I need to analyze the data I am having for return periods of extreme rainfall events. I did some research and saw some papers recommended the partial duration method of analysis as more reliable. If I were to be advised to use the partial duration method what statistical distributions would be best to fit my data? My plan is to analyse the first 25 years separately and the other 25 years separately, after obtaining the return periods of each then compare the two, that is basically what I want to achieve.
Dear Sinazo Mzili,
Why 25 years though you have 50 years of data? Use daily data for not less than 30 years for your analysis. You can estimate rainfall depth, frequency, and duration in addition to the return period. Regarding the return period, Use extreme value theory with either the Annual Maximum Series (AMS) or Partial Duration Series (PDS), also called the peaks-over-threshold (POT) approach. Good luck.
Pick up your annual maximal data and make time series, then estimate flow extreme values corresponding to the probabilities of occurrence of 0.1%, 1%, and 3% using Kritski Menkel (or other similar) formula.
If you use peaks-over- threshold (POT) data for frequency analysis, then Generalized Pareto (GP) distribution would be the right choice. It has been shown that the conditional distribution of excess follows the GP distribution.
- Badges
- Science topic
- Similar topics
- Papers