[Thread Prev][Thread Next][Index]
[ferret_users] Re: external function samplexy_curv and modulo
Hi Patrick,
What you say is quite true. For a general curvilinear grid, some of
which are very non-rectangular, the concept of modulo longitudes is
difficult to define. How will the algorithm know which lon,lat cell
should wrap to which cell across the branch cut?
The function SAMPLEXY_CURV looks from one side only, for the "closest"
location.
In fact your data shows a shortcoming of the algorighm in
samplexy_curv. For the grid in your file samplexy_curv.nc the latitudes
in LAT are 0 at index j=74. For j=74, and everywhere nearby, there are
no longitudes in LON anywhere near 0. So the function should not return
a result for samplexy_curv(votemper, lon,lat, 0,0). What the function
does is it finds where on each J line of LON the values of LON crosses
the xpoint requested, here 0. It finds that LON crosses 0 only along
the line x=51, j=149, at the top of the grid (do a SHADE LON to see what
the LON coordinate variable is like), and it then looks in LAT and finds
the location closest to (51,149) where LAT = ypt, again 0 for this
example. We need to implement a threshold, so that the function will
not return a result from halfway across the globe.
Ansley
Brockmann Patrick wrote:
Hi all,
A question on the external function samplexy_curv
which does not consider modulo aspect. If I sample at (lon,lat) (0,0)
or (360,0),
the samplexy_curv gives different results.
See scripts:
http://dods.ipsl.jussieu.fr/brocksce/ferret_sample/samplexy_curv_test.jnl
list samplexy_curv(votemper, lon,lat, 0,0)
VARIABLE : SAMPLEXY_CURV(VOTEMPER, LON,LAT, 0,0)
FILENAME : samplexy_curv.nc
X : 1
Z : 5
T : 1
0.01885
list samplexy_curv(votemper, lon,lat, 360,0)
VARIABLE : SAMPLEXY_CURV(VOTEMPER, LON,LAT, 360,0)
FILENAME : samplexy_curv.nc
X : 1
Z : 5
T : 1
30.29
What do I have to believe ?
Regards
Patrick
[Thread Prev][Thread Next][Index]
Dept of Commerce /
NOAA /
OAR /
PMEL /
TMAP
Contact Us | Privacy Policy | Disclaimer | Accessibility Statement