ukcp18 2.2 nc file problem for rolling window problem

[Clare Wang]

Hi Iris expert,

I downloaded ukcp18 2.2 climate change precipitation hourly data( see example file: http://dap.ceda.ac.uk/thredds/fileServer/badc/ukcp18/data/land-cpm/uk/2.2km/rcp85/01/pr/1hr/v20210615/pr_rcp85_land-cpm_uk_2.2km_01_1hr_19810101-19810130.nc), and I would like to do rolling window 'SUM' analysis for window size of 6, 9, 12, 24.

I have done rolling window for size 6, 9, they are all fine. However, I got error when I tried window size of  12, 

infile = "pr_rcp85_land-cpm_uk_2.2km_01_1hr_19810101-19810130.nc"

cube = iris.load_cube(infile)

rolling_cube = cube.rolling_window("time", iris.analysis.SUM, 12)

The error message is as follow. Please any one can help for pointing out what caused the problem:

djf 1981

C:\Miniforge3\envs\climatechange\lib\site-packages\iris\cube.py:4383: UserWarning: The bounds of coordinate 'time' were ignored in the rolling window operation.

  warnings.warn(

Traceback (most recent call last):

  File "C:\Users\changguiwang\source\repos\climatechange\src\rainfall_proc.py", line 109, in <module>

    main(infile)

  File "C:\Users\changguiwang\source\repos\climatechange\src\rainfall_proc.py", line 68, in main

    rolling_cube = cube.rolling_window("time", iris.analysis.SUM, 11)

  File "C:\Miniforge3\envs\climatechange\lib\site-packages\iris\cube.py", line 4442, in rolling_window

    data_result = aggregator.aggregate(

  File "C:\Miniforge3\envs\climatechange\lib\site-packages\iris\analysis\__init__.py", line 584, in aggregate

    result = self.call_func(data, axis=axis, **kwargs)

  File "C:\Miniforge3\envs\climatechange\lib\site-packages\iris\_lazy_data.py", line 33, in inner

    result = func(*args, **kwargs)

  File "C:\Miniforge3\envs\climatechange\lib\site-packages\iris\analysis\__init__.py", line 1136, in inner_stat

    dask_result = dask_stats_function(array, axis=axis, **kwargs)

  File "C:\Miniforge3\envs\climatechange\lib\site-packages\iris\analysis\__init__.py", line 1402, in _lazy_sum

    array = iris._lazy_data.as_lazy_data(array)

  File "C:\Miniforge3\envs\climatechange\lib\site-packages\iris\_lazy_data.py", line 195, in as_lazy_data

    data = da.from_array(

  File "C:\Miniforge3\envs\climatechange\lib\site-packages\dask\array\core.py", line 3169, in from_array

    chunks = normalize_chunks(

  File "C:\Miniforge3\envs\climatechange\lib\site-packages\dask\array\core.py", line 2807, in normalize_chunks

    raise ValueError(

ValueError: Empty tuples are not allowed in chunks. Express zero length dimensions with 0(s) in chunks

Clare

Many thanks

[Simon Peatman]

Dear Clare,

I've tried this with the same file and I find it runs with no errors for window sizes of 11 (as in the traceback you pasted below) or 12.  I'm using Iris version 3.0.4 and dask version 2021.07.2.

Simon