Determination of trophic level of Melen reservoir by uncertainty analysis

The causes of water pollution depend on various anthropogenic and natural factors, and in recent years various strategies have been developed for this issue. In this context, water management has been integrated with land management in order to make correct management decisions. One of the water pollution problems, which is of utmost importance, is eutrophication. The reason of eutrophication is the excess nutrients that reach to the water body from the watershed, as point and non-point sources. In order to estimate the nutrients in water, it is necessary to use watershed models and water quality models together. These administrative strategies are being implemented to remove or reduce the eutrophication problem in the reservoirs. This is achieved by the control of point and diffuse pollutants, by estimating the effects of these pollutant sources by mathematical modeling before taking measures. New and integrated approaches to water resources management (WRM) should be provided, as well as participation of sectors and public. While the WRM is concerned with a variety of complex issues, it must take into account the uncertainties that may arise in modeling or in data processing and should take these uncertainties into account in the management process. In this study, the significance of uncertainty analysis in water quality models was mentioned in the water management process and trophic level of Melen Reservoir, which is under construction, was determined using CE-QAUL-W2 water quality model by performing uncertainty analysis. In this context, the CE-QUAL-W2 model was considered as the appropriate water quality model for Melen Reservoir. The model was structured by taking into account the dam characteristics. In addition, the data obtained from the dam characteristics were entered in the model to define the dam outlet structures. Water bodies, model branches, segments and layers are defined in the development of CE-QUAL-W2 model. The model was defined as one water body (as Melen Resevoir), 4 model branches, 68 segments, and 101 layers. Layers were arranged to be one meter depth. The deepest part of the model is the 64th segment, and at the same time this segment is defined as the output segment of Melen Reservoir. In the application of the model, for initial conditions, bathymetry, water level, temperature and water quality components, boundary conditions, inputs / outputs, temperature, water quality and meteorology data are needed. These data are converted into the format requested by the model, and they are defined to the model. CE-QUAL-W2 model was used for Melen Resevoir, WASP8 model was used for the streams, and SWAT model was used as the watershed model for the representation of the watershed system. These models were linked, and in this linking process SWAT and WASP8 outputs were defined as input to CE-QUAL-W2 model. Calibration and verification processes for the model have not been performed since the reservoir is under construction; thus, no monitoring data is available yet. For this reason, sensitivity and uncertainty analyzes have been introduced for the representation of the watershed system. Parameters and intervals for the application of sensitivity analysis of these parameters were selected with the information gathered from the literature. In addition, default values for each parameter were entered to the model. When each of the selected kinetic parameters was changed, the model was run with the default values of the remaining parameters, and this process is repeated with the values of the specified intervals of each parameter. For this process, 90 model runs were conducted, and the number of selected kinetic parameters was 9. These kinetic parameters are algal growth (AG), algal respiration (AR), algal mortality (AM), algal excretion (AE), algal settling (AS), ammonium decay (NH4DK), nitrate decay (NO3DK), algal half saturation constant for nitrogen (AHSN) and algal half saturation constant for phosphorus (AHSP). A script has been developed that generates model folders for sensitivity and uncertainty analysis and runs the CE-QUAL-W2 executable in these folders without leaving the work to the user. In addition, another script has been developed that takes model result files in folders and plots the graphs of these results with boxplots. These scripts did not only disburden the workload of the user in running the model, but also facilitated the data analysis by eliminating the complexity of the two-dimensional model results. In the evaluation of the results, the model results of the 64th model segment were used for sensitivity and uncertainty analysis. In addition, while the interpretations performed in the sensitivity analysis were made on the median values, 95% confidence level values were used in the evaluation of the uncertainty analysis. With the help of the developed scripts, the models were run, model results were plotted, and the sensistive parameters among the selected ones and their sensitivity intervals were determined. As a result of sensitivity analysis, it was determined that the model results were sensitive to AG (algal growth), AR (algal respiration), AM (algal mortality), AS (algal settling), NO3DK (nitrate decay), AHSN (algal half saturation constant for nitrogen) and AHSP (algal half saturation constant for phosphorus). However, AE and NH4DK parameters have not been found to have high effects on model results. Uncertainty analysis was performed at the intervals determined previously for the seven kinetic parameters that were found to be sensitive to the model results. The remaining two kinetic parameters (AE and NH4DK) were not included to uncertainty analysis simulations. For the analysis of uncertainty, hundreds of simulations were performed with randomly selected values of the sensitive intervals of the parameters. A separate code has been developed for determining the randomly selected values for the uncertainty analysis. The values obtained by this code were also converted to the desired format as '.csv' file, and provided as input to the above mentioned multiple CE-QUAL-W2 executable code. Then, for the uncertainty analysis with multiple CE-QUAL-W2 executable code, randomly generated kinetic parameter values were defined in the model control file and the models were run. Codes were used in plotting the graphs of the results obtained from 500 simulations performed for the uncertainty analysis to determine the trophic level of Melen Reservoir. These graphs were generated for values of 5% and 95% confidence intervals for each parameter, and the outliers were removed from the graph. As a result of the water quality modeling study of Melen Reservoir it was estimated that, Algae, DO (dissolved oxygen), TN and TP concentrations will be between 0-5 mg/L, 6.1-13.85 mg/L, 0.125-2.05 mg/L ve 0.021-0.132 mg/L, respectively for 5% -95% confidence intervals. The evaluations caried out from the generated graphs and over the 95% confidence interval values for TN, TP and Agea and %5 confidence interval values for DO of these graphs. The graphs for the uncertainty analysis were generated for algae, dissolved oxygen, total nitrogen and total phosphorus parameters which are used for determining trophic levels. The results for the 95% confidence interval gathered from 500 simulations found as 5000 μg / L for Alg, 13.85 mg / L for DO, 2050 μg / L for TN and 132 μg / L for TP and 5% confidence interval value as 6,1 mg/L for DO. The results for the 5% - 95% confidence interval show that the reservoir has a hypertrophic level (values) for TP, TN, Chlorophyll-a, and secchi disc values, while DO values meet mezotrophic levels (values) when compared to surface water quality regulation. It was determined that the trophic level of Melen Reservoir would be at the hypertrophic level for the specified confidence interval. As a result of these evaluations, it is suggested that the point and diffuse loads should be reduced, which come from watershed to Melen Reservoir, by taking into account that it will be at the hypertrophic level when the construction is completed. When the construction is completed monitoring study in the reservoir should be conducted and the model developed in this study should be calibrated and verified with the observation data in order to make more robust estimations.