Probabilistic nonlinear prediction of river flows

Abstract

In the recent past the nonlinear prediction (NLP) method, initially developed in the context of nonlinear time series analysis, has been successfully applied to river flow deterministic forecasting. In this work a probabilistic approach to the NLP method is proposed, which allows one to estimate the probability distribution of the predicted discharge values and to quantify the total uncertainty related to the forecast. An ensemble technique is also proposed in order to optimize the choice of the parameter values and to provide robustness to the model calibration. The probabilistic NLP method is applied to a river flow time series, giving results that confirm the effectiveness and reliability of the proposed approach.