Uncertainty analysis for the forecast of lake level fluctuations using ensembles of ANN and ANFIS models

In this study various ANN and ANFIS models were developed to forecast the lake level fluctuations in Lake Urmia in northwest of Iran. In addition to the time series of lake levels, the time series of three most effective variables in the water budget of the lake namely, the rainfall, evaporation and inflow were also used to find the best input variables to the models. Furthermore the uncertainty due to the error in measuring the hydrological variables and also the uncertainty in the outputs of ANN and ANFIS models which stems from their sensitivity to the training sets used to train these models and also the initial configuration before commencement of training were estimated. Comparing the outputs and confidence intervals of the two types of models it was found that the results of ANFIS model are superior to those of ANN’ in that they are both more accurate and with less uncertainty.