basemap warpimage does not handle transparency in non-cylindral projections

[Artax-zanna]

Hi,

the situation is the following: I want to do a contourf on top of a background image, added with warpimage. The image is a png with transparent pixels. If the projection is cylindrical everything works fine, but in other projections (I tried only mollweide and robin) the transparency of the background is lost. As result, the plot is completely obscurated by the image.

Here a sample script to reproduce the bug. In the attachments you can find a sample background image.

import matplotlib.pyplot as plt
import numpy as num
from mpl_toolkits.basemap import Basemap

# Background image
img = 'gtrack.png'

# Create longitude/latitude grid
lon_arr = num.arange(-180.0, 181.0, 5.0)
lat_arr = num.arange(-90.0, 91.0, 5.0)
lon, lat = num.meshgrid(lon_arr, lat_arr)

# Create sample data
Z = num.sin(2.0*num.pi*lon/180.0)*num.sin(2.0*num.pi*lat/180.0)

# Plot using cylindrical projection
mp = Basemap(projection='cyl', lon_0=0.0, rsphere=1.0)
fig = plt.figure()
ax = fig.add_subplot(1,1,1)
mp.contourf(lon, lat, Z, ax=ax, latlon=True)
mp.warpimage(image=img, ax=ax, extent=[-180.0, 180.0, -90.0, 90.0],
             aspect='equal', interpolation='none', zorder=2)
ax.set_title('Cylindrical')

# Plot using Mollweide projection
mp = Basemap(projection='moll', lon_0=0.0, rsphere=1.0)
fig = plt.figure()
ax = fig.add_subplot(1,1,1)
mp.contourf(lon, lat, Z, ax=ax, latlon=True)
mp.warpimage(image=img, ax=ax, extent=[-180.0, 180.0, -90.0, 90.0],
             aspect='equal', interpolation='none', zorder=2)
ax.set_title('Mollweide')

# Plot using Hammer projection
mp = Basemap(projection='robin', lon_0=0.0, rsphere=1.0)
fig = plt.figure()
ax = fig.add_subplot(1,1,1)
mp.contourf(lon, lat, Z, ax=ax, latlon=True)
mp.warpimage(image=img, ax=ax, extent=[-180.0, 180.0, -90.0, 90.0],
             aspect='equal', interpolation='none', zorder=2)
ax.set_title('Robin')

[tacaswell]

Moved to basemap issue tracker.