THE EFFECT OF PRECIPITATION FORECASTS AND MONITORING ON HOURLY HYDROLOGICAL STATISTICAL PREDICTION MODELS

The so-called “electric sector” of Brazil, namely the set comprising electricity companies (state-owned and private) plus regulatory agency has been implementing automated telemetric systems for hydrological monitoring in the country’s main energy-producing watersheds. This is motivated both by the electricity regulatory agency (ANEEL) requirements and by the belief that better (meaning higher frequency and telemetric) monitoring will lead to more safety and better reservoir operation with regard to multiple use criteria (mainly energy-generation and floodcontrol). Once an automated monitoring system is in place, it is desirable to assess its effectiveness in providing reliable hydrological forecasts. Here, we will try to study the case of the Iguacu River Monitoring System upstream of Foz do Areia Reservoir. For that purpose, we use relatively well-known statistical forecasting tools: ARIMA models and Kalman Filter models. In both cases we compare the use of conventionally operated stream gages (with data reported by radio or telephone two times a day) to the now more readily available hourly telemetric data. We also analyze the impact of precipitation forecasts (simulated from historical records with different degrees of skill) on the quality of the streamflow forecasts. It is expected that objective ways of assessing the benefits of enhanced monitoring will play an important role in defining criteria for network project and cost-benefit analysis in some of the Country’s key water resource sectors.