As imported data are in daily values, the program has to split them into hourly values. It may do this in two ways:
-By constructing a smooth distribution proportional to the Clear Day model (that is: a clear day with reduced amplitude according to the required daily sum).
-By generating a random distribution according to the Markov matrices, in the same way as when generating synthetic data.
The second method is more realistic, but global simulation results are not very different.