Forecasting of Soil Water Content using Support Vector Regression with the Purpose of Agricultural Drought

Agricultural drought especially in the arid and semi-arid countries is one of the most important factors affecting agricultural production. Therefore, soil water content monitoring is crucial and required. Application of the Support Vector Regression (SVR) for the short-term forecast of soil water content is important for predicting and reducing the effects of drought. However, the soil water content in many countries is not a measured data on a regular basis. At the same time, there is no prediction for the future of this data as well. The main subject of this study is to fulfill for the short term forecast of the soil water content by using DVR approach. The experimental study for collecting soil water content data was done at the field of Atatürk Soil Water and Agricultural Meteorology Research Station Directorate in the Kýrklareli city of Turkey during the growing periods of winter wheat. The soil water content data (volumetric), which were measured in the depth of 0-30 cm, from January 1, 2010 to January 1, 2012, were used. Chaotic approach was used to derive the input matrix for SVR, which in recent years, has become increasingly significant in the field of time series forecasting. Achieved results are quite promising.