Streamflow simulation by a watershed model using stochastically generated weather in New York City watersheds

Recent studies have reported increasing trends in total precipitation, and in the frequency and magnitude of extreme precipitation events in the West of Hudson (WOH) watersheds of the New York City (NYC) water supply. The potential effects of these changes may pose challenges for both water quality (such as increased sediment and nutrient loading) and quantity (such as reservoir storage and management). The NYC Dept. of Environmental Protection Climate Change Integrated Modeling Project (CCIMP) is using "bottom-up" or vulnerability based methods to explore climate impacts on water resources. Stochastic weather generators (SWGs) are an integral component of the bottom-up approach. Previous work has identified and evaluated the skill of alternative stochastic weather generators of varying complexity for simulating the statistical characteristics of observed minimum and maximum daily air temperature and occurrence and amount of precipitation. This evaluation focused on the skill in representing extreme streamflow event probabilities across NYC West of Hudson (WOH) watersheds. Synthetic weather time series from the selected (skewed normal) SWG were used to drive the Generalized Watershed Loading Function (GWLF) watershed model for a 600 year long period to simulate daily streamflows for WOH watersheds under a wide range of hydrologic conditions. Long-term average daily streamflows generated using the synthetic weather time series were comparable to values generated using observed long-term (1950-2009) weather time series. This study demonstrates the ability of the selected weather generator to adequately represent the hydrologic response in WOH watersheds with respect to the total, peak, and seasonality in streamflows. Future application of SWGs in NYC watersheds will include generating multiple scenarios of changing climate to evaluate water supply system vulnerability and selection of appropriate adaptation measures.