bug in calculate_area

[fsenf]

The function calculate_area takes features, mask as argument. If mask is 3D (i.e. incl. time), calculate_area fails because internally calculates an area field (np.array) which is 2D. Both mask and area are provided to thescipy.ndimage function labeled_comprehension which however assumes that the 1st two arguments have the same shape. confict -> 2D != 3D

Here is part of the error

File [~/TROPOS/proj/2022-06-tobac-devel/tobac/tobac/analysis.py:960](https://file+.vscode-resource.vscode-cdn.net/home/senf/TROPOS/proj/2022-06-tobac-devel/tobac/examples/Basics/~/TROPOS/proj/2022-06-tobac-devel/tobac/tobac/analysis.py:960), in calculate_area(features, mask, method_area)
    957 else:
    958     raise ValueError("method undefined")
--> 960 feature_areas = labeled_comprehension(
    961     area, mask.data, features["feature"], np.sum, area.dtype, np.nan
    962 )
    964 features["area"] = feature_areas
    966 return features

File [~/TROPOS/proj/2022-06-tobac-devel/tobac-env-py38/lib/python3.8/site-packages/scipy/ndimage/_measurements.py:511](https://file+.vscode-resource.vscode-cdn.net/home/senf/TROPOS/proj/2022-06-tobac-devel/tobac/examples/Basics/~/TROPOS/proj/2022-06-tobac-devel/tobac-env-py38/lib/python3.8/site-packages/scipy/ndimage/_measurements.py:511), in labeled_comprehension(input, labels, index, func, out_dtype, default, pass_positions)
    509     input, labels = numpy.broadcast_arrays(input, labels)
    510 except ValueError as e:
--> 511     raise ValueError("input and labels must have the same shape "
    512                         "(excepting dimensions with width 1)") from e
    514 if index is None:
    515     if not pass_positions:

ValueError: input and labels must have the same shape (excepting dimensions with width 1)

A solution would be to care in the calculations that areahas always the same shape as mask.

As calculate_area is not included in the testing I guess it is not really used by the group, right?

[w-k-jones]

Good catch, we should probably allow both a 2d area calculation and volume for 3d fields

[w-k-jones]

Footprint area of 3D segments is now resolved by #378 . It may be useful to add a function to calculate volume of 3D objects in future, but I will close this for now