Refactor methods to operate on and return DataArrays

[jgerardsimcock]

• climate_toolbox version:
• Python version:
• Operating System:

Description

Right now methods for projecting from gridded to iso/hierid operate on datasets. This is kind of a problem when we have multiple climate variables in the same dataset (spline-tas, degree-days) or something like below.

Dimensions:  (lat: 720, lon: 1440, time: 365)
Coordinates:
  * time     (time) datetime64[ns] 2006-01-01T12:00:00 2006-01-02T12:00:00 ...
  * lat      (lat) float32 -89.875 -89.625 -89.375 -89.125 -88.875 -88.625 ...
  * lon      (lon) float32 0.125 0.375 0.625 0.875 1.125 1.375 1.625 1.875 ...
Data variables:
    tas      (time, lat, lon) float32 271.728 271.744 271.745 271.733 ...
    tas_2    (time, lat, lon) float32 73836.0 73844.9 73845.5 73838.8 ...
    tas_3    (time, lat, lon) float32 2.00633e+07 2.00669e+07 2.00672e+07 ...

I am uncomfortable with the current implementation where, in order to project to hierid, we would have to iterate over each variable like the following:

for var in ds.data_vars.keys():
            ds = weighted_aggregate_grid_to_regions(
                ds, var, aggwt, agglev, weights=weights)

The assignment and reassingment of datasets in xarray makes me uncomfortable.

One possible implementation:

xformd_das = []
vars = []
for var in ds.data_vars.keys():
        da = ds[var]
        da = weighted_aggregate_grid_to_regions(da, aggwt, agglev, weights
        xformd.append(da)
        vars.append(var)

da = xr.concat(xformd_das, pd.Index(vars))
...

da.to_netcdf(...)