I am trying to clarify data on weather from the Royal Netherlands Meteorological Institute (KNMI); I have incorporated the Solar Position Algorithm (SPA), such that I can calculate sunset and sunrise (by means of the zenith angle, i.e. setting the zenith to 90 degrees). What I am trying to do is get a model of the behaviour of the irradiance (joules per square centimeters J/cm2). That is, by means of the irradiance measurements per hour provided by the KNMI. The thing that struck me is that the overall irradiance is quite variable.
In the underlying figure the sinusoid is a model which should resemble the measurements on a clear sky day. The spiky curve reflects the actual measurements, and the somewhat horizontal line is cloudcover, i.e. 8 is totally cover and 0 is clear sky.
So, what strikes me is for example that the irradiance measurements of two days later (day 59) has a maximum irradiance of 46 J/cm2 with full cloud cover (measure 8).
In other words; full cloud cover on day 57 at 12:00 results in $\approx$ 65 J/cm2, whereas on day 59 with the same cloud cover and same timestamp results in $\approx$ 27 J/cm2.
It seems as if there is another relevant factor which has a significat impact on irradiance. What could it be?
METADATA (Source: KNMI):
- Cloud cover (in octants), at the time of observation (9=sky invisible)
- Global radiation (in J/cm2) during the hourly division