Skip to content

Latest commit

 

History

History
329 lines (242 loc) · 13.1 KB

README.md

File metadata and controls

329 lines (242 loc) · 13.1 KB
Raw

Harmony Browse Image Generator (HyBIG).

This Harmony backend service is designed to produce browse imagery, with default behaviour to produce browse imagery that is compatible with the NASA Global Image Browse Services (GIBS).

This means that defaults for images are selected to match the visualization generation requirements and recommendations put forth in the GIBS Interface Control Document (ICD), which can be found on Earthdata Wiki along with additional GIBS documentation.

HyBIG creates paletted PNG images and associated metadata from GeoTIFF input images. Scientific parameter raster data as well as RGB[A] raster images can be converted to browse PNGs. These browse images undergo transformation by reprojection, tiling and coloring to seamlessly integrate with GIBS.

Reprojection

GIBS expects to receive images in one of three Coordinate Reference System (CRS) projections.

Region Code Name
north polar EPSG:3413 WGS 84 / NSIDC Sea Ice Polar Stereographic North
south polar EPSG:3031 WGS 84 / Antarctic Polar Stereographic
global EPSG:4326 WGS 84 -- WGS84 - World Geodetic System 1984, used in GPS

HyBIG processing will attempt to choose a GIBS-suitable target CRS from the input image or read it from the inputs. Reprojection is done by resampling via nearest neighbor. It is important to note that HyBig outputs are not scientific data, but browse imagery and should not be used for scientific analysis.

Tiling

Large output images are divided into smaller, more manageable tiles for efficient handling and processing, as per agreement with GIBS. The maximum untiled image size generated by HyBIG is 67,108,864 cells (8,192 x 8,192). If the output image exceeds this threshold, HyBIG automatically tiles the output into multiple 4,096 x 4,096 cell images.

Tiled images are labeled with the zero-based column and row numbers inserted into the output filename before its extension. For example, VCF5KYR_1991001_001_2018224205008.r01c02.png represents the second row and third column of the output tiles. The tiles at the edges are truncated to fit the overall image dimensions. Currently, you cannot override this behavior.

Coloring

HyBIG images are colored in several ways. A palette can be included in the input STAC Item. If an Item's asset contains a value with the role of palette, it is assumed to be a reference to a remote color table, which is fetched from the asset's href and parsed as a GDAL color table.

If the STAC Item lacks color information, the Harmony message source is searched for a related URL with a "content type" of VisualizationURL and a "type" of Color Map. If found, it is presumed to be a remote color table and fetched from that location.

In the absence of remote color information, the input image itself is searched for a color map, which is used if present.

If no color information can be found, grayscale is used.

Defaults

HyBIG tries to provide GIBS-appropriate default values for the browse image outputs. When a user does not provide a target values for the output, HyBIG will try to pick an appropriate default.

Coordinate Reference System (CRS)

HyBIG selects a default CRS from the list of GIBS preferred projections. The steps followed are simple but effective:

  1. If the proj is lonlat use global (EPSG:4326)
  2. If the projection latitude of origin is above 80° N use northern (EPSG:3413)
  3. If the projection latitude of origin is below -80° N use southern (EPSG:3031)
  4. Otherwise use global (EPGS:4326)

Scale Extent (Image Bounds)

The default scale extent for an output image is computed by reprojecting the input data boundary into the target CRS. It densifies the edges by adding 21 points (rasterio's default) to each edge before reprojection to account for non-linear edges produced by the transformation ensuring inclusion of all data in the output image.

Dimensions / Scale Sizes

Output image dimensions can be explicitly included as width and height in the harmony message or computed based on the scale extent and scale size (resolution).

The dimension computations from the scale extent and scale size:

height = round((scale_extent['ymax'] - scale_extent['ymin']) / scale_size.y)
width = round((scale_extent['xmax'] - scale_extent['xmin']) / scale_size.x)

When a Harmony message contains neither dimensions nor scaleSizes a default set of dimensions is computed.

For coarse input data, the resolution (scale size) is used with the scale extent to compute the output dimensions. For high resolution data, finer than 2km per gridcell, the input resolution is used to lookup the closest GIBS preferred resolution (Table 4.1.8-1 and -2 from the ICD) and the preferred resolution along with the scale extent is used to compute the output image dimensions.

Customizations

Users can request customizations to the output images such as crs, scale_extents, or scale_sizes and dimensions (height & width) in the harmony request. However, the generated outputs may not be compatible with GIBS.

When a user customizes scale_extent or scale_size, they must also include a crs in the request. The units of the cusomized values must match the target CRS. For example, specifying a bounding box in degrees requires a target CRS also with units of degrees.

Repository structure:

|- 📂 bin
|- 📂 docker
|- 📂 docs
|- 📂 harmony_browse_image_generator
|- 📂 tests
|- CHANGELOG.md
|- CONTRIBUTING.md
|- LICENSE
|- README.md
|- conda_requirements.txt
|- dev-requirements.txt
|- legacy-CHANGELOG.md
|- pip_requirements.txt

  • bin - A directory containing utility scripts to build the service and test images. A script to extract the release notes for the most recent service version, as contained in CHANGELOG.md is also in this directory.

