Self Organising Maps of Solar Wind Structures

Two Self Organising Map neural networks have been trained to classify structures in the solar wind, one using Bz, n, and V, and the other using p and E. The classification has been done for values of the Dst index at -50, -100, -150, -200, and -250 nT, for events where the gradient of n and V exceeded a certain threshold (shocks as a result of coronal mass ejections) and for events where the gradient of the 12-hour average for n was below a certain negative threshold and the gradient for the 12-hour average for V was larger than a certain threshold (onset of high-speed streams due to coronal holes). Geomagnetic storms can be caused by coronal holes and by coronal mass ejections. The strongest geomagnetic storms occur when coronal mass ejections are superimposed upon coronal hole structures.