Forecasting Dst index using the ENLIL solar wind data through the Rice neural network prediction model

We present a new long-term prediction model of Dst index driven by the ENLIL solar wind data through the Rice artificial neural network algorithm (ANN). The model is capable of forecasting Dst out to several days in advance. The ENLIL solar wind and IMF data has been made available by NOAA's Space Weather Prediction Center through ESWDS server. Space Environment Technologies (SET)'s software framework retrieves the data, produced at 1-3 minute cadence, from ESWDS to compute hourly averages. The data is then fed to Rice's ANN-based Dst algorithm to compute the Dst at any desired epoch. The Rice ANN models (all running at Rice University, Houston, Texas) are empirical space weather prediction models that are capable of giving Kp, Dst, and AE forecasts up to 3 h ahead in near real time. Currently, the early operational version of the ”ENLIL-Rice Dst” has been running in the Rice server for over 7 months now and in the SET's server since the end of May 2014. Shortly, the operations will switch to the SET's server for prime forecasting with Rice providing the backup and redundancy. The operational outputs can be accessed from the SET's webiste at www.spacewx.com, and alternatively from the Rice University's space weather forecast page: http://mms.rice.edu/realtime/forecast.html. This paper will discuss the model architecture and the forecasting scheme, and will highlight the success of the joint venture between a government agency, private industry and an academic institution.