[pyngl-talk] isccp grid with Pyngl
Kevin Hallock
hallock at ucar.edu
Mon Oct 24 11:41:18 MDT 2016
Hi Xiaoni,
> 1) To install Pyngl, I followed the suggestion on your website:
> conda create --name pyn_test python=2
> source activate pyn_test
> conda install --channel ncar --channel khallock pynio pyngl
Everything above looks correct.
> Then I modified my .bashrc by add the new path : export PATH="/obs/xxw/anaconda2/bin:/obs/xxw/anaconda2/envs/pyn_test/bin:$PATH"
> …
> Then I tried to import Nio, but I have an error: ImportError: No module named Nio
You should not need to explicitly add the /obs/xxw/anaconda2/envs/pyn_test/bin directory to your path; running “source activate pyn_test” does that for you.
In this case, it looks like in addition to adding pyn_test/bin to $PATH manually, the root conda environment has been added to $PATH with a higher priority. Running “which python” will output “/obs/xxw/anaconda2/bin/python”, and that python binary uses the directory “/obs/xxw/anaconda2/lib/python2.7/site-packages” for external modules. Because PyNIO and PyNGL were installed in the pyn_test environment, they do not exist in that directory.
Could you please try modifying your .bashrc file and remove all conda-related commands except for export PATH="/obs/xxw/anaconda2/bin:$PATH”? After doing that, try running “source activate pyn_test” again (the other two conda commands are only necessary when creating/installing PyNIO/PyNGL into a new environment).
Please let us know if you continue to experience issues with conda.
Kevin
> On Oct 24, 2016, at 11:13 AM, Mary Haley <haley at ucar.edu> wrote:
>
> Xianoi,
>
> I've already replied offline about the conda issues, and I'm hoping Kevin can address these.
>
> Meanwhile, with regard to your python script, I think you are on the right track. However, I don't think you want to transpose the data before you plot it. If it is dimensioned latitude x longitude, with latitude being the leftmost dimension, then you don't need to transpose it. If it is dimensioned longitude by latitude, then you need to transpose it.
>
> To show the figure on the screen (we call it an "X11 window"), then change "wks_type" to be "x11" instead of "ps". You need to click on the X11 window with your left mouse button if you want the plot to go away.
>
> Also, you don't want to set these two options:
>
> resources.nglDraw = False
> resources.nglFrame = False
>
> because these settings would cause your plot to not be drawn at all. Either remove those lines or comment them out. They are both True by default.
>
> And yes, once you have all the parameters (we call them "resources") set up, you should be able to use the same ones to plot other data. The tiMainString is probably the only thing you'll need to change.
>
> However, if you decide you don't like the contour levels that PyNGL chooses for you, then you might want to change these for your plots. You can look at the cn05p.py for an example of setting contour levels using a min value, max value, and a spacing (for equally spaced contours), or ctnccl.py for example of explicitly setting contour levels to an unequally-spaced array of values.
>
> Finally, if your data is taking a long time to plot because it is a large array, then you can try setting:
>
> resources.cnFillMode = "RasterFill" # default is AreaFill
>
> Raster contouring can be noticeably faster than the default "AreaFill", which is smooth contouring. You can also set:
>
> resources.cnRasterSmoothingOn = True
>
> which will smooth your raster contours, but still should be faster than AreaFill.
>
> I will attach a Python script that plots the "topo" variable from old and partial ISCCP file that I have. The data on this file is one-dimensional, along with the lat/lon arrays. Here's what a dump of the file looks like:
>
> file global attributes:
> description : ISSCP reduced grid
> dimensions:
> nodes = 6596
> variables:
> float lon ( nodes )
> units : degrees east
> long_name : longitude
>
> float lat ( nodes )
> units : degrees north
> long_name : latitude
>
> float topo ( nodes )
> units : meters above sea level (I think)
> long_name : topology
>
>
> I'll include the script and the second and third images just in case. I use "png" as the output format.
>
> --Mary
>
> On Mon, Oct 24, 2016 at 6:50 AM, Xiaoni Wang <xiaoni.wang at obspm.fr <mailto:xiaoni.wang at obspm.fr>> wrote:
> Dear colleague,
> I tried to install Pyngl on my computer in order to draw 2D figures with ISCCP grids. Before this, I had used condo to install python2 and other python tools in /obs/xxwang/anaconda2/.
>
> 1) To install Pyngl, I followed the suggestion on your website:
> conda create --name pyn_test python=2
> source activate pyn_test
> conda install --channel ncar --channel khallock pynio pyngl
>
> Then I modified my .bashrc by add the new path : export PATH="/obs/xxw/anaconda2/bin:/obs/xxw/anaconda2/envs/pyn_test/bin:$PATH »
> I can see the environments:
> pyn_test * /obs/xxw/anaconda2/envs/pyn_test
> root /obs/xxw/anaconda2
>
> Then I tried to import Nio, but I have an error: ImportError: No module named Nio
>
> Could you help me to figure out why ?
>
> 2) I have another question about using the Pyngl for plotting data on ISCCP grids. Maybe this question would be (too) simple….
>
> I have my data on ISCCP equal area grids (0.25 degree in latitude, and varying espace in longitude, with more grids near equator and less grids near pole). My data are global data. In my test:
> ISCCP latitude : lat_isccp, 1D, (256915,) ),
> ISCCP longitude: lon_isccp, (1D, (256915,))
> data: emis (1D , (256915,)).
>
> I would like to plot emis data on 2D map. I do not know if I should define resources.mpProjection ? how to show figure on screen, or how to save this figure ? or If I have several parameters to plot, I guess that I can use these configurations for all the figures (except resources.tiMainString )?
> In the below is my example code for plotting one parameter ‘emis’ . I adapted from your SEAM example.
>
> wks_type = "ps"
> wks = Ngl.open_wks(wks_type, "isccp")
> resources = Ngl.Resources()
> resources.sfXArray = lon_isccp
> resources.sfYArray = lat_isccp
> resources.nglSpreadColorStart = 176
> resources.nglSpreadColorEnd = 2
> resources.cnFillOn = True
> resources.cnFillMode = "AreaFill"
> resources.cnFillPalette = "BlAqGrYeOrReVi200"
> resources.cnLinesOn = False
> resources.cnLineLabelsOn = False
> resources.lbBoxLinesOn = False
> resources.lbLabelFontHeightF = 0.015
> resources.tiMainString = "Emis"
> resources.nglDraw = False
> resources.nglFrame = False
>
> zt = np.transpose(emis)
> contour = Ngl.contour(wks,zt,resources)
> Ngl.end()
>
> Thank you very much for your help in advance !!
>
>
> Best regards
>
> Xiaoni W
>
>
>
>
>
>
> _______________________________________________
> pyngl-talk mailing list
> List instructions, subscriber options, unsubscribe:
> http://mailman.ucar.edu/mailman/listinfo/pyngl-talk <http://mailman.ucar.edu/mailman/listinfo/pyngl-talk>
>
>
> <isccp3.png><isccp2.png><isccp.py>_______________________________________________
> pyngl-talk mailing list
> List instructions, subscriber options, unsubscribe:
> http://mailman.ucar.edu/mailman/listinfo/pyngl-talk
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mailman.ucar.edu/pipermail/pyngl-talk/attachments/20161024/0c3eacea/attachment-0001.html
More information about the pyngl-talk
mailing list