Any views expressed within media held on this service are those of the contributors, should not be taken as approved or endorsed by the University, and do not necessarily reflect the views of the University in respect of any particular issue.

Space: weather or not

Space: weather or not

Concerning the quest to predict space weather more accurately

Continuing work on ARIMA forecast

Happy new year! I completed some more work on the ARIMA dataset before Christmas that I wanted to share.

I firstly applied the same method as I used in previous analyses, plotting the RMS error over the full forecast time to see how the different methods react to longer forecast times. Using the full 1991-2016 dataset, you can see that the error from simply predicting the average value is now an improvement on the BGS ARIMA model at the longest forecast scales (longer than ~1 day). You can also see that the simple model I created based on the persistence model and the mean in equal proportions is also an improvement at lower time scales. I would see this as evidence that the ARIMA forecast does not appear to react well to high geomagnetic indices (presumably because they are so rare).

As expected, the Interquartile range (IQR) of the forecast decreases over time, suggesting that longer forecast times will have a smaller range of values which are regularly predicted, though this effect is not as significant as I was expecting, and there still seems to be some variation in the results given at ~72hour forecast times.

When plotting the error yearly, you see that the forecast has a huge difference between the most error-prone and least error-prone years. 1991 and 2003 have the highest error.

I also plotted this information onto a colour plot to visualise more easily.

Using this, you can see that there appears to be some correlation between solar maximum and high error in the forecast. Using an approximate graph for sunspot numbers (to give some form of context), you can see that there appears to be some form of correlation.

I used linear regression on this data to produce a regression line and provide a regression coefficient.

Using this, you can see that sunspot number does have a significant effect on the BGS ARIMA forecast. I would, however, claim that there are likely to be other factors at play here, as the correlation coefficient is not very high (only ~0.4). This may be due to other factors in the data; as an example, a possible reason why 2003 was so error-prone despite the relatively lower sunspot number may be related to the Halloween 2003 storm, which may have introduced high error values which distort the trend. (Sunspot number is average in 2003 but the error is relatively high!)

Leave a reply

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>


Report this page

To report inappropriate content on this page, please use the form below. Upon receiving your report, we will be in touch as per the Take Down Policy of the service.

Please note that personal data collected through this form is used and stored for the purposes of processing this report and communication with you.

If you are unable to report a concern about content via this form please contact the Service Owner.

Please enter an email address you wish to be contacted on. Please describe the unacceptable content in sufficient detail to allow us to locate it, and why you consider it to be unacceptable.
By submitting this report, you accept that it is accurate and that fraudulent or nuisance complaints may result in action by the University.