Infrastructure Risk Visualisation Tool

This project provides interactive data visualisations of risk analysis results.

The tool presents the infrastructure systems and hazards considered in the analysis, then presents results as modelled for the whole system at a fine scale.

Other functionality:

• Summarise risk analysis at an administrative regional scale.
• Zoom in to see networks in detail.
• See an overview of hazard data.
• Inspect details of hazard layers.
• Query attributes of elements of the system.
• Range of potential economic impacts of failure, consisting of direct damages to infrastructure assets and indirect economic losses resulting from infrastructure service disruption (loss of power, loss of access).
• Explore a cost-benefit analysis (under uncertainty, with options to explore some parameters) of adaptation measures.

This README covers requirements and steps through how to prepare data for visualisation and how to run the tool.

Architecture

The tool runs as a set of containerised services:

• Traefik reverse proxy to direct requests to the other services
• Web server (nginx) ghcr.io/nismod/gri-web-server
• Vector tileserver (tileserver-gl) ghcr.io/nismod/gri-vector-tileserver
• Backend / API (bespoke Python app for vector data and raster tiles (+meta)) ghcr.io/nismod/gri-backend
• API Database (Postgres with PostGIS serves backend) (Dev only)
• Tiles Database (Postgres server with multiples terracotta metadata databases) (Dev only)

The services are orchestrated using docker compose.

N.B. The app was built with docker engine version 20.10.16 and compose version 2.5.0. It may not work with other versions.

Usage

Data preparation

The visualisation tool runs using prepared versions of analysis data and results:

• Rasters stored as Cloud-Optimised GeoTIFFs, with metadata ingested into a terracotta database, hosted within the backend API.
• Vector data stored in a PostgreSQL database, and preprocessed into Mapbox Vector Tiles

See ETL directory for details.

Data to be served from the vector and raster tileservers should be placed on the host within tileserver/<data_type>. These folders are made available to the running tileservers as docker bind mounts.

For example, in tileserver/raster/data/aqueduct there might live TIF files like these:

coastal_mangrove__rp_100__rcp_baseline__epoch_2010__conf_None.tif
coastal_mangrove__rp_25__rcp_baseline__epoch_2010__conf_None.tif
coastal_mangrove__rp_500__rcp_baseline__epoch_2010__conf_None.tif
coastal_nomangrove_minus_mangrove__rp_100__rcp_baseline__epoch_2010__conf_None.tif

And in tileserver/vector/, mbtiles files like these:

airport_runways.mbtiles
airport_terminals.mbtiles
buildings_commercial.mbtiles
buildings_industrial.mbtiles

Environment

Environment variables for the various services (and the ETL workflow) are stored in env files. Example files are given in envs/dev-example. These can be placed in envs/dev to get started.

Production env files should be placed in envs/prod.

Deploy

To deploy the stack we use the docker compose tool.

Development

The set of long-running services can include:

• traefik: Reverse proxy for other services, handles TLS
• web-server: Nginx server for the frontend code and static files
• db: PostgreSQL database holding vector data and raster metadata
• backend: API for available datasets and raster tileserver (terracotta)
• vector-tileserver: TileServer-GL for serving .mbtiles files
• redis: In-memory database for autopackage job queueing
• irv-autopkg-worker: Autopackage data processing (clipping, serialisation, etc.)
• irv-autopkg-api: Autopackage service coordination

If you're running your own FE development server, or connecting to a remotely hosted database, or not using the autopackage API, you may not need all these services.

To this end, we use profiles to define 'core' services which always run, and optional services. A bare docker compose -f docker-compose-dev.yaml up will run only the core services (those without a profiles attribute).

For example, when running your own FE development server to add a new raster layer the following should suffice: docker compose -f docker-compose-dev.yaml up. This will bring up db, tiles-db, backend and vector-tileserver.

To run the core services with a standard frontend: docker compose -f docker-compose-dev.yaml --profile web-server up.

To run the core services alongside the autopackage services: docker compose -f docker-compose-dev.yaml --profile autopkg up.

To run all of these behind traefik (every long-running service): docker compose -f docker-compose-dev.yaml --profile traefik --profile web-server --profile autopkg up.

