The Weather Company (TWC), an IBM Business, was named the most accurate forecaster in a study by ForecastWatch, the US authority in meteorological accuracy validation. The study names The Weather Company’s consumer brands The Weather Channel and Weather Underground as the most accurate forecasters overall, across diverse geographic regions and time periods covered.
“Our users and clients turn to us to help them make weather-related decisions with confidence. This study shows The Weather Company as the undisputed accuracy leader, confirming the trust of more than 250 million people who choose The Weather Company for weather information every month,” said Mary Glackin, TWC’s Senior Vice President of Science and Forecasting Operations. “Making a claim is easy, but backing it up takes continued effort and focus on who you’re serving. With IBM, we’ve stepped up investment and made a dedicated effort toward advancing our forecasting skill in 2016.”
The study evaluated the top forecasting services in three regions independently: the United States – since 2010, and Europe and Asia Pacific – since 2013. Within each region, data was broken out into three different forecast time ranges. According to ForecastWatch, TWC was the overwhelming leader, offering the most accurate 1-3 day and 3-5 day forecasts in the U.S., Europe and Asia, and the most accurate 6-9 day forecasts in the U.S. and Asia. The study collected forecasts from eleven different providers and analyzed a total of more than 139.3 million forecasts.
Creating the highest level of accuracy
- It starts with better data
Most forecast providers turn to reliable, yet sparsely placed government-provided data sources. Alternatively, TWC feeds its forecasting engine with additional sources of the richest, highest-resolution, finest granularity data available. TWC not only relies on skillful numerical weather prediction models from agencies — notably the National Weather Service in the U.S., the U.K. Met Office and the ECMWF in Europe, and more — but also from IBM’s own proprietary Deep Thunder model. The engine also includes data from surface observations, precipitation, radar, satellite, personal weather stations, lightning sources, data collected from planes every day, as well as one of the largest troves of location data available anywhere.
- Analytics helps advance weather modeling
To create a single forecast, 162 individual forecasts are generated within TWC’s analytical system combining a wide variety of government and private forecast models — including the Deep Thunder model. Machine-learning algorithms weigh factors like temperature or precipitation from each forecast based on geography, time, weather type and recent forecast accuracy. The system then blends those weighted contributions to arrive at a single synthesized forecast that provides the best possible accuracy available.
- Cloud-based infrastructure
Running these analytical models is computationally intensive, so TWC created a cloud-based platform purpose-built for all types of big data. The platform increased data-handling capacity tenfold and handles 400 terabytes of data every day, generating tens of millions of forecasts around the globe within microseconds at 15-minute intervals every day. Today it handles much more than weather data and forms an integral piece of IBM’s Watson Data Platform.
Full-spectrum Weather: Past, present and future
Now part of IBM, TWC has invested heavily and dedicated efforts toward advancing its forecasting skills and meteorological expertise. Consumers and businesses can trust TWC’s precise weather data, whether they’re looking to make critical decisions based on weather in the past, present or future.
- Forecasting the future — When the company launched Forecasts on Demand in 2013, it reinvented forecasting by bringing many forecast-generating aspects into a real-time on-demand process for 2.2 billion locations on the planet every 15 minutes. Forecasts on Demand looks ahead up to 10 days, while seasonal forecasts offering a three-month overview on what to expect in the U.S. and Europe for longer-term weather insights.
- Observing current conditions — Consumers and businesses use real-time data to plan daily. A fresher, more precise, globalized Currents on Demand system launched this past October to report current conditions for the same hyper-local grid of 2.2 billion locations globally. The upgrade uses request-time integration and can also improve forecasts by informing the first few hours in a short-term forecast.
- Analyzing the past — Comparing historical weather, sales and location data can help businesses anticipate changes when certain weather patterns may drive fluctuations, such as the need for various levels of staffing, stocking and pricing.
These advances all come together for one purpose — to help keep every person on the planet informed. TWC’s weather platform provides information in almost every country, based on its billions of forecast points worldwide. With the most precise forecast and reliable current and historical data and models available globally, Weather aims to deliver the right data to the right person at the right time to help the end user make more informed weather-related decisions.
More around this topic...
In the same section
© HPC Today 2017 - All rights reserved.
Thank you for reading HPC Today.
Most read articles