Public app: https://climate.you

A registry-driven climate data platform with:

• a Python data packaging pipeline for climate datasets (derivation, tiling, and release artifact generation)
• shared Python modules for registries, data transforms, and reusable domain utilities
• a FastAPI backend serving packaged climate metrics and map-ready payloads
• a Next.js web application for interactive climate data exploration
• operational automation for validation, benchmarking, runtime management, and deployment tasks

Landing Page	Global Temperature
Local Temperature	Local Precipitations
Local Coral	Alternative Backgrounds

• climate/: shared package code for dataset derivation, registries, tiles, packager
• climate_api/: FastAPI app, schemas, endpoint services
• data/: local data artifacts (locations, masks, caches, releases)
• deploy/: deployment assets for runtime environments (service files, reverse-proxy config, env templates)
• docs/: architecture diagrams and runbooks
• infra/: cloud infrastructure definitions (Terraform for provisioning)
• registry/: authoritative manifests (datasets, metrics, maps, layers, panels)
• scripts/: operational build, validation, benchmark, runtime, and deployment scripts
• tests/: unit/integration/end-to-end test coverage
• web/: Next.js frontend application

Pipeline diagrams: docs/project_pipeline_diagrams.md

Using Conda (Anaconda or Miniconda) is the recommended path. Create and use a dedicated environment, then run all Python scripts/backend commands from that environment.

conda create -n <your-env-name> python=3.11
conda activate <your-env-name>

Install Python dependencies (Conda alone does not provide all required packages):

pip install -e ".[api,packager,validate-all,dev]"

Set repo-local import path when running scripts:

export PYTHONPATH="$(pwd)"

You can install packages manually without Conda, but that path is less reproducible and not recommended.

Node.js is required for the web application tooling/runtime, and npm is required for package management.

cd web
npm install

Optional credentials for ERA5 downloads (~/.cdsapirc):

url: https://cds.climate.copernicus.eu/api
key: <your-key>

1. Choose a data path:

• Preferred for quick start: use the pre-packaged demo release archive (includes locations, masks, and release assets).
• Build-from-scratch path: run the full preparation pipeline (see Data Preparation Overview below).

info: The remaining Quickstart steps assume you selected the pre-packaged demo release path.

Pre-packaged archive:

• download link: climate-demo-2026_03_11.tar.gz

Extract the archive and place the data folder at the root of the repository.

2. Start backend:

./scripts/api_backend.sh

3. Start frontend:

cd web
npm install # first time only
npm run dev

• API: http://localhost:8001
• Web: http://localhost:3000

If you want to create a release from scratch instead of relying on the pre-packaged demo release, follow the runbooks below.

Before running the API/UI against local data, prepare:

• location search artifacts (city + marine autocomplete index, city-only nearest set) and ocean naming masks
• optional reef-domain masks for DHW workflows
• optional sparse-risk mask (0.25°) for mixed-grid panel bbox optimization
• packaged release artifacts (metrics/maps) from dataset caches

Practical note: full global metric production across 40+ years can take days with CDS/ERDDAP sources, especially for daily-variable aggregation workflows (for example air/sea hot-day indicators and coral reef stress indicators).

Disk-space note: a single daily variable over ~40 years is often around 20-80 GB of local cache/output footprint. A realistic multi-metric build commonly needs about 150 GB of free space.

Detailed runbooks:

• docs/runbooks/locations-and-ocean-mask.md
• docs/runbooks/reef-mask.md
• docs/runbooks/dataset-cache-and-packaging.md

This repository also includes scripts and configuration to deploy the platform to a cloud-hosted Linux environment.

• Deployment scripts in scripts/deploy/ automate bootstrap, app rollout, auth toggling, and smoke checks.
• Runtime deployment assets in deploy/ provide service templates, proxy configuration, and environment-variable examples.
• Provider-neutral deployment runbook: docs/runbooks/deployment.md
• Infrastructure examples include GCP Terraform in infra/terraform/gcp/ (you can provision with other providers as long as VM requirements in the runbook are met).

For day-to-day development, keep this README as orientation and use runbooks for operational details.

Note: dev-only web query options (?debug=on, ?texture=mobile|full|auto) are documented in the backend+frontend runbook.