There are also a few short-lived 'utility containers', which can be run to perform particular tasks:

• recreate-metadata-schema: Drop the contents of the db database, recreate with empty tables
• raster-tile-delete-entries: Delete raster entries of specified dataset in tiles-db
• raster-tile-drop-database: Drop whole database for specified dataset from tiles-db

When starting from a clean slate, the recreate-metadata-schema service must be run to create the tables in db that backend relies upon. If you find that the backend service is complaining that the raster_tile_sources database table is not available, you may need to create the appropriate tables in the db service first. To do that, bring the db service up as described above, and then run: docker-compose -f docker-compose-dev.yaml up recreate-metadata-schema to (re)create the tables. Note that this will drop any data currently in database.

Production

To run local builds of production containers we use the docker-compose-prod-build.yaml file. See [below](#Updating a service) for more details.

To deploy containers into a production environment: docker compose -f docker-compose-prod-deploy.yaml up -d

Updating a service

To update a service:

• We make the necessary changes to the container
• Build a new container
• Push it to the container repository
• Pull it on the production machine
• Deploy it

As an example, below we update the backend on a development machine:

# Edit docker-compose-prod-build.yaml image version:
#     image: ghcr.io/nismod/gri-backend:1.5.0

# Build
docker compose -f docker-compose-prod-build.yaml build backend

# Log in to the container registry
# see: https://docs.github.com/en/packages/working-with-a-github-packages-registry/working-with-the-container-registry

# Push
docker push ghcr.io/nismod/gri-backend:1.5.0

On the production remote, pull the image and restart the service:

# Pull image
docker pull ghcr.io/nismod/gri-backend:1.5.0

# Edit docker-compose-prod-deploy.yaml image version (or sync up):
#     image: ghcr.io/nismod/gri-backend:1.5.0

# Restart service
docker compose up -d backend

Adding new data layers

To add a raster data layer (for example, the iris set of tropical cyclone return period maps) see the ETL directory.

IRV AutoPackage Service

Provides API for extraction of data (and hosting of results) from various layers using pre-defined boundaries.

See irv-autopkg for more information.

Acknowledgements

This tool has been developed through several projects.

• v0.1 was developed by Oxford Infrastructure Analytics for the Government of Argentina with funding support from the World Bank Group and Global Facility for Disaster Reduction and Recovery (GFDRR).
• v0.2 was developed by Oxford Infrastructure Analytics for the Disaster Risk Financing and Insurance Program (DRFIP) of the World Bank with support from the Japan—World Bank Program for Mainstreaming DRM in Developing Countries, which is financed by the Government of Japan and managed by the Global Facility for Disaster Reduction and Recovery (GFDRR) through the Tokyo Disaster Risk Management Hub.
• v0.3 was developed by the Oxford Programme for Sustainable Infrastructure Systems (OPSIS) in the Environmental Change Institute, University of Oxford, for the Government of Jamaica (GoJ) as part of a project funded by UK Aid (FCDO). The initiative forms part of the Coalition for Climate Resilient Investment’s (CCRI) collaboration with the GoJ, which also includes analysis of nature-based approaches to build resilience in Jamaica to be procured and funded by the Green Climate Fund (GCF).
• release/caribbean was developed as part of the Jamaica project.
• release/east-africa was developed by researchers in the University of Southampton's Transportation Research Group and the Oxford Programme for Sustainable Infrastructure Systems, University of Oxford, supported by engagement with infrastructure and climate specialists and related government bodies, and funded by UKAID through the UK Foreign, Commonwealth & Development Office under the High Volume Transport Applied Research Programme, managed by DT Global.
• current work on global-scale data and analysis continues to be led by researchers in OPSIS. In part this is in collaboration with the Global Resilience Index Initiative, including the Oxford Spatial Finance Initiative and Global Earthquake Model Foundation. This work has been funded by the World Bank Group, Willis Towers Watson Insurance for Development Forum, the UK Natural Environment Research Council (NERC) through the UK Centre for Greening Finance and Investment, and UK Foreign, Commonwealth and Development Office (FCDO) through the Climate Compatible Growth (CCG) programme.