[ncl-talk] Fwd: contour plot misalignment

Thu Jan 26 15:45:37 MST 2017

Apologies. Yes, I should have explained this better.
I do not mean to merge the vectors but rather find areas where they contain common values.
I’m trying finds areas in a satellite granule where they overlap. Therefore, in sequence, granule B will contain the some of the same information already binned in granule A. I would like to remove those areas before adding granule B to the binned data.
However, the length of granule A and B may not be equal in some cases. I thought of using “conform” to make the vectors the same lengths in order to mask out the overlaps. Any tips on how to do this efficiently?

/M

On 26 Jan 2017, 23:12 +0100, Dennis Shea <shea at ucar.edu>, wrote:
This thread is very confusing [to me]. I have not looked at the script or data. It seems some 'custom processing' task is being described.

You use the word "conform" rather loosely from my perspective.

"conform a vector X of length, say 10, with another vector B of length 20"

X(10), B(20)

NCL's  'conform' and 'conform_dims' **require** that some dimension size 'match'. Sizes 10 and 20 do not conform.

Do you mean 'merge'?

X = ispan(1,10,1)
B = ispan(1,20,1)
XB = array_append_record(X, B, 0)      ; XB(30)
XB_unique = get_unique_values(XB)
print(XB_unique)

Variable: XB_unique
Type: integer
Total Size: 80 bytes
            20 values
Number of Dimensions: 1
Dimensions and sizes:   [20]
Coordinates:

(0)     1
(1)     2
(2)     3
        [snip]
(17)    18
(18)    19
(19)    20

------------

On Thu, Jan 26, 2017 at 1:41 PM, Marston Johnston <marston.johnston at gu.se<mailto:marston.johnston at gu.se>> wrote:
Hi Mary,

I understood what you meant. Apologies for the ambiguity.
My script is a long one and it is not yet a running smoothly and bug-free.
I’ve sent an attachment along with a subset of the data:

https://www.dropbox.com/sh/gac3w26dzdtpd8y/AACrzZSVc3Fj5Dinh137E9Ywa?dl=0

I’m rethinking how to best bin and average the data. There is still problem, which I sent in as a issue today and that is how to I conform a vector X of length, say 10, with another vector B of length 20, for example ( see line 154).

As you can see from the script I use the model's lat lon to for the grid on which the MHS data is binned. In this case I’m using a square grid where the lat and lon spacing are equal. The lats and lons are from the model are not changed, which is why I suspect the binning.

Of course this is not the plot script as there is another averaging script in between. This is why I was hesitant to send any data as I’m trying to eliminate much of the secondary averaging and ensure this initial binning of the data is done correctly.

I would appreciate any tips you can give on making the attached script more efficient.

One idea I have is to use the bin_sum function and just do a running mean instead of saving the daily means.

/M

Marston S. Johnston, PhD
Department of Earth Sciences
University of Gothenburg, Sweden
Email: marston.johnston at gu.se<mailto:marston.johnston at gu.se>
SkypeID: marston.johnston
Phone: +46-31-7864901<tel:+46%2031%20786%2049%2001>
Only the fruitful thing is true!

On 26 Jan 2017, 18:20 +0100, Mary Haley , wrote:
Marston,

To be clear, I don't think you should set tfDoNHiDCOverlay at all!  I was asking if you were doing this, because that's what could make your plot wrong.

I don't think you should need to regrid the data to plot it, either.  You really should be able to plot the observational values as they are.

It's hard for us to debug a problem if we can't see your script. If you can send this, then I might have a better idea of what's going on. It also helps if you can provide the data.

--Mary

On Thu, Jan 26, 2017 at 9:48 AM, GMAIL <shejo284 at gmail.com<mailto:shejo284 at gmail.com>> wrote:
Hi Mary,

I’m still working on this. I’ve not set the tfDoNDCOverlay and what I realised is that my model doesn’t have a square grid, i.e., the
spacing in the lat and lon directions are slightly different. I’m regridding instead to a square grid and then binning the MHS data to the square grid. If this fails I’ll plot the lat and lon values as you suggested and contact you again with more information for some help.
Appreciate the tip :-)

/M

On 26 Jan 2017, 06:18 +0100, Mary Haley <haley at ucar.edu<mailto:haley at ucar.edu>>, wrote:
Marston,

There could be a few things going on here.

One is simply that your lat/lon values are not quite correct.

