About Nippon Predict
We started this because we got tired of getting rained on. Three people with different skills, all riding around Tokyo, all realizing the same thing: the weather forecast was wrong because it was forecasting for the wrong place.
Who We Are
Rei Nakamura — Founder & Lead Forecaster
Rei spent six years as a regional forecaster at the Japan Meteorological Agency's Tokyo Regional Headquarters in Otemachi. He specialized in urban meteorology — specifically, how Tokyo's built environment modifies synoptic-scale forecasts. He left the JMA not because he disagreed with their work (he didn't — the JMA does world-class large-scale forecasting) but because he wanted to build something the JMA couldn't: street-level predictions for people who actually navigate the city.
Rei's also been a daily cyclist in Tokyo for fifteen years. He's ridden through every microclimate the city has to offer, from the fog pockets of early-morning Sumida to the wind tunnels of Meiji-dori. He's been soaked by storms that the forecast said would miss "Tokyo" entirely, and he's sweated through UHI peaks that didn't show up in any official numbers. Nippon Predict is, in a sense, the forecasting system he wishes he'd had on his commute.
At the JMA, Rei worked on the verification team for the Meso-Scale Model, analyzing forecast accuracy across the Kanto plain. He noticed systematic errors in the model's surface temperature predictions for dense urban areas — errors that were consistent enough to be predictable. The model ran at 5km resolution, which is excellent for synoptic forecasting but far too coarse to capture building-scale effects. He started keeping notes, then spreadsheets, then a database of correction factors. That database became the seed of Nippon Predict's spatial disaggregation model.
Rei holds a Master's degree in Atmospheric Science from the University of Tokyo. He doesn't call himself a "certified meteorologist" because that term implies JMA certification, which he's careful not to claim. He's a forecaster with operational experience, a cyclist with thousands of kilometers of urban observation, and a data analyst who knows where the models break down.
Sarah Kim — Co-Founder & Data Scientist
Sarah built the prediction models. She joined Rei after five years in climate data analytics at a Tokyo-based environmental consultancy, where she worked on building energy modeling and urban heat island mitigation strategies. She has a PhD in Civil Engineering from MIT, with a focus on urban microclimate modeling using computational fluid dynamics.
Sarah's contribution to Nippon Predict was turning Rei's empirical observations into a systematic framework. She took the correction factors he'd developed and back-tested them against five years of observation data from the JMA's AMeDAS network, validating that the spatial patterns were statistically significant and temporally consistent. She then built the automated prediction pipeline that pulls Open-Meteo data, applies the district-specific corrections, and generates the hourly forecasts you see on our dashboard.
She also built our SVG visualization system — the prediction bands that show 24 hours of temperature data in a single glance. "People don't read tables," she says. "They need to see the pattern. The bands tell you whether it's getting warmer, cooler, or stable — that's what matters for decision-making." The color gradient from violet (cold) through pink (mild) to amber (hot) was her design choice, intended to make temperature intuitive at a glance.
Sarah is particularly proud of our rain corridor model. She developed a simplified Lagrangian particle tracker that follows moisture trajectories across Tokyo's terrain, identifying which districts are likely to see precipitation from approaching storms. The model isn't perfect — convective initiation is genuinely hard to predict — but it improves our rain timing accuracy by approximately 20% compared to raw model output.
Tomoaki Sato — Contributor & Field Operations
Tomoaki was a bike messenger in Tokyo for eight years. He knows every wind tunnel, every rain-sheltered alley, and every intersection that floods in a thunderstorm. He's not a meteorologist — he's a practical expert in Tokyo's urban weather, earned through tens of thousands of kilometers of riding in every condition the city produces.
Tomoaki joined Nippon Predict as our field validation team. He rides regular routes through all six of our observation districts, carrying temperature and wind loggers, and reports back on conditions that match or contradict our predictions. His reports are qualitative but invaluable — he'll tell us that Akihabara felt 2 degrees warmer than predicted because of a temporary construction barrier, or that Ginza had unexpected wind because a building facade was being renovated, changing the surface roughness.
Tomoaki's also our user advocate. He reviews every prediction interface before it goes live, checking that the information is actually useful for someone making a go/no-go decision in 30 seconds. If a data display doesn't pass the "can I read this while waiting at a traffic light" test, it doesn't ship. He's ruthless about this, and our product is better for it.
Beyond his fieldwork, Tomoaki maintains our district knowledge base — the detailed microclimate descriptions you read on our hourly shifts and seasonal drifts pages. These aren't copied from textbooks. They're compiled from his own observations, supplemented by academic literature and verified against our measurement data. When we say that the Yaesu exit has sudden gusts, it's because Tomoaki has been knocked sideways there more than once.
What We Believe
We believe weather forecasts should be honest about their limitations. When we're wrong, we say so. When we don't know, we say that too. We don't use vague language like "scattered showers possible" — we give you a percentage chance for your specific district, and we track our accuracy so you know whether to trust us.
We believe in open data. Our predictions are derived from publicly available sources — primarily the Open-Meteo API, which provides access to JMA and ECMWF model output. We don't have proprietary weather stations or secret algorithms. What we have is better spatial interpretation, grounded in physical understanding of Tokyo's urban environment.
We believe cycling makes you a better forecaster. There's no weather shelter on a bicycle. You feel every temperature change, every wind shift, every raindrop. Our team has collectively ridden over 100,000 kilometers in Tokyo, and that embodied knowledge informs everything we build. You can't learn urban microclimate from a desk.
What We're Building
Nippon Predict started as a personal project — Rei sharing his JMA-derived correction factors with a mailing list of cyclist friends. It grew into a website, then a small business, and now a funded venture with a clear mission: build the best short-range microclimate forecasts for Tokyo, and eventually for other cities with complex urban geography.
Our roadmap includes: expanding from 6 to 20 observation districts across all 23 wards; adding real-time rain radar integration for nowcasting; building a citizen science network of temperature loggers for ground-truth validation; and developing a mobile app that delivers district-specific push notifications when conditions change.
We're not trying to replace the JMA. We're trying to do something the JMA can't do within their mandate: interpret their excellent large-scale forecasts at the street level, for the people who actually live and work and ride in the spaces between the official observation points. We think there's value in that. Our users seem to agree.
How to Reach Us
Rei handles forecast questions and media inquiries. Sarah handles technical partnerships and data licensing. Tomoaki handles field reports and user feedback. General inquiries go to forecast@nipponpredict.com and we'll route them appropriately.
USA: Nippon Predict LLC, 1918 8th Ave, Suite 400, Seattle, WA 98101. Phone: +1 (206) 413-7829
Japan: Nippon Predict Tokyo Station, 1-21-15 Shibuya, Shibuya-ku, Tokyo 150-0002. Phone: +81 (3) 5468-9234