The NASA Short-term Prediction Research and Transition Center (SPoRT) has been providing experimental hydrologic forecasts to the Melbourne, FL forecast office for a little over a year. These forecasts utilize a Long Short-term Memory (LSTM) machine learning model which was trained to predict river stage height with 6-hr temporal resolution out to 7-days based on time-lagged inputs of precipitation, soil moisture and gauge height. The developed LSTM-based forecast system, Streamflow-AI, currently provides three different forecast scenarios based on three different quantitative precipitation forecasts, which include: the NWS’s Weather Prediction Center, the NCEP Global Forecast System model and the National Blend of Models.
Within east central Florida, the Streamflow-AI provides forecasts for two smaller basin and fast responding streams, the Little Wekiva River in Altamonte Springs and Shingle Creek at Campbell in the greater Orlando, FL area. Due to their small size these basin locations do not have routine forecasts from the Southeast River Forecast Center. The SPoRT river level stage guidance forecasts were used to provide emergency management officials decision support leading up to Hurricane Ian’s historical flood impacts on east central Florida in September 2022. Critical preparedness activities were initiated, including pro-active evacuations of a retirement village area, before the onset of significant flooding in Osceola County. This presentation will focus on the use of NASA SPoRT’s Streamflow-AI products at WFO Melbourne, including during the record flooding of Hurricane Ian.