  • docker - A directory containing the Dockerfiles for the service and test images. It also contains service_version.txt, which contains the semantic version number of the service image. Any time an update is made that should have an accompanying service image release, this file should be updated.

  • docs - A directory with example usage notebooks.

  • harmony_browse_image_generator - A directory containing Python source code for the HyBIG. adapter.py contains the BrowseImageGeneratorAdapter class that is invoked by calls to the service.

  • tests - A directory containing the service unit test suite.

  • CHANGELOG.md - This file contains a record of changes applied to each new release of a service Docker image. Any release of a new service version should have a record of what was changed in this file.

  • CONTRIBUTING.md - This file contains guidance for making contributions to HyBIG, including recommended git best practices.

  • LICENSE - Required for distribution under NASA open-source approval. Details conditions for use, reproduction and distribution.

  • README.md - This file, containing guidance on developing the service.

  • conda_requirements.txt - A list of service dependencies, such as GDAL, that cannot be installed via Pip.

  • dev-requirements.txt - list of packages required for service development.

  • legacy-CHANGELOG.md - Notes for each version that was previously released internally to EOSDIS, prior to open-source publication of the code and Docker image.

  • pip_requirements.txt - A list of service Python package dependencies.

Local development:

Local testing of service functionality is best achieved via a local instance of Harmony. Please see instructions there regarding creation of a local Harmony instance.

If testing small functions locally that do not require inputs from the main Harmony application, it is recommended that you create a Python virtual environment via conda, and then install the necessary dependencies for the service within that environment via conda and pip then install the pre-commit hooks.

> conda create -name hybig-env python==3.11
> conda install --file conda_requirements.txt
> pip install -r pip_requirements.txt
> pip install -r dev-requirements.txt

> pre-commit install

Tests:

This service utilises the Python unittest package to perform unit tests on classes and functions in the service. After local development is complete, and test have been updated, they can be run in Docker via:

$ ./bin/build-image
$ ./bin/build-test
$ ./bin/run-test

The tests/run_tests.sh script will also generate a coverage report, rendered in HTML, and scan the code with pylint.

Currently, the unittest suite is run automatically within a GitHub workflow as part of a CI/CD pipeline. These tests are run for all changes made in a PR against the main branch. The tests must pass in order to merge the PR.

The unit tests are also run prior to publication of a new Docker image, when commits including changes to docker/service_version.txt are merged into the main branch. If these unit tests fail, the new version of the Docker image will not be published.

Versioning:

Service Docker images for HyBIG adhere to semantic version numbers: major.minor.patch.

  • Major increments: These are non-backwards compatible API changes.
  • Minor increments: These are backwards compatible API changes.
  • Patch increments: These updates do not affect the API to the service.

When publishing a new Docker image for the service, two files need to be updated:

  • CHANGELOG.md - Notes should be added to capture the changes to the service.
  • docker/service_version.txt - The semantic version number should be updated.

CI/CD:

The CI/CD for HyBIG is contained in GitHub workflows in the .github/workflows directory:

  • run_tests.yml - A reusable workflow that builds the service and test Docker images, then runs the Python unit test suite in an instance of the test Docker container.
  • run_tests_on_pull_requests.yml - Triggered for all PRs against the main branch. It runs the workflow in run_tests.yml to ensure all tests pass for the new code.
  • publish_docker_image.yml - Triggered either manually or for commits to the main branch that contain changes to the docker/service_version.txt file.

The publish_docker_image.yml workflow will:

  • Run the full unit test suite, to prevent publication of broken code.
  • Extract the semantic version number from docker/service_version.txt.
  • Extract the released notes for the most recent version from CHANGELOG.md.
  • Create a GitHub release that will also tag the related git commit with the semantic version number.
  • Build and deploy a this service's docker image to ghcr.io.

Before triggering a release, ensure both the docker/service_version.txt and CHANGELOG.md files are updated. The CHANGELOG.md file requires a specific format for a new release, as it looks for the following string to define the newest release of the code (starting at the top of the file).

## vX.Y.Z - YYYY-MM-DD

pre-commit hooks:

This repository uses pre-commit to enable pre-commit checking the repository for some coding standard best practices. These include:

  • Removing trailing whitespaces.
  • Removing blank lines at the end of a file.
  • JSON files have valid formats.
  • ruff Python linting checks.
  • black Python code formatting checks.

To enable these checks locally:

# Install pre-commit Python package as part of test requirements:
pip install -r tests/pip_test_requirements.txt

# Install the git hook scripts:
pre-commit install

# (Optional) Run against all files:
pre-commit run --all-files

When you try to make a new commit locally, pre-commit will automatically run. If any of the hooks detect non-compliance (e.g., trailing whitespace), that hook will state it failed, and also try to fix the issue. You will need to review and git add the changes before you can make a commit.

It is planned to implement additional hooks, possibly including tools such as mypy.

pre-commit.ci is configured such that these same hooks will be automatically run for every pull request.

Releasing a new version of the service:

Once a new Docker image has been published with a new semantic version tag, that service version can be released to a Harmony environment by following the directions in the Harmony Managing Existing Services Guide.

Get in touch:

You can reach out to the maintainers of this repository via email: