Modelling values of river macrophyte metrics using artificial neural networks

The results of field research at 230 river sections located throughout Poland were used to examine the possibility of predicting values of macrophyte metrics of ecological status. Artificial intelligence methods such as artificial neural networks were used in the modelling. The physicochemical parameters of water (alkalinity, conductivity, nitrate and ammonium nitrogen, reactive and total phosphorus, and biochemical oxygen demand) were used as the explanatory (modelling) variables. The explained (modelled) parameters were the Polish MIR (Macrophyte Index for Rivers), the British MTR (Mean Trophic Rank) and the French IBMR (River Macrophytes Biological Index). The quality of the constructed models was assessed using the normalized root mean square error (NRMSE) and the r–Pearson’s linear correlation coefficient between variables modelled by the networks and calculated on the basis of the botanical research. These analyses demonstrated that the network modelling MIR values had the highest accuracy. The lowest prediction accuracy was obtained for MTR and IBMR indices. The differences between particular models are likely to result from better adjustment of the Polish method to local rivers (particularly in terms of indicator species used).