Leveraging digital mobility data to estimate visitation in National Wildlife Refuges

Area of Science:

• Ecology
• Conservation Science
• Data Science

Background:

• Accurate visitation estimates are crucial for managing National Wildlife Refuges and National Fish Hatcheries.
• Current methods face challenges in providing reliable system-wide visitation data.
• Understanding recreation demand and ecosystem protection relies on accurate visitor counts.

Purpose of the Study:

• To evaluate methods for overcoming constraints in estimating visitation levels in US Fish and Wildlife Service managed areas.
• To develop and test a visitation modeling approach using statistical models and digital mobility data.
• To assess the reliability of various digital mobility data sources for estimating refuge visitation.

Main Methods:

• Utilized multiple linear regression to develop a visitation modeling approach.
• Incorporated predictors from eight mobility data sources: social media, community science, and mobile device location data.
• Tested the performance of integrated multi-dataset models against models using individual datasets.

Main Results:

• Observed visitor numbers correlate with mobility data volume, but individual data sources (social media, mobile location) are unreliable proxies.
• Integrated mobility datasets in models accounted for variability and outperformed single-source models.
• Refuge-level effects were the most important predictor, highlighting the need for site-specific characteristics.

Conclusions:

• Statistical models incorporating multiple digital mobility datasets can improve visitor estimate accuracy.
• These models offer potential for standardizing data collection and simplifying estimation for agency staff.
• Future models incorporating site characteristics will enhance generalizability for refuge visitation estimation.