Managing the touristic pressure: Performances prediction of an advanced biological system by means of regression trees

A regression tree model has been used to make predictions of six gross parameters (COD, CODsol, N-NH4 +, TN, Ptot and TSS) of an innovative SBBGR reactor. R2 values ranging from 0.94 to 0.97 were found for ammonia and total phosphorus, respectively. This application showed its usefulness as a decision support system for wastewater treatment plants in tourist areas which typically operate under high stress conditions due to the sharp fluctuations of wastewater flow and composition. A forecast of the bioreactor's performance would help the plant manager to put in place the required practices and procedures. The regression tree model could be part of the automation and control system of the SBBGR plant, allowing the change of operating conditions to be carried out automatically and in an effective way to face the touristic stress issue.