Machine Learning in Water Resources Management: Paddy Rice Irrigation Case Study

Authors : Ünal KIZIL, Hakkı Fırat ALTINBİLEK

Pages : 112-122

Doi:10.33202/comuagri.1245421

View : 68 | Download : 72

Publication Date : 2023-07-19

Article Type : Research Paper

Abstract :Paddy rice irrigation takes an important role in water consumption. Therefore, the savings to be made in paddy rice irrigation will have significant impacts. In the sustainable use of water resources, both the irrigation methods and the methods to be used in the planning of water resources are critical. Hence, the use of drip irrigation should be expanded. On the other hand, the use of modern satellite technologies and machine learning models should be used while planning irrigation. In this study, Google Earth Engine insert ignore into journalissuearticles values(GEE);, which is a cloud-based image processing platform was employed in the calculation of paddy rice cultivation areas. Random Forest insert ignore into journalissuearticles values(RF); and Support Vector Machines insert ignore into journalissuearticles values(SVM); machine learning algorithms were applied. The results showed that RF algorithm can calculate the paddy cultivation areas with an accuracy of 97%. A difference of 27.69 km2 was found between the officially declared cultivation areas and the calculated area. This can yield a miscalculation of water requirement with an error of 33.8, 38.1 and 155 million m3, in subsurface drip irrigation, drip irrigation and basin irrigation methods, respectively. Results showed that accurate calculation of paddy rice cultivation areas and drip irrigation will both minimize this error and allow 4 times more area to be irrigated.
Keywords : Uzaktan algılama, Makine öğrenmesi, Rassal orman, Destek vektör makineleri, Damla sulama, Çeltik