[ncl-talk] joint-histogram
Adam Phillips
asphilli at ucar.edu
Fri Nov 17 13:56:15 MST 2017
Hi Marston,
Same issue I believe, with large differences between the deltas between
neighboring values.
I got it to work by changing your x array from this:
x =
(/0.0,0.01,1.2,3.0,5.0,7.0,10.0,15.0,20.0,25.0,30.0,40.0,50.0,60.0,80.0,81.0/)
to this:
x =
(/-5.0,0.01,1.2,3.0,5.0,7.0,10.0,15.0,20.0,25.0,30.0,40.0,50.0,60.0,80.0,85.0/)
In this case, altering the x-coordinate doesn't really matter as you are
also altering/setting the tmXBLabels, so x can be anything as long as the
XBLabels are accurate:
res at tmXBLabels =
(/"0.0","0.01","1.2","3.0","5.0","7.0","10.0","15.0","20.0",\
"25.0","30.0","40.0","50.0","60.0","80.0"," "/)
Hope that helps!
Adam
On Fri, Nov 17, 2017 at 9:23 AM, Marston Johnston <shejo284 at gmail.com>
wrote:
> Hi again Adam,
>
> Apologies for revisiting this issue, but it seems I need another solution
> for the problem you identified, i.e., spacing of the axis values being too
> wide.
> I’ve developed the script to now plot the backscattering ratio of CALIPSO
> as a function of height. In this case, the y-axis is regularly spaced, but
> the x-axis is
> more problematic than before. It seems that simply adjusting the end
> positions to just slightly beyond the previous point isn’t enough.
> This is the final of the histograms with irregular spaced axis values.
>
> There is new data (see link) and my script is attached.
> https://www.dropbox.com/s/vliftbkqfk549wd/hist_data.zip?dl=0
>
> Appreciate your thoughts/help on/with this,
>
> /M
>
> Marston S. Ward
>
> On 16 Nov 2017, 18:56 +0100, Adam Phillips <asphilli at ucar.edu>, wrote:
>
> Hi Marston,
> When I change your y definition to this:
> y := (/-2.0,0.0,0.5,1.0,1.5,2.0,2.5,3.0,4.0,5.0,7.0,9.0,11.0,13.
> 0,15.0,17.0,19.0/)
> all is well. NCL seems to have trouble when you set the first value to -99
> and the last value to 99, and I'm not clear as to why. Setting them to
> values closer to the other values in the array seemingly works. I have
> attached the resulting plot. Note that the space between each of those
> values on the Y-axis is the same in the misr_fq.png figure. If you
> set gsnYAxisIrregular2Linear = True the space will be proportional to the
> differences between the y values. See the attached misr_fq.2.png figure.
> Hope that helps!
> Adam
>
> On Thu, Nov 16, 2017 at 10:30 AM, Marston Johnston <shejo284 at gmail.com>
> wrote:
>
>> Hi again Adam,
>>
>> Thanks to your help/advice/explanation, I got the isccp and modis plots
>> exactly how I want them :-)
>> But I still cannot get misr to function properly (see attached). For the
>> misr y axis, I want the labels to
>> be: (/"NR","0","0.5","1.0","1.5","2.0","2.5","3.0","4.0","5.0","
>> 7.0","9.0","11.0","13.0","15.0","17.0"," “/), which is 1 greater than
>> the y-coord in the misr array.
>> Sadly, I still get the same error:
>>
>> "warning:_NhlCreateSplineCoordApprox: Attempt to create spline
>> approximation for Y axis failed: consider adjusting trYTensionF value
>> warning:IrTransInitialize: error creating spline approximation for
>> trYCoordPoints; defaulting to linear"
>>
>> In this case, y is 1 greater than the I’m&hgt. What’s going on here that
>> is different from the other 2 cases?
>>
>> /M
>>
>> Marston S. Ward
>>
>> On 16 Nov 2017, 17:40 +0100, Adam Phillips <asphilli at ucar.edu>, wrote:
>>
>> Hi Marston,
>> I downloaded your dropbox file, but didn't see your script within the zip
>> file or attached to your email. Can you send that?
>>
>> With regards to your questions about what sections of code are doing:
>> res at sfXArray = im&tau
>> res at sfYArray = im&pres
>>
>> Normally NCL uses coordinate variables attached to the array input into
>> the graphical functions to determine where to plot the data in the
>> graphical space. However, in the raster_plot.ncl script that Mary sent you
>> the coordinate variables (x and y) are of size +1 greater than those
>> dimensions of the im array. This necessitates the use of sfXArray/sfYArray
>> to tell NCL how to plot the data. If the dimension sizes of your data
>> matches those of the desired coordinate variables then you do not need to
>> use sfXArray/sfYArray, assuming the coordinate variables are properly
>> attached to the array.
