KeyError when applying xgcm.Grid.cumsum on the LAYERS package output

I have a xgcm.grid object that is recognizing the layers grid:

import xgcm.grid as Grid
grid = Grid(ds, periodic=['X'])
grid
--------
<xgcm.Grid>
Z Axis (not periodic):
  * center   Z (40) --> left
  * left     Zl (40) --> center
  * outer    Zp1 (41) --> center
  * right    Zu (40) --> center
Y Axis (not periodic):
  * center   YC (200) --> left
  * left     YG (200) --> center
X Axis (periodic):
  * center   XC (100) --> left
  * left     XG (100) --> center
1TH Axis (not periodic):
  * center   layer_1TH_center (42) --> outer
  * outer    layer_1TH_bounds (43) --> center
  * inner    layer_1TH_interface (41) --> center
T Axis (not periodic):
  * center   time (3600)

However, when I try to apply the grid.cumsum function over the 1TH Axis, I get the following KeyError:

vname = 'LaVH1TH' 
mean_thkns = ds[vname].mean(['time','XC'])
print(mean_thkns)
grid.cumsum(mean_thkns[::-1], 'layer_1TH_center')
--------
<xarray.DataArray 'LaVH1TH' (layer_1TH_center: 42, YG: 200)>
dask.array<shape=(42, 200), dtype=float32, chunksize=(42, 200)>
Coordinates:
  * YG                (YG) >f4 0.0 10000.0 20000.0 30000.0 40000.0 50000.0 ...
  * layer_1TH_center  (layer_1TH_center) float32 -0.1 0.1 0.3 0.5 0.7 0.9 ...
---------------------------------------------------------------------------
KeyError                                  Traceback (most recent call last)
<ipython-input-32-7d428bd4e21f> in <module>()
      2 mean_thkns = ds[vname].mean(['time','XC'])
      3 print(mean_thkns)
----> 4 grid.cumsum(mean_thkns, 'layer_1TH_center')

~/xgcm/xgcm/grid.py in cumsum(self, da, axis, **kwargs)
    552         """
    553 
--> 554         ax = self.axes[axis]
    555         return ax.cumsum(da, **kwargs)
    556 

KeyError: 'layer_1TH_center'

Is there a work around for this?

[rabernat]

Do you understand how I debugged this? The important step is to understand the nature of this error

--> 554         ax = self.axes[axis]
    555         return ax.cumsum(da, **kwargs)
    556 

KeyError: 'layer_1TH_center'

You could have examined grid.axes yourself to verify that ‘layer_1TH_center’ was not a valid key.

[rabernat]

The name of the axis is 1TH, not layer_1TH_center. layer_1TH_center is the name of a coordinate, not an axis. So you should do

grid.cumsum(mean_thkns, '1TH')

(Also, do not reverse the data. that will not work with xgcm.)