Home >> Forums >> Number of model runs required

Number of model runs required

Dear Sir/Ma'am,
Since OSMOSE model has the provision to incorporate a stochastic mortality algorithm, we have the following query regarding the same.
 
How do we as an end-user decide upon the number of model runs required if we choose to set nature of mortality algorithm as "stochastic"? Is there a criteria that the end-user needs to be aware of while deciding upon the number of times the model needs to be run?
 
We look forward to your response regarding the same.
 
Kind regards,
Tarun Joseph.

 

 

Dear Tarun,

Good question and there is no general answer, it all depends on one's configuration. How many replicated simulations should you run to make sure that the outputs are statistically meaningful ? As much as needed so that the standard deviation of the averaged Osmose outputs does not vary anymore. Which Osmose outputs should you take into consideration ? I'd say the one you judge that are relevant for capturing the variability on your simulations : biomass, catches, mortality ?..

Let me know if you'd like me to clarify or elaborate.

Cheers, Philippe