>>
>> res at tmXBMode = "Explicit"
>> res at tmXBValues = im&tau
>> res at tmXBLabels = ""+im&tau
>>
>> res at tmYLMode = "Explicit"
>> res at tmYLValues = im&pres
>> res at tmYLLabels = im&pres+""
>>
>> Keep in mind that the above ONLY adjusts the tick marks and their labels.
>> It has nothing to do with how the data is plotted in space. In the above,
>> you are explicitly telling NCL where to put the major tickmarks
>> (tmXBValues/tmYLValues) as well as how to label those major tickmarks.
>> (tmXBLabels/tmYLLabels).
>>
>> With regards to your error message: I cannot tell what is going on
>> without looking at the script. In my experience when I encounter a
>> graphical error that doesn't look like it should be occurring, I will try
>> to simplify the resource lists and/or the input data to see where the error
>> is coming from, eliminating resources until the error message disappears,
>> allowing me to see what is causing the issue.
>> Hope that helps!
>> Adam
>>
>>
>> On Wed, Nov 15, 2017 at 5:32 PM, Marston Johnston <shejo284 at gmail.com>
>> wrote:
>>
>>> Hi Mary,
>>>
>>> I appreciate your help with this and fixing that error. I got a bit
>>> frustrated at the end and therefore a bit sloppy.
>>> I must say that I still do not fully understand the following features
>>> of the code, especially the addition of a string to the labels.
>>>
>>> res at sfXArray = im&tau
>>> res at sfYArray = im&pres
>>>
>>> res at tmXBMode = "Explicit"
>>> res at tmXBValues = im&tau
>>> res at tmXBLabels = ""+im&tau
>>> res at tiXAxisString = "Cloud Optical Depth"
>>>
>>> res at tmYLMode = "Explicit"
>>> res at tmYLValues = im&pres
>>> res at tmYLLabels = im&pres+""
>>>
>>> In the attached code, I’m calculating the weighted-mean of modis, isccp,
>>> and misr joint-histograms.
>>> With your help, I’m able to get the isccp and modis figures to behave
>>> better but they are still off.
>>> With these two figures, the contour doesn’t want to line up with the
>>> major tick marks.
>>> The 3rd misr fig has a different set of y-axis values, and when I try to
>>> plot these I get:
>>>
>>> "warning:_NhlCreateSplineCoordApprox: Attempt to create spline
>>> approximation for Y axis failed: consider adjusting trYTensionF value
>>> warning:IrTransInitialize: error creating spline approximation for
>>> trYCoordPoints; defaulting to linear”
>>>
>>> From what I’ve read of this error, it should not occur in my case.
>>>
>>> I could really use your help in nailing this down properly. Why is the
>>> contour off and my misr figure causing this error when I’m plotting the
>>> y-axis explicitly?
>>> I’ve attached the faulty figures and included a link to the data below.
>>> There are a lot of small files but on my mac it goes quite fast (few
>>> seconds).
>>>
>>> https://www.dropbox.com/s/ian474obfp3amu1/fq_data.zip?dl=0
>>>
>>> Appreciate the help,
>>>
>>> /M
>>>
>>> Marston S. Ward
>>>
>>> On 14 Nov 2017, 18:05 +0100, Mary Haley <haley at ucar.edu>, wrote:
>>>
>>> Hi Marston,
>>>
>>> The reason you are seeing what looks like log values is because the Y
>>> coordinate array is irregularly spaced, and NCL by default will take your
>>> irregular Y axis values and place them linearly on the Y axis.
>>>
>>> I don't know if this is intentional, but you have:
>>>
>>> im!0 ="x"
>>> im!1 ="y"
>>>
>>> Dimension 0 is associated with the Y axis, and dimension 1 is the X
>>> axis. Just wanted to point that out, since you are calling dimension 0
>>> "x", and dimension 1 "y". I changed this to:
>>>
>>> im!1 ="x"
>>> im!0 ="y"
>>>
>>> Which seems to make your plot match better with the one you attached.
>>> You can't make x a coordinate array, because it has one more element
>>> than the X dimension of your data.
