I am trying to do a frequency analysis of extreme rainfall data using Peak-Over-Threshold method. From the extreme value theory, GPD is used for the POT data and, GEV distribution for the block maxima. However, in an exceptional case if GP is not a good fit at any threshold, is it justifiable to use GEV for the data?.
I would really appreciate your answers and if possible a source of literature to support the argument.
Thank you
Dear Rohith,
I think you can this as long as you demonstrate that the GEV provides a reasonable fit.
I think we followed this procedure in:
Future high and low flow estimations for Central Vietnam: A hydrometeorological modeling chain approach, July 2017, Hydrological Sciences Journal/Journal des Sciences Hydrologiques, DOI: 10.1080/02626667.2017.1353696
Please check this out.
Good luck!
Patrick
Dear Patrick Laux ,
Thank you very much for answering the question, I too feel the same way that we can use GEV if it is a good fit. However, I am unable to find a literature to support the argument.
I have quickly glanced through your paper and noticed that you have used GEV for the annual maximum and annual minimum data. I was hooping for an example where GEV is used for peak-over-threshold data.
Thank you
Rohith
Not only GP can be used to represent a POT model; you have other options such as exponential, gamma, or log-normal distributions. For the threshold you are considering, you should test your POT model for the distributions mentioned above and choose one based on a Goodness of fit (GoF) technique. The latter can be done via visual inspection and/or through statistical GoF tests.
Check out this paper:
https://www.tandfonline.com/doi/pdf/10.1080/02626667.2013.831174
- Badges
- Science topic
- Similar topics
- Medicine
- Physiology
- Biosignals
- Biomedical Signal Processing
- Statistical Analysis