Non-stationarity and forecast to support reservoir operations

Knowledge of the amount of inflow and water quality from sub-watersheds draining to reservoirs is required to simulate storage and operation in large water supply systems such as the New York City Water Supply System. Often models and statistical approaches based on an assumption of stationary hydroclimatological statistics are used to help evaluate operating options. However, regional studies have found trends in historical records of various hydroclimatic variables which when combined with climate change projections add more complexity to the problem. During extreme events this process is exacerbated by the unique character of each event and associated antecedent conditions and our limited ability to forecast with a certain degree of accuracy pertinent information to guide operations. In this presentation we discuss the importance of hydroclimatic forecast for reservoir system operations by comparing approaches that include historical data based conditioned and non-conditioned statistical approaches, and hydrologic modeling. Our objective is to highlight a discussion looking at advantages and disadvantages of each method and look into alternatives to improve our capability for addressing the challenging issue of non-stationarity. 1 Institute for Sustainable Cities, Hunter College, City University of New York, New York, NY. 2 Bureau of Water Supply, New York City Environmental Protection, Kingston, NY. 3 Nova Consulting, New York, NY