I have a little confusing with some aspect related to the Peak Over Thresold Series and differencing the methodology of analysis from that of the Maximum Annual Series.

For example, for a daily series can one does this?: 1. Order the data (several years or a non-integer number of them) from the highest to the lower asigning the rank (1, 2, 3...) in an extra column, keeping the duplicate values. 2. Obtain the exceedence frequency with one of the well known formulas (California (order/number of observations or i/n), Weibull (i/(n+1)), Cunane, Gringorton...). 3. Calculate the return period for the exceedence probability as T=1/(1-F(x)), or maybe T=n/t(1-F(x)), where n is the number of years, t is the number of observations and F(x) is the exceedence probability.