Rasterized dataset cannot utilize the time dimension in holoviews/geoviews

Datashader
1

I published a cookbook with project pythia utilizing geoviews and the datashader. When you plot a regular quadmesh, the time slider works when specifying that time is a dimension in the dataset, however, when you rasterize the dataset, the time slider no longer works. The example code and cookbook can be found here: 02_InteractiveVisualization Part 1: Geoviews — ARCO ERA-5 Interactive Visualization (projectpythia.org)

2

I can’t run the following cell:

msl93 = msl.sel(time=slice('1993-03-13T18:00:00','1993-03-14T00:00:00')).pipe(mirror_point_at_360)
t2m93 = t2m.sel(time=slice('1993-03-13T18:00:00','1993-03-14T00:00:00')).pipe(mirror_point_at_360)

Getting this traceback:

KeyError: 'Indexing with a boolean dask array is not allowed. This will result in a dask array of unknown shape. Such arrays are unsupported by Xarray.Please compute the indexer first using .compute()'
3

@Hoxbro Yes, this has crept into recent versions of Xarray. In our Project Pythia Cookbook we pinned xarray 2022.12.0.

4

This works fine in the notebook:

There has been a regression in Jupyterlab>4, which could cause the problem. This is fixed in Panel 1.1.1.

5

YES!! @MikeBWx and I had just hypothesized that something was different with Jupyterlab 4. I can confirm that the time slider functionality works now on the rasterized quadmesh.

We will update the Project Pythia Cookbook with panel 1.1.1 in the environment.yml file.

Thanks!

1 Like