Are you setting tfDoNDCOverlay to True by any chance? If you are trying to use a native projection and you don't have the native projection parameters set exactly right, then the plot can be slightly misaligned.

One thing I would suggest is plotting the observational lat/lon values as markers on your second plot to see if they are also misaligned with the coastlines. You should be able to use gsn_add_polymarker to plot the markers.

Finally, as usual, if you can provide the data for the observational plot, and a script, then I can take a look.

--Mary

On Wed, Jan 25, 2017 at 3:17 AM, Marston Johnston <marston.johnston at gu.se<mailto:marston.johnston at gu.se>> wrote:
Hi,

I’ve binned (bin_sum) some MHS data to a model grid using rational lats and lons values (see linked fig<https://www.dropbox.com/s/fi6s577iio47u4w/mhs_t511_si_contour.pdf?dl=0>).
The upper panel in the figure is the model data but the bottom figure is from the observations. As you can see the
coastlines are slightly off and I cannot figure out why. Does anyone have any idea why this is happening to the bin_sum figure?
I’ve used the same set of lats and lons consistently throughout the process.

Any hint/help is appreciated.

/M

Marston S. Johnston, PhD
Department of Earth Sciences
University of Gothenburg, Sweden
Email: marston.johnston at gu.se<mailto:marston.johnston at gu.se>
SkypeID: marston.johnston
Phone: +46-31-7864901<tel:+46%2031%20786%2049%2001>
Only the fruitful thing is true!

_______________________________________________
ncl-talk mailing list
ncl-talk at ucar.edu<mailto:ncl-talk at ucar.edu>
List instructions, subscriber options, unsubscribe:
http://mailman.ucar.edu/mailman/listinfo/ncl-talk

_______________________________________________
ncl-talk mailing list
ncl-talk at ucar.edu<mailto:ncl-talk at ucar.edu>
List instructions, subscriber options, unsubscribe:
http://mailman.ucar.edu/mailman/listinfo/ncl-talk

On 26 Jan 2017, 06:18 +0100, Mary Haley <haley at ucar.edu<mailto:haley at ucar.edu>>, wrote:
Marston,

There could be a few things going on here.

One is simply that your lat/lon values are not quite correct.

Are you setting tfDoNDCOverlay to True by any chance? If you are trying to use a native projection and you don't have the native projection parameters set exactly right, then the plot can be slightly misaligned.

One thing I would suggest is plotting the observational lat/lon values as markers on your second plot to see if they are also misaligned with the coastlines. You should be able to use gsn_add_polymarker to plot the markers.

Finally, as usual, if you can provide the data for the observational plot, and a script, then I can take a look.

--Mary

On Wed, Jan 25, 2017 at 3:17 AM, Marston Johnston <marston.johnston at gu.se<mailto:marston.johnston at gu.se>> wrote:
Hi,

I’ve binned (bin_sum) some MHS data to a model grid using rational lats and lons values (see linked fig<https://www.dropbox.com/s/fi6s577iio47u4w/mhs_t511_si_contour.pdf?dl=0>).
The upper panel in the figure is the model data but the bottom figure is from the observations. As you can see the
coastlines are slightly off and I cannot figure out why. Does anyone have any idea why this is happening to the bin_sum figure?
I’ve used the same set of lats and lons consistently throughout the process.

Any hint/help is appreciated.

/M

Marston S. Johnston, PhD
Department of Earth Sciences
University of Gothenburg, Sweden
Email: marston.johnston at gu.se<mailto:marston.johnston at gu.se>
SkypeID: marston.johnston
Phone: +46-31-7864901<tel:+46%2031%20786%2049%2001>
Only the fruitful thing is true!

_______________________________________________
ncl-talk mailing list
ncl-talk at ucar.edu<mailto:ncl-talk at ucar.edu>
List instructions, subscriber options, unsubscribe:
http://mailman.ucar.edu/mailman/listinfo/ncl-talk

_______________________________________________
ncl-talk mailing list
ncl-talk at ucar.edu<mailto:ncl-talk at ucar.edu>
List instructions, subscriber options, unsubscribe:
http://mailman.ucar.edu/mailman/listinfo/ncl-talk

_______________________________________________
ncl-talk mailing list
ncl-talk at ucar.edu<mailto:ncl-talk at ucar.edu>
List instructions, subscriber options, unsubscribe:
http://mailman.ucar.edu/mailman/listinfo/ncl-talk

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mailman.ucar.edu/pipermail/ncl-talk/attachments/20170126/c89fd955/attachment.html 