Regional precipitation forecasting with double fuzzy inference system (DOFIS) model

Abstract

Abstract This paper proposes a double fuzzy inference system (DOFIS) as a way to estimate point hydrological values ​​at any desired point, based on records from easting and northing fuzzy sets. For a given set of station locations, hydrological quantities are added as fuzzy set core with a membership degree (MD) equal to one and support equal to zero MD. The application of the methodology is presented on the basis of annual precipitation amounts at 10 meteorological stations for the southeastern province of Turkey. A comparison of the methodology against the radial basis function (RBF), Kriging (KRG) and inverse distance square (IDS) interpolation techniques is presented. The mean relative error percentages for the DOFIS, RBF, KRG and IDS methods are 0.078, -7.61, -7.80 and -7.72, respectively. All are within ±10% acceptable error limits in practice. This point represents a significant improvement in spatial estimation based on the DOFIS methodology.