Home >> Forums >> Calibration of Osmose model

Calibration of Osmose model

Dear Sir/Ma'am,
I have come across a few scientific papers that reveal as to why they have done calibration, while some documents reveal working examples using calibrar package (for example: PredatorPrey Model).
I would like to know if there is a guide/User's manual exclusively for calibration of Osmose model (version 3 update 2).
If not, is there any document,which you can share, that contains a working example (or demo) for calibraton of Osmose model using calibrar package?
Are there any alternatives to calibrar package when it comes to calibration of OSMOSE?
Kind regards,
Tarun Joseph

Dear Tarun,

The calibration process is still work-in-progress, which means that the tool is not ready to be shared publically (not user friendly yet) and the documentation is not finalized though we have a draft I can share by email.

Ricardo Oliveros Ramos spend his PhD working on the calibration aspects of the Osmose model and he dig into the matter of calibrating complex ecosytem model quite deep. You may have other packages available in R (though I cannot recommend any, by lack of personnal knowledge on the subject) but I have no idea whether they would give satisfaction for calibrating the Osmose model.

You may contact Ricardo Oliveros Ramos directly and ask him whether there is any updated version of the package for calibrating Osmose. Otherwise we do have older versions of the code and a bit of documentation that may help for a basic calibration of your configuration.

Cheers, Philippe

Hi Philippe,
Could you share your insight with respect to the following queries?
Query 1:
Since the calibrationEA and OSMOSE2R packages are currently work in progress, could you tell me about the options that are available to the end users of OSMOSE to start off with calibration process (given that we have the required observed data to do so) after having downloaded OSMOSE model (version 3 update 2)?
Query 2:
From what I have read so far, calibration of OSMOSE model allows the user to estimate parameters such as larval mortality, plankton availability coefficients, etc from the observed data for a given year (Kindly do correct me if I'm wrong here). Just to make ensure that I have not missed out on any important detail, I would like to know if there are any parameters (other than larval mortality, plankton availability coefficients) estimated through the calibration process (like species relative fecundity)?  Would it be possible for you to provide a template for the output file generated through calibration process?
Query 3:
Would it be possible for you to share (through a mail) a template for the observed data (which the calibration code requires) along with the older version of code and it's documentation available for basic calibration of OSMOSE (which you have mentioned in your earlier response that it's available)?
Query 4:
Afer the calibration process, we would have a list of values for parameters such as larval mortality,plankton availability coefficients, etc.  I understand that for calibration to be labelled as a success, we need to have the simulated values (of catch, biomass, etc using the values for larval mortality, plankton availability coefficients obtained from calibration) to match with the observed data. However, I would like to know if the end-user should be aware any important detail (such as any specific criteria based on which simulated values should be weighed against the observed values) when it comes to matching simulated values with observed data to know for sure if our calibration process has been successful or not.
Kind regards,
Tarun Joseph

Hello Tarun,

You are asking very relevant questions about the calibration process and I may not be able to answer everything in detail as I explained I mostly work on Osmose whereas the calibration was mostly developped by Ricardo O. and Yunne S.

I will send shortly a template of the calibration. It should be easy enough, with a bit of guidance to tweak it in a way that you may estimate as well relative fecundity. Just be aware that the more parameter you want to estimate the more data (at leat biomass and catches estimates) you need for driving the EA toward a satisfactory solution.

The output file of the calibration is a CSV file that provides values for the estimated parameters at each generation. By the end of the calibration process, the estimated parameters should lead to simulations with levels of biomass and catches that are close enough to the ones provided in the DATA/ folder.

Cheers, Philippe