Vol. 3 No. 2 (2023)
Articles

Forecasting of Water Levels by Artificial Neural Networks Technique in Lake Michigan-Huron

PDF
  • Mehmet Fehmi Yıldız,
  • Vahdettin Demir

Published 2023-09-30

How to Cite

Yıldız, M. F., & Demir, V. . (2023). Forecasting of Water Levels by Artificial Neural Networks Technique in Lake Michigan-Huron. Advanced Geomatics, 3(2), 63–71. Retrieved from https://publish.mersin.edu.tr/index.php/geomatics/article/view/1097

Copyright (c) 2023 Advanced Geomatics

Creative Commons License

This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.

Abstract

Water is an indispensable resource for all living things on Earth. Therefore, it is important to pay attention to current water consumption and to comply with safety precautions. Many water sources in the world experience ups and downs in the water level. Lake Michigan-Huron is an 8 km long body of water formed by the merging of Lake Michigan and Huron. The Huron and Michigan hydrological description is a single lake because the water from the Strait of Mackinac, which connects these lakes, balances what it expects. The flow is generally eastward, but the water moves in both directions depending on the local structure. Lake Michigan-Huron combined is the largest freshwater lake in the world. The aim of this study is to estimate the changes in water levels of Lake Michigan-Huron in the USA. In this study, the estimation of water levels on a monthly basis was investigated by using three different artificial neural network (ANN) models in order to predict the Michigan-Huron Lake water levels one month in advance. The ANN models used are Multilayer ANN (MANN), Radial Based ANN (RBANN) and Generalized Regression ANN (GRANN). The data sample consists of a 104-year (1918-2021) record of mean lake water level. 75% of all data were used for the training phase and 25% for the testing phase. Mean Absolute Error (MAE), Root Mean Square Error (RMSE) and coefficient of determination (R2) were used as evaluation criteria. When the results are examined, all models give very good predictions during the training and testing phases. However, according to the test results, the model algorithms that give the most successful results are RBANN, MANN and GRANN, respectively.

In this study, the estimation of water levels on a monthly basis was investigated by using three different ANN models in order to predict the Michigan-Huron Lake water levels one month in advance. The ANN models used are Multilayer ANN (MANN)), Radial Based ANN (RBANN) and Generalized Regression ANN (GRANN). The data sample consists of a 104-year (1918-2021) record of mean lake water level. 75% of all data were used for the training phase and 25% for the testing phase. Mean absolute error (MAE), root mean square error (RMSE) and coefficient of determination (R2) were used as evaluation criteria.

When the results are examined, it is seen that all models make very good predictions during the training and testing phases. However, according to the test results, the model algorithms that give the most successful results are RBANN, MANN and GRANN, respectively.

