I have a 5 years database of 1-minute time basis. I need to calculate the hourly irradiance of a day that represents an entire month of each year . Then I want to compare it.
I apply the Roesch-Long QC filter to the database (BSRN regular methodology) so I obtained a filtered database (with blanks). I need to know how to compare statistically correct the averages obtained (see attached file)
You may try to do some trapezoidal integration of your curve with time, then divide by the total time interval — as an approximation of the average of a function f over interval [a,b] defined by 1/(b-a) sumabf(t) dt.
This handles holes, inequal sampling times, different durations… and in the ideal case of no hole and equally spaced sampling times, will almost give the arithmetic mean of your data (the higher the number of times, the closest the two values).
- Badges
- Science topic