Grantee Research Project Results

Final Report: Recreator crowdsourcing of particle levels during wildfires

EPA Grant Number: SU840571
Title: Recreator crowdsourcing of particle levels during wildfires
Investigators: Clark, Kayla , Galley, Annalee , Helm, Colter , Ludwig, Logan
Institution: University of Wyoming
EPA Project Officer: Page, Angela
Phase: I
Project Period: August 1, 2023 through July 31, 2024
Project Amount: $25,000
RFA: 19th Annual P3 Awards: A National Student Design Competition Focusing on People, Prosperity and the Planet Request for Applications (RFA) (2022) RFA Text |  Recipients Lists
Research Category: P3 Awards , P3 Challenge Area - Air Quality

Objective:

The overarching goal of this project is to improve the availability and targeted distribution of real-time air quality information at outdoor recreation sites, particularly those vulnerable to rapid fluctuations in particulate matter (PM₂.₅) levels during wildfire events. In alignment with the EPA’s P3 call and its commitment to advancing public health, environmental protection, and community resilience, our project leverages a novel, cost-effective, and scalable approach to monitor air quality in remote and underserved regions.

At its core, the project addresses a longstanding technical challenge: the prohibitive costs and logistical constraints associated with deploying traditional air quality monitoring sensors in rural, high-recreation areas. Rather than relying solely on expensive hardware, our project integrates a low-tech visibility analysis with high-resolution image data collected by cellular devices. Outdoor recreators are provided with standardized signage and a dedicated “cell phone post” equipped with a holder at a single beta test backcountry recreation site in Albany County, Wyoming (Fig. 1, left). This installation ensures that images captured by participants are consistently framed with clearly identifiable geologic landmarks at known distances (e.g., ¼ mile, 1 mile, 3 miles, 10 miles). By correlating these visibility measurements with simultaneously recorded PM₂.₅ data from a co-located, precisely calibrated sensor, we will develop and refine a predictive model that reliably estimates local particulate concentrations.

The innovation of our approach lies in its bridging of high-tech and low-tech methods. Traditional methods based on algorithmic image analysis and sophisticated machine learning models have proven to be technically demanding and challenging to operationalize, particularly within constrained budgets and timelines. Instead, by harnessing the well-documented analytical relationship between atmospheric visibility and particulate matter concentrations (as described by Wang et al. 2019 and Liu et al. 2017), our project sidesteps the pitfalls associated with more complex computational models (Fig. 1, right). The simplicity of our method enhances the precision of our visibility assessments even as air quality conditions fluctuate.

A key aspect of the project is the development of a mobile application prototype designed to support image uploads and facilitate direct communication between outdoor recreators and our central data repository, hosted at the University of Wyoming. During Phase I, while the full back-end functionality of the application is deferred to Phase II, the design and front-end interface are critical for piloting the image crowdsourcing effort. In parallel, the project will pilot alternative methods—such as direct text messaging and hashtag-based submissions—to ensure robust engagement and troubleshoot potential data transmission challenges. The initial focus on front-end design enables us to clearly define the application’s functionality, appearance, and user experience, thereby providing comprehensive guidelines for future development.

Complementing the technical innovations is the integration of a “digital passport” within the mobile application. This feature not only tracks the outdoor recreation site visited by participants but also incentivizes engagement through community-sponsored awards and rewards. Local small businesses, which play a critical role in the region’s tourism and recreation economy, are key partners in this initiative. By linking user participation with tangible economic incentives, the project fosters community buy-in and enhances the sustainability of the monitoring network. This symbiotic relationship is particularly significant in the Rocky Mountain West, where economic diversification is imperative as traditional industries such as fossil fuel extraction give way to outdoor recreation and tourism.

To accomplish this goal, the project has the following six objectives:

•       Design a mobile application interface that supports image upload and transmission to a central server hosted at the University of Wyoming.

•       Integrate the mobile application with community partners to support distribution to outdoor recreators.

•       Select one beta test backcountry recreation site and install project signage and instructions for participating in the crowdsourcing of viewshed images. 10 beta sites were initially proposed, but one monitoring site was selected due to technological constraints and landowner agreements. 

•       Deploy PM₂.₅ monitoring sensors at the beta test site.

•       Model the relationship between distance-based “visibility” from crowdsourced images and recorded PM₂.₅ levels.

•       Communicate air quality predictions to mobile application users during wildfire events and/or cases of elevated PM₂.₅ to support informed recreation decision-making.

 

Fig 1. Example signage and "phone pocket" that will be used to standardize image collection at the selected study area recreation sites (left). Traditional approaches to approximating particulate matter from image analysis rely on local algorithmic computations (e.g., machine learning) that have stalled during the development phase. The proposed approach tracks a known analytical relationship between visibility distance and locally observed particulate matter (low tech) to predict and communicate PM_2.5 risk levels to a network of users when local air sensors are removed (right). 

Ultimately, this project is designed to address a critical gap in current air quality monitoring approaches. By providing precise, location-specific air quality information, we enable outdoor recreators to make informed decisions that minimize exposure to harmful pollutants. This not only protects individual health but also supports the economic viability of the outdoor recreation and tourism sectors in regions increasingly affected by wildfire smoke and related air quality challenges. In doing so, the project embodies the EPA’s mission to “protect human health and the environment” and reinforces the principles of the P3 framework—enhancing people’s lives, bolstering economic prosperity, and safeguarding the planet.

Summary/Accomplishments (Outputs/Outcomes):

Fig 2. Mobile prototype interface prompting users to phtograph a landscape. 

