When we have temperatures in decades (ten days) for example for 5 years, could Anova be adequate to compare sums of tempreatures to know whether there is difference between these sums over the years or there is a stability?
I'm not sure ANOVA is the answer for this problem. It's a time series and what you need to do is a trend analysis, taking into account variations due to seasonality etc.. Seasonality can be corrected by moving average or Exponential smoothing techniques.
Hope this helps and that I understood your question.
As Mahdi, suggested ANOVA can give you limited answer. However a trend analysis would be better option, as recommended by Omar. All depends, what you want to study.