Driving the World’s Most Accurate Weather Forecasts

Share this post:

Weather forecasting is a data challenge on the scale of few others. Weather is perhaps the single largest external swing factor in business performance today, responsible for an annual economic impact of nearly half a trillion dollars in the U.S. alone. We all need help anticipating weather’s effects on us whether it’s a simple question of taking an umbrella to work or the slightly higher-stakes multi-million-dollar energy market trade.

The Weather Company, an IBM company, is using data science, analytics, and machine learning to help tens of millions of people make more informed, confident decisions every day.

Today, a new independent third-party research study recognized The Weather Company as the provider of the industry’s most accurate, reliable forecasts. According to the report by ForecastWatch, The Weather Company offers the most accurate 1-3-day and 3-5-day forecasts in the U.S., Europe, and Asia, and 6-9-day forecasts in the U.S. and Asia.

Over the past six years, The Weather Company was rated most accurate 78% of the times/regions/periods studied while the next closest competitor won only 14%. Behind the scenes, continuous advances in data science, analytics, and machine learning have helped make this leadership possible.

A Behind the Scenes Look at the Science of Weather Forecasts

Data: We start with the richest, highest-resolution, most granular data available from numerical weather prediction (NWP) models around the world, our own Deep Thunder NWP model, and other conventional weather sources such as radar, weather balloons, buoys, and satellites. In all, we process 400 TB of data daily. That’s about 27X the data equivalent of all the books in the US Library of Congress.

Analytics and Machine Learning: Rather than delivering these forecasts directly to users, we use machine learning to improve the result further. To generate a single final forecast, we take input from 162 individual forecast sources and use advanced machine-learning algorithms to continuously optimize each forecast element (e.g. temperature, precipitation, wind direction and speed, humidity and pressure) based on geography, time, weather type and recent accuracy. These optimized forecast elements are then assembled into intelligent complete forecast products to arrive at the single final forecast we deliver to customers.

Massive Scale: All of this is done in real time on a worldwide forecast grid at 500-meter increments, or 2.2 billion locations worldwide every 15 minutes.

Weather Forecasting Drives Business Decisions

Weather directly impacts how we all feel, act, and purchase. We help people make better business decisions with confidence and have earned the trust of more than 55 M unique monthly consumers but also business users like energy traders making daily multi-million-dollar positions and the pilots of more than 55,000 daily flights. At The Weather Company, we forecast purchasing patterns, staffing levels, worker safety, energy usage, turbulence and foot traffic. Combining this forecast data and an understanding of how it affects markets with IBM outcome and decision analytics serves as the foundation of trust clients have in these forecasting services for their businesses.

When you’re looking to make more confident business decisions that hit your bottom line, our forecasts are there to help. For more information on how to make our forecasts work for you, contact us today for a short demonstration.