Changes between Version 9 and Version 10 of 2016WP/Noc Wad

2016-06-09T10:32:04+02:00 (4 years ago)


  • 2016WP/Noc Wad

    v9 v10  
    3213211. Main points 
     323A. History and Objectives: Jason and Hedong have previous experience of WAD in POLCOMS and FVCOM respectively. They have been working on WAD in NEMO since 2012. WAD is a key capability for NOC and it would be valuable for NOC to have it in NEMO. The Met Office needs WAD to work in NEMO for an upgrade to CMEMS north-west European shelf system expected to be required within 1-2 years. The aim is to implement  a non intrusive, accurate, stable and efficient formulation of WAD within NEMO. In order to 
     324be non intrusive a scheme using a minimum depth of water (@ 5-10 cm) at all potentially wet point is being implemented (similar to POM and ROMS implementation).  
     325We should publish results of our implementation. To do this we will need to compare results with other schemes (e.g. POM and ROMS implementations) for  
     326simple test cases.       
     328B. Review of WAD code: The main changes to an ocean model and issues associated with an implementation of WAD are:  
     330- a time-varying land-sea mask (wdmask) is required; the initial set up of the grid needs to allow negative depths (domgr) 
     332- the depth of water at ocean points cannot be allowed to go negative; some care with flux limiters is needed to ensure this. The splitting of the barotropic and baroclinic timesteps is complicated by cells wetting and drying within the baroclinic time-step 
     334- the thin film of water at dry cells results in an artificial horizontal pressure gradient (hpg) along a sloping bottom 
     336- some implementations (e.g. POM) have adjusted the bottom friction  
     338- the 3D equations can be unstable in a very thin layer of water (with many levels); e.g. problems with CFL for vertical advection or diffusion  
     340C. The WAD implementation by ROMS is particularly simple. The flux limiter simply specifies that there is no flux out of a dry cell. This allows 
     341wetting of dry cells and (in the no wind case) ensures that the artificial hpgs have no impact on the model  
     343D. The time-step splitting could give stability problems. In order to separate issues it would be valuable to run without time-step splitting.  
     344We can do this by removing the updates calculated from the barotropic time-steps that are made on the baroclinic time-step. 
     346E. Andrew presented results using Hedong's code for a tidally driven sloping channel. It seems that when the tide is going out, strong horizontal 
     347sea surface height gradients develop (before the layers dry out). Either the pressure gradients are being mis-calculated or the outward 
     348fluxes of water are being mis-calculated.  
    3253502. Actions