This paper evaluates a suite of forecasts for Chesapeake Bay water temperature, salinity, and dissolved oxygen created using a numerical model. By comparing the model forecasts with observations, we show that the model forecasts for temperature and salinity are more accurate than reference forecasts of previously observed conditions or the long-term mean; in other words, the forecasts are skillful. In general, the forecasts are skillful for at least 2 weeks into the future. Improvements to our forecasting system, such as predicting future river discharge into Chesapeake Bay, would likely improve the forecast skill even more. By showing that accurate, skillful forecasts are possible for a much longer time frame than previously considered, this paper takes an important step toward applying forecasts to improve water quality and fisheries management and to prepare for the impacts of extreme events like hurricanes and heat waves.
Supported by Grant# NA18OAR4320123