Water extent exercise (Session 6) broken

[eefaye]

Landsat data ("new" collection 2 data) and WOfS ("old" GA-based WOfS) have both started showing only half of the extent selected.

This breaks the exercise and plotting options.

It also changes the water area detected:

Left - how it is now, notice the yaxis only goes to 300
Right - how it was originally/is displayed on the Training website currently

Having trouble isolating the cause, but will have a closer look tomorrow. Would appreciate some extra eyes on this.

[nanaboamah89]

Hi @eefaye, I was able to go through the session six.

Landsat data ("new" collection 2 data) and WOfS ("old" GA-based WOfS) have both started showing only half of the extent selected.
I had the same image showing half and not full imagery, maybe could be an issue of cloud.

With the chart showing the extent of water body I had the same as the one in the exercise as shown below:

Few things I noticed the timestep now loaded 43 times for the Landsat when I run it and, on the website is 44 timesteps.
Training Page

Sandbox

dc load for WOFS has the square bracket include for loading the WOFS and it breaks the code

The above are what I also noticed from the session 6.

[andrewdhicks]

I tried this for loading WOfS:

ds_wofs = dc.load(product="wofs_ls",
                  fuse_func=wofs_fuser,
                  dask_chunks={'time': 1},
                  platform='landsat-8',
                  **query
                 )

And got back 46 timestamps.

This way we cut out the LS7 data that will distort the area count.

[eefaye]

It only works intermittently for me - it fixed itself but then broke again!

I have pushed some updates to the website. If @nanaboamah89 or @andrewdhicks , you could confirm for me the notebook here actually runs: https://training.digitalearthafrica.org/en/latest/_static/session_6/Water_extent_exercise-solution.ipynb