It appears that this might be a result of a bug in matplotlib. Jeff Whitaker took a look and said it looked right and I tried to reproduce this behavior without using Basemap and I was able to.
It seems that the aspect of the data values might cause the output image to be the wrong size.
Here's some code that shows the problem. Sorry for the false alarm.
# rectangle.py --
import matplotlib.pyplot as plt
def create_image():
fig = plt.figure(1, figsize=(8, 14), frameon=False, dpi=100)
fig.add_axes([0, 0, 1, 1])
ax = plt.gca()
# This isn't necessary to create the issue unless you want to see the
# transparent pixels at bottom.
# for spine in ax.spines.values():
# spine.set_linewidth(0.0)
limb = ax.axesPatch
limb.set_facecolor('#6587ad')
x1 = 0.0
y1 = 0.0
x2 = 16.
# Use this line and get what I was expecting:
# y2 = 27.999999999999994671 # produces 800 x 1400 image
# Use this line and get the wrong size
y2 = 27.999999999999994670 # produces (wrong?) 800 x 1399 image
corners = ((x1, y1), (x2, y2))
ax.update_datalim(corners)
ax.set_xlim((x1, x2))
ax.set_ylim((y1, y2))
ax.set_aspect('equal', anchor='C')
ax.set_xticks([])
ax.set_yticks([])
plt.savefig('rectangle.png', pad_inches=0.0, bbox_inches='tight')
# If you use this below, the file size is correct, but there is a single
# line transparent pixels along the bottom of the image if you set the
# linewidth to zero...
# plt.savefig('rectangle.png', pad_inches=0.0)
if __name__ == '__main__':
create_image()