Collapse a dataset from 2d or 3D into a 1D time series ?

[ledm]

The most common calculation that I do with model data is generate a 1D time series from 2D or 3D data.

Basically, to do this, just take an area-weighted (or sometimes volume-weighted) spatial average of some part of the model domain.

In numpy, it's something like this:

time_series = np.average(nc.variable['field'][:], weights = nc.variables['areacello'][:], axes=[1,2])

However, it needs to be cleverer that that, as sometimes there's no time axes in data, and sometimes you need the cell area.

In iris, it's called the collapse method: https://scitools-iris.readthedocs.io/en/latest/userguide/cube_statistics.html#area-averaging

new_cube = cube.collapsed(['grid_longitude', 'grid_latitude'], iris.analysis.MEAN, weights=grid_areas)

Does something like this function exist in coast?