Pipeline overview

See Why volcano cameras? for dataset context. This demo pins to one Te Kaha Te Kaha
Te Kaha (TKAH.01) — the single GeoNet volcano camera this demo ingests. West-facing site on the Bay of Plenty coast, Aotearoa New Zealand; captures a JPEG every 10 minutes with Whakaari/White Island on the horizon. camera (TKAH.01); the pipeline below is generic.

End-to-end flow in ap-southeast-2:

1. Source — GeoNet Aotearoa New Zealand Data GeoNet Aotearoa New Zealand Data
   GeoNet — Aotearoa New Zealand's geological hazard monitoring programme. Sensors, cameras, and open data (including s3://geonet-open-data via the AWS Open Data Registry). (AWS Open Data Registry, s3://geonet-open-data) publishes Te Kaha Te Kaha
   Te Kaha (TKAH.01) — the single GeoNet volcano camera this demo ingests. West-facing site on the Bay of Plenty coast, Aotearoa New Zealand; captures a JPEG every 10 minutes with Whakaari/White Island on the horizon. volcano camera JPEGs (no account required to read).
2. Ingest — volcano-ingest Lambda Lambda
   AWS Lambda — serverless compute. This demo uses volcano-ingest (copy + annotate) and volcano-api (query + presign). lists the last 7 UTC days of keys, copies each JPEG into a private bucket, derives simple tags from the key path and pixel brightness, and writes an S3 Annotation S3 Annotation
   Amazon S3 Annotations — attach structured JSON metadata to S3 objects via named namespaces (put_object_annotation / get_object_annotation). ( environment environment
   S3 Annotations namespace on each object in this demo (ANNOTATION_NAMESPACE). Holds the JSON metadata payload. namespace) on the same object.
3. Store — The private bucket is the source of truth: each JPEG carries its JSON metadata via S3 Annotations. An optional DynamoDB DynamoDB
   Amazon DynamoDB — optional denormalized mirror of annotation fields for faster reads at larger scale; S3 remains canonical. table mirrors annotation fields for faster reads at larger scale.
4. Query — volcano-api Lambda (behind API Gateway API Gateway
   Amazon API Gateway HTTP API — exposes GET /images with CORS; invokes the query Lambda. GET /images) reads annotations, applies filters (day_phase, utc_day, visibility), paginates with offset + limit, and returns presigned image URLs.
5. Gallery — Amplify Amplify
   AWS Amplify Hosting — serves the React gallery SPA that filters images using annotation-backed metadata. -hosted React SPA calls the API, shows a result count, paginates, and renders tag chips from the same annotation JSON you see in the S3 console.

Components

Component	AWS service	Role
GeoNet open bucket	S3 (public)	Source JPEGs — camera/volcano/images/.../TKAH/TKAH.01/...
EventBridge Scheduler	EventBridge EventBridge Amazon EventBridge Scheduler — optional one-time trigger to run the ingest Lambda at a chosen UTC time.	Optional one-time ingest trigger (enable_scheduler)
Ingest Lambda	Lambda (Python 3.14)	Copy images, put_object_annotation
Private bucket	S3	Images + annotations (canonical metadata)
DynamoDB	DynamoDB	Optional denormalized mirror (enable_dynamodb)
API Lambda	Lambda (Python 3.14)	get_object_annotation, filter, presign
HTTP API	API Gateway v2	GET /images with CORS
Gallery	Amplify Hosting	React UI — filters, count, pagination

Data source prefix: camera/volcano/images/{YYYY}/TKAH/TKAH.01/{YYYY.DDD}/*.jpg — TKAH TKAH
Volcano site code for Te Kaha in GeoNet object key paths (TKAH/TKAH.01). This demo filters ingest to this prefix only. site code for the Te Kaha Te Kaha
Te Kaha (TKAH.01) — the single GeoNet volcano camera this demo ingests. West-facing site on the Bay of Plenty coast, Aotearoa New Zealand; captures a JPEG every 10 minutes with Whakaari/White Island on the horizon. camera (TKAH.01).

Terraform outputs

After apply:

cd terraform
terraform output amplify_app_url
terraform output api_invoke_url
terraform output private_bucket_name

Or from the demo repository root:

terraform -chdir=terraform output amplify_app_url