ABSTRACT. Lake Water Level Estimation with Artificial Intelligence Methods.
With the decrease in water resources due to climate change, dam reservoir level estimation is important in terms of the construction, operation, design and safety of dams. In this study, average air temperature (T), relative humidity (SR), and precipitation (P) parameters were used for lake water level estimation. Thurmond Lake in McCormick County, South Carolina, USA was selected as the study area. 1286 daily data measured in real time between 2017-2023 were used as the study data. M5 Decision Tree (M5 Tree), Adaptive Neuro Fuzzy Inference System (ANFIS) and Artificial Neural Network (ANN) models were selected for lake water level estimation and model results were compared with real observation results. In the comparison of the prediction models, performance criteria such as coefficient of determination (R2), Mean Absolute Error (MAE) and Root Mean Square Error (RMSE) were used. When the model results were examined; it was determined that artificial intelligence methods performed well in predicting the lake water level change.
Keywords: Fuzzy Logic, Modeling, Artificial Intelligence, Multiple Linear Regression, Prediction.
INTRODUCTION
The most crucial element for maintaining human life is water. However, the growing water issues and their solutions have become more significant, particularly in light of the recent global climate change. Because of the significant drop in water consumption, some precautions are now unavoidable. To understand the potential of current water resources and to use them more effectively, precautions must be taken (Kucukerdem, 2019). Estimating how much water enters and exits the system during the design and operation of water structures is a significant task. These data have recently been estimated using well-known artificial intelligence models (Salam, 2018). Numerous studies have been published in the literature that use artificial intelligence techniques to construct prediction models about hydrology and water structures. Terkos Dam Lake's dam reservoir level was calculated by Aydemir (2020) using the Fuzzy Logic. Using water values from 2001 to 2012 in Terkos Dam, he established a modeling mechanism using the Adaptive Neuro Fuzzy Inference System method to investigate the most accurate estimation of future watervalues, lines. (2010) used artificial neural networks to assess the level of the dam reservoir. One artificial intelligence technology, the Artificial Neural Network method, was used by Une§ et al. (2018) to estimate the quantity of evaporation in the Cambridge reservoir basin. Millers Ferry Dam on the Alaba River in the United States was chosen by Une§ et al. (2019) to evaluate the dam reservoir level changes. It has been noted that Adaptive Neuro Fuzzy Inference System models outperform other artificial and classical models. Using data from the General Directorate of State Hydraulic Works (DSI) in the Kumlu district of the Amik Plain of Hatay, Demirci et al. (2017) used the Artificial Neural Network to estimate the earth dam reservoir level. Demirci et al. (2018) used multiple linear regression and Adaptive Neuro Fuzzy Inference System models to estimate the reservoir capacity of the Brook Dam in Massachusetts, USA. Results from multiple linear regression and Adaptive Neuro Fuzzy Inference System were compared.
In this study, M5 Decision Tree (M5 Tree), Adaptive Neuro Fuzzy Inference System (ANFIS) and Artificial Neural Network (ANN) models were developed for lake water level estimation by using the relative humidity, air temperature, precipitation and lagged lake water level data of Thurmond Lake at South Carolina, USA between the years 2017-2023 as input parameters. In addition, model results were compared with each other.
METHODOLOGY
Study area
In this study, 1608 daily meteorological and hydrological data belonging to station 02193900 on Thurmond Lake at South Carolina, USA, between the years
2017-2023 were used. In the study, 80% of all data was allocated for training and 20% for testing. 1286 days of data were used for training and 322 days of measurement data were used for testing. In these model applications, the relative humidity, air temperature, precipitation and lagged lake water level data obtained from the United States Geological Survey (USGS) were used to estimate the amount of lake water level. The study area and lake water level were given respectively in Fig. 1 and Fig. 2.
M5 Decision Tree (M5 Tree)
The M5 Tree was first proposed by Quinlan (1992). This method provides the estimated value of the dependent variable in a timely, useful, and understandable manner. This paradigm of logic is adaptable. It acts as a guide for dealing with missing values and numerical data. It produces understandable, extremely precise outputs at very high rates and functions relatively quickly. This situation can be explained by decision tree learning's robust and flexible operation, which can manage the demands of real-world data sets. The M5T technique creates a regression series by repeatedly dividing the sample space with tests on a single feature that maximizes the variance in the target space. There are two types of decision trees:
(1) A classification tree is the most widely used symbolic class for estimating the value of a numerical property.
