YAPAY SİNİR AĞLARINI KULLANARAK GÜNLÜK YAĞIŞ MİKTARININ TAHMİNİ

Öz

Su kaynaklarının kullanılması ve su yapılarının planlanması gibi pek çok konuda analizlerin sağlıklı bir biçimde yapılabilmesi için birçok parametrenin (yağış, akış, sızma, buharlaşma ve terleme v.s) doğru ve uzun süreli ölçümünün yapılabilmesi son derece önemlidir. Yağış verisi ise bu parametrelerin en önemlilerinden bir tanesidir. Geçmişe yönelik yağış verileri ile ilgili sağlıklı ve uzun süreli bilgi iyi bir analiz için büyük önem taşımaktadır. Geçmişe yönelik ölçümü bulunmayan istasyonların verileri aynı havza içerisinde bulunan ve hidrometeorolojik olarak benzer diğer istasyonların verileri ile tahmin edilebilmektedir. Elde edilen sonuçların güvenirliği açısından tahmin yönteminin doğru seçilmesi önemli olmaktadır. Bir istasyonda mevcut ölçüm verileri yardımıyla, geçmişe yönelik verilerin elde edilip edilemeyeceği veya gelecekte meydana gelebilecek veri eksikliklerinin giderilip giderilemeyeceği irdelenmelidir. Bu çalışmada Yapay Sinir Ağları (YSA) yöntemi kullanılarak yağış tahmini için bir model teklif edilmiştir. Bu yöntem, Amerika Birleşik Devletlerinin Portland bölgesinde bulunan 121, 120, 21 ve 107 nolu yağış gözlem istasyonlarında ölçülmüş günlük yağış verilerine uygulanmıştır. Modelleme, 2000 – 2009 yılları arasında her bir istasyondan alınan 3170 adet günlük yağış verisi için uygulanmıştır. Elde edilen sonuçlar literatürde mevcut olan ağırlıklı ve harmonik ortalama metotlarından elde edilen sonuçlar ile karşılaştırılmıştır. Anahtar Kelimeler- Yağış, Yapay Sinir Ağları, Tahmin, Portland Abstract For the proper utilization of water resources and properly planned water structures to be built and conducted rightly, it is necessary that many parameters (precipitation, runoff, infiltration, evapotranspiration, etc.) and their influences should be analyzed and examined properly. Precipitation is one of the most important parameters in the hydrology field. The rainfall data is very important for the level of water resources conduct and running. While planning, recorded data has great importance. Data of the ungauged stations may be estimated with those of similar stations hydrometeorogically in the same/similar watershed. Method of forecast is very important from the view point of the reliability of results obtained. It must be investigated, in detail, whether the data in the past is obtained or the lacked data, would be in the future, is completed. In this study a model is proposed for estimating daily rainfall amount by using the artificial neural network (ANN) method. This method has been tested for the daily precipitation data obtained from the precipitation stations of 121, 120, 21 and 107 in Portland province of USA. Number of 3170 data in the years of 2000-2009 belonging to these stations, has been used in this model. Furthermore, this model formed is compared with the commonly used weighted average, and harmonic average methods. Keywords- Precipitation, Artificial Neural Network, Prediction, Portland

Anahtar Kelimeler

• Yağış
• Yapay Sinir Ağları
• Tahmin
• Portland

Predicting Of Daily Precipitation Using Artificial Neural Network

Öz

For the proper utilization of water resources and properly planned water structures to be built and conducted rightly, it is necessary that many parameters (precipitation, runoff, infiltration, evapotranspiration, etc.) and their influences should be analyzed and examined properly. Precipitation is one of the most important parameters in the hydrology field. The rainfall data is very important for the level of water resources conduct and running. While planning, recorded data has great importance. Data of the ungauged stations may be estimated with those of similar stations hydrometeorogically in the same/similar watershed. Method of forecast is very important from the view point of the reliability of results obtained. It must be investigated, in detail, whether the data in the past is obtained or the lacked data, would be in the future, is completed. In this study a model is proposed for estimating daily rainfall amount by using the artificial neural network (ANN) method. This method has been tested for the daily precipitation data obtained from the precipitation stations of 121, 120, 21 and 107 in Portland province of USA. Number of 3170 data in the years of 2000-2009 belonging to these stations, has been used in this model. Furthermore, this model formed is compared with the commonly used weighted average, and harmonic average methods. 

