Effects of SPEs on cloud cover
I am currently working on the concept of using the EPIC camera to try to look at SPEs and their immediate effect on cloud cover. At current, it appears that the camera does not have sufficient data so I may have to find a geostationary station instead. The effect of the Forbush Decrease appears to be well studied; I do not know if it is worth doing further reading here.
The first two images here show the effect of k-means clustering on an EPIC image.
The Elbow Plot for the k-means clustering looks like this:
I chose a cluster number of 2, and it seems to reproduce cloud cover well. I think, however, that using simple thresholds for brightness of pixels might be the more efficient method like so:
I would also like to show some problems with DSCOVR’s EPIC; Firstly, the part of the earth most effected by the SPE is quickly moved away from; secondly, I have found a lack of data around certain events (two SPEs in September 2017 have a lack of data to look at:
Finally, data is only available over a small number of SPEs, making statistical analysis difficult.