Ensemble prediction of estuary set-up and set down using operational WRF winds
The Indian River Lagoon (IRL), which extends over 175 miles north-to-south along the central east coast of Florida from Ponce de Leon Inlet to Jupiter Inlet, is a shallow estuary with limited salt water input (see Fig. 1). The lagoon system consists of connected tributaries including the Mosquito Lagoon, the Banana River, Hobe Sound, and associated marshes, and mangrove forests. Outside of dredge areas, the IRL is generally less than a meter deep and given the limited ocean access, tidal ranges are generally small (away from inlets). As a result, wind-driven tides are generally more important. Furthermore, the National Weather Service Melbourne office currently uses rules-of-thumb to forecast wave conditions along the IRL in Brevard and Indian River counties here in Florida. Hence, our primary goal is to provide the wind forcing necessary to generate ensemble hydrodynamic forecasts of set-up and set-down in the Indian River Lagoon (IRL) for high wind events. An additional by-product of this work is an evaluation of the WRF model wind forecasts.
The proposed work is a collaborative effort between the PI, a second PI who is responsible for the hydrodynamic simulations and the local NWS here in Melbourne. The atmospheric component consists of a one-year evaluation of the Melbourne NWS archival WRF model wind output (at 3 km horizontal resolution and 1 h temporal) at locations along/near the Indian River Lagoon (IRL) that coincide with surface observations (see Fig.). The WRF model forecast wind errors will be used to force hydrodynamic ensembles in which permutations in the forecast wind are introduced using the composited errors. PDFs of the WRF model wind error estimates, stratified by forecast time, will be randomly sampled to generate an ensemble of forcing data sets that vary about the WRF model forecast. The temporal variations will be mapped to the spatial domain through the weighting of the time-dependent error by the forecast wind magnitude. The proposed forecast window of interest is on the order of 24 h. A couple of high wind events will be selected to serve as a test bed for evaluation.