Depending on who you ask, weather forecasting has been around for hundreds or even thousands of years. Having an idea of whether it rains on a particular day is endlessly convenient, yes — but it’s also hard to overstate the importance of weather tech for many modern industries. Although it may seem that weather forecasting is solely a consumer-focused product, many businesses and government institutions depend on it, and an inaccurate forecast can easily cost millions.
In spite of the long history of weather forecasting, the accuracy of most modern methods leaves a lot to be desired. That’s why we’d looked at the available solutions, noted their weak spots, and built Meteum — the most powerful forecasting platform to date.
Meteum Weather forecasting technology
We don’t collect someone else’s weather data only to relay it in a pretty package. Instead, we aggregate it from several providers, who themselves collect their data from a vast array of sources, such as weather balloons, aircrafts, satellites, on-ground stations, and radars. Then we let our models analyze this data to derive the ultimate forecasts, which we transmit to more than 50 million of our end-users.
Here’s how it works: Meteum parses a number of different forecasts — one conducted by our own team of meteorologists and the rest incoming from third-party providers. All of them are generated using various meteorological models, both proprietary and open-source. But no model is completely fail-safe, and the astounding volume of data we collect makes it practically impossible for humans to handle. The good news is that our own AI algorithms can find patterns by juxtaposing past forecasts with actual weather history and output the optimal combination where the forecasts all align with each other.
In addition to AI-generated forecasts for the week, Meteum also provides precise information about precipitation expected within two hours, split into ten-minute intervals. This info is presented to users in map form; the app sends rain alerts when appropriate. The assessments are made with the help of machine learning, satellite imagery, and radar data. Meteum considers various factors, such as the height of the Sun above the horizon, humidity, and altitude.
It’s no easy feat for a weather app to correctly reflect the conditions you see from the window: weather stations are typically sparsely located and follow a strict schedule when taking measurements. Conditions can vary drastically even within the same city: the Earth’s atmosphere is prone to unexpected shifts, which are insanely difficult to predict. It doesn’t help that seeing this stark contrast between a “perfectly clear” report on one’s phone screen and the raging downpour outside raises questions about the forecast’s reliability. But there is a way to display current weather: meteorology buffs refer to this as nowcasting. Meteum uses a mix of mathematical models and recent data from weather stations — but this alone doesn’t quite hit the accuracy mark we’re striving for.
That’s where we innovate: Meteum is the world’s first weather system to base its forecasts not only on measurements made with professional equipment but also on user reports. In our weather app, users can provide instant feedback: for instance, to confirm whether it’s still raining at their location. Every day, Meteum receives up to 3 million user reports from around the globe, and we use this feedback to generate a precipitation map that shows the rain zones with near-perfect accuracy. The reports are also used in training our algorithms to make future now- and forecasts even better.
We’ve come a long way with Meteum and now have a robust system capable of delivering remarkable forecasts — some of the most accurate you’ll find on the market.
But the sky’s the limit, and Meteum continues to improve, adding new data sources and perpetually collecting user reports to train our algorithms. With Meteum, you can weatherproof your decisions with state-of-the-art forecasting solutions tailored to your industry’s or your business’s needs.