Happy Earth Day. As of today, TerraFuture's community science programs are ten years old. The first sensor we deployed was a single PurpleAir node, zip-tied to a street sign in East Portland in April 2016, operated by a team of four volunteers and funded by a foundation grant that lasted six months. I remember the morning we got it running because I spent most of it trying to keep the web dashboard from crashing every time someone loaded it.
What has grown from that starting point has been larger than anything we imagined at the outset. I want to use Earth Day 2026 to do something we do not often do: step back and look at the shape of a decade. The specific numbers. The patterns that held up. The things we were wrong about. What we are building next.
The Decade in Numbers
A few summary statistics from 2016 through the end of 2025:
43,200 community scientists participated in at least one TerraFuture data collection activity. That number includes one-time volunteers, long-term participants, and classroom programs.
9.3 million data points were contributed to our public monitoring networks across air quality, water quality, biodiversity, urban heat, and soil health programs.
312 environmental sensors are currently operating across our active networks. In 2016, that number was 1.
47 neighborhoods are now within the primary coverage area of at least three of our monitoring programs. In 2016, the answer was zero.
186 peer-reviewed papers have cited our public datasets, based on the most recent Google Scholar search we ran in March.
54 policy documents at the state, county, or municipal level reference TerraFuture data directly. This number is harder to verify fully — we track it by search — but it is the best estimate we have.
$11.2 million in community benefits has been generated through programs made possible by our data — including rebates, grants, and direct investments in neighborhoods documented through our programs as exposure-elevated. The underlying accounting methodology is published in our 2024 impact report.
These are large numbers. They are also, in my view, the least interesting part of the retrospective. What has been more informative is the patterns that have shown up underneath them.
Five Things That Held Up
Community science data is credible when the methodology is rigorous. When we started, many of the scientists we talked to were skeptical of community-collected environmental data. The reasonable concern was quality control. Ten years later, with an extensive paper trail of validation studies, instrument calibration records, and published comparisons against reference networks, that skepticism has largely dissolved. Our datasets are used by research universities, state agencies, and municipal planners, including in legal and regulatory contexts. The pattern was consistent: rigor earned trust, and trust expanded the uses the data could be put to.
Hyperlocal data changes policy conversations. Census tract and ZIP code-level environmental data, which was the norm in 2016, systematically obscures the neighborhoods with the most acute exposures. Our decade-long investment in neighborhood and sub-neighborhood resolution has changed how several municipal programs target investments. The same policy conversations held against block-level data go very differently than against ZIP-level data.
Indigenous and community knowledge are not supplementary to instrumental data. They are often earlier signals, and they capture patterns our instruments miss. Our research team's collaboration with the natural resources departments of the Confederated Tribes of Warm Springs, Grand Ronde, and Siletz Indians has repeatedly identified ecosystem shifts before they showed up in our sensor records. We began those collaborations expecting to share our data; we have ended every year having learned more than we contributed.
Long-term datasets compound in value. The single most cited piece of our work is our 2019–2025 urban heat time series, not because any individual year's data was revelatory, but because seven years of consistent measurement let researchers isolate signal from interannual noise. Three years of data would not have produced that. Ten years will be significantly more valuable than seven.
Community scientists stay when the work is real. Volunteer retention in community science is hard. Our retention numbers improved substantially when we stopped treating volunteer programming as outreach and started treating it as research. Volunteers who understood that their data was going into datasets that state agencies actually used kept showing up. Volunteers who were led to believe their contribution was symbolic did not. The honesty matters.
Three Things We Were Wrong About
We underestimated how much of the impact would come from youth programming. In our early years, youth programming was a small side project, funded with whatever foundation money happened to support it. Today, our youth fellows program produces research outputs and policy engagement that rivals the rest of our programming. Nearly a third of our current community scientists first engaged with us as youth participants. We should have invested here earlier.
We overestimated how quickly data would translate into policy. In our first five years, we built a dataset, published it, and waited. Not much happened. The missing piece was the relationship work — the slow, unglamorous, meeting-heavy work of being in the rooms where policy is drafted, with the data ready when it was useful. We now invest roughly as much in policy relationship-building as in data collection itself. If we were starting over, we would do this from year one.
We assumed a consistent federal partner. Our programmatic structure through 2020 implicitly assumed that federal climate policy and funding would provide a stable platform. It did not and does not. Organizations that depend on a single funding source or a single policy environment are fragile. We have spent the past five years diversifying both, and the work is genuinely stronger for it. We should have started sooner.
What We Are Building Next
Three commitments for the next decade, published here for public accountability.
Expand the sensor network to 500 nodes by 2028. This would bring nearly every neighborhood in our primary service area to direct measurement for air quality, and expand coverage to Gresham, Milwaukie, and Beaverton. Funding is secured for 380 nodes; the remaining 120 depend on continued community and foundation support.
Formalize our partnership with tribal natural resources departments. Our current collaborations have been project-based. We are working toward a multi-year framework that formalizes data sharing, ensures appropriate data sovereignty protections, and supports tribal priorities as well as shared ones. First memorandum is in drafting with the Warm Springs tribes; we expect to sign by the end of 2026.
Double youth fellows capacity. We will grow from 18 fellows this year to 36 by 2028, with a new cohort specifically focused on tribal youth in partnership with the tribal education departments we work with. Fellowship stipends will increase to match living wage standards for the fellows' ages; our current stipends are below where they should be.
A Personal Word
I am writing this on a Wednesday morning, which happens to be Earth Day. I spent yesterday afternoon watching a fellow teach thirty ninth-graders how to read a temperature map of their own neighborhood. A couple of the students asked her smart questions. One of them asked a genuinely better question than I would have come up with, which is a familiar feeling at this point in this work.
Ten years is a long time. It is also not very long. Most of the measurable trends in our data are still establishing themselves. Most of the policy wins are still provisional. Most of the neighborhoods most affected by our changing climate are still catching up to the exposures they already face.
But the shape of the work is durable now in a way it was not in 2016. Our data is trusted. Our youth programming produces real leaders. Our community science networks are embedded in places that did not know us a decade ago. The infrastructure is real.
Thank you to the 43,200 people who showed up at some point over ten years. Thank you to the funders who believed in the early, small version. Thank you to every volunteer who stood in a creek at 9 a.m. on a Saturday in March because it mattered to do so.
We are just getting started. Here is to the next ten.
