Prediction at ungauged site with topological kriging and modified group method of data handling

Abstract

Since the number of ungauged basins is increasing over the world, there is a need for reliable streamflow prediction. This research aims to develop an accurate and reliable instrument for streamflow prediction in ungauged basins. Modified Group Method of Data Handling (MGMDH) is proposed for flood streamflow prediction. The MGMDH is an enhancement from conventional Group Method of Data Handling (GMDH) model. Previously, GMDH model is using second order polynomial as a function to describe the input and output variable relationship. In MGMDH model, tangent sigmoid function is used as a function to describe the relationship between input and output variable. Topological kriging, a geostatistical technique is then used to build a region in Peninsular Malaysia based on rainfall data on gauged station. The result indicates that the proposed approach MGMDH based top-kriging delivers the best performance in terms of prediction accuracy.