>>>
>>> Instead, in order to associate the x and y values with your plot, use
>>> these special resources, which are mentioned in the raster_5.ncl example on
>>> the same page:
>>>
>>> res at sfXArray = x
>>> res at sfYArray = y
>>>
>>> Finally, you are going to get weird labels on your axes, because NCL is
>>> simply trying to put labels along the X axis where it thinks it makes
>>> sense. If you want the same labels on your axes as your actual x values,
>>> then set:
>>>
>>> res at tmXBMode = "Explicit"
>>> res at tmXBValues = x
>>> res at tmXBLabels = ""+x ; labels have to be strings
>>>
>>> See attached.
>>>
>>> --Mary
>>>
>>>
>>>
>>>
>>>
>>> On Mon, Nov 13, 2017 at 6:04 PM, Marston Johnston <shejo284 at gmail.com>
>>> wrote:
>>>
>>>> Hi Mary,
>>>>
>>>> Thanks for the tips in the table. However, there is a better example:
>>>>
>>>> https://www.ncl.ucar.edu/Applications/raster.shtml
>>>>
>>>> However, I’m having some trouble with the finer details of the plot. It
>>>> involves getting the tick labels just in the correct place.
>>>>
>>>> My code below is basically example 6, which looks very much like the
>>>> same coordinate values of one of the iSCCP products:
>>>>
>>>> I’m trying to compare my model output with the above figure but there are some issue:
>>>>
>>>> 1.) In the ncl example 6, the figure dimension is 7x6 while mine is 7x7. When I plot it as below I get log x-axis despite my values not being log.
>>>>
>>>> 2.) None of the labels ended up on the major axis.
>>>>
>>>> I’m not sure how NCL axis works. Could I get some tips on how to force the axis to fall on the grid edges as in the figure above?
>>>>
>>>> im =
>>>> (/(/0.60 0.81 1.86 3.83 3.09 0.78 0.01/), (/0.45 0.63 1.47 4.96 5.45 2.27 0.28/), \ (/0.60 0.66 0.88 2.03 2.45 1.56 0.50/),(/0.76 0.91 1.02 1.17 1.46 1.44 0.63/), \
>>>>
>>>> (/0.85 1.41 1.32 0.88 0.91 1.07 0.72/),(/1.55 1.84 1.24 0.70 0.41 0.46 0.36/), (/7.19 1.73 0.60 0.28 0.14 0.08 0.05/)/)
>>>>
>>>> x = (/0.0,0.3,1.3,3.6,9.4,23.0,60.0,100.0/)
>>>> y = (/1000.0,800.0,680.0,560.0,440.0,310.0,180.0/)
>>>>
>>>> im!0=“x"
>>>> im!1 =“y”
>>>> im&x = x
>>>> im&y = y
>>>>
>>>> wks = gsn_open_wks("png","raster") ; send graphics to PNG file
>>>> res = True ; plot mods desired
>>>> res at cnFillOn = True ; Turn on contour fill
>>>> res at cnFillMode = "RasterFill" ; Turn on raster fill
>>>> res at cnLinesOn = False ; Turn off contour lines
>>>> res at cnLineLabelsOn = False ; turns off contour line
>>>> labels
>>>> res at cnFillPalette = "BlAqGrYeOrReVi200";MPL_viridis"
>>>> res at lbLabelBarOn = True
>>>> res at lbOrientation = "vertical" ; vertical label bar
>>>>
>>>> plot = gsn_csm_contour(wks,im,res) ; contour the variable
>>>>
>>>> Best,
>>>> /M
>>>>
>>>> *Marston S. Ward*
>>>>
>>>> On 16 Oct 2017, 21:31 +0200, Mary Haley <haley at ucar.edu>, wrote:
>>>>
>>>> Marston,
>>>>
>>>> Take a look at this page:
>>>>
>>>> http://www.ncl.ucar.edu/Applications/table.shtml
>>>>
>>>> These are examples showing how to draw filed cells (square or triangle)
>>>> based on some criteria. I think table_4.ncl colors the cells by ranges of
>>>> values.
>>>>
>>>> --Mary
>>>>
>>>>
>>>> On Mon, Oct 16, 2017 at 1:03 PM, Marston Johnston <shejo284 at gmail.com>
>>>> wrote:
>>>>
>>>>> Hi Dennis,
>>>>>
>>>>>
>>>>>
>>>>> At first glance it would appear so, but example 6 calculates the
>>>>> percentage. In the ucar example:
>>>>>
>>>>> https://climatedataguide.ucar.edu/climate-data/isccp-interna
>>>>> tional-satellite-cloud-climatology-project
>>>>>
>>>>>
>>>>>
>>>>> The ISCCP cloud fraction is calculated for each bin instead of density
>>>>> of the frequency of occurrence. So in example 6 there are 2 datasets that
>>>>> are used as input data, but in the figure in question seems to use 3.
