The Community Coordinated Modeling Center (CCMC) from NASA has already a space weather services over GoogleEarth ( http://ccmc.gsfc.nasa.gov/downloads/googleearth.php ) which is able to visualise auroras models. However, we have aimed for this challenge to visualise near real-time data.
Our group in Dublin had to search for suitable data to visualise. We found that there used to be a spacecraft that took images of aurora from space. The UVI, on-board the POLAR satellite, (http://tideuvira.nsstc.nasa.gov/uvi/) took ultraviolet images , but the spacecraft has been dead since 2008. These images were used to validate many of the models used now-a-days (e.g. http://helios.swpc.noaa.gov/ovation/ ).
To our knowledge there are just two other sources used to model/visualise real-time aurora.
One comes from the measurement of the Earth's magnetic field (a parameter called Kp index - http://en.wikipedia.org/wiki/K-index ), this can be measured at Earth, or estimated from space observation. This method is used by the Geophysics Institute at the University of Alaska Fairbanks (http://www.gi.alaska.edu/AuroraForecast/NorthPolar/ ) and it is well documented by Sigernes et al. (2011; http://dx.doi.org/10.1051/swsc/2011003).
The second method consists of a extrapolation of the Total Energy Detected by the POES satellites from NOAA. This is used by NOAA itself (http://www.swpc.noaa.gov/pmap/) as an aurora real-time visualisation. Though we have found a detailed description of the data and how it has become a tool to visualise auroras by Dave S. Evans (http://www.bcdxc.org/noaa_poes_essay.htm), it does not explain exactly which method is used for extrapolate the data itself.
Let in the weekend, it has came to our attention that NOAA has also created an aurora visualisation tool with this data (accessible from: http://www.swpc.noaa.gov/pmap/GEpmap/).
Both of the available GoogleEarth visualisation tools mentioned above are made through image warping on the Earth globe which makes it not accessible from other tools like Marble (an open source GoogleEarth like tool; http://edu.kde.org/marble/). For this reason we have tried to achieve this challenge by creating kml polygons that visualize the aurora data in any of both programs.
On one hand we have developed a Python program that generates the Aurora oval using "The Feldstein-Starkov method" (described on the paper referenced above) into a kml polygon from the most recent Kp value (http://www.swpc.noaa.gov/wingkp/wingkp_list.txt). On the other hand, in Java we wrote a program to download and process data from POES satellites to visualised the Total Energy Detected and compare it with the model generated previously.
The processing required for the POES data is quite challenging, there are 6 working POES satellites at the moment orbiting Earth. Since the aurora oval is stronger and broader on the Earth's night side we cannot use a whole day of data. This introduces a new problem, which is a large empty area (where the satellites have not passed) to interpolate.
The final task was to generate a dynamic kml file that changes every 4 hours showing the latest prediction and the latest observations. This may be done through a cron job running on a server which updates the kml file remotely.
When the Java program starts it downloads the most recent POES data from (http://satdat.ngdc.noaa.gov/sem/poes/data/), the data file format is described in the servers readme.txt file. The Date/Time, Latitude/Longitude, Total Energy Detector (TED) values are extracted averaged and collected into a form for future processing and rendering.
Over the weekend TOPCAT (http://www.starlink.ac.uk/topcat/) was extensively used to visualise and understand the POES data. Pylab(http://www.scipy.org/PyLab) and simplekml(http://code.google.com/p/simplekml/) python libraries were used to calculate and create the aurora model.
Still to do: 1. import data points to kml files to visualize on Google Earth 2. Run programs automatically on the webserver to update users' kml files.