Skip to content

Getting Started with OCR

This guide helps you get started with accessing and using OCR's wind-adjusted fire risk data for the continental United States.

What is OCR?

Open Climate Risk (OCR) is CarbonPlan's platform for analyzing building-level wildfire risk across CONUS. OCR provides:

  • Building-level fire risk for ~160 million structures
  • Wind-adjusted fire spread modeling that accounts for directional fire propagation
  • Multiple output formats: Interactive web maps, downloadable datasets, and cloud-native data access
  • Present and future scenarios: Current conditions (circa 2011) and future projections (circa 2047)

Quick Access Options

Option 1: Explore the Web Tool

The fastest way to explore OCR data is through our interactive web map. The web tool allows you to:

  • Search for specific addresses or locations
  • View building-level risk scores on a 1-10 scale
  • Explore county, census tract, and census block aggregations

Option 2: Access Production Data

If you want to analyze OCR data programmatically, you can access our production datasets directly from cloud storage using Python.

Accessing Production Data

OCR's output data is stored in Icechunk, a versioned, cloud-native data format that works seamlessly with Xarray and Zarr.

Prerequisites

You'll need Python with a few packages installed:

pip install xarray icechunk

Load the Dataset

Here's a minimal example to load OCR's wind-adjusted fire risk data:

import icechunk
import xarray as xr

# Connect to OCR's production Icechunk repository
version = 'v0.12.0'  # Check GitHub releases for latest version
storage = icechunk.s3_storage(
    bucket='us-west-2.opendata.source.coop',
    prefix=f'carbonplan/carbonplan-ocr/output/fire-risk/tensor/production/{version}/ocr.icechunk',
    anonymous=True,
)

repo = icechunk.Repository.open(storage)
session = repo.readonly_session('main')

# Open the dataset
ds = xr.open_dataset(session.store, engine='zarr', chunks={})
ds

This gives you access to:

  • Raster datasets: 30m resolution risk surfaces
  • Risk scores (RPS): Risk to Potential Structures values
  • Spatial coverage: Full CONUS extent
  • Multiple variables: Burn probability, conditional risk, wind-adjusted metrics

Understanding the Data

The dataset contains several key variables:

  • rps: Risk to Potential Structures (expected net value change per year)
  • bp: Burn Probability (annual likelihood of burning)
  • crps: Conditional Risk to Potential Structures (damage if fire occurs)
  • Risk scores are for a "generic" or "potential" structure at each location

Important Limitation

Risk scores represent a hypothetical structure and do NOT account for building-specific factors like construction materials, retrofits, or defensible space management.

Next Steps

For Data Users

For Researchers & Analysts

For Developers

Support

Available Data Versions

Check our GitHub Releases page for:

  • Latest data version numbers
  • Release notes and changelogs
  • Known issues and fixes
  • Data format changes

Ready to dive deeper? Check out the Working with Data notebook for hands-on examples.