pcolor with non-rectangular pixels?

Community matplotlib-users
1

I'd like to make pcolor images using non-'cyl' projections, but I want the pcolor pixels to be trapezoidal rather than rectangular. Is this possible?

Mike

Example script"

# example data
lonin = numpy.arange(0.,360.,2.5)
latin = numpy.arange(-90.,92.5,2.5)
field = numpy.ones([len(latin),len(lonin)]) + 10.0*numpy.random.random([len(latin),len(lonin)])

# adapt for plot
#field,lons = addcyclic(field,lonin)
#field,lons = shiftgrid(180.,field,lons,start=False)
# offset for pcolor
field = 0.5*(field[:-1,:-1]+field[1:,1:])

fig = figure()
canvas = FigureCanvas(fig)
the_map = Basemap(resolution='c',projection='lcc',llcrnrlon=-75.,
                   llcrnrlat=15.,urcrnrlon=40.,urcrnrlat=60.,lat_1=50.,
                   lon_0=-45.)

# convert to projected coordinates
dx = 2.*pi*the_map.rmajor/len(lons)
nx = int((the_map.xmax-the_map.xmin)/dx)+1
ny = int((the_map.ymax-the_map.ymin)/dx)+1
z,x,y = the_map.transform_scalar(field,lons,lats,nx,ny,returnxy=True,order=0)

# creates seemingly correct image in term of grid placement, but with rectangular pixels
the_image = the_map.pcolor(x,y,z,shading='flat')

2

Michael Bauer wrote:

I'd like to make pcolor images using non-'cyl' projections, but I want the pcolor pixels to be trapezoidal rather than rectangular. Is this possible?

Mike

Example script"

# example data
lonin = numpy.arange(0.,360.,2.5)
latin = numpy.arange(-90.,92.5,2.5)
field = numpy.ones([len(latin),len(lonin)]) + 10.0*numpy.random.random([len(latin),len(lonin)])

# adapt for plot
#field,lons = addcyclic(field,lonin)
#field,lons = shiftgrid(180.,field,lons,start=False)
# offset for pcolor
field = 0.5*(field[:-1,:-1]+field[1:,1:])

fig = figure()
canvas = FigureCanvas(fig)
the_map = Basemap(resolution='c',projection='lcc',llcrnrlon=-75.,
                   llcrnrlat=15.,urcrnrlon=40.,urcrnrlat=60.,lat_1=50.,
                   lon_0=-45.)

# convert to projected coordinates
dx = 2.*pi*the_map.rmajor/len(lons)
nx = int((the_map.xmax-the_map.xmin)/dx)+1
ny = int((the_map.ymax-the_map.ymin)/dx)+1
z,x,y = the_map.transform_scalar(field,lons,lats,nx,ny,returnxy=True,order=0)

# creates seemingly correct image in term of grid placement, but with rectangular pixels
the_image = the_map.pcolor(x,y,z,shading='flat')

Mike: There's no need to do the interpolation to map projection coordinates. Just pass pcolor the vertices of the lat/lon grid in map projection coordinates, i.e.

x,y = the_map(lons, lats)
the_image = the_map.pcolor(x,y,field)

The resulting pixels will not be square, but will have the shape of lat/lon boxes in projected coordinates.

-Jeff

···

--
Jeffrey S. Whitaker Phone : (303)497-6313
NOAA/OAR/CDC R/PSD1 FAX : (303)497-6449
325 Broadway Boulder, CO, USA 80305-3328