LAUSR

The Mississippi-Atchafalaya River Basin delivers large amounts of fresh water and nutrients to the northern Gulf of Mexico promoting the development of a large hypoxic zone every summer. Statistical and semi-empirical models have long been used to provide seasonal forecasts of the mid-summer hypoxic extent using historic time series of spring nutrient load and mid-summer hypoxic extent. These forecasts consist of a scalar estimate of hypoxic area with uncertainty but do not include spatial distributions or temporal evolution of hypoxic conditions. Three-dimensional (3D) circulation-biogeochemical models of the coastal ocean simulate the temporal evolution of hypoxia in a spatially explicit manner but have not yet been used for seasonal hypoxia forecasting. Here we present a hybrid method for seasonal, spatially explicit, time-evolving forecasts of the hypoxic zone that combines statistical forecasting with information from a 3D biogeochemical model. The hybrid method uses spring nitrate load and a multiyear (1985-2018) 3D hindcast simulation to produce a seasonal forecast. Validation shows that the method explains up to 76% of the observed year-to-year variability in hypoxic area. The forecasts suggest that the maximum seasonal extent of hypoxia is reached, on average, on August 13, two weeks after the completion of the annual cruise. An analysis of month-to-month variations in hypoxia forecasts due to variability in wind speed and freshwater discharge allows estimates of weather-related uncertainties in the forecast.