Autonomous VPS-Based Manganese Prediction System for Sub-Tropical Water Reservoirs

Abstract

An early warning system has been set up in order to forecast soluble manganese concentrations in the epilimnion of Advancetown Lake, Australia, 7 days ahead. The input data are automatically collected mainly from a in-situ vertical profiler and transferred to a data-driven model that was able to predict the 2012 and 2013 winter turnovers and manganese critical concentrations with high accuracy. Eventually, the outputs are displayed with the use of a graphical user-friendly interface. The created early warning systems brought to a drastic reduction of expensive water samplings and laboratory analysis and to a more efficient water treatment management.