Gut reaction, to gain valuable daily data considering the monthly seasonality you need more than one year of data. Otherwise add all of them together and divide the sum by 365. 

If you have five years of data, you can figure out seasonality, distribution, range, etc. Understanding these trends would allow you to have better estimates.

You can use the method of 'rolling forecast' or another forecasting method. Consider the first month as your 'fact.' In this month your daily mean is 0.05/31. The next month, (0.05+0.14)/2/28. The following month (0.05+.14+0.46)/3/31 is your daily mean. From this point only use the last three months to create the next estimate (0.14+0.46+1.5)/30.

You can apply the same method by days instead of applying the same mean for each day of the month. The problem with that, we cannot find any trend in one-year data, so it is not sure how the following days differ from each other.

Thanks Mr, Svetoslav Anev for the excel files and the interpolation model. I see that the program can make a linear interpolation. I will take this model into consideration

Dear Robert Istvan Radics

Thank you for your answer. It is a interesting idea, I will take that as a possible solution for my problem

Regards