lgolston

Yes, I'd like to echo this issue. Simply running: ``` from owslib.wmts import WebMapTileService wmts = WebMapTileService("http://map1c.vis.earthdata.nasa.gov/wmts-geo/wmts.cgi") ``` Gives `KeyError: 'Content metadata for layer "MODIS_Combined_MAIAC_L2G_AerosolOpticalDepth" already exists'` We are seeing...

I suspect this is a bug. Using your example, setting the first term in set_extent to be between 0 and 1 (for instance, 0.6 in the image below), progressively adjusts...

A work around is to use data coordinates instead: ``` import numpy as np import matplotlib.pyplot as plt from matplotlib.path import Path import cartopy.crs as ccrs from cartopy.feature import LAND,...

Unfortunately I think some loss of accuracy associated with the speedup from transform_first=True is expected - this is noted at the contour [docs](https://scitools.org.uk/cartopy/docs/latest/reference/generated/cartopy.mpl.geoaxes.GeoAxes.html#cartopy.mpl.geoaxes.GeoAxes.contourf) and [associated example](https://scitools.org.uk/cartopy/docs/latest/gallery/scalar_data/contour_transforms.html), although the differences are...

Great point @rcomer! I have investigated a little bit and that is definitely causing a problem - at low and high projected (and sorted) x values, y is jumping back...

I don't think there is an inherent reason why the fast version couldn't perform better here. Looking again, with sorting turned off there is a large discontinuity in x around...

Thanks for the report - a bit hard to figure out what is going on here. I have not found the issue but did notice that the Mollweide plot generally...

After seeing #1076, this problem appears to be similar. Was able to create some test cases based on your example to better show what is going on. Each case isolates...

Working through the plotting functions, the problem arises from _project_linear_ring in crs.py and ultimately project_linear in trace.pyx. Here is a more reduced example, which replicates the behavior of test case...

Okay, I was able to track this down through trace (notably seeing POINT_OUT states in bisect once the problem starts happening) and ultimately back up to the definition of the...