I am very new in hydrodynamic modeling. Recently, I have done a simulation in SMS-SRH 2D. Now, I want to perform calibration and validation of my model. But, I have no basic on calibration and validation. Can anyone give me some documents or tutorial or guidelines on calibration and validation of numerical model which will, not only build my basic but also make me able to perform on my existing simulation?
Some specific questions:
1) What are the parameters need to be used for fine tuning of model during calibration?
2) If I have limited observed data, will model give satisfactory result?
3) During validation, shall I use different cross section of another time period for my model domain?
You need to use Matlab for fitting your experimental and simulation data. Extract your data first and then check it with your simulation. RMSE can be helpful for evaluating your modelling.
Hamidreza Rahimi , thanks for your reply but I want to know more detail information about calibration and validation of hydrodynamic modeling.
I would strongly recommend reading this excellent textbook on 2D modelling by Prof. Pasternack from UC Davis:
https://www.amazon.com/Modeling-Ecohydraulic-Analysis-Gregory-Pasternack/dp/1466320095
There you will find many useful insights and real-world examples of how to calibrate and validate your model.
Six main public domain water quality models which are presently available for Rivers and streams are being captured in this article. These main models could produce important results if they are used in the correct manner, because they are different in terms of assumptions, strength and weaknesses, processes they represent, modeling capability and data input requirements. The Model review discussed includes, water quality analysis simulation program (WASP7), simulation catchment (SIMCAT), quality simulation along Rivers (QUASAR), and the temporal overall model for catchment (TOMCAT), QUAL2KW, QUAL2EU. The models are described individually according to a consistent set of criteria-conceptualization, model capability, model strengths, limitations, input data and how it utilized. The outcome showed that TOMCAT and SIMCAT are important in ASSESSING effect of point sources in a very simple way. The QUAL2KW, unlike the QUAL2EU where macrophytes play a major interaction, it can convert algal death to carbonaceous Biochemical Oxygen Demand (CBOD), thereby making it more suitable. In addition to the extensive requirement of data, it is expensive and time consuming to set up these complex models such as QUASAR and WASP7. Therefore, one model cannot be used for all the required functionalities. Choosing a model would depend on a specific application, financial cost and time availability. This article may be of help in choosing a suitable model for a specific water quality problem.
Article Water Quality Model for Streams: A Review
Hi, you are talking of hydrodynamic mathematical 2D model. For calibration and / or verification studies, you would require a few prototype hydraulic observations, like velocities with direction, depths of flow, water levels at different locations. Corresponding discharges are of critical importance. If gauge discharge curves are available within the reach reproduced in the model, then it is of immense value.
You will require discharge hydrograph for control of discharge at the upstream end and a gauge discharge at the downstream end of the model for water level control.
As you are at the "Verification stage", it is assumed that these are already available and used for the studies.
Initially, you are required to run the model with the specified discharge values at the upstream end and water level control curve at the downstream end.
The first stage is to reproduce the average water surface slopes for the different discharge stages. For calibration of the same, the Mannings' value on n needs to be corrected / adjusted.
After getting satisfactory results, for the specific discharges (Depending up on the available prototype data) you may have to look at the flow conditions (Velocities with direction, depth of flow, etc) for the specific discharges.
Most of these verification studies are problem-dependent. Each and every parameter observed at the site may or may not be correctly reproduced all the time. A description of the modeller is many times required.
Yes, Ravindra Oak is absolutely right.
To calibrate a fixed bed hydrodynamic model implemented in SMS, you must take care of the roughness coefficient and you have to spend time in shaping the flexible mesh in order to represent the right topography as much as you can.
You should simulate a past event, and if available, find the survey of that event representing the extension of the flood and some water depth spread in the floodplain. This data is truly useful when you are dealing with 2D modelling.
The need to calibrate any phisical variable data from the field to comparate with the model results. You should select some variable and move it in range till have same results in simulation with the data in the field.
You can also consult the tutorials provided by SMS, they have some practical examples of calibration based on roughness and the water depth. See for example the tutorial on applying observation coverages:
https://www.xmswiki.com/wiki/SMS:Observation_Coverage
Thanks, Jeffrey Tuhtan, for your generous reply and help.
Gregory Pasternack gave me some tutorials made by him based on his book mentioned by you before. Those are very helpful. Thank you again.
Since you are doing 2D modeling, you may think about the advantages of 2D over 1D modeling, and then set you calibration and validation strategy to confirm/verify those advantages. For instance, a 2D model will give you the transverse variation of water level and velocity; you might want to check how well the 2D model performs in this regard.
You may use the Manning's roughness coefficient over the 2D domain (rather than the reach length) as the calibration parameter.
One should use a separate time domain for validation.
The calibration of the model is divided to flow and sediment calibration:
** First Flow calibration by two ways
1- Compare the water surface profile (water level) from the model with the observed (measured) water level
This based on bed roughness (Manning Coefficient "n" )
2- Compare velocity distributions from the model with the observed (measured) velocity distributions
* You can compare the Flow for different discharges (different scenarios) low, average and high flood
** Second Sediment calibration
1- First enter the measured cross sections for specific year for example (2010)
2- Compare the output cross sections from the model (simulated) with the measured cross sections for a specific year for example (2020)
This based on sediment equation
- Badges
- Science topic
- Similar topics
- Computer Science and Engineering
- Computer Graphics and Animation
- 3D