Tiler module

[lillythomas]

Tiling module that generates 256x256 tiled xarray datasets from the larger scene-level datacube generated in datacube.py.

To do:

• add cloud filter
• refine extent to tile within to valid data region
• add main script to call the datacube module followed by the tile module
• decide on method for storing the tiled xarray datasets
• add some docstrings
• plot tiles to make sure they look good

[yellowcap]

FYI I am using the following snippet to plot the merged data for visual cross-checks:

def plot(merged):
    """
    Plot a tile in RGB and vh, and DEM
    """
    import numpy
    from matplotlib import pyplot as plt

    # data = merged.coarsen(x=100, y=100, boundary="trim").mean()
    data = merged

    fig, axs = plt.subplots(3)

    rgb = (
        numpy.array(
            [
                data.get("B04"),
                data.get("B03"),
                data.get("B02"),
            ]
        ).clip(0, 3000)
        / 3000
    )
    axs[0].imshow(rgb.transpose(1, 2, 0))
    axs[1].imshow(numpy.log(data.get("vh")))
    axs[2].imshow(data.get("DEM"))

    plt.show()

[yellowcap]

Haven't had time to test this, was focused on fixing the cube. But looks good from a high level.

[lillythomas]

The tiler function seems to work nicely. Here is a plotted result [RGB, VH, DEM] in order from top to bottom.

[lillythomas]

Added a catch for sampled S2 reference dates which don't have S1 scenes within the +/- 3 day surrounding interval. This should prevent any failures due to unavailable matchup as it just triggers a new sampling for S2.

[lillythomas]

Going to merge this to keep things moving. Will address the tile saving method in a follow-up PR.