Re: White band at edge of domain

[Steve Cousins <cousins@xxxxxxxxxxxxxxxxxxxxxxx>]

On Wed, 28 Jan 2004, Jonathan Callahan wrote:

> Steve,
> 
> What version of Ferret are you using?

Hi Jon,

 ! NOAA/PMEL TMAP
 ! FERRET v5.51
 ! Linux 2.4.3-12smp - 03/05/03

I need to reboot one of these months!  I've updated the kernel a number of
times but haven't done a reboot in a while.

Steve


> 
> 
> -- Jon
> 
> 
> Steve Cousins wrote:
> 
> 
> 
> On Tue, 27 Jan 2004, Jonathan Callahan wrote:
> 
>   
> 
> 
> 
> 
> Steve,
> 
> The first thing that occurs to me is that the modulo flag might not be 
> set for the longitude axis in the netCDF file.  This would cause the 
> behavior you describe.
>     
> 
> 
> 
> 
> Hi Jon,
> 
> I've changed the data set to have:
> 
> dimensions:
> 	time = UNLIMITED ; // (12 currently)
> 	latitude = 116 ;
> 	longitude = 100 ;
> 	depth = 25 ;
> 	latitude_u = 116 ;
> 	longitude_u = 100 ;
> variables:
> 	float time(time) ;
> 		time:long_name = "time" ;
> 		time:units = "months since 15-DEC-2003 00:00:00" ;
> 		time:time_origin = "15-DEC-2003 00:00:00" ;
> 		time:modulo = " " ;	
> 	float latitude(latitude) ;
> 		latitude:units = "degrees" ;
> 		latitude:long_name = "Latitude" ;
> 	float longitude(longitude) ;
> 		longitude:unitsL = "degrees" ;
> 		longitude:long_name = "Longitude" ;
> 		longitude:modulo = " " ;
> 	float latitude_u(latitude_u) ;
> 		latitude_u:units = "degrees" ;
> 	float longitude_u(longitude_u) ;
> 		longitude_u:units = "degrees" ;
> 		longitude_u:modulo = " " ;
> 
> etc.
> 
> Unfortunately it doesn't help as far as Ferret goes.  Ansley suggested a
> quick test doing:
> 
> 	define axis/modulo axisname
> 
> I tried a number of variations of this with no luck, however, I was
> successful with:
> 
> 	set axis/modulo longitude
> 
> So, that works with FerretL.  The data I'm working with is actually on a
> restricted LAS server and LAS doesn't seem to be picking up the
> information about the modulo longitude variable (I did addXml.pl and then
> make) or something else is going on because LAS still shows the White
> line. 
> 
> With the modulo attribute set in the netCDF file should I have to do
> anything else in Ferret (define axis, or set axis) in order to have ferret
> overlap the longitudes correctly?  With modulo set should I be able to do:
> 
> use l.nc
> fill/X=-180:180/k=5/Lt=1 DIAT
> 
> and not have the line under Africa?  Currently this works only if I also
> do: set axis/modulo longitude.  This would be fine if I was just using
> Ferret, but with LAS it might complicate things a little.  Do I need to
> put something in the las.xml file for this?  Depending on the answers I
> get above I'll move my questions to the LAS email list.
> 
> Thanks very much,
> 
> Steve
> 
> 
>   
> 
> 
> 
> 
> 
> -- Jon
> 
> 
> Steve Cousins wrote:
> 
>     
> 
> 
> 
> 
> 
> 
> Hi,
> 
> I have a netCDF file that has longitudes from 23.4 degrees to 379.8
> degrees stepping by 3.6 degrees.
> 
