The use of GAPI model for flood forecasting

Among many different existing forecasting methods it is difficult to obtain one which is accurate, flexible and able to describe the physics of transformation accurately. A new rainfall - runoff model has been developed in Hungary, known as GAPI model, the model receives its name from the so called Antecedent Precipitation Index (API) which represent the soil moisture conditions of the catchment. The model was tested over five catchments from different geographical locations, ranging in size from 2,350 Q2 to 61,780 km2. The principle governing the analysis in this study is to deal with GAPI as lumped, time-invariant non-linear model in updating mode, and to modify the components of the model so that it can work in simulation mode. The results of application indicate the unsuitability of GAPI model on large catchments. The results of applying the autoregressive model for error updating are quite satisfactory. In simulation mode the model performance is generally poor. In real-time use the model can provide us with an accurate forecast up to 4 days ahead if the input data are reliable and sufficient. The results of this study have confirmed the general experience that the main problems in hydrological modeling are related to unreliability and insufficiency of input data.