latitudes north of the 47th parallel

[alan.wa...@hycom.org]

Most people who would normally answer are at the LOM meeting this week. I can answer on Friday once I am back in the office. One way to find the array location is to download a full globe netcdf file, e.g. a 2D file since they are smaller. Any full globe netcdf file will include lat and Lon arrays. Or wait until Friday.

Alan.

[alan.wa...@hycom.org]

Starting with a 4500 by 3297 array and counting from 1 (i.e. the Fortran convention),  The box you want is 3199,2051 to 3574,2575.  Note that netCDF might put the 3297 dimension first (i.e. it might flip these indexes) and might count from 0.

Alan.

On Monday, May 20, 2013 9:54:35 AM UTC-5, Rebecca Ross wrote:

Hi All,

I am another beginner and am only interfacing with Hycom outputs rather than custom runs. I am hoping for some guidance on using hycom in areas north of the 47th parallel. I realise this falls within the "bipolar patch" and was wondering if anyone could help me work out how to find my area of latitude as it is represented in the hycom netcdf?

Initially I am looking to just find my coordinates to download subsets of the netcdf files through OPeNDAP: my area is the north east atlantic 40N to 65N and -30E (330) to 0E (360). 

a) can I extract this at all easily from the HYCOM + NCODA Global 1/12° Analysis netcdf outputs (I am flexible on year as at the moment I am trialling this), or am I mistaken to think this area is included?

b) if it is there (as I think it is due to visualising the global netcdf and managing to grab my area through the netcdf subset service) then how do I target this latitude using x/y indices as required by OPeNDAP/ how do I figure out the latitude translation?

Like I say I have managed to get it through NSS, and can use a wget script if I have to, but I am using a software which will interface directly through OPeNDAP so it seems worthwhile to figure out how to use that service in this context. It would be mighty helpful if the lat long to ij tool were made available independent of the source code for we output users - as I will need to, down the line, simulate the release of particles from particular locations which I will also have to locate on the grid, and i am sure there are many others like me with similar requirements.

Many thanks for your help,

All the best,

Bex

[alan.wa...@hycom.org]

Using the Fortran conventions (4500  by 3297 counting from 1), the box you want is 3199,2051 to 3574,2575.  You may have to flip the indexes and subtract 1 for netCDF's C-convention arrays.

[Nathan Putman]

Thanks for the question Bex and the reply Alan!

Along this same line of thought is there a method to transform the spherical coordinates back to Mercator? The particle-tracking routine I am using seems to need a consistent coordinate system to work. So, for instance, I can track particles south of 47 N or north of 47 N but not across this line. It seems like there should be a simple solution for this, e.g., grab the desired lat-lon area following Alan's methods into an array and then do some computation to find and move your northern longitudes into alignment with the southern ones. 

Any suggestions on this (or even just a better description of the problem) would be appreciated.

Thanks!

Nathan

[alan.wa...@hycom.org]

There are two classes of solution to this.  The first is to interpolate the curvi-linear grid to a recti-linear grid and the second is to stay on the native curvi-linear grid.

For GOFS 3.01, i.e. for 90.9, we do provide most fields interplolated to a uniform cylindrical (i.e. constant lat and lon grid increment) 2/25 degree global grid.  These are the only fields being served by the Navy to its intenal users.  These are available (i.e. publically released), but are not being served by hycom.org.  This is an example of the first approach.  Their major advantage is a very simple recti-linear layout, at the cost of some loss in horizontal resolution.

We don't make the 2nd approach very easy, because we rotate the velocities from x-wards,y-wards (needed for native-grid particle tracking) to eastwards,northwards above 47N.  However, in principle all you need is HYCOM's "metrics" file (arrays of lon,lat and grid spacing and grid angle w.r.t. eastwards) to implement curvi-linear native-grid tracking by array lookup.  Since HYCOM's tripole grid is based on three invertable projections merged together in a consistent manner (uniform cylindical south of 66S, Mercator from 66S to 47N and Arctic dipole north of 47N), it is in principle possible to implement native grid tracking without using an array (metrics) lookup approach but this isn't straight forward.

Alan.

[Rebecca Ross]

Brilliant, many thanks for your help Alan, and best of luck to you Nathan!

Bex

[Nathan Putman]

Much appreciated, Alan.

Thanks, 

Nathan