Volunteer Heat Sensors Reveal Hidden Hot Blocks
A summer grid of low-cost sensors shows temperature gaps that official stations miss.
A volunteer-deployed heat sensor grid found evening temperature differences of several degrees between tree-lined ridges and asphalt-heavy commercial strips, according to preliminary maps from the climate resilience office.
The largest gap recorded so far came on the evening of August 4 last year, when a sensor on a porch rail in the Larkspur Ridge blocks read 24.1 degrees Celsius at 9 p.m. while a unit mounted under a shop eave on the Fourth Basin approach read 30.8. Six-point-seven degrees, four kilometers apart, at the same minute. The nearest official weather station, at the regional airfield eleven kilometers inland, reported 25.9 and is the number that appeared in most forecasts that night.
Sensors hang on porches and shop eaves under a common protocol. Data is open nightly. Quality flags mark units that drifted or lost shade reference. The grid ran 214 units last summer across 38 census blocks, of which 187 produced records complete enough to use. The protocol requires each unit to be mounted between two and three meters above ground, at least one meter from a wall, under a radiation shield, with a photograph of the installation filed at deployment and again at the end of the season.
Public works staff say the maps will influence tree-planting priority lists more than any single elegant master plan.
“We plant where the dots are red,” said forester Maya Ruiz. Her department’s planting budget is $412,000 a year, enough for roughly 700 street trees including three years of establishment watering, against a backlog she puts at 9,000 vacant planting sites citywide. Before the grid, the priority list was built from canopy-cover percentages derived from aerial imagery, which she says systematically flattered blocks with a few very large trees and penalized blocks with many small ones. “Canopy percentage tells you what the sky sees,” Ruiz said. “It does not tell you what a person waiting at a bus stop feels at seven in the evening.”
## The instrumentation gap nobody planned
Equity advocates note that rental buildings were harder to instrument because permission chains are longer. A second wave targets multifamily with landlord agreements. The numbers make the problem legible: 71 percent of the deployed sensors ended up on owner-occupied properties, in a city where 44 percent of households rent. In three blocks along the Cordage corridor, all of them majority-renter, the grid managed a single unit.
Tomasz Idrizi, who organizes with a tenants’ coalition in the basin neighborhoods, joined the project’s advisory group after the first season and does not soften his assessment.
“The map has a hole exactly where the heat is,” he said. “That is not a small methodological footnote. If you take this map to a council budget hearing, you are going to plant trees on the streets of people who already had trees, and you will do it with a chart that looks scientific.” He supports the project, he said, which is why he is unwilling to let the second season repeat the first.
The response is procedural rather than technical. The team drafted a one-page landlord consent form reviewed by a housing attorney, negotiated blanket agreements with two property managers covering 41 buildings, and shifted the mounting standard to allow fire-escape and balcony-rail installations that the original protocol excluded on siting-quality grounds. That last change is a real compromise: balcony-mounted units sit closer to a building’s own thermal mass and will read warm. The team plans to publish those units under a separate flag rather than blend them.
Field technicians emphasized that sensors fail in ordinary ways: bird strikes, battery sag, firmware quirks. Twenty-seven units were lost or destroyed last season. One was taken by a gull on the Second Basin. Four went down in the same week in July when a firmware update introduced a timestamp bug that shifted readings by an hour, which the team caught only because a volunteer in the Steepbank blocks noticed her porch’s afternoon peak had moved and emailed to ask why.
Students on the project are writing methods sections that double as training manuals for the next cohort. Continuity is a research output when grants turn over every two years. The current grant, $148,000 from a regional foundation, runs through March 2027.
Public communication remains a deliberate practice. Figures are released with uncertainty bands, and press officers have stopped promising single-number headlines the data cannot support. The office’s own summary of last season leads with a range - 3.1 to 6.7 degrees of evening differential depending on block pair and date - rather than the single dramatic number, a choice that cost it at least two television segments.
Replication plans are already in motion with a partner lab inland, which will run a 60-unit grid on the same protocol in a river city with a different building stock. Independent confirmation is the price of any municipal interest in the work, researchers said, and the resilience office has been explicit that it will not fund a capital program off a single unreplicated season.
The project ends at first frost, with a public report due midwinter. Ruiz has already committed the coming autumn’s planting - 180 trees, mostly on four blocks the grid flagged - and the crews will be in the ground by November.
She keeps a printout of the August 4 map folded in her truck. On it, someone has circled a bus shelter at the Fourth Basin approach and written a note in the margin: no tree pit here, sidewalk is four feet. It is the kind of thing a map cannot show and a person standing on the corner cannot miss.
Reporting for this story was prepared for The Harbor Ledger’s science desk. Tips:newsroom@theharborledger.com