(2) Prediction using regression trees (Witten and Frank, 2002). The standard deviation reduction (SDR) can be calculated mathematically using Equation 1.
Adaptive Neuro Fuzzy Inference System (ANFIS)
Adaptive Neuro Fuzzy Inference System (ANFIS) is a hybrid artificial intelligence method that uses the ability of parallel neural networks to calculate and learn artificial neural networks and the inference of fuzzy logic. The ANFIS model developed by Jang (1993) uses the fuzzy inference model and Hybrid learning algorithm. Adaptive networks consist of directly connected nodes. Each node represents a processing unit. The connections between the nodes indicate an undetermined interest (weight) between them. All or part of the nodes can be adaptive. ANFIS is a universal approximation methodology and is capable of approximating any real continuous function on a compact set to any degree of accuracy. ANFIS with the first-order Sugeno fuzzy model was used in this study. For more information, researchers can access Jang (1993).
Artificial Neural Network (ANN)
Human brain cells are represented by a computational model called an artificial neural network (ANN). In terms of prediction capacity, it performs better than other traditional regression models. Layers make up an ANN model. The input layers receive the starting data. Weighted connections and activation functions are used by hidden layers to process inputs. The final product is produced by the output layer.
RESULTS AND DISCUSSIONS
The following statistical criteria were used in this work to compare the M5 Tree, ANFIS and ANN approaches. The variables used in the estimating models were relative humidity(RH, %), air temperature(T,°C), precipitation (P, mm) and lagged lake water level(LWL) data.
Each model was examined and contrasted using statistical metrics such the determination coefficient (R2), mean absolute error (MAE), and root mean square error (RMSE). Table 1 provides the MAE, RMSE, and R2 values for the M5 Tree, ANFIS and ANN.
The M5 Tree model's performance is shown in Figure 3. The distribution and scatter graphs show a good connection between the estimated and observed values. The analysis of Table l's MAE, RMSE, and R2 (0.032, 0.047 and 0.998). The scatterand distribution graphs of the ANFIS models are shown in Figure 4. When the distribution and scatter patterns are examined, the computed values closely resemble the actual values. Table 1 shows the benchmark values for MAE, RMSE and R2 (0.035, 0.049 and 0.997).
The ANFIS model gave results close to the M5 Tree model. The determination value being close to 1 and the errors being close to 0 indicates that the model performance is good. When considering the ANN models, Figure 5 demonstrates that the determination coefficient is high and the dam lake level estimates are fairly close to the actual data. The ANN1 model had low error and strong correlation based on the MAE, RMSE, and R2 (0.032, 0.047 and 0.998) criteria listed in Table 1.
CONCLUSION In this study, the lake water level of the Thurmond Lake in South Carolina, USA, was estimated using the M5 Decision Tree (M5 Tree), Adaptive Neuro Fuzzy Inference System (ANFIS) and Artificial Neural Network (ANN) models. Model inputs included discharge parameters, precipitation and water temperature. In terms of lake level prediction, M5 Tree, ANFIS and ANN models perform better than any other model due to their strong determination coefficient (R2) and low error ratios (RMSE, MAE). However, comparisons indicate that solutions from the artificial intelligence models may provide a more accurate representation of the observed data. Artificial intelligence systems in this study are for lake water level prediction because their structure takes into account the non-linear dynamics of the problem throughout the complete data set.
REFERENCES
1. Kucukerdem TS., Kilit M., Saplioglu K. Bulamk cikanm sistemlerinde kullamlan kiime sayilarinm K- ortalamalar ile belirlenmesi ve baraj hacmi modellenmesi: Kestel baraji ornegi. Pamukkale Universitesi Miihendislik Bilimleri Dergisi 2019; 25(8): 962-967.
2. Salam ZKAA., Keskin ME. Yapay sinir aglan ile Dibis baraji'nm seviye tahmini. Miihendislik Bilimleri ve Tasanm Dergisi 2018: 6(4): 564-569. 3. Aydemir M. Yapay sinir aglan ile biitce gelirlerinin tahmini. inonii Universitesi Sosyal Bilimleri Enstitusu, Yiiksek Lisans Tezi, 2020.
4. tines F. Dam reservoir level modeling by neural network approach: A case study. Neural Network World 2010; 20(4): 461 (in Turkish). 5. Unes. F., Dogan S., Ta§ar B., Kaya YZ., Demirci M. The evaluation and comparison of daily reference evapotranspiration with ANN and empirical methods. Natural and Engineering Sciences 2018; 3(3): 54-64 (in Turkish).
