Save the T90P as the nc file, the type value is numpy.timedelta64.

[CGL5230]

Generic Issue

• xclim version:0.46.0
• Python version:3.11
• Operating System:Linux

Description

I save the TG90P index as nc format. But I open the nc file and find the type of TG90P is numpy.timedelta64. The expected type is numpy.

What I Did

I calculated the TG90P by ssp585_tas_p90 = xclim.core.calendar.percentile_doy(tas_AWI_CM_ssp585.tas,per=90) AWI_CM_ssp585_TG90P = xclim.indicators.atmos.tg90p(tas='tas',tas_per=ssp585_tas_p90,freq='YS',ds=tas_AWI_CM_ssp585) and save as the nc format as ssp585_tas_p90.to_netcdf('/home/workdir/WHs_Climate/future/cmip6/indexs/AWI_CM_ssp585_tg90P_p90.nc') AWI_CM_ssp585_TG90P.to_netcdf('/home/workdir/WHs_Climate/future/cmip6/indexs/AWI_CM_ssp585_TG90P.nc')

What I Received

This value is not intuitive, it should be a "number of days" greater than a certain threshold.

Code of Conduct

• I agree to follow this project's Code of Conduct

[CGL5230]

I read the description of TG90P . I think the type of tg90P should be "number".

.Returns:tg90p (DataArray) – Number of days with mean temperature above the 90th percentile (days_with_air_temperature_above_threshold) [days],

[aulemahal]

That's a feature of xarray. It reads that the units are "days" and it thinks it will help you by giving you "timedelta" objects.

The workaround is to open the netcdf with decode_timedelta=False. See options on https://docs.xarray.dev/en/stable/generated/xarray.open_dataset.html#xarray.open_dataset.

xr.open_dataset('/home/workdir/WHs_Climate/future/cmip6/indexs/AWI_CM_ssp585_TG90P.nc', decode_timedelta=False)