Reproducible
geospatial infrastructure

From satellite to auditable insight.

What you can build

Real workflows, fully reproducible. Every result traces back to source.

Renewable Energy Siting

"Where in Nevada can we site a 200MW solar farm?"

→ 14 viable sites

14 contiguous parcels over 1,000 acres with high suitability scores.

IRA tailwinds weighted_overlay site_selection

dem: stac.usgs/3dep-10m
land_cover: stac.mrlc/nlcd-2021
solar: stac.nrel/nsrdb-ghi
grid: osm/transmission-lines
bbox: [-120, 38, -117, 42]

slope: dem | slope
aspect: dem | aspect
exclude: land_cover in [water, urban, forest]
buffer: grid | distance_to_features

weighted_overlay:
  solar_resource: 0.4
  slope_score: 0.3
  grid_proximity: 0.3
rank: by suitability desc

layout: split-map
layers: [suitability, exclusions]
chart: site-ranking-table
export: cog, geojson, pdf

Environmental Compliance

"Has the permitted area been disturbed since baseline?"

→ 3 clearings

12.4 ha unauthorized disturbance detected. Audit-ready provenance chain.

CSRD / TNFD change_detection provenance

baseline: stac.sentinel/s2-2023-Q1
current: stac.sentinel/s2-2025-Q4
permit: upload/permit-boundary.geojson
nlcd: stac.mrlc/nlcd-2021

classify: [baseline, current] | ndvi
clip: permit | buffer(500m)
detect: change_detection(baseline, current)

flag: forest → non-forest within permit
zonal_stats:
  by: permit_zone
  sum: disturbed_hectares

layout: before-after
layers: [baseline, changes, permit]
report: compliance-summary
provenance: full-chain
export: gpkg, pdf

Climate Risk Analytics

"What's the multi-hazard risk score for this property portfolio?"

→ 340 scored

Flood, wildfire, and heat risk. Property-level scores with scenario sensitivity.

insurance weighted_overlay multi-hazard

flood: stac.deltares/gfd-v4
fire: stac.usfs/whp
heat: stac.landsat/c2l2-st
properties: upload/portfolio.geojson
dem: stac.usgs/3dep-10m

normalize: [flood, fire, heat] | 0..1
slope: dem | slope
buffer: properties | buffer(1km)
clip: [flood, fire, heat] | clip(buffer)

weighted_overlay:
  flood: 0.4
  wildfire: 0.35
  heat: 0.25
aggregate: by property, by region

layout: dashboard
layers: [risk_scores, hazard_zones]
chart: portfolio-heatmap
export: geoparquet, pdf

Terrain & Viewshed Analysis

"Can these three tower locations see the whole valley?"

→ 94% coverage

Tower 2 relocated 300m east for full valley line-of-sight.

viewshed slope optimize

dem: stac.usgs/3dep-10m
towers: upload/tower-locations.geojson
valley: upload/study-area.geojson
observer_height: 30m

slope: dem | slope
aspect: dem | aspect
viewshed: towers | viewshed(dem, 30m)
merge: union(viewshed_1, viewshed_2, viewshed_3)

coverage: merge / valley_area
gaps: valley - merge
optimize:
  target: max_coverage
  constraint: within 500m of original

layout: split-map
layers: [coverage, gaps, towers]
chart: coverage-summary
export: cog, geojson, pdf

Who it's for

From compliance to exploration

Whether you're meeting a regulatory deadline or asking a new question about the planet.

Sustainability & ESG teams

TNFD and CSRD compliance with auditable geospatial evidence. Reproducible analysis your auditor can verify.

Renewable energy developers

Multi-constraint site selection in days, not months. Weighted suitability scoring with full provenance.

GIS analysts & developers

The composable alternative to scripts and GUIs. Declarative specs, standard formats, inspect every step.

Research teams

Reproducible spatial analysis. Process satellite imagery, publish results, share pipelines as code.

Architecture

The composable stack

Four layers, standard formats. Use the full stack or eject at any layer.

Source

Federated data catalogs

STAC COG Zarr Parquet

Transform

Declarative pipelines

slope reproject tile

Analyze

Compute on demand

zonal viewshed query

Deploy

Render anywhere

MapView Chart Card

Capabilities

Modern geo stack

Cloud-native formats. Federated catalogs. Run anywhere.

◇

Federated catalog

Index data where it lives. STAC pointers to Planetary Computer, AWS, your own buckets. Never duplicate.

⟳

Run anywhere

Same pipeline runs locally or as parallelized batch compute. Develop on your laptop, scale to the cloud.

Cost-aware

Estimate before you execute. Transparent pricing, compare providers, no surprise bills.

≡

Cache by content

Same inputs + steps = cached output. Content-addressed storage. Never recompute what already exists.

☁

Cloud-native output

PMTiles, COG, GeoParquet, Zarr. Optimized for streaming, not downloading. HTTP range requests everywhere.

⬡

MCP-native

Expose everything via Model Context Protocol. Any AI client can query, estimate, and execute pipelines.

Reproduciblegeospatial infrastructure