![]() |
||||||||||
![]() |
||||||||||
![]() |
||||||||||
![]() |
||||||||||
![]() |
![]() |
|
![]() |
|||||||
![]() |
||||||||||
![]() |
![]() |
![]() |
Weather information is becoming a key factor in big data processes. And with it, companies are generating predictive models for their business and planning practices. Repsol uses weather forecasts, for example, for managing renewable electricity facilities, to offer its customers advice on energy efficiency at home or to anticipate incidents on its trading business shipping routes. «Weather factors affect our activity, climate change is an example of this», explains Julia Díaz, head of Data Science at Repsol's Data & Analytics & Artificial Intelligence Hub. «That's why at Repsol, we have to be open to the advancement of a science that is constantly innovating» and that increasingly provides greater precision in its forecasts on weather conditions, which are highly complex. Meteorological data for decision making |
![]() |
||||||
![]() |
The Data & Analytics ![]() |
The energy company has a Data & Analytics & Artificial Intelligence Hub, in which 52 specialists in data analytics work with different profiles (translator analytics, engineers specialized in data, data scientists, and data managers). These specialists collaborate with 400 professionals from the different business units to extract value from the more than three million gigabytes of data that Repsol handles daily. «We use weather data provided by specialized agencies which, in turn, we combine with data from our own activity» and from other external sources, such as geolocated information or sociodemographic studies from the National Institute of Statistics (INE). The challenge is to incorporate such diverse sources into mathematical models so that «they ‘speak to us’ in the simplest way possible terms and serve as support in decision-making», continues Díaz. |
![]() |
|||||||
![]() |
||||||||||
![]() |
![]() |
![]() |
||||||||
![]() |
![]() |
|
![]() |
|||||||
![]() |
||||||||||
![]() |
![]() |
For these advanced analytical techniques to be successful, «it is essential that the data for the calculations is of high quality». That is why Repsol collaborates with the Spanish State Meteorological Agency (AEMET), «a reference in our country», that also markets the global forecasts of the European Center for Medium-Range Weather Forecasts (ECMWF). This body is made up of 34 European weather agencies and is internationally recognized for its forecast quality. On the technological side, Repsol has a strategic partnership with Microsoft that includes hosting its Data & Analytics platform on Azure, the American company's 'cloud'. Effects on consumption patterns
|
||||||||
![]() |
Weather is ![]() |
Repsol uses the 'Degree day' concept, based on the AEMET temperature forecast and business analysis, to «approximate trends that respond to our customers' needs». This tool helps to anticipate, for example, the consumption of LPG (butane and propane) to heat homes and prepare the logistics chain. Weather forecasts are also used to offer tailor-made products for adverse weather conditions such as anti-freeze systems for agriculture. The Data & Analytics & AI Hub is working on a project to issue technical energy efficiency recommendations to its power and gas customers. They will include smart guidelines, «to adjust your refrigerator's energy consumption according to the temperature or use more or less light depending on how cloudy it is», Díaz assures, based on the results of algorithms that use meteorology. |
![]() |
|||||||
![]() |
![]() |
![]() |
||||||||
![]() |
![]() |
|
![]() |
|||||||
![]() |
||||||||||
![]() |
Management of renewable energies
|
![]() |
||||||||
![]() |
Machine Learning ![]() |
One line of ongoing work that will make meteorology evolve is the application of Machine Learning and Artificial Intelligence techniques «similar to those we are already using at Repsol for our business scenarios». These are cognitive technologies that train computers using past predictions and behaviors so that they learn automatically and provide more precise calculations. Today, meteorological services provide a highly detailed 48-hour forecast and a reliable seven-day forecast, with more uncertainty the further removed in time. Meteorologists estimate that each day of improvement implies 10 years of research, and they also warn that 100% accurate predictions will never be possible because the atmosphere is, by nature, a chaotic system. Hardware also poses a challenge due to the capacity requirements and computing costs for modeling these physical phenomena. In the next few years, the first quantum computers could arrive with a calculation speed 100 times greater than that of current supercomputers, «an emerging technology that responds to computing requirements and an ever-increasing data volume which will demand more precision, but also more speed», concludes Díaz. ![]() |
![]() |
|||||||
![]() |
||||||||||
![]() |
|
![]() |