GeoTrellis Chattanooga model demo

This is a demo of GeoTrellis functionality. The demo consists of two parts: the tile ingest process and demo server to query ingested data.

Dependencies

• Vagrant 1.9.5
• VirtualBox 5.1+
• AWS CLI 1.11+
• AWS Account (to access S3)

Getting Started

To provision a VM and fetch our pre-ingested demo data:

$ ./scripts/setup
$ vagrant ssh

This will download data into ./service/geotrellis/data/chatta-demo. See the ingest sections for information about ingesting data manually using either the local filesystem or geodocker.

Scripts

Helper and development scripts are located in the ./scripts directory at the root of this project. These scripts are designed to encapsulate and perform commonly used actions such as starting a development server, accessing a development console, or running tests.

Script Name	Purpose
update	Pulls/builds necessary containers
setup	Provisions the VM, fetch ingest data.
server	Starts a development server that listens at http://localhost:8777. Use the --geodocker flag to run the server with an accumulo backend.
console	Gives access to a running container via docker-compose run. Use the --geodocker flag to run with an accumulo backend
test	Runs tests for project
cibuild	Invoked by CI server and makes use of test.
cipublish	Build JAR and publish container images to container image repositories.

Testing

Run all the tests:

$ ./scripts/test

Details

The demo covers Chattanooga with different Byte tiles. (In fact each tile is essentially of type Bit because they only contain the values {0, 1}). Each tile is ingests into it's own layer, and the resulting map consists of layers which consist of combinations of differently-weighted source layers (a weighted overlay).

API routes:

• gt/colors - Color Ramps
• gt/breaks - Color Breaks
• gt/tms/{zoom}/{x}/{y} - Weighted Overlay
• gt/sum - Zonal Summary

Color Ramps

List of available color ramps to color weighted overlay:

• blue-to-orange
• green-to-orange
• blue-to-red
• green-to-red-orange
• light-to-dark-sunset
• light-to-dark-green
• yellow-to-red-heatmap
• blue-to-yellow-to-red-heatmap
• dark-red-to-yellow-heatmap
• purple-to-dark-purple-to-white-heatmap
• bold-land-use-qualitative
• muted-terrain-qualitative

Color Breaks

Get Parameters: layers, weights, numBreaks.

Calculates breaks for combined layers by weights with specified breaks amount.

Weighted Overlay:

Get Parameters: layers, weights, breaks, bbox, colors: [default: 4], colorRamp: [default: "blue-to-red"], mask.

It is a TMS layer service that gets {zoom}/{x}/{y}, passed a series of layer names and weights, and returns PNG TMS tiles of the weighted overlay. It also takes the breaks that were computed using the gt/breaks service. If the mask option is set to a polygon, {zoom}/{x}/{y} tiles masked by that polygon would be returned.

Zonal Summary:

Get Parameters: polygon, layers, weights.

This service takes layers, weights and a polygon. It will compute a weighted summary of the area under the polygon.

Ingesting Data

Quick clarification:

• Ingest requires Spark usage.
• Server works without Spark (uses GeoTrellis Collections API).

This section includes instructions on how to do local filesystem and Geodocker ingests to either the local filesystem, or to Accumulo (using Geodocker).

Local ingest

In the event that you need to run a local ingest, the gt-chatta-ingest container will run a spark-submit job that writes ingest data to the local filesystem. Make sure the Chatta Demo JAR has been built, then run the container:

$ docker-compose run --rm gt-chatta assembly
$ docker-compose build gt-chatta-ingest
$ docker-compose run --rm gt-chatta-ingest

Data will be installed into ./service/geotrellis/data/chatta-demo, which is mounted at /data/chatta-demo inside of the gt-chatta container.

Geodocker Ingests

To simulate running this demo in a distibuted environment, we prepared a Geodocker cluster including Hadoop, Accumulo and Spark. The application is configured for the geodocker setup using the application.conf file in the geodocker/ folder. Make sure you build an accumulo-configured JAR with make build-geodocker before attempting a Geodocker ingest.

• To run cluster:

    docker-compose up

  To check that cluster is operating normally check the availability of these pages:

  • Hadoop http://localhost:50070/
  • Accumulo http://localhost:50095/
  • Spark http://localhost:8080/

To check containers status, use the following command:

docker-compose -f docker-compose.geodocker.yml ps

• Install and run this demo using GeoDocker cluster

  • Running a Geodocker ingest will require more memory for the Vagrant VM. Before running an ingest, set GT_CHATTA_VM_MEMORY=6144 (or a higher value) and run vagrant reload.

  • Build the accumulo-configured geotrellis JAR, run an ingest, then start the server.

      make build-geodocker
      make ingest-geodocker
      ./scripts/server --geodocker

The demo catalog will be available through the accumulo backend. More information avaible is available in the GeoDocker cluster repo.