calculate maximum intensity-depth-duration table for the storm
Answers
Usually they say you can develop some of these factors for experimental watersheds on how they respond to storms and neighboring watersheds in about ten years. But rainfall is so different in its variation in location, direction and intensity of storms, various types of storms, etc. Rainfall also has cycles of wet and dry, perhaps on 20 to 50 year cycles. We had a storm in South Carolina that for some areas exceeded the 1000 year event based on IDF with probably over 100 years of data. So there is never going to be having too much data. I suppose you might start developing rainfall IDF after 20 years with a tentative concern if the years are affected by a drought or several uncharacteristic storms, but with the understanding this should be ongoing and continuing work. You may even feel good with your success after maybe 50 years. But for some areas of the world near warm moist air masses such as coastal areas, that occasional interact with cold weather frontal movements or cyclonic storms, the difficulty increases in providing reliable IDFs and more and more data may be needed to do this subject justice. Climate change concerns further complicates the issues as even more extremes are suggested. The IDF we used in my college days was call Technical Paper 40 put out for the USA by the National Weather Bureau or Service back then. I am pretty sure there are models on how to make the varied calculations today when you have a data set, but calculations were much harder in those early days. Basically the model would search out the most rainfall in a 15 minute, 30 minute, 45 minute, 60 minute, etc. to perhaps 3 days or more, for each year. Then when you have those values by year, your order them from low to high and we used to plot them by hand. So if you had 50 years of data and were addressing the 60 minute extreme, the highest value is more or less the 50 year, the 20th value is the 20 year. It usually plots well on semilog paper with years on x axis. That is the old fashioned way. There is more to it for sure, but this is the basics. If you decide to do this with your data, I would try to find a approved model or computer programming script that can do the calculations from your data. It gets more complex when you do it for a nation with hundreds or thousands of precipitation recording sites, various periods of record, some missing data, etc.