Merging Short and Long Lead-time Dst Probabilistic Forecasts

The Dst index serves as a useful indicator of space weather activity and riskto many susceptible technological systems. Accurate prediction and forecastof the Dst index has been a focus of effort for decades.For example, predictions of the Dst index are used to determine risk of geomagnetic storm occurrence and severity, and they are used as an input for real-time thermosphere density models. Traditionally, forecasts of the Dst index have been deterministic. However, operations to research (O2R) needs of the space weather community require probabilistic forecasts for both short lead-time and long lead-time Dst.Recently, efforts to add probability to Dst prediction usingmachine learning methods have yielded promising results. Two different probabilistic, machine learned models of Dst prediction---onefor short lead-time (1-6 hours) and one for long lead-time (1-6 days)---havebeen proposed using solar wind parameters and solar disk images as predictors, respectively.We discuss and show the results of O2Refforts to merge the two proposed models into a single forecast product.The unified probability model of short and long lead-time Dst prediction will allow space weather forecasters to provide a more realistic range of possible Dst values. This will allow space weather forecast users to constrain their risk assessments, limiting the scope of their required actionsbased on the improved probabilistic Dst forecasts.