Posted by James Watson on 04 May, 2018
I've just uploaded a few changes to the SSN Data Help Page that I hope explain the range of prediction products available from the SIDC a little better.
Recommendation ITU-R P.371-8 (1999) states that ITURHFProp should be used in conjunction with SSN values produced using the Standard Curves (SC) model. However, the SIDC acknowledge that the prediction method, which produces very good results during the beginning to mid-point of a cycle, becomes less effective as we approach the minima.
The SC model uses on the number of sunspots as an input parameter and compares these with historical data. A fundamental limitation of this approach is that it treats each cycle in isolation so is unable to predict the onset of the next cycle. The Combined Model (CM) uses both sunspots and the 'aa' geomagnetic index as inputs. As noted by the SIDC website;
"In order to obtain a reliable prediction around the time of cycle minima (from 1.5 year before to up to the next maximum), the fit uses the aa geomagnetic index as precursor index for predicting the maximum value of the next cycle. The least-square fit then uses a combination of the last observed monthly sunspot number and this aa-based prediction of the next maximum value (until it is actually reached)."
Users may therefore wish to experiment with the Combined Model (both with and without the Kalman Filter) in a bid to improve prediction accuracy.