METHODS

How it works

We combine machine learning with wildlife biology to predict habitat quality across the landscape. Here's what goes into it.

The Process

Aggregate Data

We align vegetation type, cover, height, elevation, slope, aspect, and road proximity from LANDFIRE and NLCD — seven channels at 30-meter resolution.

Train Models

A ResUNet-34 CNN learns habitat patterns for each species, trained per USGS region against GAP habitat maps.

Generate Predictions

Models score every 30m cell across CONUS for habitat suitability — over 8 billion cells per species.

Rank by Percentile

Raw scores become percentiles so you can instantly see what's in the top 1%, 5%, or 10% — calibrated both within your state and nationally.

Data Sources

Our models are built entirely on publicly available federal geospatial data.

• LANDFIRE 2024 Existing Vegetation Type, Cover, and Height (EVT/EVC/EVH)

• LANDFIRE 2020 Elevation DEM (30m)

• NLCD 2024 Impervious Surface Descriptor (road proximity)

• USGS GAP Species Habitat Maps (training labels)

• PADUS public land boundaries (map overlay)

Understanding Percentiles

Habitat quality is shown as percentile ranks within your selected state. If an area is in the top 5%, it means 95% of habitat in that state scores lower.

Elite (Top 1%)

Exceptional habitat - rare and valuable

Prime (Top 5%)

Excellent habitat - high confidence

Strong (Top 10%)

Quality habitat - worth investigating

Who it's for

Serious Hunters

Find the best habitat faster. Spend less time scouting unproductive ground.

Land Managers

Understand habitat quality across your property. Identify improvement opportunities.

Conservation Groups

Target acquisition and restoration efforts where they'll have the most impact.

What we can't do

We believe in transparency about our limitations.

• Models predict habitat, not animal locations. Good habitat doesn't guarantee animals.

• Predictions are based on landscape features. They don't know about recent changes or hunting pressure.

• Our data has resolution limits. Features smaller than 30 meters may not be captured.

• Public data has gaps. Some regions have better coverage than others.

Ready to scout?

Launch Scout

Go Pro

Home Scout Pro Methods Contact

Privacy

Terms