I have to decide for the use of a big size grid with a high resolution or the use of nesting with two /three grids to simulate wind forced circulation and upwelling in an area of 400 km x 400 km approximately.
I think to use QuickScat data, and Da Silva (winds) and World Ocean Atlas (T and S) climatologies.
Is there a criterion to choose the grid parameters for nesting (specially resolution and size / extension of each grid) ?
Bye
Is there a nesting criterion ?
Do you have a target resolution in mind? You want your smallest domain to be 400 km x 400 km at a few km? For the forcing and boundary files, there are issues of both spatial and temporal resolution. To get the best possible upwelling, you want fine space and time scales in your forcing. For the boundary conditions, we've been having not too bad luck using 1 degree global fields every four days for our 10 km simulation. I used to have much more trouble using monthly Levitus T,S fields, but I don't know which of the many changes to credit for our improved success - consistent u,v boundary fields, higher temporal resolution, or ROMS changes.
The current product we are going to try is SODA from Jim Carton's lab. This is a global model simulation with data assimilation saved at 5-day intervals. We would hope to use this to drive a 10 km domain, with nests of about 3 km inside. The atmospheric experience is that a nest ratio of 3-4 is about as far as you want to go and we are just lucky to be getting away with the drastic ratio of 10 on our outermost domain.
You do want some buffer zone between the domain boundary and the region of interest. The size will be determined by what you can afford, the timescales for things to get from one to the other, the timescales of interest.
The current product we are going to try is SODA from Jim Carton's lab. This is a global model simulation with data assimilation saved at 5-day intervals. We would hope to use this to drive a 10 km domain, with nests of about 3 km inside. The atmospheric experience is that a nest ratio of 3-4 is about as far as you want to go and we are just lucky to be getting away with the drastic ratio of 10 on our outermost domain.
You do want some buffer zone between the domain boundary and the region of interest. The size will be determined by what you can afford, the timescales for things to get from one to the other, the timescales of interest.
Kate,
Really I have thought in three different domains of interest:
1. a first of 2000 km x 1000 km (resolution ~ 20 km)
2. a second of 400 km x 400 km (resolution ~ 5 km)
3. a third of smaller with resolution ~ 1 km
to get a ocean picture in regional and meso scales, and then to decide for nesting.
Although I have decided for climatological T-S fields and dayly QuickScat winds certainly I could do changes to improve the results choosing other sources of data.
Really I have thought in three different domains of interest:
1. a first of 2000 km x 1000 km (resolution ~ 20 km)
2. a second of 400 km x 400 km (resolution ~ 5 km)
3. a third of smaller with resolution ~ 1 km
to get a ocean picture in regional and meso scales, and then to decide for nesting.
Although I have decided for climatological T-S fields and dayly QuickScat winds certainly I could do changes to improve the results choosing other sources of data.