Emily Berndt (ST11) has been leading the development of a NASA/NOAA GOES-16 satellite-based dust machine learning product (i.e., DustTracker-AI). Members of the Short-term Prediction Research and Transition (SPoRT) Center, Rob Junod (UAH) focuses on model development, Kevin Fuell (UAH) conducts user engagement, and Sebastian Harkema (UAH) helps identify dust cases. The dust machine learning model was put into near real-time production by Paul Meyer during Winter 2022 and the product was subsequently assessed by six National Weather Service forecasters during Spring 2022.
DustTracker-AI is running in near-real time production for the identification of hazardous dust events that typically occur in the spring and fall in the southwest U.S. Stakeholder feedback indicated that DustTracker-AI added confidence to interpreting dust in satellite imagery and in 75% of responses the product increased the amount of time the dust plume could be tracked. In one event on 3/17/22, DustTracker-AI was highly impactful in the user’s decision to issue a dust advisory and brief emergency managers.
As a result of the successful stakeholder assessment Emily Berndt and Kevin Fuell (UAH) were invited to give a presentation to the National Weather Service Science and Operations Officers in the Western U.S. region as part of their SOOConWest event during the last week of June 2022. The presentation highlighted the model development as well as results from the Spring 2022 forecaster product assessment. The GOES-16 satellite-based DustTracker-AI product has the potential to increase confidence in identifying and warning on dust events that are especially difficult to monitor at night, benefiting NWS forecasting operations.