PDF

References

  1. Aksoy, A. S., Aksoy, Y. R., Keskin, M. E., & Yılmazkoç, B., (2017). Göl Seviye Tahmini: Eğirdir Gölü: Mühendislik Bilimleri ve Tasarım Dergisi., 5(3), 601-608.
  2. Albek, E. A., Albek, M. & Göncü, S. (2017). Trend Analysis of Burdur, Eğirdir, Sapanca and Tuz Lake Water Levels Using Nonparametric Statistical Methods. Afyon Kocatepe University Journal of Sciences and Engineering, 17(2), 555–570.
  3. Arslan, N., Çiçekli, S., Dönmez, C., Şekertekin, A., (2018). Adana Seyhan Baraj Gölü Alanının Mevsimsel Değişiminin Yapay Sinir Ağları ile Analizi. Uzaktan Algılama ve Coğrafi Bilgi Sistemi Sempozumu, Ankara.
  4. Chen, T., Cheng, P., He, H., Yan, Y., (2019). Tree Height Estimation of Forest Plantation in Mountainous Terrain from Bare-Earth Points Using a DoG-Coupled Radial Basis Function Neural Network., 11,1271.
  5. Ciliz, M. K., & Işik, C., (1997). On-Line Learning Control of Manipulators Based On Artificial Neural Network Models. Robotica, 15(3), 293–304. doi:10.1017/s0263574797000337
  6. Çalım, M.M., (2008). Yapay Sinir Ağları Yöntemi ile Baraj Hazne Kotu Tahmini., Yüksek Lisans Tezi, Mustafa Kemal Üniversitesi, Fen Bilimleri Enstitüsü, Hatay,61p.
  7. Çubukçu, E. A., Yılmaz C. B., Demir, V., & Sevimli, M. F. (2021). Forecasting Of Monthly Average Lake Levels Of Lake Michigan With Artificial Neural Networks. 1st Advanced Engineering Days, 4-7.
  8. D’Addona D.M. (2014). Neural Network, in: CIRP Encycl. Prod. Eng., Springer Berlin Heidelberg, Berlin, Heidelberg, 2014: pp. 911–918. https://doi.org/10.1007/978-3-642-20617-7_6563.
  9. Demir, V. (2022) Enhancing Monthly Lake Levels Forecasting Using Heuristic Regression Techniques with Periodicity Data Component: Application Of Lake Michigan. Theor Appl Climatol 148, 915–929. https://doi.org/10.1007/s00704-022-03982-0
  10. Demir, V., Yaseen, Z. M., (2022). Neurocomputing Intelligence Models for Lakes Water Level Forecasting: A Comprehensive Review. Neural Computing and Applications 2023 (35), 303–343
  11. Desmukh, T., Tanty, R., (2015). Application of Artificial Neural Network in Hydrology- A Review. International Journal of Engineering Research and Technology, 4(6), 184-188.
  12. Dikbaş, F., Fırat, M., (2006). Göllerde Üç Boyutlu Hidrodinamik Modellemede POM ve Yapay Sinir Ağları Yöntemlerinin Kullanılması: Gökpınar Baraj Gölü Örneği: Mühendislik Bilimleri Dergisi., 1(12), 43-50.
  13. Kılıç, E., Özbalcı, Ü., & Özçalık, H. R. (2012). Comparison of MLP and RBF Structures in Modeling of Nonlinear Dynamic Systems with Artificial Neural Networks, Elektrik - Elektronik ve Bilgisayar Mühendisliği Sempozyumu, 29 Kasım - 01 Aralık 2012, Bursa.
  14. Koca, Y. (2014). Rize İyidere Alt Havzası İkİzdere Kesiti İçin Birim Hidrografın Belirlenmesi. Uzmanlık tezi, Orman ve Su İşleri Bakanlığı, Ankara, 68p.
  15. Michigan-Huron (2023), Lake Michigan–Huron https://en.wikipedia.org/wiki/Lake_Michigan%E2%80%93Huron, Access date: 25.05.2023
  16. Okkan, U. & Dalkılıç, H. (2012). Radyal Tabanlı Yapay Sinir Ağları ile Kemer Barajı Aylık Akımlarının Modellenmesi. Teknik Dergi , 23 (112), 5957-5966.
  17. Özaydın, Ö., (2009). ARMAX Modelleri ve Porsuk Barajı Su Seviyesinin Öngörüsü. Doktora Tezi, Eskişehir Osmangazi Üniversitesi, Fen Bilimleri Enstitüsü, İstatistik Anabilim Dalı, Eskişehir.
  18. Shamseldin A. Y., (2010), ANN Model for River Flow Forecasting in A Developing Country, Journal of Hydroin formatics, 12 (1), 22–35.
  19. Teltik İ., Aksoy H., Ünal N.E., (2008). Van Gölü Su Seviyesi Stokastik Modelleri, Van Gölü Hidrolojisi ve Kirliliği Konferansı, 21-22 Ağustos 2008, DSİ XVII. Bölge Müdürlüğü, Van, s. 74-81.
  20. Ustaoglu B., Cigizoglu H. K., Karaca M., (2008) Forecast of Daily Mean, Maximum and Minimum Temperature Time Series by Three Artificial Neural Network Methods, Meteorological Applications 15, 431–445.
  21. Yarar, A., Onüçyıldız, M., (2009). Yapay Sinir Ağlari ile Beyşehir Gölü Su Seviyesi Değişimlerinin Belirlenmesi., Selçuk Üniversitesi Mühendislik Mimarlık Fakültesi Dergisi, 24(2), 21-30.