Update repository with recent DigitalEarthAfrica notebooks and code

[robbibt]

In the recent ODC Hackathon 2020 there were some major additions and updates to code in the DEAfrica notebooks repository (https://github.com/digitalearthafrica/deafrica-sandbox-notebooks).

We should keep a list of relevant things that were updated in DEAfrica notebooks/code so that we can port them back over to dea-notebooks.

@cbur24, @caitlinadams, if you get a chance can you comment with any new notebooks/functionality that was developed during the hack that you think would also be useful to include in dea-notebooks? I'm thinking things like the Dask notebook I saw that would work just as well in the Australian context.

[caitlinadams]

• [x] We're most of the way through an introductory notebook to explain Dask: https://github.com/digitalearthafrica/deafrica-sandbox-notebooks/pull/38
• Potentially the Sentinel-1 notebook: https://github.com/digitalearthafrica/deafrica-sandbox-notebooks/blob/master/Datasets/Sentinel_1.ipynb
• Possibly the Urbanization case study I started: https://github.com/digitalearthafrica/deafrica-sandbox-notebooks/pull/37 -- However, I think this should probably be reworked to focus less on calculating a sustainable development goal, and potentially more on an Australian industry/monitoring kind of question

[cbur24]

• The format/structure of the Landsat Collections notebook could work well for Aus. Would need to switch out the USGS Landsat collections C1 & C2 for the Aus variants C3 & C2.
• Wetlands Insight Tool notebook under Real_world_examples. Some adjustments would need to be made to the deafrica_wetlandstools.py script to make it work for Aus.
• Changes to the deafrica_Classificationtools/get_training_data_for_shp to allow the function to handle time series data.
• Loading fractional cover with the load_ard function. FC hasn't been generated for C3 yet but presumably its not too far away

[robbibt]

Another one I noticed come up in Github emails: was there a bug fix to mostcommon_crs to allow it to work when certain query/data load params are passed to it? @cbur24

[cbur24]

Yes, that's right @robbibt . Pretty basic, I just removed the offending params:

query = deepcopy(query)
if 'dask_chunks' in query:
      query.pop('dask_chunks', None)

if 'align' in query:
      query.pop('align', None)

A better fix might be to take the query input and only select the params that mostcommon_crs requires (time, lat, lon)

[cbur24]

A soil moisture notebook has recently been added to the 'Datasets' folder in the DE Africa repo. Uses a global dataset at 10km resolution so should be useful for Australia too

[caitlinadams]

I've added a notebook on generating median, mean, min and max composites to go along with the geomedian notebook. This largely replaces the functionality offered by the DCAL mosaic notebook.

See the PR for details, as I also changed the title of the existing geomedians notebook to group them together.

[cbur24]

• A notebook on downloading and anlaysing ERA5 reanalysis climate data
• Major update to the training data function(s) in dea_classificationtools.py, runs in parallel now and can accept a custom function. Line 305 to 650 here

[andrewdhicks]

Added ds.sel(method='...') examples in Frequently_used_code/Working with time in xarray notebook in this PR.

[cbur24]

New Frequently_used_code notebook demonstrating a PCA on multispectral data

[cbur24]

Two new Beginners_guide notebooks:

• Intro_to_numpy
• Intro_to_xarray

[cbur24]

New notebook Vegetation_phenology, and accompanying script deafrica_temporal_statistics

[robbibt]

This is done, but we should revisit in the future