DEA_Wetlands_Insight_Tool product notebook conversion to odc-stac

[colourmeamused]

Generalised checklist for notebook updates:

• odc-stac load() and configure_s3_access imported
• pystac_client Client imported
• environment setup includes catalog, stac_client and configure_s3_access()
• cell that searched a collection for items with pystac-client
• cell that loads discovered items into an xarray with odc-stac

This notebook offers an example of loading data with STAC to use as a guide: https://github.com/caitlinadams/de-sar-sample-data/blob/main/demo_notebooks/03_loading_with_stac.ipynb

[brooks-eco]

coincidentally I converted this notebook for STAC today. the load() tripped me up a few times but here's an example of where I landed -might help someone.

search bounding box and time. Specifying the bounding box in the search to find the coincident data, then again in the load because the chunking expands the extent to the edges of the chunks. I found having the same size arrays to check against the open datacube version was useful.

# Specifying coordinate reference system of the polygon.
gpgon = Geometry(poly.geometry[0], CRS(str(poly.crs)))

# Convert to WGS84 for STAC search
bbox_wgs84 = gpgon.to_crs("EPSG:4326").boundingbox

# Select time period. Consistent data is available from 01-09-1987.
time = ("2014-01-01", "2015-01-01")

# Convert time span to STAC datetime format
datetime_str = f"{time[0]}/{time[1]}"

wo_items = catalog.search(
    collections=["ga_ls_wo_3"],
    bbox=bbox_wgs84,
    datetime=datetime_str
).item_collection()

# Load Water Observations data
ds_wo = odc.stac.load(
    wo_items,
    bbox=bbox_wgs84,
    crs="EPSG:3577",
    resolution=30,
    groupby="solar_day",
    chunks={"time": 1, "x": 2048, "y": 2048}
)

another trip was in the Masking I got an error because my load of WoFS water data above created float32 and didn't support bitwise operations so convert to integer first

mask = (ds_wo.water.astype(int) & 0b01100011) == 0