Alexander G. Contreras Proposed Statement of Research
Kirt A. Costello
Rice University

I have just recently finished the work on my masters thesis entitled, "Retraining Neural Networks for the Prediction of Dst in the Rice Magnetospheric Specification and Forecast Model." My Ph. D. research will involve moving the Rice MSFM into a true forecast mode. Due to the processing times involved with many of the ground based parameters used for inputs to the MSFM the current "forecast" runs about 3 hours behind using an Air Force Kp and no Dst. (Dst processing time is about a month). By switching front end and forecast modules over to solar wind driven algorithms we hope to free up this processing time and provide a true forecast. Satellite operators can make use of MSFM model fluxes run in retrospective modes to confirm whether space weather conditions played a role in a satellite upset, but in order to prevent or avoid damaging spacecraft they must be supplied with a reliable, true forecast.

In June I will be attending the NOAA Space Models Evaluation Meeting to present a paper on our ring current particle flux results and get a better idea of what standards to use to quantify the performance of the MSFM, and I will be attending the GEM workshop in Snowmass to increase my knowledge about the physical processes involved in the magnetospheric storms that the MSFM is used to model.

A true forecast from the Rice MSFM should aid the satellite operator in avoiding risky maneuvers or strategic command sequences and hopefully prevent the loss of Earth orbiting spacecraft. Below is the abstract for my propsed Ph.D research.


Moving the Rice MSFM into a Real-time Forecast Mode Using Solar Wind Driven Front-end Models and Forecast Modules

The Rice Magnetospheric Specification and Forecast Model (MSFM) will be upgraded to remove the need for the ground based parameters Kp and Dst through the use of solar wind driven analytical and artificial neural network models specifying and predicting these parameters. The ground based parameters require long preprocessing times before being made available to the U.S. Air Force Space Weather Forecasting Center and the installed version of the MSFM. This delay hinders the usefulness of the specification of the magnetosphere and subsequent forecasts. By replacing these parameters with values specified by solar wind driven models this processing time can be reduced and the MSFM can achieve a true forecast. The effects of a higher time resolution quasi-Kp index from the MEB inversion algorithms on the MSFM equatorial particle refluxes will also be investigated.


Wednesday, 26-Mar-2003 21:49:50 CST