>>>>>
>>>>>
>>>>>
>>>>> /M
>>>>>
>>>>>
>>>>>
>>>>> ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
>>>>>
>>>>> Marston S. Ward, PhD
>>>>>
>>>>> Department of Earth Sciences
>>>>>
>>>>> University of Gothenburg, Sweden
>>>>>
>>>>> Email: marston.johnston at gu.se
>>>>>
>>>>> SkypeID: marston.johnston
>>>>>
>>>>> Phone: +46-31-7864901 <+46%2031%20786%2049%2001>
>>>>>
>>>>> Only the fruitful thing is true!
>>>>>
>>>>> ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>
>>>>> *From:* ncl-talk <ncl-talk-bounces at ucar.edu> on behalf of Dennis Shea
>>>>> <shea at ucar.edu>
>>>>> *Date:* Monday, 16 October 2017 at 20:53
>>>>> *To:* Marston Ward <marston.johnston at gu.se>
>>>>> *Cc:* "ncl-talk at ucar.edu" <ncl-talk at ucar.edu>
>>>>> *Subject:* Re: [ncl-talk] joint-histogram
>>>>>
>>>>>
>>>>>
>>>>> I am not a graphics person.
>>>>>
>>>>> Looks like example 6
>>>>> https://www.ncl.ucar.edu/Applications/pdf.shtml
>>>>>
>>>>> ===
>>>>>
>>>>> Really, a contour plot with raster mode.
>>>>>
>>>>>
>>>>>
>>>>> On Mon, Oct 16, 2017 at 12:46 PM, Marston Ward <marston.johnston at gu.se>
>>>>> wrote:
>>>>>
>>>>> Hi guys!
>>>>>
>>>>>
>>>>>
>>>>> I’m trying to develop a NCL script to produce the attached figure,
>>>>> which is a joint-histogram, but a bit different from what pdfxy produces.
>>>>>
>>>>> I have a good idea of how this image is produced; but why reinvent the
>>>>> wheel? I don’t think that NCL has such a plot function, or am I mistaken?
>>>>>
>>>>> Does anyone have such a script – and would like to share?
>>>>>
>>>>> This is something I can definitely think of contributing to the NCL.
>>>>>
>>>>>
>>>>>
>>>>> Best,
>>>>>
>>>>> /M
>>>>>
>>>>> ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
>>>>>
>>>>> Marston S. Ward, PhD
>>>>>
>>>>> Department of Earth Sciences
>>>>>
>>>>> University of Gothenburg, Sweden
>>>>>
>>>>> Email: marston.johnston at gu.se
>>>>>
>>>>> SkypeID: marston.johnston
>>>>>
>>>>> Phone: +46-31-7864901 <+46%2031%20786%2049%2001>
>>>>>
>>>>> Only the fruitful thing is true!
>>>>>
>>>>> ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
>>>>>
>>>>>
>>>>>
>>>>>
>>>>> _______________________________________________
>>>>> ncl-talk mailing list
>>>>> 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 List instructions, subscriber options, unsubscribe:
>>>>> http://mailman.ucar.edu/mailman/listinfo/ncl-talk
>>>>>
>>>>> _______________________________________________
>>>>> ncl-talk mailing list
>>>>> 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
>>> List instructions, subscriber options, unsubscribe:
>>> http://mailman.ucar.edu/mailman/listinfo/ncl-talk
>>>
>>>
>>
>>
>> --
>> Adam Phillips
>> Associate Scientist, Climate and Global Dynamics Laboratory, NCAR
>> www.cgd.ucar.edu/staff/asphilli/ 303-497-1726 <(303)%20497-1726>
>>
>> <http://www.cgd.ucar.edu/staff/asphilli>
>>
>>
>
>
> --
> Adam Phillips
> Associate Scientist, Climate and Global Dynamics Laboratory, NCAR
> www.cgd.ucar.edu/staff/asphilli/ 303-497-1726 <(303)%20497-1726>
>
> <http://www.cgd.ucar.edu/staff/asphilli>
>
>
--
Adam Phillips
Associate Scientist, Climate and Global Dynamics Laboratory, NCAR
www.cgd.ucar.edu/staff/asphilli/ 303-497-1726
<http://www.cgd.ucar.edu/staff/asphilli>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.ucar.edu/pipermail/ncl-talk/attachments/20171117/0e32230e/attachment.html>
More information about the ncl-talk
mailing list