Time Series Models for Treatment of Surface Waters

Abstract

The use of probability models derived from a long‐term (30 yr) color record forms an approach of forecasting raw water color and assigning an alum dose for treatment of surface waters. Three probabilistic models are developed which incorporate the uncertainties associated with forecasting raw water color and alum dose. These forecasts are based upon the risk or probability associated with underdosing. The effect of seasonal and daily color variation is incorporated into one time series model. In the other time series model, daily color variation is considered exclusively. All models are compared for treatment performance. The methods of model identification, parameter estimation, and forecasting are presented for each model. The results indicate that the two time series models are feasible methods because they lead to the judicious use of alum without excessive overdosing.