How much does timestep matter?

[Tucker Furniss]

Hi folks,

In what ways does running succession at a 1-yr timestep differ from running at 5, 10, or 20 year timesteps? Has anyone directly evaluated the effect of timestep on model behavior? 

Generally speaking, I understand that the succession extension (NECN, in this case) will compensate for a timestep greater than 1 year by multiplying growth, mortality, and dispersal by the desired timestep. In other words, biomass for a cohort at year 50 should be approximately equivalent if succession at either 1 yr timesteps or 10 year timesteps. Has anyone tested this? And what about dispersal and establishment of new cohorts? I can imagine 1-yr timesteps may simulate more fine-scale recruitment dynamics between competing cohorts if their growth/mortality is re-calculated every year, compared to a 10-yr sum. 

If anyone has experience with this, I'd appreciate hearing about what discovered!

Thanks,

Tucker

[Robert Scheller]

Tucker,

Time step has huge consequences, particularly for succession.  More frequent time steps means more regeneration, and therefore more CPU required to evaluate neighborhood seed sources, and more regeneration means more cohorts (more RAM required and more growth calculations).  This will be true for any succession extension.

Our original 2007 paper explored the effect of time steps; maybe still useful?

Cheers,

Rob