I need to transform a daily average temperatures forecast into an hourly forecast based on the hourly temperature profile observed historically.
I work in Python.
I have found ways of forecasting the temperatures based on historical values (either by fitting a polynomial or using designated models like Prophet), without using my average daily temperatures forecast, which I need to include.
I don’t know how to forecast the hourly temperatures from the daily forecast. Using the hourly forecast curves based on the fitted models mentioned before, I have tried calculating the percentage difference of each hourly observation to its daily mean and then tried to reapply that same percentage difference to the average daily forecast I already have. Unfortunately, that isn’t working well because I end up with “shocks” in the data each time the daily average changes.
Do you have a suggestion on how to infer an hourly profile from the known average daily temperatures forecast and historical hourly temperatures that would create a smooth hourly curve?
Thank you.
