The person behind the site
Jawad Khan — developer and maintainer.
This site is built and run by a working software developer, not by a newsroom or a meteorology practice. The calculator, the scoring model, the weather integration and every page on the site are the work of one person, and the methodology is published in full so that the reasoning can be checked rather than taken on trust.
There is no editorial board here and no team of writers. Describing one would be more flattering and less true. What there is instead is a codebase you can reason about from the outside: a published scoring model, named data sources, and pages that state their own limits rather than burying them.
What is actually built
The site is a statically generated application. Every page — the homepage calculator, the focused ice, cold and college calculators, the country and regional hubs, and each city page — is rendered to plain HTML at build time and served as a static file. The forecast itself is fetched in your browser when a page loads, which is why results are current rather than baked in at build time.
- Writing and maintaining the scoring model in the site's source code
- Choosing and integrating the weather and official-alert data sources
- Keeping the published methodology matched to what the code actually does
- Reviewing location pages and correcting anything reported as inaccurate
- Responding to bug reports, accessibility issues and factual corrections
Which data sources are used, and why
Forecast data comes from Open-Meteo. The browser requests daily snowfall, rain, total precipitation, precipitation probability, minimum and maximum temperature and maximum wind, plus hourly snowfall so the evening, overnight and morning bus windows can be separated. Open-Meteo was chosen because its variable definitions are documented publicly, it needs no account, and it serves coordinates across both the United States and Canada consistently.
Official United States alerts come from the National Weather Service API. For supported U.S. coordinates the browser checks for active winter alerts and shows them beside the result with a link to the official local forecast. That check is deliberately non-blocking and deliberately kept out of the score, because the model has never been calibrated against a verified dataset of alert status versus actual school decisions. Folding alerts into the percentage would look more sophisticated and mean less.
Canadian alerts are linked, not embedded. Environment and Climate Change Canada issues Canada's official warnings. The automated check does not cover them, so Canadian pages say so plainly and link to the national alert index instead of leaving an empty panel to be misread as “no warnings”.
Location context is hand-written, not generated. The geography and local closure factors on each city page were written per city, and they intentionally contain no snowfall averages, closure counts or district thresholds. Those numbers exist in real sources, but reproducing them from memory is how invented statistics enter the world.
How the published method is kept honest
The methodology page lists six factors with weights of 35, 18, 15, 12, 12 and 8 percent. Those are the numbers in the source code, not a simplified version written for readers. When a threshold changes in the code, the methodology page is updated in the same change.
The build also refuses to ship certain classes of mistake. Automated checks run before every deployment and fail the build on a duplicate page title, a duplicate meta description, a duplicate canonical URL, a page with more or fewer than one H1, a broken internal link, a missing image description, invalid structured data, a canonical URL missing from the sitemap, or near-duplicate location prose above a similarity threshold. That last check is the reason there is no filler text on the city pages: it would not pass.
What this site does not claim
No meteorology credential. Nobody involved is a meteorologist, and no such qualification is claimed anywhere on the site. The model applies published, inspectable rules to public forecast data.
- No accuracy percentage, because verifying one would require a complete history of forecasts and decisions for every district covered.
- No access to any district's roads, staffing, buildings or internal decisions.
- No claim to be the most accurate calculator, or to be an official notice of any kind.
- No snowfall threshold for any named school system, because those are operational judgements rather than published figures.
- No reviews and no ratings, invented or otherwise.
The accuracy and limitations page goes further into what the estimate can and cannot support.
Updates and corrections
Location pages carry the date their local context was last reviewed, and guides carry their own updated dates. Reviews happen when the model changes, when a data source changes behaviour, and when something is reported as wrong.
If you find a calculation bug, a broken link, an accessibility problem or a factual error — particularly on a city page, where local knowledge beats mine — please report it through the contact page. Include the page URL and enough detail to reproduce the issue. Corrections to local context are genuinely welcome; do not send student or school records.