matplotlib-users Mailing List for matplotlib

I am encountering the same problem with fillcontinents that has been
discussed here in the past:
https://sourceforge.net/mailarchive/message.php?msg_id=22351355
I am using Basemap 1.0 in Python 2.4.3.
fillcontinents sometimes fails and the entire map is the assigned ocean
color, aqua. Jeff Whitaker suggested that I try drawlsmask or bluemarble
methods of Basemap. Neither of these have worked for me, as explained below
...
We do not have PIL (Python Image Library) installed and it is required for
bluemarble and I don't have authority to install, so for now bluemarble
cannot be used.
Drawlsmask raises "ValueError: lons and lats must be increasing!". But they
are increasing - here is an attempt to make a global map:
m = Basemap(llcrnrlat=-85.0,llcrnrlon=0.0,urcrnrlat=85.0,urcrnrlon=360.0)
m.drawlsmask()
And this throws ValueError (so do other proper lat-long boundaries).
So back to the problem with fillcontinents ...
I have no problem on my development system - all maps are so far good. But
on our operational system, some combinations of lat-long and figure size
cause fillcontinents (or something else?) to fail.
The original code reads like this:
m = Basemap(llcrnrlat=-85.0,llcrnrlon=0.0,urcrnrlat=85.0,urcrnrlon=360.0) #
global map, it is OK
m.drawcoastlines()
m.drawcountries(linewidth=1)
m.drawmapboundary(fill_color='aqua')
m.fillcontinents(color='coral')
BTW, here are values that make it fail for me (north Atlantic region):
m = Basemap(llcrnrlat=10.0,llcrnrlon=-100.0,urcrnrlat=70.0,urcrnrlon=15.0)
and
plt.figure(figsize=(2.0,1.043478))
If I shrink lat-lon boundaries by 1 degree (11,-99)(69,14) or change the
figsize somewhat (but maintain aspect ratio), the continents will appear.
If fillcontinents is the problem, then if I comment it out, shouldn't I get
coastlines and countries (line drawings shouldn't be a problem)? I do on my
dev system, but I still get solid 'aqua' on the operational system.

#----------------------------
import matplotlib.pyplot as plt
ax = plt.subplot(1,1,1)
x = [0.1, 0.3, 2.6, 3.3]
y = [25.4, 18.9, 13.5, 12.6]
ax.plot(x, y, marker="s", label='foo', linestyle="None")
#ax.plot(x, y, marker="s", label='foo')
#leg = ax.legend(numpoints=1, title='legend title')
leg = ax.legend(title='legend title')
print(leg.numpoints)
plt.show()
#----------------------------
Why does legend.numpoints default to 2? It looks much better when it's 1.
I'm using matplotlib version 0.99.3.
Steve

#----------------------------
import matplotlib.pyplot as plt
ax = plt.subplot(1,1,1)
ax.plot(1, 2, label='foo')
ax.legend(numpoints=0)
#----------------------------
Traceback (most recent call last):
 File "<stdin>", line 1, in <module>
 File "/usr/lib/pymodules/python2.7/matplotlib/axes.py", line 4044, in legend
 self.legend_ = mlegend.Legend(self, handles, labels, **kwargs)
 File "/usr/lib/pymodules/python2.7/matplotlib/legend.py", line 226,
in __init__
 raise ValueError("numpoints must be >= 0; it was %d"% numpoints)
ValueError: numpoints must be >= 0; it was 0
The exception message doesn't match the condition that raises it. I
think it should say "numpoints must be > 0". I'm using matplotlib
version 0.99.3.
Steve

On Sunday, June 19, 2011, Lukmanul Hakim <plg...@ya...> wrote:
> Hello,
>
> I would like to ask for some hints
> and help. I am currently trying to plot a "pseudo contour" over a Basemap. This contour is confined by borders obtained from a shapefile. I can generate the contour, I can retrieve the shapefile and put them on a basemap. What I have not been able to do is to limit the contour plot such that it only fills the area defined by a shapefile (in my case it is bordered by Province Lampung). You can use different shapefile for example, as the point of my question is how to limit contour fill by a shapefile.
>
> Shapefile can be downloaded from http://www.gadm.org/data/shp/IDN_adm.zip
>
> The generated figure can be found here:
> https://lh3.googleusercontent.com/-hMlwe6MAjdc/Tf6HlxYxn1I/AAAAAAAAAEY/exdzSv30ZL4/s640/to_matplotlib_userlists.png
>
> Thanks for the help!
> --
> Lukmanul Hakim
>
> Here's the code:
> ----------------
> from pylab import cm
> import matplotlib.pyplot as plt
> import numpy as np
> from mpl_toolkits.basemap import Basemap
>
> #Define area about Province Lampung
> ulat, llat, ulon, llon = -2.5, -6.5, 107, 103
> m = Basemap(projection='merc', lon_0=0.0, llcrnrlon=llon,
>           llcrnrlat=llat, urcrnrlon=ulon, urcrnrlat=ulat,
>           resolution='i')
>
>
> #Read shapefile of Province Lampung
> s = m.readshapefile('E:/Works/UNILA/Research/IDN_adm/LampungMap/LampungMap', 'lampung')
>
> #Define area for contour plot
> llon1,ulon1 = 103,106
> llat1,ulat1 = -6,-3.5
>
> #Generate random data
> nx,ny=5,5
> data2 = np.random.sample((ny,nx))
> x = np.linspace(llon1, ulon1, nx)
> y = np.linspace(llat1, ulat1, ny)
> X,Y = np.meshgrid(x,y)
> px,py = m(X,Y)
> m.contourf(px, py, data2, cmap=cm.jet)
> m.drawcoastlines()#(linewidth=0.5)
> m.drawparallels(np.arange(-6.5,-2.5,1.),labels=[1,0,0,0],color='black',dashes=[1,0],labelstyle='+/-',linewidth=0.2) # draw parallels
> m.drawmeridians(np.arange(102.,108.,1.),labels=[0,0,0,1],color='black',dashes=[1,0],labelstyle='+/-',linewidth=0.2) # draw meridians
>
> plt.show()
Not exactly sure how to do this, but if you can get a true/false mask
of the region in the same shape as the input data, then you can have a
masked array that goes into contourf. Any area where the mask was
true will be blanked.
The hard part, though is getting the mask.
I hope that helps!
Ben Root