6. tines. F., Demirci M., Ta§ar B., Kaya YZ., Varcin H. Estimating dam reservoir level fluctuations using data-driven techniques. Polish Journal of Environ Mentals Studie 2019; 28(5): 3451-3462.
7. Demirci, M, Unes, F., Kaya, Y. Z., Mamak, M., Tasar, B., & Ispir, E. (2017, March). Estimation of groundwater level using artificial neural networks: a case study of Hatay-Turkey. In 10th International Conference Environmental Engineering ".
8. Demirci, M, Unes, F., Kaya, Y. Z., Tasar, B., & Varcin, H. (2018). Modeling of dam reservoir volume using adaptive neuro fuzzy method. Aerul si Apa. Componente ale Mediului, 145-152. 9. USGS.gov Science for a changing world. - https://www.usgs.gov/.
10. Quinlan, J. R. (1992): Learning with continuous classes. - 5th Australian Joint Conference on Artificial Intelligence 92: 343-348. 11. Witten, I. H., & Frank, E. (2002). Data mining: practical machine learning tools and techniques with Java implementations. Acm Sigmod Record, 31(1), 76-77.
12. Jang, J. S. (1993). ANFIS: adaptive-network-based fuzzy inference system. IEEE transactions on systems, man, and cybernetics, 23(3), 665-685.
You have requested "on-the-fly" machine translation of selected content from our databases. This functionality is provided solely for your convenience and is in no way intended to replace human translation. Show full disclaimer
Neither ProQuest nor its licensors make any representations or warranties with respect to the translations. The translations are automatically generated "AS IS" and "AS AVAILABLE" and are not retained in our systems. PROQUEST AND ITS LICENSORS SPECIFICALLY DISCLAIM ANY AND ALL EXPRESS OR IMPLIED WARRANTIES, INCLUDING WITHOUT LIMITATION, ANY WARRANTIES FOR AVAILABILITY, ACCURACY, TIMELINESS, COMPLETENESS, NON-INFRINGMENT, MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE. Your use of the translations is subject to all use restrictions contained in your Electronic Products License Agreement and by using the translation functionality you agree to forgo any and all claims against ProQuest or its licensors for your use of the translation functionality and any output derived there from. Hide full disclaimer
© 2025. This work is published under https://creativecommons.org/licenses/by/4.0/ (the "License"). Notwithstanding the ProQuest Terms and Conditions, you may use this content in accordance with the terms of the License.
Abstract
Lake Water Level Estimation with Artificial Intelligence Methods.
With the decrease in water resources due to climate change, dam reservoir level estimation is important in terms of the construction, operation, design and safety of dams. In this study, average air temperature (T), relative humidity (SR), and precipitation (P) parameters were used for lake water level estimation. Thurmond Lake in McCormick County, South Carolina, USA was selected as the study area. 1286 daily data measured in real time between 2017-2023 were used as the study data. M5 Decision Tree (M5 Tree), Adaptive Neuro Fuzzy Inference System (ANFIS) and Artificial Neural Network (ANN) models were selected for lake water level estimation and model results were compared with real observation results. In the comparison of the prediction models, performance criteria such as coefficient of determination (R2), Mean Absolute Error (MAE) and Root Mean Square Error (RMSE) were used. When the model results were examined; it was determined that artificial intelligence methods performed well in predicting the lake water level change.
You have requested "on-the-fly" machine translation of selected content from our databases. This functionality is provided solely for your convenience and is in no way intended to replace human translation. Show full disclaimer
Neither ProQuest nor its licensors make any representations or warranties with respect to the translations. The translations are automatically generated "AS IS" and "AS AVAILABLE" and are not retained in our systems. PROQUEST AND ITS LICENSORS SPECIFICALLY DISCLAIM ANY AND ALL EXPRESS OR IMPLIED WARRANTIES, INCLUDING WITHOUT LIMITATION, ANY WARRANTIES FOR AVAILABILITY, ACCURACY, TIMELINESS, COMPLETENESS, NON-INFRINGMENT, MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE. Your use of the translations is subject to all use restrictions contained in your Electronic Products License Agreement and by using the translation functionality you agree to forgo any and all claims against ProQuest or its licensors for your use of the translation functionality and any output derived there from. Hide full disclaimer
Details
1 Iskenderun Technical University, Hatay – TURKEY