Water-bodies example

The test suite selects tools and the workflow from this release document:

https://github.com/eoap/mastering-app-package/releases/download/1.1.0/app-water-bodies-cloud-native.1.1.0.cwl

A fragment such as #crop selects a process in that CWL document. The end-to-end case selects #water-bodies.

Processing stages

flowchart TD Item[STAC item] --> Green[Crop green band] Item --> NIR[Crop near-infrared band] Green --> NDWI[Normalized difference] NIR --> NDWI NDWI --> Otsu[Otsu thresholding] Otsu --> STAC[STAC catalog and water mask] Item --> STAC

The crop jobs use the bounding box -121.409,39.824,-120.73,40.482 and the coordinate reference system EPSG:4326. Their band parameter selects green or nir. The item fixture identifies a Sentinel-2B scene acquired on 13 July 2021 and provides asset links for those inputs.

Normalized difference combines the two cropped rasters into NDWI. Otsu thresholding converts that intermediate raster into a binary mask. The final step associates the mask with item metadata and writes a STAC catalog.

Existing cases

ID	CWL fragment	Job	Output assertion
`crop-green`	`#crop`	`inputs-01.yaml`	`cropped`: File, flexible location
`crop-nir`	`#crop`	`inputs-02.yaml`	`cropped`: File, flexible location
`normalized-difference`	`#norm_diff`	`inputs-03.yaml`	`ndwi`: File, flexible location
`otsu-threshold`	`#otsu`	`inputs-04.yaml`	`binary_mask_item`: File, flexible location
`stac-generation`	`#stac`	`inputs-05.yaml`	`stac_catalog`: Directory with sizes and checksums
`workflow-end-to-end`	`#water-bodies`	`inputs-e2e.yaml`	The same deterministic catalog expectation

The five unit cases execute application steps through the CWL runner. They can still require containers and network access. The normalized-difference, thresholding, and catalog cases use saved intermediate rasters, allowing each step to be investigated independently of upstream execution.

End-to-end inputs

The full workflow uses this job structure:

stac_items:
  - https://github.com/eoap/testing-framework/raw/refs/heads/main/tests/data/item.json
aoi: "-121.409,39.824,-120.73,40.482"
epsg: "EPSG:4326"
bands:
  - green
  - nir

Expected catalog

Both catalog tests check three files, including the water-mask raster:

catalog.json
S2B_10TFK_20210713_0_L2A/
  S2B_10TFK_20210713_0_L2A.json
  otsu.tif

The expected sizes in the suite are 363 bytes for the catalog, 2,755 bytes for the item, and 6,918 bytes for the mask. The suite also records a SHA-1 checksum for each file.

These comparisons detect changes to the baseline artifacts. They do not independently validate the STAC schemas or judge whether the detected water bodies are scientifically correct. See writing tests for choosing assertions and reproducibility for maintaining the baseline.