meteohub.de

The excel algorithm is in set-up simple, but it's programming complex.
My weatherdata is logged per 30 minutes, so 48 records per day.
What I do is:
- Open the weather station CSV yearfile as excelsheet.
- Create a new tab for the desired calculation of day averages, max & min per measured parameter.
- In this tab I calculate in one row for each day the desired averages and extremes.
- First, using row numbers, I calculate the start row and the end row for each day in the CSV tab.
- Then using a combination of the INDIRECT and CONCATE functions I create a reference to a celrange for that day.
- With this cel reference one can calculate average, Tmax, Tmin etc.
e.g for Tmax: =MAX(INDIRECT(CONCATENATE("'CSV-tab'!D";RK1;":D";RK2)))

Thanks for this suggestion. But

My weatherdata is logged per 30 minutes, so 48 records per day.
What I do is:

is not suitable for my situation, I have only raw files with 16 million records per year. I did load 5 years now, and stay with this history for now.

Excel can only hold a million lines, so indeed that does not really help crunching your 16 million per year.
The only other I can think of is to write a program that reads the lines to be averaged into an array and convert 'm to a single line with average, max & min.