Abstract
It is the purpose of this paper to propose an urban runoff on-line prediction model and evaluate the model by using the real storm-water data. In modeling, it is necessary to overcome the shortcoming, which AR model has, that the predicted runoff has its peak behind the real runoff. We think that a model describing physical aspects of urban runoff is effective in order to overcome the shortcoming. The proposed on-line prediction model is based on a better physical understanding of the urban runoff process. The concept incorporating effective rainfall in impervious areas, time-area curve, and storage routing is used in this model.
We apply it to a basin of 1000ha. As a result of simulation using the real drainage system data and observation data (rainfall rate 90mm/day), we obtained good result. The proposed model was verifyed to be better than AR model of order 2 and to overcome the above shortcoming.
