Utilization of numerical weather prediction models for flood planning and operational studies in datascarce basins


Thesis Type: Doctorate

Institution Of The Thesis: Middle East Technical University, Turkey

Approval Date: 2020

Student: Okan Çağrı Bozkurt

Co-Supervisor: SEVDA ZUHAL AKYÜREK, NURİ MERZİ

Abstract:

With the increasing rates of urbanization, floods have become a more critical problem day by day. Consequently, the perception arises that the flood control structures have become insufficient. Well-calibrated weather forecasting and hydrological models are important complementary tools to reduce the damage caused by floods. Such applications can yield results as useful as structural measures. In this study, the effects of the use of numerical weather forecasting and hydrological models on structural and non-structural based flood protection activities were examined. Lake Mogan basin in Ankara is chosen as the study area. In the analyses, the outputs of 2 different numerical weather forecast models (Weather Research Forecasting (WRF) and Transition Step Between ALadin and AROme (ALARO, ALR), which are used operationally by the Turkish State General Directorate of Meteorology (Meteoroloji Genel Müdürlüğü, MGM), were taken from the MGM for 6 different dates. The accuracy of the precipitation forecasts was measured by various statistical methods on a point and areal basis. The results obtained showed that the areal based forecasts give better results than the point-based forecasts. However, it has been observed that the parameters that the MGM uses to run the WRF and ALR numerical weather models are not successful in predicting the convectional precipitation in this basin. There is no stream gauge station with a long-term measurement in the study area. Besides, in recent years, water transfer to the Mogan Lake basin has been provided by pipelines outside the basin, the exact amount and time of which cannot be determined precisely. For these reasons, the hydrological model constructed for the basin could be calibrated on an event-based basis with water level measurement values of Lake Mogan. Flood operation simulations have been carried out in 2 main scenarios. In the first scenario, it was assumed that Mogan Detention Pond, located upstream of Mogan Lake, is in operation, and the water level of Lake Mogan is at current operating values (973.42 m., value on 14.03.2016). In the second scenario, it was assumed that the detention pond is not built, and the normal operating level of the Lake Mogan is 972.00 m. as given in the planning reports of the State Hydraulic Works (Devlet Su İşleri Genel Müdürlüğü, DSİ). Under these two main scenarios, different cases have been created in which 50-, 100- and 500-year storms can occur, and the capacity of the canal located at the exit of Lake Mogan is taken as either 15 or 30 m3/s. Precipitation forecasts of WRF and ALR models regarding 50, 100, and 500 years design hydrographs were created fictitiously using observed forecasts of the 14.03.2016 event. According to the results of the flood operation studies, it is seen that the 500-year flood reached 974,850 m. in the 1st scenario and 974,909 m. in the 2nd scenario in the most disadvantageous cases. These results are thought to contribute to decision-makers regarding structural and non-structural solution alternatives in flood planning studies.