided by seven ensemble prediction systems of the THORPEX Interactive Grand Global Ensemble (TIGGE) archive is evaluated. A flood forecast of this multi-model ensemble has been computed for case study the October 2007 floods in Romania using the LISFLOOD model of the European Flood Forecasting System (EFAS) as hydrological component. None of the forecast centres (ECMWF, UKMO, JMA, NCEP, CMA, CMC, BOM and simple multi-model ensemble) predict the distribution of precipitation observations exactly, with most centres exhibiting an over prediction at day nine. The distribution of discharge predictions for observations and forecasts is similar at the lower flow. The percentage of precipitation and discharge forecasts above and below the 10th and 90th percentile of the predicted EPS distributions is very high. The UKMO, ECMWF and the multi-model ensemble forecast perform favourably in terms of root mean squared error of the ensemble mean discharge and precipitation predictions. All forecasts (apart from the one issued by BOM) would have lead to a correct flood warning about 8 days in advance. It can be demonstrates that the Multi-model ensemble has the best average properties, followed by the forecast of ECMWF and UKMO. The increased quality achieved by the multi-model ensemble is explained by the fact that it provides a better approximation at the tails of the distribution. All analysis is based on a set of criteria specific for this case study and might be different in more general cases.
URL https://www.ecmwf.int/en/elibrary/75947-tigge-medium-range-multi-model-weather-forecast-ensembles-flood-forecasting-case
DOI 10.21957/sa3bkjdiy
Download citation