Scheduled Downtime
On Friday 21 April 2023 @ 5pm MT, this website will be down for maintenance and expected to return online the morning of 24 April 2023 at the latest

Questions about time step

This post was from a previous version of the WRF&MPAS-A Support Forum. New replies have been disabled and if you have follow up questions related to this post, then please start a new thread from the forum home page.


Hi WRF Help,

I have a few questions about the WRF model time step.

I am running a two-way nested real-case simulation with three domains. The coarsest domain has a spatial resolution of 9 km. I set my parent_grid_ratio to = 1, 3, 3, so my two nested domains have horizontal resolutions of 3 km and 1 km respectively. However, because I also have a very fine vertical resolution in the lower levels (~20 m), I set my time step to be 12 sec. to keep my simulation numerically stable. I also set the parent_time_ step_ratio to = 1, 3, 2, so the time steps for d01, d02, d03 are 12 sec., 4 sec. and 2 sec., respectively.

My questions are:
1) Is using an even value (e.g., 2 for the third domain as described above) for the time step ratio a problem? I believe it is generally recommended to use odd ratios (3 or 5) for the parent_grid_ratio, but I read that the time step ratio can be different from the parent grid ratio. Though, I am just unsure if using an even ratio is OK.

2) I realize that the 12/4/2 seconds time steps are probably more than small enough for my domains, but in my understanding, there is no lower limit on the time step. Is this correct?

My simulation completed fine. I was just curious about the answers.

Many thanks,

Hi David,
Those are good questions!
1. Yes, it is okay to use an even value for the time step ratio. The reason why we recommend an odd ratio (with feedback on) for the spatial ratio is so that the center of each parent grid point will coincide with a center point of a child grid, if that makes sense. Picture a single square (parent grid cell) with 9 grids inside (3x3). There would be a center grid cell that would have a center point that would be right on top of the very center of the single parent cell. As for timing, even numbers are perfectly fine.

2. That is correct. As long as you are okay with the slowness of the simulation with a small time step, there is no problem doing it this way.