Gerrit Holl

```python import hdf5plugin from glob import glob from satpy.multiscene import MultiScene, timeseries from satpy.utils import debug_on; debug_on() ms = MultiScene.from_files( glob("/media/nas/x21308/MTG_test_data/2022_05_MTG_Testdata/RC007[012]/*BODY*.nc"), reader="fci_l1c_nc", group_keys=["repeat_cycle_in_day"]) ms.load(["temporal_rgb_vis06"]) sc = ms.blend(blend_function=timeseries) ls = sc.resample("eurol")...

Good question. Maybe something about the different start times (nominal, actual, etc.). The time matching as I've currently implemented it needs to be exact, which would only work with nominal...

We probably need a way to add tolerance or slicing capability, "search between T-65 and T-55" or so.

I found that `isinstance(t1, metpy.units.units.Quantity))` is False, whereas the same is true for `t2`. This leads to a divergence in behaviour in `metpy.units._mutate_arguments`: https://github.com/Unidata/MetPy/blob/feeca2672d99656834684770ad0cf7c166040110/src/metpy/units.py#L204-L217 What I don't know yet, is...

If I instead calculate `t3 = xr.DataArray(da.array([300]), attrs={"units": "K"}).metpy.quantify().data[0].compute()`, changing the order of operations, then t3 belongs to the correct unit registry and the calculation is successful.

Good to hear I can avoid using `.metpy.quantify()` entirely. The reason I used it in the first place is because I have dask arrays, but not all metpy functions are...

My workaround is a little harder as I'm trying to replace `parcel_profile(p[::-1], t[-1], d[-1])` (works, but computes the dask arrays 18 (!) times), with `dask.delayed(parcel_profile)(p[::-1], t[-1], d[-1]).compute()` which fails with...

I'm using numpy 1.16.1. The working directory is clean: ``` (py36) gholl@lce00:~/checkouts/python-hdf4> git status # On branch master nothing to commit (working directory clean) (py36) gholl@lce00:~/checkouts/python-hdf4> ls -A AUTHORS doc...

The reader seems to work, but maybe some metadata is not being set correctly, because resampling fails with `IndexError`. ```python from satpy import Scene from satpy.utils import debug_on; debug_on() fn...

Applying the scale factors solves the resampling problem.