Calibration of a one-way coupled atmospheric-hydrologic model for the study of extreme events over Cyprus

18:00 Tuesday 28 May




Corrado Camera (Italy) 1; Georgios Zittis (Cyprus) 2; Adriana Bruggeman (Cyprus) 2; Ioannis Sofokleous (Cyprus) 2

1 - Dipartimento di Scienze della Terra - Università degli Studi di Milano; 2 - Energy Environment and Water Research Center - The Cyprus Institute

Coupled atmospheric-hydrologic simulations are a very promising tool to investigate hydrologic processes and manage water resources under climate change. The WRF hydrological extension (WRF-Hydro) represents a complete model environment for the analysis of atmospheric-hydrologic processes. The study’s objectives were to calibrate the one-way coupled WRF-hydro model for simulating extreme events in Cyprus from observed precipitation, and evaluate the model performance when forced with WRF downscaled (1 x 1 km2) re-analysis precipitation data. The focus was on January 1989 and November 1994. For both events, 15 days were modelled over 22 watersheds. Both observed and WRF-modelled precipitation were fed to the model with a 1-hour time step. The model version adopted allows spatially distributed values for model factors controlling rainfall partitioning, saturated hydraulic conductivity, and deep drainage. Optional routines (lake/reservoir and baseflow) were activated too. Calibration was performed manually, following a trial and error procedure, on four parameters (the rainfall partitioning coefficient, the reference hydraulic conductivity parameter, the retention-depth routing factor, and the overland-roughness routing factor).

The performance of WRF-Hydro was evaluated on daily streamflow. The factor controlling rainfall partitioning (REFKDT) was found to be the most sensitive parameter. In five watersheds (Xeros, Peristerona, Akaki, Agios Onoufrious and Pedieos), Nash-Sutcliffe Effieciency Coefficient (NSE) values larger than 0.4 were obtained for both events. However, REFKDT values leading to the best NSEs were usually different, for the same watershed, under the wet conditions of January 1989 and the dry conditions of November 1994. The WRF-modelled rainfall, evaluated on daily data, showed a NSE of 0.71 and 0.42 for the events of 1989 and 1994, respectively. Nevertheless, simulations of the two events with the calibrated WRF-Hydro and the modelled rainfall returned negative NSE values on all watersheds. However, the streamflow volumes for simulations forced with the observed and with the modelled rainfall are comparable. This result indicates that the modelled chain is reliable and could be used to predict streamflow resulting from modelled future extreme rainfall events.