> longitude = 23.4, 27, 30.6, 34.2, 37.8, 41.4, 45, 48.6, 52.2, 55.8, 59.4,
> 63, 66.6, 70.2, 73.8, 77.4, 81, 84.6, 88.2, 91.8, 95.4, 99, 102.6, 106.2,
> 109.8, 113.4, 117, 120.6, 124.2, 127.8, 131.4, 135, 138.6, 142.2, 145.8,
> 149.4, 153, 156.6, 160.2, 163.8, 167.4, 171, 174.6,178.2, 181.8, 185.4,
> 189, 192.6, 196.2, 199.8, 203.4, 207, 210.6, 214.2, 217.8, 221.4, 225,
> 228.6, 232.2, 235.8, 239.4, 243, 246.6,L 250.2, 253.8, 257.4, 261, 264.6,
> 268.2, 271.8, 275.4, 279, 282.6, 286.2, 289.8, 293.4, 297, 300.6, 304.2,
> 307.8, 311.4, 315, 318.6, 322.2, 325.8, 329.4, 333, 336.6, 340.2, 343.8,
> 347.4, 351, 354.6, 358.2, 361.8, 365.4, 369, 372.6, 376.2, 379.8 ;
> 	
> If I do:
> 
> 	fill/k=1/l=1 SSH
> 
> It looks fine.  If I do:
> 
> 	fill/i=21:120/k=1/l=1 SSH
> 
> to shift the the plot then there is a vertical white line that goes
> through Africa.  You can see this at:
> 
> 	[ http://rocky.umeoce.maine.edu/images/problem.gif ]http://rocky.umeoce.maine.edu/images/problem.gif
> 
> I assume this has to do with some overlap problem but I
> haven't seen it before with other similar data sets.  I've checked the
> data and there is data in the last column so it shouldn't be interpreted
> as land.  For instance:
> 
>  379.4641, 382.4534, 358.5003, 356.5043, 353.662, 359.2267, 365.5977, 
>    369.4947, 392.3723, 374.757, 372.0594, 389.124, 390.5706, 391.9053, 
>    393.023, 397.463, 375.3581, 371.0126, 372.924, 392.4577, 373.5735, 
>    364.2137, 364.4178, 372.6771, 392.912, 378.2903, 369.8222, 391.9157L, 
>    370.6215, 371.5577, 371.0256, 377.0153, 374.279, 386.7569, 390.8015, 
>    371.6572, 365.4334, 366.7746, 364.0185, 362.7578, 379.4369, 377.3554, 
>    375.8331, 376.2615, 376.9933, 378.4856, 379.1119, 369.9149, 373.0526, 
>    376.518, 378.6648, 381.4642, 382.6173, 383.2715, 384.5013, 384.4879, 
>    384.6168, 386.2058, 386.3334, 384.5129, 383.7413, 384.0238, 384.1737, 
>    384.5389, 384.4421, 382.0946, 380.4161, 384.2753, 386.06, 384.2455, 
>    383.0246, 381.7079, 383.1534, 385.2602, 386.4101, 370.0849, -1e+3L4, 
>    368.7702, 362.1729, 358.6587, 360.2878, 362.5882, 363.6682, 367.4212, 
>    374.6625, 380.5357, 384.4504, 386.2372, 387.3375, 386.3895, 383.0319, 
>    381.321, 381.4418, 383.392, 382.2667, 382.2667, 380.2295, 376.7909, 
>    379.5762, 378.9742,
> 
> This is at 66 degrees South which has valid entries for almost all of the
> horizontal grid points.  -1e+34 is the land value.
> 
> Has anyone seen this before?  Any ideas about how to bridge the gap?  This
> is version 5.51 on a Redhat Linux 7.3 server.
> 
> Thanks,
> 
> Steve
> L_____________________________________________________________
> Steve Cousins                 Email: [ mailto:cousins@umit.maine.edu ]cousins@umit.maine.edu
> Research Associate            Phone: (207) 581-4302
> Ocean Modeling Group
> School of Marine Sciences     208 Libby Hall
> University of Maine           Orono, Maine 04469
> 
> 
>  
> 
>         
> 
> 
> 
> 
>       
> 
> 
> 
> 
> 
> 
> 
> 
> 
>   
> 
> 
> 
>