问题描述
这是对 防止乱码的pcolor(mesh)数据虚假水平线的后续问题 和 为什么带有蒙版数组的pcolor仍然填充连接到蒙版点的四边形，我该如何停止呢? .在常规坐标中，当我同时遮罩坐标和数据时，我可以在两部分中为环绕的坐标(例如经度)绘制pcolor，现在，在常规坐标中，我成功地避免了不必要的四边形的出现.但是，当我将其转换为地图坐标时，此解决方案失败:
This is a followup question to preventing spurious horizontal lines for ungridded pcolor(mesh) data and why does pcolor with masked array still fill quadrangles connecting to masked points, and how do I stop this?. In regular coordinates, when I mask both the coordinates and the data, I can plot a pcolor for coordinates that wrap around, such as longitudes, in two parts, and now I succeed to not get undesired quadrangles when in regular coordinates. However, when I transform it to map coordinates, this solution fails:
#!/usr/bin/env python3.6
from numpy import array, ma
from matplotlib.pyplot import figure, pcolor, savefig, axes
lons = array([[ 100., 120., 140., 160., 180.],
[ 120., 140., 160., 180., -160.],
[ 140., 160., 180., -160., -140.],
[ 160., 180., -160., -140., -120.],
[ 180., -160., -140., -120., -100.],
[-160., -140., -120., -100., -80.]])
lats = array([[ 0., 10., 20., 30., 40.],
[ 0., 10., 20., 30., 40.],
[ 0., 10., 20., 30., 40.],
[ 0., 10., 20., 30., 40.],
[ 0., 10., 20., 30., 40.],
[ 0., 10., 20., 30., 40.]])
bts = array([[ 0, 1, 2, 3, 4],
[ 5, 6, 7, 8, 9],
[10, 11, 12, 13, 14],
[15, 16, 17, 18, 19],
[20, 21, 22, 23, 24],
[25, 26, 27, 28, 29]])
figure()
pcolor(ma.masked_where(lons>0, lons), ma.masked_where(lons>0, lats), bts)
pcolor(ma.masked_where(lons<0, lons), ma.masked_where(lons<0, lats), bts)
savefig("/tmp/ok.png")
# now with cartopy
import cartopy.crs as ccrs
proj = ccrs.Mollweide(central_longitude=0)
trans = proj.transform_points(ccrs.Geodetic(), lons, lats)
figure()
ax = axes(projection=proj)
ax.pcolormesh(ma.masked_where(lons>0, trans[:, :, 0]), ma.masked_where(lons>0, trans[:, :, 1]), ma.masked_where(lons>0, bts), transform=proj)
ax.pcolormesh(ma.masked_where(lons<0, trans[:, :, 0]), ma.masked_where(lons<0, trans[:, :, 1]), ma.masked_where(lons<0, bts), transform=proj)
savefig("/tmp/not_ok.png")
在常规坐标中，根据需要:
In regular coordinates, as desired:
在地图坐标中，不需要的四边形又回来了:
In map coordinates, the undesired quadrangles are back:
请注意，因为当前投影的中心经度为零，所以任何正经度都映射到任何正向地图坐标，反之亦然.当我另外掩盖经度等于±180的经度时，我仍然会遇到相同的情况.因此问题出在其他地方.在投影地图坐标中如何将pcolor分为两部分绘制?
Note that any positive longitude maps to any positive map coordinate and vice versa, because the central longitude for the current projection is zero. When I additionally mask longitudes equal to ±180 I still get the same situation. So the problem lies elsewhere. How can I plot the pcolor in two parts while in projected map coordinates?
推荐答案
我的印象是，该代码旨在作为一种解决方法，用于根据实际上并没有很好地工作/根本没有?(?).这段代码试图做类似的事情来掩盖不同的区域，但是以某种方式不能产生期望的结果.
I have the impression that the code that is meant to be a workaround for wrapping coordinates around the limits of the projection which was introduced into cartopy according to this issue is not actually working well/at all(?). This code tries to do a similar thing of masking the different regions, but somehow does not produce the desired result.
现在，另一方面，遍布全球的小蜜蜂问题仅存在于pcolormesh中，而不存在于pcolor中.可能是由于两种情况下使用的网格划分都不相同.
因此，当使用pcolor时，该图看起来像期望的那样.
Now, on the other hand the issue of facets beeing wrapped around the globe is anyways only present in pcolormesh, not in pcolor; probably due to the different meshing used in both cases.
Therefore when using pcolor the plot looks as desired.
import cartopy.crs as ccrs
proj = ccrs.Mollweide(central_longitude=0)
trans = proj.transform_points(ccrs.Geodetic(), lons, lats)
plt.figure()
ax = plt.axes(projection=proj)
ax.pcolor(ma.masked_where(trans[:, :, 0]>0, trans[:, :, 0]), ma.masked_where(trans[:, :, 0]>0, trans[:, :, 1]), ma.masked_where(trans[:, :, 0]>0, bts), transform=proj)
ax.pcolor(ma.masked_where(trans[:, :, 0]<0, trans[:, :, 0]), ma.masked_where(trans[:, :, 0]<0, trans[:, :, 1]), ma.masked_where(trans[:, :, 0]<0, bts), transform=proj)
plt.show()
这篇关于当投影在Cartopy坐标中时，为什么带有蒙版数组的pcolor会填充不想要的四边形?的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持！