nested mesh level 1 and 2

Question

Hello, where can I find information about creating a nested system 1 and 2. They told me that in a level 1 nested system I should run the model only propagating the tide and then in a level 2 nested system the tide information is received and there I add wind, currents and temperature. But I don't know where to start to do such a thing.

Luis Fernandes · Answer

Hello, 
 Thank you for your reply. Please keep in mind, that to perform a downscaling for future projections you'll need boundary conditions for that period (2006-2100), including meteorological forcing and biogeochemical data. 
 You mention that tide is important. Does the MPI-ESM-MR solution includes tide, or just general low frequency circulation (density gradients and low frequency atmospheric forcing)? If it does not, then the "sea level height" you mention is actually "sea level anomaly", right? 
 To be able to use MOHID Water in OpenFlows FLOOD to downscale from a global solution with no tide, and add tidal forcing to the solution, you'll first need to setup and validate a tide model of the gulf of California. Have you done yet? 
 You can use the instructions described in this article . 
 I'll wait for your feedback. 
 Kind regards, Luis

Luis Fernandes · Answer

Hello, 
 The "WaterPoints3D" dataset in the Grid group of the HDF5 file is 3D array with all zeros. This dataset is a land/water mask. It must be 0 for grid cells that are land, and 1 for grid cells that are water. 
 I would recommend that before you setup the biogeochemical model you focus on getting the 3D hydrodynamics correctly simulated. If the hydrodynamics is not well simulated there's no point in coupling the biogeochemical model. 
 Kind regards, Luis

Luis Fernandes · Answer

Hello, 
 I thought you had created the file using your matlab tool. I see now you're using ConvertToHDF5. 
 I believe the problem is related with the way ConvertToHDF5 derives the land mask from the netcdf data. 
 You have set in the grid settings these options: 
 NETCDF_NAME_MAPPING : chl MAPPING_LIMIT : 1 
 Keep in mind that NETCDF_NAME_MAPPING is the name of the variable ( chl) in the netcdf file you're going to use as a reference to compute the land mask and MAPPING_LIMIT : 1 means that the land mask will be created using the grid cells where chl values are lower than 1. If you set this to zero it will probably work correctly. 
 Let me know please. 
 Thanks and regards, Luis

Luis Fernandes · Answer

Hello, 
 Before adding any other type of forcing, you should validate your tide model. If your "downscaling" of FES2014 to your model domain is not correct, there's no point in making the model 3D and adding more forcing on the boundaries (open boundaries and surface/meteo boundary). 
 Have you calibrated/validated your tide model comparing the model with observations taken at different locations inside the domain? 
 You should select and run a period of at least 15 days to compare with observations. 
 When you have confidence in your results you can then think of adding more forcings. 
 Keep in mind, that you should use the slow start option in the Hydrodynamic module tide options. Set a value around 1-2 days (i.e 86400/172800 seconds) to ensure the "warm-up" part of the simulation does not generate too much numerical noise that the model may then have difficulty dissipating. 
 Kind regards, Luis

Luis Fernandes · Answer

Hello, 
 I don't know the accuracy of JTides (I can't access the webpage), but for validating the model here's things you should pay attention to: 
 1) check the time reference you define in the Tide input options. FES2014 is set in UTC so you need to set -6 or -7 if you want to compare to local time data. 
 2) the open boundary condition of your model should be placed outside the continental shelf, in deeper areas 
 3) As the gulf is not very wide, make sure the model resolution is fine enough to represent how circulation occurs. Check the real coastline together with the "gridded coastline" and check for inconsistencies. 
 4) Also when comparing with data make sure the tidal gauge where the observations are taken is well represented in the model. From experience I know that if the model is coarse and you compare with tidal data in La Paz, if the bay is not well represented in the model then the results will not be good, because of the local circulation in the bay. 
 Can you show some of the results comparison? You can also upload the model (without the FES2014.hdf5 and without results) for us to have a look of what might be wrong. 
 Thanks and regards, Luis

Luis Fernandes · Answer

Hello, 
 When comparing modelled and observed sea levels you must be aware of what's the bathymetry referential and what's the gauge referential. 
 Your bathymetry, I assume, is derived from a global database and reference to global mean sea level. But the tidal gauge assumes a different referential. At the scale of your model (0.1 degrees ~10-11 km), the approximations you are assuming in terms of bathymetry are not important in terms of simulating the sea level. Nevertheless you should select a reference level. If you're comparing model results with different gauges it may be that different gauges have different reference levels, so you'll need to post-process your results and add or subtract the difference between the selected reference level and the gauge reference level. 
 In your case, you had selected 2.4m from Mazatl&aacute;n, but JTide for La Paz is using zero. You'll need to adjust so that they match. 
 As I had mentioned, in your model you only have 3-4 cells to describe the bay of La Paz and that may be insufficient to model how tide propagates there. I would suggest increasing your model resolution. It does not mean the model is entirely wrong. Other gauges along the coast may show better results because they don't have the recirculation patterns of La Paz. But to get it right at La Paz you may need a finer grid (or a submodel). 
 Does the right-hand side graph shows both JTide and MOHID2.4? Do they match? It's not very clear what the graphs represent. 
 Finally I would suggest that you expand the grid south (~20&ordm; N) so that the open boundary is not directly facing and very close to Cabo San Lucas (red square in the image below) and also avoiding possible less accurate data in used tidal harmonics database in the area around the red circle in the image below. 
 Kind regards, Luis

Luis Fernandes · Answer

Hi Metzli, 
 Yes, that's it. You need the data from CMIP5 to completely cover the extent of your domain. 
 Thanks and regards, Luis