A simple ZXY tile server for GeoTrellis tilesets

Viewer is a tool for GeoTrellis administrators and developers to inspect ingested datasets in GeoTrellis. It is also a fairly simple GeoTrellis application and can hopefully serve as a starting point for developers building custom services with GeoTrellis.

In distributed computing, a general performance guideline is to keep the data where the processing is going to happen - if you have 100 nodes but only one of them has the numbers that you want to crunch, then a lot of the benefit of those extra nodes is wasted on getting the data to them for processing. So, a first step in deploying GeoTrellis is to ingest data - transforming it from a contiguous format like a collection of big geotiffs to something more amenable to distributed processing, often a network accessible store with fast access to small tiles. GeoTrellis can do this work for you, but once your raster is in a GeoTrellis-specific format you need a GeoTrellis-specific viewer to inspect it.

Before you can use Viewer you need to ingest some data with GeoTrellis, so please consult the GeoTrellis documentation about the ingest process if you need to. With data ingested, you can deploy Viewer to have a simple tile service, amenable to viewing with a browser-based library such as Leaflet or OpenLayers.

Viewer needs to be configured with a few details such as the Amazon S3 bucket where data tiles are stored. Please see src/main/resources/application.conf for details.

All dynamic endpoints are under the gt prefix to avoid collisions with static resources.

• gt/breaks/{layer} gives class breaks for the named layer. The required parameter numBreaks specifies a maximum number of classes to return.

• gt/tms/{layer}/{z}/{x}/{y} gives a PNG tile rendered from the layer. The optional parameter breaks gives a list of class breaks for rendering. The optional parameter colorRamp names a color ramp to colorize the classes. Names for color ramps are defined in ColorRampMap.scala. Every tile also has a few sub-resources that give views into the data behind the rendered tile.

  • grid gives a textual representation of the values of all the cells in the tile
  • type gives the name of the class representing the tile (useful to verify cell type)
  • breaks gives class breaks considering only the specified tile rather than the entire dataset
  • histo gives a histogram computed over the specified tile
  • stats gives a geotrellis Statistics report with mean, median, mode, min, max, standard deviation.

For reference, here is the command (slightly obfuscated) that I've been using to rebuild and upload to the EC2 VM that I have been testing on:

$ ./sbt assembly && rsync -e 'ssh -i ~/geotrellis.pem' target/scala-2.10/geotrellis-ingest-test-assembly-0.1.0.jar user@<machine>.amazonaws.com: -P

And to run on the VM:

$ java -cp /usr/lib/spark/lib/spark-assembly.jar:geotrellis-ingest-test-assembly-0.1.0.jar geotrellis.ingest.test.Server