Estimation of missing temperature data by Artificial Neural Network (ANN)

Year 2021, Volume: 12 Issue: 2, 431 - 438, 30.03.2021

Okan Mert Katipoğlu , Reşat Acar

https://doi.org/10.24012/dumf.852821

Cited By: 4

Abstract

Ensuring more reliable and quality meteorological and climatological studies by providing data continuity and widening the data range. For this reason, missing values in meteorological data such as temperature, precipitation, evaporation must be completed. In this study, an artificial neural network (ANN) model was used to complete missing temperature data in the Horasan meteorology station. To establish the ANN model, monthly average temperature values of neighboring stations having similar climatic characteristics and altitude with Horasan were used as input. The monthly average temperature values of the Horasan station were used as output. Approximately 70% of the data was used for training, about 15% for testing, and about 15% for verification in the ANN model. Various statistical parameters were compared to determine the best network architecture and best model. As a result, the model's high determination coefficient (R2 = 0.99) and low mean absolute error (MAE = 0.61) showed that the ANN model can be used effectively in estimating missing temperature data.

Keywords

Missing data, Temperature, Artificial Neural Network (ANN), Network training, Network testing and validation

Eksik sıcaklık verilerinin Yapay Sinir Ağları (YSA) ile tahmin edilmesi

Abstract

Veri sürekliliğinin sağlanması ve aralığın genişletilmesi ile meteorolojik ve klimatolojik çalışmaların daha güvenilir ve kaliteli olmasını sağlamaktadır. Bu nedenle sıcaklık, yağış, buharlaşma gibi meteorolojik verilerde eksik olan değerlerin tamamlanması gerekmektedir. Bu çalışmada, Horasan meteoroloji istasyonundaki eksik sıcaklık verilerini tamamlamak için Yapay sinir ağı (YSA) modeli kullanılmıştır. YSA modelinin kurulması için Horasan ile benzer iklim özelliklerine ve rakıma sahip komşu istasyonların aylık ortalama sıcaklık değerleri girdi olarak kullanılmıştır. Horasan istasyonunun aylık ortalama sıcaklık değerleri ise çıkış olarak kullanılmıştır. YSA modelinde verilerin yaklaşık% 70'i eğitim için, yaklaşık% 15'i test için ve yaklaşık% 15'i doğrulama için kullanılmıştır. En iyi ağ mimarisini ve en iyi modeli belirlemek için çeşitli istatistiksel parametreler karşılaştırılmıştır. Sonuç olarak, modelin yüksek belirlilik katsayısı (R2 = 0.99) ve düşük ortalama mutlak hataya (OMH = 0.61) sahip olması YSA modelinin eksik sıcaklık verilerini tahmin etmede etkin bir şekilde kullanılabileceğini göstermiştir.

Keywords

Eksik veri, Sıcaklık, Yapay Sinir Ağları (YSA), Ağın eğitimi, Ağın test edilmesi ve doğrulanması

References

• Referans1. Şen, Z. Artificial neural networks principles. Water Foundation, 2004.
• Referans2. Pielke, R.A.; Cotton, W.R., Walko, R.E.A., Tremback, C.J., Lyons, W.A., Grasso, L.D., ... and Copeland, J.H. A comprehensive meteorological modeling system—RAMS. Meteorology and atmospheric Physics 1992, 49(1-4), 69-91.
• Referans3. Yeşilnacar, Y.O. Estimation of wind speed, pressure and temperature for Bilecik city using artificial neural networks. Bilecik University, Institute of science and technology, Bilecik, 2011.
• Referans4. Venkadesh, S.; Hoogenboom, G., Potter, W., and McClendon, R. A genetic algorithm to refine input data selection for air temperature prediction using artificial neural networks. Applied Soft Computing 2013, 13(5), 2253-2260.
• Referans5. Güç, R. Solar energy analysis and temperature forecast with artificial neural networks for bilecik province, Bilecik Şeyh Edebali University, Institute of science and technology, Bilecik, 2016.
• Referans6. Aslay, F.; and Ozen, U. Estimating soil temperature with artificial neural networks using meteorological parameters. Journal of Polytechnic-Politeknik Dergisi 2013,16(4), 139-145.
• Referans7. Terzi, Ö. Estimation of water temperature of lake eğirdir using artificial neural networks method. Süleyman Demirel University Journal of the Institute of Science 2006, vol. 10, no. 2, pp. 297-302.
• Referans8. Taşar, B.; Üneş, F., Demirci, M. and Kaya, Y.Z. Evaporation amount estimation using artificial neural networks method, Dicle University Journal of Engineering 2018 vol. 9, no. 1, pp. 543-551.
• Referans9. Perez-Llera, C.; Fernandez-Baizan, M.C., Feito, J.L., and González del Valle, V. Local short-term prediction of wind speed: a neural network analysis, 2002.
• Referans10. Yıldıran A. and Kandemir, S.Y. Estimation of Rainfall Amount with Artificial Neural Networks, Bilecik Şeyh Edebali University Journal of Science 2018, vol. 5, no. 2, pp. 97-104.
• Referans11. Kişi, Ö. River flow forecasting and estimation using different artificial neural network techniques, Hydrology Research 2008, vol. 39, no. 1, pp. 27-40.
• Referans12. Cigizoglu, H.K. Estimation, forecasting and extrapolation of river flows by artificial neural networks. Hydrological Sciences Journal 2003, 48(3), 349-361.
• Referans13. Kızılaslan, M.; Sağın, F., Doğan, E., and Sönmez, O. Estimation of lower Sakarya River flow using artificial neural networks," Sakarya University Journal of the Institute of Science 2014 vol. 18, no. 2, pp. 99-103.
• Referans14. Minns A. and Hall, M. Artificial neural networks as rainfall-runoff models, Hydrological sciences journal 1996, vol. 41, no. 3, pp. 399-417.
• Referans15. Bishop, C.M. Neural networks and their applications, Review of scientific instruments 1994, vol. 65, no. 6, pp. 1803-1832.
• Referans16. Campolo, M.; Andreussi, P. and Soldati, A. River flood forecasting with a neural network model, Water resources research 1999 vol. 35, no. 4, pp. 1191-1197.
• Referans17. Ilie, C.; Ilie, M. Melnic, L. and Topalu, A.-M., Estimating the Romanian Economic Sentiment Indicator Using Artificial Intelligence Techniques. Journal of Eastern Europe Research in Business & Economics 2012, 1.
• Referans18. Hocking, R. R.; A Biometrics invited paper. The analysis and selection of variables in linear regression. Biometrics, 1976, 32(1), 1-49.
• Referans19. Lindley, D. V. Regression and correlation analysis. In Time Series and Statistics Palgrave Macmillan, London. 1990; pp. 237-243.
• Referans20 Dombaycı, Ö. A.; and Gölcü, M. Daily means ambient temperature prediction using artificial neural network method: A case study of Turkey. Renewable Energy, 2009, 34(4), 1158-1161.