Closed brey closed 1 year ago
times = pd.date_range("2023-12-30", "2024-01-02", freq="h")
ds = xr.Dataset(
{
"time": (("time"), times),
}
)
ds.resample(time="1D")
@pmav99 It works. Since we have no gaps there is no problem with resample
. Thanks!
By the way, since from our tests we saw that split by day gives some performance boost, should we perhaps enable it by default?
I don't think so. If there is a gap in time, resample
will interpolate, thus increasing the wall time. Also, if the file is small (regional window) there might not be a benefit. We should document this well and let the user decide.
Currently we have a way to split the 3 day meteo dataset in daily chunks with
which outputs 4 daily chunks.
However, there is a problem with the
groupby
functionality.One can see it below:
Using
dayofyear
solves it for the monthly occurrence but we get the same for cross years.@pmav99 @tomsail Any ideas?