About Tokyo Temp

Why We Do This

Tokyo Temp started with a bicycle and an infrared thermometer. In the summer of 2019, Kenji Mori — then a graduate student in materials science at the University of Tokyo — cycled a 200-kilometer loop through all 23 wards, pointing an IR gun at every surface he passed. Asphalt at 61°C. Concrete at 49°C. Grass at 31°C. The water in the Imperial Palace moat at 26°C. The numbers told a story that the weather forecast didn't: the city itself was a thermal system, storing and releasing energy on a schedule all its own.

That ride produced 1,200 surface temperature readings and a master's thesis that few people read. But the data wouldn't leave him alone. The 30-degree difference between asphalt and grass over a distance of 10 meters. The way Shinjuku at midnight was still 4 degrees warmer than Fuchu at noon on the same day. The September afternoons that felt like July, the March mornings that felt like January. These weren't weather quirks. They were physics. And they were affecting millions of people who didn't know why.

Tokyo Temp exists to tell that story. We track thermal lag because it's the most tangible way to show how the built environment shapes the lived experience of weather. We publish our data because we think urban thermal science should be public infrastructure, not an academic specialty. And we write about it the way we do — methodical, specific, occasionally obsessive about material properties — because we believe the best science writing respects its readers' intelligence.

Kenji Mori — Founder & Lead Researcher

Kenji grew up in Sapporo, where winter means something different than it does in Tokyo. He came to the capital for university and spent his first September wondering why everyone was still using air conditioning when autumn had supposedly arrived. The question stuck with him through four years of materials science coursework and eventually became his MSc thesis: "Thermal Storage in Urban Concrete Structures and Its Effect on Seasonal Temperature Cycles in the Tokyo Metropolitan Area."

The thesis involved building a 1:100 scale thermal model of a Tokyo city block using concrete samples with embedded thermocouples, then exposing it to simulated solar input in a climate chamber. The model predicted a 32-day thermal lag for dense urban configurations — a finding that matched JMA station data well enough to earn a publication in the Journal of Applied Meteorology and Climatology. But Kenji found academic publishing too slow for a problem that was affecting people in real time. He wanted to put the data in front of residents, planners, and policymakers while it was still warm. Literally.

After graduating in 2020, he spent a year as a research assistant at the National Institute for Environmental Studies in Tsukuba, studying urban heat island mitigation strategies. In 2021, he registered Tokyo Temp Research LLC in New Jersey (his mother is American, which made the paperwork straightforward) and launched this site. He still does most of the fieldwork — the quarterly bicycle surveys, the sensor maintenance, the IR camera transects — and writes the majority of the content.

When he's not thinking about thermal mass, Kenji rides long-distance cycling events (he's completed the 400km Brevet Randonneur three times) and grows tomatoes on his apartment balcony in Nakano, where he tracks the microclimate effects of his building's concrete walls on fruit ripening times. The tomatoes ripen 4–5 days later than identical plants at his mother's house in Tama. He finds this satisfying rather than frustrating.

Yuki Tanabe — Co-Founder & Meteorology Lead

Yuki met Kenji at a JMA open house in 2020, where they were both examining the AMeDAS station exhibits with slightly too much enthusiasm. At the time, Yuki was working as a weather forecaster for a private meteorological company, translating JMA model output into forecasts for construction and logistics clients. She'd noticed that her temperature forecasts for central Tokyo were consistently off by 1–2°C in the autumn transition period, and she'd started to suspect that the urban heat island models she was using didn't account for thermal lag properly.

She was right. The standard UHI model — which adds a constant temperature offset based on population density — doesn't capture the temporal dynamics of heat storage and release. Yuki's breakthrough insight was that thermal lag varies seasonally, which means the same urban configuration produces different temperature anomalies at different times of year. A simple offset can't represent that. You need a full thermal model with heat capacity terms.

Yuki brought meteorological rigor to Tokyo Temp. She built the cross-correlation analysis pipeline, established the JMA data partnerships, and designed the sensor network layout. Her background in synoptic-scale meteorology — she holds a BSc from Hokkaido University — means she can distinguish UHI effects from large-scale weather patterns, which is essential for isolating the thermal lag signal from the noise of passing fronts and pressure systems.

Before Tokyo Temp, Yuki tracked weather for 50 Tokyo building rooftops as part of a thermal mass study commissioned by a major real estate developer. The developer wanted to know how rooftop gardens affected cooling loads. Yuki wanted to know how the entire building interacted with the atmosphere. She collected three years of data, produced a report that the developer filed away, and then joined Kenji in building something that would actually get read.

Yuki lives in Meguro with her cat, who has learned to sleep on the warmest surface available regardless of the season. She says this makes him a natural thermal sensor, though his calibration is questionable.

Rei Nakamura — Contributor & Data Engineer

Rei built the thermal-lag spiral SVG. That alone would justify their place on this team, but they've done considerably more. Rei is a data scientist by training — BSc in Statistics from Keio University, followed by three years at a fintech company building real-time data visualization dashboards. They joined Tokyo Temp in 2022 after seeing Kenji present at a Tokyo data meetup and thinking, "Your visualizations are terrible. Let me help."

The spiral diagram on our homepage is Rei's masterwork. It encodes 12 months of lag data in a single radial plot, with color gradients that communicate temperature associations without requiring a legend. The SVG is hand-coded — no D3, no Chart.js, no visualization library — because Rei believes that understanding your data means understanding how it's drawn. Every path element, every coordinate calculation, every color stop is deliberate.

Rei also built our data pipeline infrastructure: the scripts that pull JMA AMeDAS data, the quality-control checks that flag anomalous sensor readings, the cross-correlation computation that runs on a $35 Raspberry Pi in Kenji's closet. They wrote the thermal-mass bar chart renderer, the weather API integration, and the cookie banner. Their code style matches the site: straightforward, commented, slightly messy in the way that working code always is.

When not coding for Tokyo Temp, Rei works as a freelance data visualization consultant and teaches an evening course on "Data and the City" at a community college in Shinjuku. They're interested in making urban data legible to non-specialists — the same mission that drives this site. They live in Koenji and can often be found at coffee shops with their laptop, drawing SVG paths by hand while listening to ambient music.

Our Approach

We run Tokyo Temp on a shoestring. The annual budget is approximately $8,000, which covers sensor hardware, web hosting, the FLIR camera rental, and the occasional train fare for fieldwork. Nobody draws a salary. Kenji supports himself through consulting work for architecture firms who want thermal analysis for their projects. Yuki still takes forecasting contracts during typhoon season. Rei has their freelance visualization clients.

This financial independence matters to us. We're not funded by a government agency, a corporate sponsor, or an academic institution. We don't have to please a grant committee or align with a policy agenda. We can publish what we find, when we find it, in the form we think is most useful. Our only obligation is to the data — to measure accurately, analyze honestly, and present clearly.

We make one product: the Tokyo Thermal Analysis Guide, a 45-page PDF that compiles our annual findings into a reference document for urban planners, architects, and researchers. It costs $14. Sales cover about 40% of our operating costs. The other 60% comes out of our pockets, which we're okay with because this work matters to us.

If you find our work useful, the best way to support us is to share it with someone who needs it — a planner, a resident, a student, anyone who's ever wondered why Tokyo's seasons don't feel like they should. And if you have data, corrections, or ideas, get in touch. We're always listening.