Sediment Rating Curve Calculation and Considerations

Note: We now have a tool in HEC-RAS that helps with these analyses (mentioned at the end of the video). I posted a demo video here: • HEC RAS Sediment Ratin...
This video covers the motivation behind developing a sediment rating curve, walks through the steps of fitting a power function, and then discusses five important considerations that you should address while developing your rating curve, including:
1. Stationary (9:29)
2. Hysteresis (10:26)
3. Transform Bias (13:40)
4. Supply Limitation (18:50)
5. Serialized Correlation (21:16)
The USGS statistics manual is here: pubs.er.usgs.gov/publication/...
And the USGS tool to evaluate Serialized Correlation is here: water.usgs.gov/osw/SALT/SAID
I posted the dataset here: drive.google.com/file/d/14dbu...

[@stanfordgibson]

Note: We now have a tool in HEC-RAS that helps with these analyses (mentioned at the end of the video). I posted a demo video here: th-cam.com/video/278wsFaVtBc/w-d-xo.html

[@rammahmohammed1840]

You don't know how helpful your video was to me, thank you so much!

[@mesertbekele3951]

the only tutor in youtube regarding sediment rating curve

[@stanfordgibson]

Thanks Mesert. The next version of RAS (which will be out very soon) will include a tool to help with these analyses: www.hec.usace.army.mil/confluence/rasdocs/rasum/latest/hydraulic-design-hd-calculators/sediment-rating-curve-analysis-tool

[@rodrigoperdigao9074]

That's a great video! Thanks for sharing!

[@SanyamGhimire]

Thanks for the information. Please upload other session of the workshop as well. . Your fan and tutee...

[@stanfordgibson]

Sure Sanyam. I'll try to record a couple of the live sessions and put them up.

[@gebiawayele9062]

Thank you

[@richman1905]

Why go through the hustle of log-transforming the data and using a complicated equation to avoid a transformation bias? Why don't people fit a non-linear model in euclidian space from the get-go? Is it just that log-sapce is better for visualization or is it a historic thing?

[@angelitoagrimor5059]

hello sir do we need to gather actual data in our river to determine the load concentration. i have a problem gathering data about sediment loading and concentration. Any method to suggest. Thank you

[@modellazioneidraulicagis]

Thank you Stanford for video! Now, If I have a River that not have sediment transport measures, there is a way to create sediment transport rating curve? In USA you have a great dataset about many rivers, for the most part of countries not! You have ever studied this condition? Best Regards

[@stanfordgibson]

Yeah, that is harder. Though, you see with our "good data" there is still an order of magnitude of scatter. So even rating curves based on data are still often calibration parameters.

[@stanfordgibson]

You have to start with a sediment budget to constrain the answer. You have to estimate what the annual load per year is, either by either looking at sediment load data or reservoir deposition in similar watersheds. Then start with a simple power function (something like Load=aQ^2 unless you have a better idea for the power) apply it to your flow series, and solve for "a" - so that your rating curve generates the annual yield from your sediment budget.

[@stanfordgibson]

You can also run RAS with your flow series and the equilibrium sediment boundary condition. Then plot Mass In (on a daily, cumulative interval - or just do Concentration which isn't time dependent) verses flow at the upstream XS. This will give you a second estimate of your rating curve. You can try it with several applicable transport functions. However, this will be very sensitive to the hydraulics of that US XS and the bed gradation at that XS...which is why we don't recommend using the Equilibrium BC.

[@stanfordgibson]

But these are all just starting points. The real answer is that if you don't know the sediment flux, it is almost certainly your most sensitive and least certain input, which makes it your calibration parameter. Go ahead and do whatever you need to do to guess the rating curve, but then adjust it to match observations of bed change or downstream concentration.

[@stanfordgibson]

I will do a video at some point about calibration. But if you don't have calibration data a sediment model is pretty speculative...but you might be able to do a "credibility test". Run a long time series and adjust your rating curve until you get the approximate, historical, bed change trajectory (e.g. deposition, equilibrium, erosion). But without a rating curve or a quantitative calibration, you can only really count on a sediment model for qualitative trend analysis.