Authors: T. Olason, D. Hurdowar-Castro1, J. Huysentruyt, P. Kirshen
Abstract
Inflow forecasts are one of the key inputs in the short-term scheduling of hydroelectric generation, and therefore an inflow forecast module is an essential part of an operational Decision Support System (DSS) for hydro scheduling. This paper presents the basis of the operational flow forecasting module of the New England Power DSS implemented by Acres International Ltd for the Upper Connecticut River. The hydraulic connectivity and routing through river reaches is usually modeled in the DSS hydro scheduler as part of the hydro system formulation and inflow forecasts are required for natural or unregulated inflows into the various lakes, reservoirs and river sections of interest The forecast time step and forecast lead time (time horizon) are determined by the dine step and lead time used in the scheduling and are commonly one hour and up to two weeks into the future, respectively.
Short-term operational forecasting involves a number of tasks, including acquisition and processing of real-time data, acquisition of meteorological forecasts, review and monitoring of hydro meteorological data, updating of the model's state variables, the flow forecast process and review and adjustment of the various subwatershed flow forecasts.
Real-time meteorological and hydrometric data from a large number of gauges as well as forecasts for precipitation and temperature are required. Real-time data are downloaded via the Internet from the National Weather Service (NWS) once and hour and processed. With the exception of gauged tributaries, inflows have to be derived using a process commonly referred to as 'backrouting'. In this process, the measured outflow, reservoir storage change are used to compute the local inflow.
Meteorological forecasts of precipitation and temperature are also a necessary input to the flow forecasting process. A meteorological forecast service provides forecasts of precipitation and temperature for a two week horizon. The state variables of the inflow forecast model are updated based on recently observed data and forecasts are subsequently calculated using forecasts of precipitation and temperature.
