The Annual Maximum Series, (AMS) rainfall amount measured 24 hourly (daily) can be reduced to shorter duration of 0.166, 0.1333, 0.5, 1, 2,...hours as may be required for analysis. Rainfall values (shorter duration) can be obtained via Indian Meteorological Dept. (IMD) empirical formula for reduction proposed by Ramaseshan (1996) as given below:

Rt = R24(t/24)0.33

where: Rt = rainfall depth in mm; and R24 = daily rainfall depth, mm and t = required duration in hrs.

Also, we have Non-stationary Extreme Value Analysis (NEVA) Software Package by Center for Hydrology and Remote Sensing, University of California, Irvine, 2014.

Regards

Prof. Ify L. Nwaogazie

The Annual Maximum Series, (AMS) rainfall amount measured 24 hourly (daily) can be reduced to shorter duration of 0.166, 0.1333, 0.5, 1, 2,...hours as may be required for analysis. Rainfall values (shorter duration) can be obtained via Indian Meteorological Dept. (IMD) empirical formula for reduction proposed by Ramaseshan (1996) as given below:

Rt = R24(t/24)0.33

where: Rt = rainfall depth in mm; and R24 = daily rainfall depth, mm and t = required duration in hrs.

Also, we have Non-stationary Extreme Value Analysis (NEVA) Software Package by Center for Hydrology and Remote Sensing, University of California, Irvine, 2014.

Regards

Prof. Ify L. Nwaogazie

This is a question that has appeared many times on Researchgate.

The short answer is that you cannot calculate the X(λt) if you do not have short duration rainfall intensities.

To get short duration rainfall intensities for a range of return period (annual exceedance probabilities) the steps are as follows:

1. Carry out an extreme value analysis on your annual maximum 24 hour rainfall time series (I’m assuming you have a sufficiently long data set for this i.e. > 10 years). This could be done by fitting a General Extreme Value (GEV) distribution to your data

2. Calculate the 24 hour extreme value rainfalls for the return periods you’re interested in

3. Disaggregate the 24 hour rainfall data for different return periods to shorter durations

The problematic part is step 3. How to disaggregate the daily rainfall data? It’s far from ideal to use this method but if you have no short duration rainfall this is probably the best way. If you have a defined rainfall intensity - duration – frequency (IDF) curve for a meteorologically “similar” region to where your 24 hour rainfall data has been collected you could use this. There are other methods via which you can disaggregate the daily rainfall but in my opinion most of them are far from ideal. It’s best to have observed short duration rainfall.

These papers may be of help with disaggregation ideas:

https://www.researchgate.net/publication/260719291_Intensity-Duration-Frequency_IDF_Rainfall_Curves_for_Data_Series_and_Climate_Projection_in_African_Cities

https://link.springer.com/article/10.1186/s40562-019-0147-x

https://core.ac.uk/download/pdf/61630780.pdf

http://www.mhwm.pl/Generation-of-Rainfall-Intensity-Duration-Frequency-curves-for-the-Barak-River-Basin,79175,0,2.html

Thank you so much Professor Ify LAWRENCE Nwaogazie and Engineer Darren Lumbroso for your insights on this topic. I really appreciate it.

I have a quick question to Darren Lumbroso: if I have rainfall data of 24-hour duration (data available for 60 years) and 3-hour duration (data available for 15 years), then can I use the scaling procedure to calculate X(λt) as proposed by Prof. Van Nguyen for other shorter duration up to 5 minutes. I have a feeling that X(λt) for duration less than an hour would not work using the data of 24-hour and 3-hour. Scaling coefficients are different for (i) duration ranging from an hour to 5 minutes and (ii) duration ranging from 24-hour to 1 hour.

GEV distribution model