xr$interp problem with translation from posXct to np.datetime64

[jmarkloew]

I tried to interpolate a xr.dataarray in R with this code:

polRatio <- ds$VV/ds$VH
polRatio$attrs$units= "VV/VH"

load_sq= seq.Date(from = as.Date("01/10/2019", format= "%m/%d/%Y"), to= as.Date("12/31/2019", format= "%m/%d/%Y"), by= "6 day")
intp_sq= seq.POSIXt(from = as.POSIXct("01/10/2019", format= "%m/%d/%Y"), to= as.POSIXct("12/31/2019", format= "%m/%d/%Y"), by= "day")

intp_sq= intp_sq[-which(intp_sq %in% load_sq)]

CR_intp= polRatio$interp(time= intp_sq[1:length(intp_sq)])

But it gave me the following error:

Error in py_call_impl(callable, dots$args, dots$keywords): KeyError: numpy.datetime64('NaT')

Detailed traceback:
  File "/home/datacube/anaconda3/envs/jupyterhub/lib/python3.9/site-packages/xarray/core/dataarray.py", line 1725, in interp
    ds = self._to_temp_dataset().interp(
  File "/home/datacube/anaconda3/envs/jupyterhub/lib/python3.9/site-packages/xarray/core/dataset.py", line 3131, in interp
    obj, newidx = missing._localize(obj, {k: v})
  File "/home/datacube/anaconda3/envs/jupyterhub/lib/python3.9/site-packages/xarray/core/missing.py", line 559, in _localize
    imin = index.get_loc(minval, method="nearest")
  File "/home/datacube/anaconda3/envs/jupyterhub/lib/python3.9/site-packages/pandas/core/indexes/datetimes.py", line 705, in get_loc
    raise KeyError(orig_key) from err

Traceback:

1. polRatio$interp(time = intp_sq[1:length(intp_sq)])
2. py_call_impl(callable, dots$args, dots$keywords)

I also checked the type of the timestamp object of the xr.dataarray I was trying to interpolate and it is posxct
I am definitely no expert but could it point to a problem with the translation from posixct to np.datetime64?

Thanks and best regards

[16EAGLE]

Without having investigated your example further, I noticed that intp_sq is of length 0. I guess that is unintended, right?

When running intp_sq[-which(intp_sq %in% load_sq)], intp_sq becomes POSIXct of length 0. Could you try changing your example and see whether the interpolation error persists? If yes I will look into the date/time conversion.

[jmarkloew]

There was a mixup between Date and POSIX formats. If I use a POSIXt sequence for int_p and load_sq xr$interp works fine. However, if a Date sequence is given in int_p the error says it needs a datetime object. It would be nice if an exception is added that catches the Date format or at least an error message which points to the required POSIX format. dc_load works fine with Date format so it would be more comfortable if it is some how handled. I also added the original error message that occurrs when using Date instead of POSIX format:

Error in py_call_impl(callable, dots$args, dots$keywords): TypeError: When interpolating over a datetime-like coordinate, the coordinates to interpolate to must be either datetime strings or datetimes. Instead got
<xarray.IndexVariable 'time' (time: 296)>
array([datetime.date(2019, 1, 11), datetime.date(2019, 1, 12),
       datetime.date(2019, 1, 13), ..., datetime.date(2019, 12, 28),
       datetime.date(2019, 12, 29), datetime.date(2019, 12, 31)], dtype=object)
Attributes:
    units:    seconds since 1970-01-01 00:00:00

Detailed traceback:
  File "/home/datacube/anaconda3/envs/jupyterhub/lib/python3.9/site-packages/xarray/core/dataarray.py", line 1725, in interp
    ds = self._to_temp_dataset().interp(
  File "/home/datacube/anaconda3/envs/jupyterhub/lib/python3.9/site-packages/xarray/core/dataset.py", line 3123, in interp
    validated_indexers = {
  File "/home/datacube/anaconda3/envs/jupyterhub/lib/python3.9/site-packages/xarray/core/dataset.py", line 3124, in <dictcomp>
    k: _validate_interp_indexer(maybe_variable(obj, k), v)
  File "/home/datacube/anaconda3/envs/jupyterhub/lib/python3.9/site-packages/xarray/core/dataset.py", line 3114, in _validate_interp_indexer
    raise TypeError(

Traceback:

1. polRatio$interp(time = intp_sq[1:length(intp_sq)])