Regional Analysis of Low Flow Using L-Moments for The Firat and Dicle Basins in Turkey Using L-Moments Approach and Plotting-Position Method
Thesis Type: Postgraduate
Institution Of The Thesis: Cukurova University, Fen Bilimleri Enstitüsü, İnşaat Mühendisliği, Turkey
Approval Date: 2024
Thesis Language: English
Student: MOHAMMAD MOSTO
Supervisor: Recep Yurtal
Open Archive Collection: AVESIS Open Access Collection
Abstract:
Analyzing low flows is a new study in water resources engineering, especially with the occurrence of hydrological droughts over the last 20 years. The frequency analysis method is regularly used in estimating droughts and possibilities of low currents by finding a suitable distribution of recorded data about river flows.
This study aimed to determine regional frequency analysis for two catchment areas in Turkey; the first one was the Firat Basin, which contains about 20 rivers. The second one was the Dicle Basin, which has about 15 rivers. In both areas, two methods were applied for getting the L-moments to depend on the unbiased and plotting position methods. The heterogeneity measures, computed by carrying out 500 simulations using the data, showed that the two basins were identified as heterogeneous. Therefore, Firat basin was divided into three sub-regions and Dicle basin was divided into two sub-regions. It was also found that Gen. Logistic distribution was the most appropriate for most Firat and all Dicle basins.
Unbiased estimators demonstrated a slight advantage over plotting position estimators for quantile estimates in the Dicle and Firat basins, as reflected by RMSE values. Specifically, the average RMSE for the Firat region using unbiased estimators was 0.69, compared to 0.77 for plotting position estimators. The average RMSE for the Dicle region using unbiased estimators was 0.87, compared to 0.92 for plotting position estimators. This advantage of using Unbiased estimator was more evident in the upper quantile. However, due to being easier to compute the parameters depending on plotting position estimators, it may require further investigation to establish a critical threshold where they become preferable to unbiased estimators in larger sample studies.