[ncl-talk] Problem aligning data with map

Thu Mar 29 14:26:37 MDT 2018

Thanks Dave,

This is very helpful!  I’ll send a note to the folks about Unidata to see if this is a “feature” in the IDV.

Thanks again!

Best wishes,

Chris

--

Dr. Christopher G. Herbster
Associate Professor
Director of Science and Technology
for the ERAU Weather Center
Applied Aviation Sciences
Embry-Riddle Aeronautical Univ.
600 S. Clyde Morris Blvd.
Daytona Beach, FL 32114-3900
 386.226.6444 Office
386.226.6446 Weather Center
http://wx.erau.edu/

Schedule at:  http://wx.erau.edu/faculty/herbster/Schedules/

From: Dave Allured - NOAA Affiliate <dave.allured at noaa.gov>
Sent: Tuesday, March 27, 2018 7:38 PM
To: Herbster, Christopher G. <herbstec at erau.edu>
Cc: ncl-talk at ucar.edu
Subject: Re: [ncl-talk] Problem aligning data with map

Chris,

Let me add this.  I took a quick look at the two netcdf files in your original message, narr-a and D-Value.  Both files appear to be correctly constructed for 2-D NARR coordinates, in conformance with CF conventions.  In particular, both lat and lon 2-D coordinate variables are independently included in each file.  Also each data variable includes a properly formed attribute for "auxiliary coordinates", e.g.:

    PRES_221_SFC:coordinates = "gridlat_221 gridlon_221" ;

This is sufficient for the ncview utility (yet another viewing tool) to properly geo-locate the data over maps, as far as I can tell by eye.  I confirmed this on several variables in each data file.  The Matlab image in your original message also seems to be correctly geo-located, but I did not confirm that.

My guess is that your current IDV version does not know how to decode this type of coordinates.  Someone else will have to answer about that.

--Dave

On Tue, Mar 27, 2018 at 4:58 PM, Mary Haley <haley at ucar.edu<mailto:haley at ucar.edu>> wrote:
Hi Chris,

You were almost there.

The issue is that you are subsetting your data by selecting a lat/lon area of interest, but then trying to plot this data using the projection defined for the original data. When you set the "tfDoNDCOverlay" resource to True, this signals to NCL that you are plotting data in its "native" projection, and hence you need to set the map parameters exactly right for the data you are plotting.

If you are plotting the original data (see narr_plot.ncl), then you can use the map projection parameters on the file to plot the data. This script shows how to plot the data on both a cylindrical equidistant projection using the 2D lat/lon arrays read off the file, and in the native projection using map projection information attached to the lat/lon arrays. I've attached the PNGs from this script.

If you want to zoom in on the data, then unless you know the exact projection parameters for the zoomed data, you will have to subset the 2D lat/lon arrays in the same fashion as the data, and use these for plotting.  See the attached narr_plot_zoom.ncl and image.

Finally, I took your original script and applied the same concepts I used in narr_plot_zoom.ncl. Instead of using tfDoOverlay, it uses the subsetted lat/lon arrays to get the overlay correct:

  res at sfXArray               = lon2d_small
  res at sfYArray               = lat2d_small

Note that in narr_plot_zoom.ncl, I didn't use sfXArray / sfYArray.  Instead, I used:

  var_subset at lat2d = lat_subset
  var_subset at lon2d = lon_subset

These are essentially the same thing. I was just lazy and didn't want to have to attach the special "lat2d" and "lon2d" attributes to every variable being plotted.

For more descriptive information on all this, you may want to visit our "plotting data on a map" page:

http://www.ncl.ucar.edu/Applications/plot_data_on_map.shtml

Let me know if you have any questions about what I did.

