I'll like to work on this @TomNicholas, where do I start from?

Hello:

I have to find maximum precipitation of each year (for example: 2007 and 2008, Dataset link are: 2007 and 2008). I have done this using resample method (i.e. .resample(time='Y').max()) after concatenating it along time dimension.

Following along SO, I am wondering if I can use preprocess to find maximum (or minimum or average) for each file first and then concatenate it using time dimension. I tried the following code and was not successful. Can someone help me with this?

```import dask.array as da import numpy as np import xarray as xr

from dask.distributed import Client client = Client() client

def preprocess_func(ds): '''Get maximum (or minimum or average) from each file and concatenate along time''' return ds.precip.max('time')

prec_ds=xr.open_mfdataset([prec_2007,prec_2008], chunks={"lat": 25,"lon": 25,"time": -1,}, preprocess=preprocess_func, concat_dim='time')```

Seconding @dcherian's comment in #4901 on an example for .encoding['source']. Working off @raybellwaves' example, something like this would have been useful to me:

```

import xarray as xr import numpy as np model1 = xr.DataArray(np.arange(2), coords=[np.arange(2)], name="f") model1.to_dataset().to_netcdf("model1.nc") model2 = xr.DataArray(np.arange(2), coords=[np.arange(2)], name="f") model2.to_dataset().to_netcdf("model2.nc") ds = xr.open_mfdataset( ... ["model1.nc", "model2.nc"], ... preprocess=lambda ds: ds.expand_dims( ... {"model_name": [ds.encoding["source"].split("/")[-1].split(".")[0]]} ... ), ... ) ds <xarray.Dataset> Dimensions: (dim_0: 2, model_name: 2) Coordinates: * dim_0 (dim_0) int64 0 1 * model_name (model_name) object 'model1' 'model2' Data variables: f (model_name, dim_0) int64 dask.array<chunksize=(1, 2), meta=np.ndarray> ```

On that note, the example above seems to work with some slight changes: ```

import numpy as np import xarray as xr

f1 = xr.DataArray(np.arange(2), coords=[np.arange(2)], dims=["a"], name="f1") f1 = f1.assign_coords(t='t0') f1.to_dataset().to_netcdf("f1.nc")

f2 = xr.DataArray(np.arange(2), coords=[np.arange(2)], dims=["a"], name="f2") f2 = f2.assign_coords(t='t1') f2.to_dataset().to_netcdf("f2.nc")

Concat along t

def preprocess(ds): ... return ds.expand_dims("t") ...

ds = xr.open_mfdataset(["f1.nc", "f2.nc"], concat_dim="t", preprocess=preprocess) ds <xarray.Dataset> Dimensions: (a: 2, t: 2) Coordinates: * t (t) object 't0' 't1' * a (a) int64 0 1 Data variables: f1 (t, a) float64 dask.array<chunksize=(2, 2), meta=np.ndarray> f2 (t, a) float64 dask.array<chunksize=(2, 2), meta=np.ndarray> ```