A test of the method on historical data

The method has been tested using historical GOES observations for 1975-2003, and the results are briefly presented here. Predictions were made for each day of the GOES observations for the occurrence of M and X flares within a day, based on a year of data prior to the day. The resulting predictions were compared with the actual historical record of flaring. The test is described in the Space Weather article (Wheatland 2005, Space Weather, Vol. 3, No. 7, S07003 10.1029/2004SW000131). A version is available from astro-ph.

Figure 1 shows year by year plots of the number of days on which M-X class events occurred, i.e. events above the M level but below X level (histogram, upper panel), and of the number of days on which X class events occurred (histogram, lower panel), for 1976-2003. The numbers of event days predicted by the method are shown by diamonds.


Figure 1: Top panel: Numbers of observed (histogram) and predicted (diamonds) event days for M-X flares. Bottom: The same, but for X flares.

Figure 2 shows "reliability plots" for the prediction of M-X events (upper panel), and of X events (lower panel), for 1976-2003, together with associated statistics. An explanation of these plots is given here. Briefly the predictions are on the horizontal axis and the observations are on the vertical axis, and the 45 degree solid line is perfect prediction. In these plots (and in Figure 1) it is evident that there is some over-prediction, in particular of M-X events.

These plots may be compared with similar plots for predictions issued by the US National Oceanic and Atmospheric Administration, available here and here.

Total days: 10226
Event days: 2600
Mean (f): 0.282
Mean (x): 0.254
Median (f): 0.218
Std dev (f): 0.251
Std dev (x): 0.435
Mean (f|x=1): 0.508
Mean (f|x=0): 0.204
Median (f|x=1): 0.565
Median (f|x=0): 0.116
Std dev (f|x=1): 0.216
Std dev (f|x=0): 0.212
Discrimination: 0.304
ME: 0.027
MAE: 0.277
MSE: 0.138
Linear assoc: 0.528
Skill: 0.272

Total days: 10226
Event days: 365
Mean (f): 0.040
Mean (x): 0.036
Median (f): 0.017
Std dev (f): 0.058
Std dev (x): 0.186
Mean (f|x=1): 0.120
Mean (f|x=0): 0.038
Median (f|x=1): 0.105
Median (f|x=0): 0.016
Std dev (f|x=1): 0.084
Std dev (f|x=0): 0.055
Discrimination: 0.082
ME: 0.005
MAE: 0.068
MSE: 0.032
Linear assoc: 0.262
Skill: 0.066

Figure 2: Reliability plots for M-X event and X event prediction, and associated statistics.


Additional information:

Main flare prediction page
More detail on how today's predictions were made
Summary of all predictions made to date
Automatic check of the reliability of the predictions
Prediction for 4 November 2003
Links to other pages related to flare prediction
Mike Wheatland's home page


Acknowledgement: The predictions given here are based on information from the Space Environment Center, Boulder, CO, National Oceanic and Atmospheric Administration (NOAA), US Dept. of Commerce.


Page maintained by m.wheatland@physics.usyd.edu.au Page last updated Thursday, 18-Aug-2005 10:41:25 AEST