The Kelly cosine value appears to be an experimentally determined modification of the usual cosine law to accommodate the finding that on the cells under test at the time, the cosine law didn't apply very well at high angles of incidence (from the normal). However the actual curve will depend on a large number of factors such as surface finish of the cover material, surface finish of the cells, height of the sun in the sky (as this changes its spectral power density which will change the percentage of light being reflected / absorbed in the covering layers). And so on. I would advise researchers not to spend a lot of time on these calculations as (1) they are quite installation specific and (2) the total energy you get from your panel when the sun is more than 45 deg off normal is far smaller than the daily variability due to weather conditions. It really isn't worth worrying about.