I need to run two models with different time step. The output from cyclic model A (with shorter time step) is used as input in model B (longer time step). Intuitively, it feels that there must be a way of running them in concert...Would appreciate any tips of how to.
Thanks/Elena
Dear Elena, I know nothing about Stella but the problem is common in other fields. I guess it suffices to take average of A outputs over the longer sampling period and use it as input to B. The end point on a straight line fitted to A outputs is another, a bit finer, option.
MK
Thanks, Miroslav. Yes, I understand that this is a rather usual task but I was wondering what would be the way to solve it in Stella... the issue is two (or, potentially, more) time scales that need to run simultaneously. The "finer" option is a very nice idea, I have to think about that.
I expect you will solve it to your satisfaction. Do not hesitate to contact me if you will need a further help.
Miroslav
You have probably solved the problem or moved on, but I will make a suggestion anyway. I use STELLA quite a bit, and have faced the same issue sometimes. I just accumulate the faster time-step and when it reaches the longer time step, I increment a counter - I use a stock with one inflow which equals 1 at all times and the outflow becomes the value of the entire stock when it reaches the number of shorter time-steps that make up the longer time-step. At each of the finer time-steps, I check to see if the outflow of the stock is > 0; if it is, then I execute the second 'model', otherwise nothing happens. In that way, the second model only executes when the counter increments. [I hope you can follow this explanation - it is a little difficult to put into words. I could do up a simple model if the above is not easy to follow.]. Joe
Hi Joe, thanks a LOT! No, I have not solved it yet (at least in an intelligent way) and put this model on hold:) Will do what you say (sounds great and I should be able to do that), but in a few weeks...I am on a fieldwork at sea right now. Can I get then back to you by mail if I get stuck?
I have had a similar issue before and I approached it by creating a module that fed the data as a conveyor. The module was run at one time step with the data being fed into the model during the modeled time step.
- Badges
- Science topic
- Similar topics
- Geoscience
- Atmospheric Sciences
- Climate Science
- Climate Change