Katipoğlu, Okan Mert; Sarıgöl, Metin

Accurate and reliable peak discharge prediction is of great importance in water resource management and flood control studies. The aim of this study was to predict monthly peak discharge by combining various signal decomposition processes and machine learning models. For this purpose, monthly peak discharge data were decomposed into subsignals utilizing Daubechies 3, Coiflet 5, discrete Meyer main wavelets, and empirical mode decomposition methods, which are the ones most commonly used in hydrological studies. The separated signals were subjected to correlation analysis, and the sum of the highly correlated signals was presented as input to the machine learning models. Support vector machines, regression trees, ensemble trees, and adaptive neuro-fuzzy inference system models were used for peak discharge forecasting. The performance of the established models was evaluated with the help of statistical indicators such as mean absolute error, root mean squared error, and determination coefficient, and graphically through Taylor diagrams. At the end of the study, the most effective results were obtained with the hybrid model established by the Daubechies 3 wavelet-4 decomposition levels and combined with the coarse Gaussian support vector machine.

Akar, Funda; Katipoğlu, Okan Mert; Yeşilyurt, Sefa Nur; Han Taş, Mehmet Bilge

In this study, Extreme Gradient Boosting (XGBoost), Gradient Boosting Machine (GBM), Random Forest (RF), Bagged Trees (BT), and Custom Deep Learning methods were used to estimate the potential evapotranspiration (PET) values at Diyarbakir airport station in the Tigris basin. In establishing the models, the average temperature, maximum temperature, minimum temperature, maximum wind speed, relative humidity, average wind speed, and total precipitation values in the monthly time period were chosen as inputs, and PET values were used as output. The data set is divided into 70% training and 30% testing. 10-fold cross-validation to avoid overfitting problems. Training and test data were randomly selected. The prediction performances of the models were evaluated according to the statistical criteria of determination coefficient (R2), root mean square error (RMSE), mean absolute error (MAE), and rank analysis. The best PET estimates were obtained using the inputs of mean, min, maximum temperature, relative humidity, total precipitation, average, and maximum wind speed. It was also concluded that XGBoost was the highest performance. When the R2 values were examined, it was seen that the Deep Learning model had higher performance. But for RMSE and MAE, XGBoost did better. As a result of the rank analysis, it was seen that XGBoost got a higher score.

Sarigöl, Metin; Yesilyurt, Sefa Nur

Flood routing analysis is of great importance in predicting floods and taking all necessary precautions in the area where the flood occurred. By applying different machine learning algorithms to historical flood data, the success of the established models in flood routing calculations has been measured. In this study, flood hydrograph data on 05.05.2014 was passed through the training and testing stages using the Support Vector Machine (SVM), Gaussian Process Regression (GPR), Regression Tree Ensembles (RTE), and Artificial Neural Networks (ANN) methods, then routing calculations were made by applying the flood data on 03.06.2015 to these models. The results were compared with either ANN, SVM, GPR, RTE, or measured values. Root Mean Square Errors (RMSE) and Correlation Coefficient (R) values were calculated at the end of the comparative analysis. In conclusion, it has been determined that verifying the flood routing results performed by the SVM is the best result.

Kurt, Zeynep Ö.; Kankal, Murat; Yüksek, Ömer

The aim of this study is to investigate the effect of building-based assessment on flood hazards. Değirmendere, which is one of the most important basins of the Eastern Black Sea Region of Turkey, was selected as the study area. Flood hazards for discharge values with different return periods of this region were found using; the damage percentages table recommended by the Huntington Civil Engineers Association (HCEA), the Van Eck and Kok depth-loss curves, and the equation obtained by Pistrika and Jonkman. The calculations were performed both for building-based and for the region-based and compared. Furthermore, the damage rates obtained for each building on a building-based were graded and shown on the map. It was determined that the damage on individual building-based estimation for Q1,000 calculated more than to the one on region-based 154% and 17% for the damage percentages table recommended by the HCEA, Van Eck and Kok depth-loss curves, respectively. Otherwise, it gave 11% less damage estimation for the equation obtained by Pistrika and Jonkman. Furthermore, it was concluded that the increase in damage from Q50 to Q100, from Q100 to Q500, and from Q500 to Q1,000 was between 16%-30% according to building-based damage estimation.

Ertugay, Neşe; Acar, Filiz Nuran

This study investigated color and COD removal of azo dye Direct Blue 71 (DB71) by a Fenton-like process in zero-valent iron (Fe0 = ZVI). The UV–vis spectra of DB71 exhibit three main absorption bands – two in the UV region and one in the visible region. UV–vis spectra showed that the azo linkages in the molecule structure of the dyes were destroyed during the decolorization. The influence of various operational parameters such as pH, ZVI dosages, hydrogen peroxide (H2 O2) concentrations and temperature are studied to determine the optimum conditions for color and COD removal. pH is highly effective parameter of decolorization and COD destruction in this work. However, H2 O2 concentration and ZVI dosages were not only effective in the removal of COD for 100 mgL-1 initial DB71 concentration. Therefore, the optimum ZVI:H2 O2 ratio is based on determining the value of COD removal. Optimum conditions were accepted to be pH of 2.5, ZVI:H2 O2 ratio 4:1 and 20ºC temperature. The result indicated that 100% of decolorization efficiency was achieved within 20 min in the range of 150-400 mg/L initial DB71 concentrations by using optimal conditions. Residual COD values were 15.38, 22.35, 22.58, 27.04, 31.86 mg/L for 150, 200, 250, 300, 400 mg/L initial DB71 concentrations at the end of reaction time, respectively. Additionally, increasing the reaction temperature from 20 to 40ºC negatively affected the decolorization and degradation rate DB71. The experimental kinetic data are fitted very well the Behnajady–Modirshahla–Ghanbery model for color removal of DB71 dye.

Osma, Etem; Serin, Memduh; Leblebici, Zeliha; Aksoy, Ahmet

This study measured levels of cadmium, chromium, copper, nickel, lead, and zinc in fruits of Capsicum annuum L., Phaseolus vulgaris L., and Solanum melongena L. during the 2007 vegetation period, to determine levels of metal pollution in Istanbul Province. Plant and soil samples were collected from six sites in Istanbul (Brook Coast area, inner city, industrial area, suburban, roadside, and rural - control - areas). Unwashed and washed leaf and soil samples were analyzed using inductively coupled plasma optical emission spectrometry. The highest and lowest values were as follows: Cd (0.28-0.89 μg·g-1), Cr (5.33-14.04 μg·g-1), Cu (1.47-5.19 μg·g-1), Ni (3.06-13.65 μg·g-1), Pb (29.28-86.20 μg·g-1), and Zn (3.70-5.74 μg·g-1). The unwashed samples were more contaminated than the washed samples. Contamination was higher in the vegetables grown in industrial areas and along roadsides. The overall metal concentration pattern in vegetables was Pb > Cr > Ni > Zn > Cu > Cd.