Zarr Cloud-Native Workflow for Water Body Detection

This repository provides a cloud-native Earth-Observation Application Package for detecting water bodies over time using Sentinel-2 data.

The workflow leverages Common Workflow Language (CWL) and outputs data in the Zarr format, described using the STAC Datacube Extension.

Features

Zarr Output: Outputs water body data as Zarr files for cloud-native geospatial processing.
STAC Metadata: Includes STAC metadata for integration with EO datalakes.
Scalable processing: Designed to process multiple Sentinel-2 STAC items in parallel.

Workflow Overview

The workflow is based on one of the workflows of the https://github.com/eoap#mastering-earth-observation-application-packaging-with-cwl module and includes the following steps:

Cropping: Crops Sentinel-2 imagery to the Area of Interest (AOI).
Normalized Difference Water Index (NDWI): Computes NDWI to identify water bodies.
Otsu Thresholding: Applies Otsu's thresholding method to binarize NDWI values.
STAC Metadata Creation: Generates STAC items for the detected water bodies.
Zarr Conversion: Converts the results into a Zarr dataset with Datacube metadata.

Running the Workflow

Use the approach described in https://github.com/eoap/dev-platform-eoap to run this module on Minikube using skaffold