Camera-Trap Occupancy Survey

Survey Design

Camera-trap density 40%

Habitat suitability threshold 0.50

Per-occasion detection prob. p 0.30

Survey speed 1.0×

Live Estimates

Occasions run (K)

Camera sites (S)

True occupancy Ψ (ground truth, camera sites)

—

Naive occupancy (≥1 detection / S)

—

Detection-corrected Ψ̂ estimate

—

Legend

Suitable habitat (high quality)

Unsuitable habitat

Deployed camera trap

Camera flash = detection this occasion

Wandering animal (true presence)

How it works

This is the classic single-season occupancy model used to design real camera-trap, acoustic and eDNA monitoring networks (MacKenzie et al. 2002). A species truly occupies a fraction Ψ of sites, but a camera only records a detection at an occupied site with probability p on any given visit ("occasion") — so raw counts always under-estimate the truth:

Site occupied (fixed for the season): z_i ~ Bernoulli(Ψ)
Detection on occasion k, given occupied: y_ik ~ Bernoulli(p)
P(species missed at an occupied site
   across all K occasions) = (1 − p)^K

The naive occupancy estimate — the fraction of camera sites with at least one detection — is biased low, because some truly-occupied sites are missed on every single visit. Correcting for that non-detection probability gives the standard closed-form estimator:

Ψ̂_naive = (sites with ≥1 detection) / (camera sites)
p̂ = mean per-occasion detections at sites with ≥1 hit
Ψ̂ = Ψ̂_naive / [1 − (1 − p̂)^K]

Habitat threshold — raises or lowers how much of the landscape actually qualifies as suitable, i.e. the true Ψ ecologists are trying to recover.
Camera density — how many grid cells get a sensor; more cameras means a larger, more representative sample (S).
Detection probability p — sensor sensitivity / animal activity level. Low p means you need many occasions before naive counts catch up to the truth.
Watch the gap between the naive count and true Ψ shrink toward the corrected Ψ̂ as K grows — this is exactly why field ecologists insist on repeat visits rather than a single pass.

Real-world relevance: this is the statistical backbone of national biodiversity monitoring programs (e.g. Snapshot USA, TEAM Network camera-trap grids) and underlies decisions about reserve design and population trend estimates where the species itself is almost never seen directly.

Next occasion in --s
Last occasion detections: 0

⚠ Couldn't load the 3D engineThree.js failed to load from the CDN. Check your connection and reload.