open TELEMAC-MASCARET The mathematically superior suite of solvers

hi everyone..


there's something make me wondering what is the different about calibration and validation?? can someone explain in detail about this.

and if i got a cycle of tidal data from field work (spring and neap - 2 weeks) or historical data... should i used it for calibration or validation? and if i used it for calibration how can i validate my telemac2d model?


hope someone here can explain it detail..

thanks in advance..


regards,


shahirah

Hello,

I am not sure to be right, but I would say that first there is verification, to show with analytical solutions that you correctly solve the equations. Then validation is to show that the physics of your application is correctly represented by the chosen equations. This is the case with Telemac-2D and tides, since tides are long waves correctly represented by Shallow Water Equations. Then calibration consists in finding unknown parameters in the equations, e.g. the bottom friction, with the help of measurements.

With best regards,

Jean-Michel Hervouet

Hi
Another point of view could be:
Calibration, you adjust some parameters to calibrate your model with some measurements.
Validation is just a verification of the results of your model by comparison with some other measurements. This will validate the accuracy of your model in general.

You calibrate on a first event and you validate on another...

Regards

Dear shahirah,

these are technical definitions:

Calibration: Process of adjusting parameters in the model to obtain agreement between modelled values and measured values.

Validation: After calibration the modelled values are compared with measured data which was not used in the calibration process in order to confirm the model’s ability to reproduce the phenomena required.

Verification: Process where the ability of the numerical schemes to correctly represent the underlying mathematical description of the phenomenon being modelled is established.


So in modelling (e.g in TELEMAC), I often do a lot of work on calibration of the model, then calibration. These steps are very important to know the accurate level of the model, what problems from the model.... so that we know how to improve the model later on..

Hope this helps

Best Regards

Huy

thanks guys..


but am i right if i say that, (eg telemac2d)
1. we calibrate our model with historical data so from the percentage error we can adjust the parameter so our model will be same like real situations?

2. we validate our model with current data / site measurement data in confirmation that our model we had calibrate is correct, am i right?

and one more thing that i need a comfirmation, the data for calibration or validation (ie tides, current flow and discharge of river/ flow rates, Q date (for estuary system model))are we need those data in same date and time. for the accuracy of the model? or we can use different date time of data?


thanks in advance..
thanks guys (Jean-Michel Hervouet, c.coulet and Huy) for your time and explanation.. hopefully my question will also help other beginners too..


regards
shahirah