--Mary

On Tue, Mar 27, 2018 at 3:02 PM, Herbster, Christopher G. <herbstec at erau.edu<mailto:herbstec at erau.edu>> wrote:
Hi folks,

Here is a sequence of events that we are using to download some NARR data from NOMADS at NCDC:
We use the get-httpsubset.pl<http://get-httpsubset.pl> script from
https://nomads.ncdc.noaa.gov/guide/index.php?name=advanced#adv-httpsubset and the Perl scripts from
http://www.cpc.ncep.noaa.gov/products/wesley/fast_downloading_grib.html to grab a handful of parameters from the NARR dataset.

This gives us GRIB files like:
http://wx.erau.edu/faculty/herbster/demo/.hidden/NCL/narr-a_221_20101201_1200_000.sub.grb

I then used: ncl_convert2nc narr-a_221_20101201_1200_000.sub.grb -nc4c -l
To make
http://wx.erau.edu/faculty/herbster/demo/.hidden/NCL/narr-a_221_20101201_1200_000.sub.nc4

I did this for 30 years of data ending 12-31-2010.

We then used an NCL script to do some calculations on a month of data, and write the results into a new file.
Here is the script:
http://wx.erau.edu/faculty/herbster/demo/.hidden/NCL/read_narr_ncar-help.ncl
My apologies to those of you who actually know how to code in NCL.  I am strictly an amateur at this.

Here is the file we created:
http://wx.erau.edu/faculty/herbster/demo/.hidden/NCL/D-Value_Monthly_Stats.2010.12.averages.1200.nc

Our problem seems to be aligning the data to the map background, as I can show with the following images:

The grey area in this image is supposed to be where the terrain is higher than the calculated value:
http://wx.erau.edu/faculty/herbster/demo/.hidden/NCL/narr-paper.000001.png

The data mask area is too far to the west, even extending out over the ocean.

When we plot it in the Unidata IDV we get what we expect:
http://wx.erau.edu/faculty/herbster/demo/.hidden/NCL/IDV_4K_Mask_D-Val.png

Then we decided to try plotting it in Matlab as another test.  We got:
http://wx.erau.edu/faculty/herbster/demo/.hidden/NCL/Matlab_Plot_NARR_Mask.jpg

While I suspect that I am not doing something properly with the mapping of the data, I am also wondering about the results of making the NetCDF file from the ncl_convert command.

First, is the compression lossy?  I realized that the default compression is zero, so this step did not do much for disk space!  (Actually made bigger files, but never mind.)

When I load the original GRIB file into the IDV it displays the region properly without any problem.  When I try to load the original NetCDF file, the IDV does not seem to know the geographic region of the data.

This can be seen in the following images:
http://wx.erau.edu/faculty/herbster/demo/.hidden/NCL/load-both-idv.png shows the Mean Sea Level Pressure data from both files.  They appear to match.  However, when I added the second panel (NetCDF file), the IDV did not display the data properly, and did so only after I selected a predefined CONUS projection (as shown).

When I open the NetCDF file and plotted MSLP in a window by itself, I got this:
http://wx.erau.edu/faculty/herbster/demo/.hidden/NCL/load-netcdf-idv.png

I had to zoom way out to display this data.  It knows where to put the data, but the IDV thinks the data should be in the white box area, I think.  (I’m not really sure about this interpretation.)

The questions I have are:

1)      Why does this happen?

2)      Is this related to my data masking and mapping problem, or are they separate issues?

Thanks in advance for any help on any of these problems/questions.

Cheers,

Chris

PS  If you want to test our script, there is a month of data in this tar file:
http://wx.erau.edu/faculty/herbster/demo/.hidden/NCL/sample-narr.tgz
--

Dr. Christopher G. Herbster
Associate Professor
Director of Science and Technology
for the ERAU Weather Center
Applied Aviation Sciences
Embry-Riddle Aeronautical Univ.
600 S. Clyde Morris Blvd.
Daytona Beach, FL 32114-3900
 386.226.6444<tel:(386)%20226-6444> Office
386.226.6446<tel:(386)%20226-6446> Weather Center
http://wx.erau.edu/
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.ucar.edu/pipermail/ncl-talk/attachments/20180329/37e581af/attachment.html>