Convert Precipitation to IDF 3

[Hqdo]

Dear all experts,

I'm civil engineer an have very basic understanding of hydrology. In my past with hydrology/hydraulic cals, IDF were given. This time, I work on a project oversea that requires some basic drainage cals but all I have is precipitation data from this website:

http://www.mandalay.climatemps.com/precipitation.php

My question is from looking at the data do you think I can create IDF curve for 1 yr, 5,10,100 yr? or do you think I can use the precipitation and divide certain time period to get intensity? And how do you interpretate some of the probability, dry and wet month...

Thanks.

 

[psmart]

To create a useful IDF curve you would typically need depth or intensity data for rainfall durations as low as 5 minutes, and ranging up to 24 hours. The site you reference doesn't contain any of this data.


Peter Smart
HydroCAD Software

http://www.hydrocad.net

 

[Hqdo]

Thanks Peter, I will ask local government for more rainfall amount from 5min to 24hr. The worst case do you think 60min is okay to use it as 1hr design point rain fall? What about frequency? do you think the probability of rainny day comes with the data set will give us safe assumption about what frequency they are?

Thanks

 

[beej67]

If you can get isohyetal maps of the area for 2 year and 100 year return periods, at 5, 15, and 60 minute durations, you can probably feel confident applying the procedure within NWS Hydro-35 to interpolate the rest.

http://www.nws.noaa.gov/oh/hdsc/currentpf.htm

The entire discussion of this paper would be useful for you to read and understand, as well as the discussions in TP-40 and some other documents at that link about how they processed rainfall data into information useful for design engineers.

Hopefully that gives you enough material to go on. Then it comes down to what data is available. In the end, you may have to make some assumptions.

Hydrology, Drainage Analysis, Flood Studies, and Complex Stormwater Litigation for Atlanta and the South East -

http://www.campbellcivil.com

 

[Hqdo]

Thanks Beej, until when I get more data, for now I was wondering if I can make the following rough assumptions:
1. The data is 1-yr storm event
2. Heaviest rain is in Oct of 7.3 inch, 7 rainy days in the month -> 7.3/7 = 1.1 inch -> 1-Yr storm, 24 hr storm duration

Thanks

 

[psmart]

Speaking very roughly, your reasoning would produce 7 days in a single month that meet the 1-year depth, so there is an inherent contradiction. I think you've established a reasonable minimum for the 1-year 25-hour depth, but the actual 1-year depth will be somewhat higher. How much higher we don't know - there simply isn't enough information. The correct 1-year depth could be several times that amount.

Peter Smart
HydroCAD Software

http://www.hydrocad.net

 

[beej67]

Make sure you understand the importance of duration with this task.

In a very general sense, in order to establish the 100 year 24 hour storm, you need 100 years worth of daily (24-hr) storm data.

Likewise, in a very general sense, in order to establish the 100 year 30 minute storm, you need 100 years worth of tipping bucket rain gage data at 30 minute intervals. If you don't have that, which nobody does, you have to interpret it using some pretty sophisticated stochastics.

Hydrology, Drainage Analysis, Flood Studies, and Complex Stormwater Litigation for Atlanta and the South East -

http://www.campbellcivil.com

 

[Hqdo]

Thanks all, now I understand what to explain and request from Client.

 

[cvg]

no, you do not need 100 years of data to establish the 100-year storm. And, no - you do not have one 100-year storm every 100 years.

For instance, NOAA has used as little as 13 years of data, however that results in significantly higher error in the estimated rainfall. In NOAA-14, the average station used in the analysis is about 37 years and none have 100 years of data

http://www.nws.noaa.gov/oh/hdsc/PF_documents/Atlas14_Volume1.pdf

 

[beej67]

@cvg, yes, exactly.

If you don't have that, which nobody does, you have to interpret it using some pretty sophisticated stochastics.

What the big boys do, if I'm recalling properly, is grab as much data as possible and then fit it to a Gumbel distribution.

https://en.wikipedia.org/wiki/Gumbel_distribution

Your statistics pencil needs to be very sharp when attempting this, and it's thankfully not something most design engineers have to fool with. We trust that NOAA's pencils were sharp when they did it. I would probably have to spend days reviewing the appropriate textbooks before I would feel confident tackling this problem in a comprehensive way. I might be tempted to cheat in some manner, like saying for instance, "Well, Foreign Country X has a climate about like Arizona, sooo......"


Hydrology, Drainage Analysis, Flood Studies, and Complex Stormwater Litigation for Atlanta and the South East -

http://www.campbellcivil.com

 

[Hqdo]

Yes Beej, you got it, we don't have that luxury to get every data we need, not even basic set of data. In the worst scenario, I will use nearby country available IDF, pick out an area with similar geographic/climate characteristics.

 

[TerryScan]

Have you tried science based web searches? I find a good bit of hits for "Mandalay, Myanmar hydrology" and "Mandalay, Myanmar precipitation" using scholar.google.com compared to regular google.

While I did not find the exact information you are requesting in the few documents I viewed, I suspect one could contact some of the references in the documents and obtain some decent data.

 

[Hqdo]

Cool TS, I found some great info. !there is a very old document called rainfall in Burma in 1965, wondering how reliable it is after 50 years, but it will definitely good supporting doc for a very rough assumptions.

Thanks

 

[beej67]

That's what factors of safety are for, Hqdo.

Hydrology, Drainage Analysis, Flood Studies, and Complex Stormwater Litigation for Atlanta and the South East -

http://www.campbellcivil.com