XPSWMM Model Calibration (suitable storm event)

[DavidBender]

Hi all,

A brief introduction, I am a civil engineer currently managing a multiple phase design project. The second and third phase of the project is basically a civil design project of roads and a upgrade of an existing drainage network (Urban). However the first phase of the project is predominantly flood modelling to assess the best design for the stormwater upgrade. Since my background is as a civil and structural engineer I am highly dependent on the technical advice of the flooding lead.

We were asked to update an 'existing' XPSWMM model to assess several design options. The brief clearly stated that no historical data is available for calibration. The Council which we are working for had a change in management and asked whether we could collect data and calibrate the model in a short time frame. My engineers informed me that we need to collect a significant storm event (5 ARI +) to provide meaningful and accurate calibration. This made sense for a purely logical standpoint, that it would be difficult to extrapolate conditions for very minor floods to 100 ARI events. I informed Council that it was practical to collect and calibrate the model without major changes to the project duration.

That brings me to my question, what duration events are commonly used by other engineers on this forum for 2d flood model calibration?

Sorry for the small essay!

 

[beej67]

I informed Council that it was practical to collect and calibrate the model without major changes to the project duration.

Sounds to me like you're stuck picking the biggest storm that happens inside your "project duration."

Pray there's not a drought and hope for the best?

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

http://www.campbellcivil.com

 

[DavidBender]

We are in Autum moving into winter so the chances of a significant storm event are low(Sub Tropics).

My question is even if we were to put flow gauges in etc and measure a 1 year ARI storm event (hopefully higher), would this add any addtional value in the model ? and do people commonly do this in flood models ?

To meet it only seems logical that you capture significant overlands flows to be able to calibrate the overland flow behaviour of your XPSWMM model.

 

[froude]

I think any calibration that you can do is probably helpful; at least you have some real data to back up a theoretical model.

However, is just one cycle worth of storm data representative enough? It seems that you would need years worth of data to get a good idea of the basin response.

Also, calibrating to just a 1-year (or another low) event may not fully represent watershed storage adequately (this is based solely on the topography of my area, where most low events stay within the bank-full area).

 

[DavidBender]

Froude,

I fully agree, we have multiple areas of storage (basins, parks etc) which are not utilized in storms smaller than 2-5 years.

I can also see how small storm events would be very sensitive to modelling assumptions and the readings could also be distorted by blockagaes etc. After doing some light reading of flood studies from the UK, I can see they have utilized mutiple data sets (different storm events) to allow them to 'fit' overland flow and channel/ pipe flows, some even match flood extents with ariel photos.

If Council want a model they can rely on they will need to do a propoer flood study (including data collection) or else they need to accept there are limitations to the model and we can progress with the Civil design component.

 

[cvg]

it is quite rare to ever calibrate a model for a site with actual data from the site. only a fraction of the worlds watersheds are gaged. Most calibrations use data from a hydrologically similar site that has been collected over at least 10 years and hopefully more. Find a site, nearby or in a similar geographic / climatic location and use that data. You will need to evaluate the similar site for all factors that influence rainfall or runoff, depending on what type of data you are able to gather. these factors might include latitude, elevation, ground cover, soil type, slopes, proximity to water bodies etc.

 

[TerryScan]

I have seen an approach to model calibration which does not rely directly upon local historical data.

The source of discharge data for FEMA studies follows in order of preference:
-Statistical analysis of guage data
-Regression analysis
-Hydrologic modeling
-flows from another study

While historical guage data is rare for every site, Regression Equations are widely available (a handy tool is: Streamstats)

http://water.usgs.gov/osw/streamstats/ssonline.html)

University of Maryland with the Marlyland Highway Administration uses the regression equations as a target to calibrate TR-20 models in their GISHYDRO application.

http://www.gishydro.umd.edu/

So, it may be possible to calibrate your SWMM model to regression equations for your stream. It may be better than a short duruation guage study.

 

[DavidBender]

Thanks for the responses.

This defiantely adds some more clarifty to the issue.

 

[DavidBender]

Sorry for the typos above, teaches me not to respond in a hurry especially with my typing skills!!