Anahtar Kelimeler

• Precipitation
• Artificial Neural Network
• Prediction
• Portland

Kaynakça

1. Zaw, W., T., and Naing, T., T., ‘‘Empirical Statistical Modeling of Rainfall Prediction over Myanmar’’, Proceedıngs of World Academy of Scıence, Engıneerıng and Technology Volume 36 December, 2070-3740, 2008.
2. Sen, N., ‘‘New forecast models for Indian south-west Monsoon season Rainfall”, in Current Science, vol. 84, No. 10, May, pp.1290-1291, 2003.
3. Singhrattna, N., Rajagopalan, B., Clark, M., Kumar K. K.,“Seasonal Forecasting of Thailand Summer Monsoon Rainfall”, in International Journal of Climatology, Meteorological Society, , pp. 649-664, 2005. Issue 5, American
4. Bayazıt, M., Hidroloji, İstanbul Teknik Üniversitesi İnşaat Fakültesi Matbaası, 1995.
5. Şen, Z., Yapay Sinir Ağları İlkeleri, Su Vakfı Yayınları, 2004.
6. McCulloch, S. W., and Pitts, H. W., ‘‘A Logical Calculus of the Ideas Immanent in Neural Net’’, Bulletin of Mathematical Biophysics, Volume 5. 1943.
7. Wang, W., Stochasticity, Nonlinearity and Forecasting of Streamflow Processes PhD Thesis, 2006.
8. Widrow, Bernard, and Hoff, Marcian, ‘‘daptive Switching Convention Record, Part 4. 1960. 1960 IRE WESCON

9. Hopfield, J., ‘‘Neural Networks and Physical Systems with Abilitie’’,Proceedings of the National Academy of Sciences, Volume 79. 1982. Computational
10. Fernando, D.A.K., Jayawardena, A.W., ‘‘Runoff forecasting using RBF Networks with OLS algorithm’’, Journal of Hydrologic Engineering 3 (3). 203-209. 1998.
11. Cigizoglu, H.K., Alp, M., Rainall_runoff modelling using three neural network methods. Lecture Notes in Artificical Intelligence (Lectur Notes in Computer Science). Springer-Verlag, pp. 166-171. 2004.
12. Tokar, A.S., Johnson, P.A., ‘‘Rainfall-runoff modelling using artificial neural Networks’’, Journal of Hydrological Engineering, ASCE 5(2), 180-189, 1999.
13. Newham, L.T.H., Norton, J.P., Prosser, I.P., Croke, B.F.W., Jakeman, A.J., ‘‘Sensitivity analysis for assesing the behaviour of a landscape- based sediment soruce and transport model. Enviromental Modelling and Software’’, 18 (8-9), 741-751, 2003.
14. Keskin, M.E., Terzi, Ö., ‘‘Artificial Neural Network Models of Daily Pan Evaporation’’, Journal of Hydrologic Engineering, 11(1), 65-70. 2006.
15. Cığızoğlu, H.K., ‘‘Estimation and forecasting of daily suspended sediment data by multi-layer perceptrons’’, Advances in Water Resources 27, 185-195. 2004.
16. Kişi, Ö., ‘‘Multi-layer perceptrons with Levenberg- Marquardt sediment concentration prediction and estimation’’, Hydrological Sciences Journal, 49(6), 1025-1040. 2004. for suspended
17. Agarwal, A., Singh, R.D., Mishra, S.K., Bhunya, P.K., ‘‘ANN-based sediment yield models for Vamsadhara riverbasin’’ (India). Water S.A. 31(1), 95-100. 2005.
18. http://or.water.usgs.gov/non
19. usgs/bes/raingage_info/clickmap.html 12/12/2009
20. Nash, J. E. and J. V. Sutcliffe , ‘‘River flow forecasting through conceptual models’’ part I — A discussion of principles, Journal of Hydrology, 10 (3), 282–290, 1970.