Building on Sections 1(c), 2, and 3 of the original proposal, our beta test at a single outdoor recreation site in Albany County confirmed both the feasibility and initial effectiveness of this low‐tech, userfriendly approach to crowdsourced particulate matter (PM₂.₅) monitoring. The following summarizes our progress for each of the six objectives, now with visual references to the accompanying figures:

1. Design a mobile application interface that supports image upload and transmission

A prototype mobile interface was developed for standardized submission of crowdsourced images (Fig. 2). This interface timestamps each photo and records device information (e.g., phone model), enabling robust control for resolution differences. Out of 81 total image submissions, 13 were invalid because the reference monitor was temporarily offline or the photograph’s timestamp could not be matched to PM₂.₅ data. These early lessons underscore the value of app‐driven automation in future development phases.

2.      Integrate the mobile application with community partners to support distribution This effort will be allocated in Phase II, if funded. The existing mobile application prototype reserves space for local business promotion, creating a relationship between local economies and recreation spaces. 

3.      Select one beta test backcountry recreation site and install signage

We placed our instructional signage and a fixed phone-holder at a popular trailhead (Fig. 3) located approximately 1,800 feet from the Laramie SLAMS PM₂.₅ monitor (Fig.4). This proximity allowed us to pair crowdsourced images with highly reliable pollutant readings. As illustrated in Figure 3, volunteers placed their phones in the black holder and photographed the landscape through the hole in the sign, ensuring consistent image framing and minimizing user error.

Fig 3. Instructional signage including fixed phone-holder (black) located at beta site.  

 

4.      Deploy PM₂.₅ monitoring sensors at the beta site

Instead of installing new hardware onsite, we tapped into the existing Laramie SLAMS monitor (a part of Wyoming’s State and Local Air Monitoring Stations network). This setup provided continuous PM₂.₅ data (24-hour rolling average, measured in μg/m³) from March 19, 2024, to May 22, 2024. Integrating a public monitor both reduced costs and validated our crowdsourced visibility data with a trusted baseline.

Fig 4. Location of Laramie SLAMS PM_2.5 monitor in geographic comparison to the signage placed at the beta site where photos were collected.

5.      Model the relationship between distancebased ‘visibility’ and recorded PM₂.₅ levels

An Ordinary Least Squares (OLS) regression includes time from sunrise, a quadratic term for daylight, phone model fixed effects, and opacity were used as the primary predictors of measured PM₂.₅. Each crowdsourced photo was matched to a series of baseline images (ranging from 0% to 50% digitally applied opacity) to derive a visibility-based measure. Preliminary results indicate a 12.8% increase in matched opacity corresponds to a 1% increase in PM₂.₅ (significant at the 10% level), and the model explains roughly 77% of the variation in PM₂.₅ readings.

6.      Communicate air quality predictions to mobile application users

As the visual matching mechanism was refined throughout Phase I of the project, several potential alert mechanisms were explored to various success. The most effective application involved prompting the user to upload their photo, then try matching the image against baseline images with opacity treatment replicating atmospheric conditions applied. The true air quality result would be immediately relayed to the user, alerting them of consequential levels of PM₂.₅ if applicable. This self-directed communication feature emphasizes transparency and understanding of how the data is collected and interpreted and would be further explored in Phase II. 

In sum, the pilot confirms that visually based citizen science can reliably capture localized PM₂.₅ trends without resorting to costly or ambiguous “black box” algorithms. The project’s success at the beta site highlights its broader potential to meet air quality data gaps in remote recreation areas across Wyoming and beyond.

Conclusions:

The results of this beta phase validate our core premise: simple, transparent, and communitydriven methods can effectively monitor air quality in recreational settings. By pairing a public PM₂.₅ sensor network (Laramie SLAMS) with standardized image-capture protocols, we established a robust link between visibility and particulate pollution. This approach balances methodological rigor with broad accessibility, an ideal outcome for environmental citizen science.

Moreover, the ease of integrating local stakeholders confirms that people and prosperity can jointly benefit when scientific tools are designed with user-friendliness and scalability in mind. Although our pilot occurred during a period of relatively stable air quality, it nonetheless demonstrated strong predictive relationships. This suggests that performance may only improve under more dramatic wildfire-driven variability in PM_2.5.

Looking ahead, our Phase II plans include:

•       Scaling to Additional Sites: Expanding our coverage to multiple recreation areas to capture a wider range of pollution levels and user demographics.

•       Refining the Application: Moving beyond a prototype to a fully functional app that automates data pairing, provides real-time air quality alerts, and accommodates diverse daylight conditions.

•       Leveraging Partnerships: Collaborating with land managers, small businesses, and potentially national park administrators to integrate these low‐cost monitoring techniques into existing information systems.

Ultimately, this work exemplifies the P3 program’s commitment to human health, sustainable economies, and environmental protection. By continuing to engage citizen scientists directly in data collection, we foster broader public awareness of air quality challenges and cultivate a practical, people-powered solution for recreation safety—particularly vital in regions prone to wildfires.

Phase II will expand the monitoring network to multiple high-use recreation sites, refine the app for automated data pairing and real-time alerts, and integrate the digital passport concept for broader community engagement. The app will be moved from prototype to development. We will partner with land managers and local businesses to enhance outreach, adopt educational programs that train citizen scientists, and measure health outcomes during wildfire events. Additionally, we will explore how simple, accessible design tools like our opacity filter can be applied in other settings where costly technology fails and clear, transparent processes matter. This ensures a scalable system that keeps local communities informed and empowered.

Supplemental Keywords:

Low-tech innovation, visibility analysis, citizen science