Collect a sample set of tiles and pull corresponding Landsat 8 to test 3.1.2

[pahbs]

This is a test of the Landsat compositing across a variety of extents in the boreal domain. Number of extents to test depends on cost. At least 5 other locations in addition to the one we've been using, distributed across AK and Canada?

[wildintellect]

It's was about $0.5 for the 1 degree test area. So yes we can easily do $5 of more data. Just make a list of Albers tile ids to use, which are actually smaller than 1 degree tiles previously used.

[nmt28]

lets go with

• 40818, 40819, 40350 (these create a 2x2 grid with the tile we already have to also allow us to test tile sizes and RAM limits etc)
• 36937 (canada)
• 11633 (Norway-ish)
• 18822 (Siberia/Russia)

[wildintellect]

The data for 40818, 40819, 40350 may already be covered by the existing sample2, @nmt28 can you check?

Question: should we convert the notebook to a python script for this stage of testing, to start aligning it with how a DPS job would be run?

[nmt28]

Will take a look now. Happy to start using python scripts.

[wildintellect]

After fixing a projection issue in the tileid layer, the new tileids corresponding are listed below. We can probably reduce that number a bit, as these tiles are a slightly different size. sample3.geojson.txt 1073 1074 1075 1132 1133 17400 17401 17613 17614 17615 21961 22199 22200 30542 30543 30821 30822 30823 31099 31100 31101 34371 34650

[wildintellect]

@pahbs The core testing block is to try in DPS covered by pre-cached sample2 30542 30543 30821 30822 30823

[wildintellect]

Noting that 34371,34650 are on the dateline and will be set aside to test handling of reprojection on the edge.

[nmt28]

Current test area tiles are:

• 30543
• 30542
• 30823
• 30822

[nmt28]

@wildintellect Do you think we can pull the data for 30821? That way we have nice quadrant of data to use as the test area?

[wildintellect]

Closing, in deference to trying to get direct AWS access in Ops now. Should not need to pull any more sample data.