How Does the Technology Behind Tomorrow.io Work?
At Tomorrow.io, we have a different approach to the weather.
The world’s first Weather Intelligence Platform, Tomorrow.io is fully customizable to any industry impacted by the weather. Customers around the world including the NFL and Uber use Tomorrow.io to dramatically improve operational efficiency and better adapt to the weather.
Tomorrow.io was built from the ground up to help teams predict the business impact of weather, streamline team communication and action plans, improve productivity, and optimize revenue margins.
But how does it work?
If you’re interested in the technical details behind the Tomorrow.io platform, let us explain. Learn how our NowCast, 1 Forecast, Insights Dashboard, Geospacial features, and more all work together to enable weather intelligence at scale.
NowCast is Tomorrow.io’s proprietary short-term quantitative precipitation forecasting tool. Tomorrow.io’s proprietary methods produce the most accurate, hyperlocal, short-term forecasts updated every minute globally.
This is a simple description of the approach:
- Collect a series of images that describe how precipitation is changing over a region during a period of time e.g. a series of radar images.
- Analyze where and how precipitation is “moving” over those images. This is called an optical flow problem and from which Tomorrow.io creates a “motion field” i.e. another image that explains where and how things are moving.
- Using the “motion field”, estimate where and how the precipitation will move over time. This is called an advection problem.
Using this technique, Tomorrow.io can produce accurate (at short lead times) precipitation forecasts, at very high spatial resolution over very large areas, quickly. We use the latest advances in scientific programming (just-in-time compilers, advanced numerical techniques, etc.) to perform this rapid process running a 2-hour simulation at 1 km resolution minute-by-minute.
1 Forecast Global Coverage
Tomorrow.io’s 1 Forecast (1F) is a novel and proprietary platform producing the most accurate hyperlocal weather forecasts globally.
Here is how it works:
- Aggregates in real-time data from existing numerical weather prediction models including those from NOAA and ECMWF, as well as Tomorrow.io’s proprietary numerical weather prediction model CBAM (Tomorrow.io Bespoke Atmospheric Model) in an automated and seamless way. CBAM is the world’s highest resolution forecasting model incorporating millions of virtual sensors, and can be customized to resolution, coverage, parameters, and finely-tuned to any region globally.
- Having access to multiple forecast products and observations allows Tomorrow.io’s proprietary 1F data product to leverage a vast amount of information while intelligently extracting only the most valuable i.e., accurate variables from each. The benefits of said approach include:
- 1F doesn’t rely on any one forecast model; instead, it uses any/all models available
- Perform robust calibration of future forecasts
- Includes a dynamic weighting of the input numerical weather prediction members, allowing Tomorrow.io to selectively weight the models that have demonstrated superior skill over a past time window of 3 days
- 1F can be used to generate forecast parameters that are not inherently included in any of the underlying numerical weather prediction system outputs, such as local visibility/fog, probabilities of thunderstorms and any other parameter, or application-specific products
- 1F gives Tomorrow.io a platform to build and use machine learning to continually improve the accuracy and capabilities of a truly proprietary forecast product
Insights – Complex Event Processing
Tomorrow.io’s API supports and processes Computational Logic Statements. Simply stated, this means that users can specify mathematical operations and expressions which then automatically produce business insights and actions based on the various ever-changing minute-to-minute states of weather.
Here is an example:
- For any given geographic location
- Either in real-time (NowCast) and/or at a forecasted time
- Apply to any of 30+ weather variables (wind, precipitation, fire index, air quality etc.)
- Logic combining the variables (low humidity, high temperature, high wind speeds)
- Rules regarding weather variable states (thresholds)
Business insights are then automatically produced and actions triggered by ever-changing, dynamic weather variable states with Tomorrow.io’s API and proprietary technology. Utilizing high-frequency intervals at high resolution, Tomorrow.io creates millions upon millions of weather variables which are automatically analyzed at a hyperlocal level and translated into actionable business insights for users.
The Tomorrow.io platform incorporates geospatial features and functions (GeoJSON Geometries) that enable numerous, unique applications of weather in business operations such as weather in vehicle routing and navigation, location-specific alerts and business insights, weather advisories updated in real-time along routes, electric vehicle range calculations, etc. with these representative functions:
- Monitors: Setting, editing, and receiving alerts on predefined polygon locations.
- Geofence: Define asset areas, and get live weather conditions as well as alerts to ensure operational safety and efficiency.
- Routing: Data throughout a route broken into timelines of each leg represented by a polyline/polygon/point geometry.
Weather for AI (Historical Weather Data)
Tomorrow.io’s Weather for AI includes 20 years of highly accurate historical weather information for millions of hyperlocal locations worldwide. This data is used to generate AI-driven insights to train machine learning models and glean insights into methods to improve present and predictive business operations.
Air Quality + Weather
Tomorrow.io is the only company in the world to provide hyperlocal weather and air quality solutions using proprietary technology. Tomorrow.io’s air quality offering includes both US and China EPA index parameters, along with PM<10 (dust), PM<2.5 (not visible), Ozone (fuel+radiation ), CO, SO, NO2, and SO2 levels.
Focused on impact, Tomorrow.io’s predictive air quality product incorporates high resolution and hyperlocal granularity, dynamic wind layers, and air quality risk impact.
The Tomorrow.io competitive advantage:
- Weather + air quality
- Global coverage vs. limited
- 30+ API parameters
- Navigational routing
- Monitoring feature
- Customized templates
- Business insights vs. raw data
Tracking more than 25 species, Tomorrow.io uses a hybrid approach to pollen, blending a mixture of statistical and physical/chemical modeling. With observations scarce, satellite imagery is leveraged to estimate areas where plants are germinating and producing pollen that may be lofted into the air. Using meteorological data (both real-time and forecast), Tomorrow.io can estimate where the winds will transport the lofted pollen — and more importantly, where it will rain out or settle to the ground. With benefits across the healthcare industry and more, the approach is informed by state-of-the-science advances in both ecology and environmental modeling.
Proactive and active fire management:
Tomorrow.io’s All-in-One Weather Intelligence Platform
Unlike other weather data providers, Tomorrow.io offers so much for than just data. With all of the technology and features above, Tomorrow.io can serve almost any business that is impacted by the weather.
Want to learn more? – Get a demo of Tomorrow.io.