Geo Domain

The foundational domain for geospatial data. Provides raster and vector operations, terrain analysis, spatial indexing, and imagery processing.

Decided in ADR-0006

Schema Extension

The geo: block on a layer declares geospatial metadata.

Field	Type	Required	Description
bbox	[number x 4]	No	Bounding box in EPSG:4326 as [west, south, east, north]
crs	string	No	Coordinate reference system. Default: EPSG:4326
query_method	string	No	How point queries are resolved: spatial_lookup, point_sample, or nearest

layers:
  dem:
    uri: data/wasatch-dem-10m.tif
    type: raster
    geo:
      bbox: [-111.9, 40.5, -111.6, 40.8]
      crs: EPSG:4326

Operation Categories

The geo domain organizes 47 operations into 9 categories:

Category	Count	Description
terrain	7	Slope, aspect, hillshade, curvature, TPI, TRI, viewshed
raster	9	NDVI, reclassify, clip, mosaic, reproject, normalize, calc, mask, rasterize
vector	11	Buffer, dissolve, intersection, union, simplify, clip, centroid, harmonize, and more
analysis	8	Zonal stats, weighted overlay, point sample, change detection, histogram, H3 index, distance, summary
imagery	3	Pansharpening, radiometric indices, optical calibration
classification	2	Train, predict
hydrology	2	Catchment network, tiered catchment network
composite	1	Treeline
utility	4	Raster-to-PMTiles, PMTiles merge, Terrarium encode/decode

Key Operations

Operation	Type	Description
terrain_slope	raster → raster	Slope angle from DEM (Horn's method). Output: float32 degrees.
terrain_aspect	raster → raster	Aspect direction (0-360 degrees clockwise from north).
terrain_hillshade	raster → raster	Hillshade illumination from configurable light source.
raster_ndvi	raster → raster	Normalized Difference Vegetation Index from NIR and Red bands.
raster_reclassify	raster → raster	Remap values to classes (manual breaks, equal interval, quantile, value map).
raster_clip	raster + vector → raster	Clip raster to polygon boundary.
raster_calc	raster → raster	Band math expressions using numpy functions.
analysis_zonal_stats	raster + vector → table	Statistics (mean, min, max, etc.) per polygon zone.
analysis_weighted_overlay	raster[] → raster	Multi-criteria weighted combination with normalization.
analysis_change_detection	raster + raster → raster	Pixel-wise difference with optional classification.
analysis_h3_index	raster → table	Index data onto H3 hexagonal grid at configurable resolution.
vector_buffer	vector → vector	Buffer features by distance in meters.
vector_dissolve	vector → vector	Merge features, optionally grouped by attribute.
analysis_point_sample	raster + vector → table	Sample raster values at point locations.

Compute

Geo operations run on GDAL, rasterio, and numpy in the local and cloud compute tiers. The platform routes automatically based on data size and location - local files run locally, cloud-hosted data runs on K8s workers.

Google Earth Engine is available as an external compute option for cloud-scale processing via backend: gee.

Tier	Libraries	Use Case
Local	GDAL, rasterio, numpy	Local files, small to medium rasters
Cloud	GDAL, rasterio, numpy (on K8s workers)	Remote data, large-scale batch processing
External (GEE)	Google Earth Engine	Explicit backend: gee, planetary-scale analysis

Decided in ADR-0007 - engines, connectors, and drivers are three orthogonal concerns.

Example: Terrain Analysis Workspace

layers:
  source/dem:
    uri: data/dem.tif
    type: raster
    geo:
      bbox: [-111.9, 40.5, -111.6, 40.8]
      crs: EPSG:4326

  terrain/slope:
    type: raster
    compute:
      op: terrain_slope
      inputs:
        dem: { layer: source/dem }
      params:
        smooth: true
        smooth_sigma: 1.5
    style:
      palette: viridis
      rescale: "0,60"
      info:
        fields: [slope_degrees]
        format: "{value}°"

  terrain/hillshade:
    type: raster
    compute:
      op: terrain_hillshade
      inputs:
        dem: { layer: source/dem }
      params:
        azimuth: 315
        altitude: 45
    style:
      palette: grayscale
      opacity: 0.7