Big Data Solutions Based on Weather Analysis and Prediction
Use of predictive data analysis techniques for applications in which weather and atmospheric dynamics are key to decision-making. Prediction of renewable energy generation and anticipation of energy consumption. Prediction and analysis of severe weather events.
TECHNOLOGY DESCRIPTION
Predictive analysis of meteorological/atmospheric data is carried out using big data techniques to support more efficient business decision-making in environments that are sensitive to weather conditions, such as those related to energy resource management or risk management in the event of adverse phenomena.
By handling highly heterogeneous databases and using both unsupervised methods and knowledge of atmospheric dynamics, the relevant variables to be used in predictive analysis are identified. This analysis, with forecasting horizons of less than three days, is based on different big data techniques: machine learning methods such as neural networks, as well as other techniques including a novel analogue-based procedure that we have developed, which combines large- and medium-scale atmospheric dynamics with local weather information from the locations of interest.
TECHNOLOGY APPLICATIONS
Weather predictive analytics has a direct impact on the bottom line of various sectors:
-
Renewable Energy Generation: Accurate prediction of wind and solar power generation for integration into the electricity market.
-
Demand Management: Anticipation or postponement of industrial processes in companies with high electricity consumption based on forecasts of energy costs and availability.
-
Precision Agronomy: Scheduling harvesting or crop protection treatments based on accurate forecasts of rainfall and temperatures.
-
Insurance Sector (InsurTech): Preventive assessment and forensic analysis of damage associated with severe weather events (DANA, hail, extreme winds).
ADVANTAGES AND INNOVATIVE ASPECTS
Weather predictive analysis for decision-making is an emerging field in which various groups and companies already have some experience, although this remains limited. When business planning and outcomes are sensitive to weather conditions, users or clients can have access to better information tailored to their specific needs; access to customised forecasts that enable them to assess risks, costs and benefits, and develop contingency plans supported by proven data and algorithms.
Our group not only uses various statistical and machine learning techniques that have proven useful in the field of big data, but has also developed proprietary methods such as the Lagrangian analogue-based method. Unlike many other groups, we have been combining expertise in data science with knowledge of atmospheric dynamics for many years.
DEVELOPMENT STAGE
The knowledge and algorithmic tools are ready for immediate application. The group has established experience in transferring predictive models to the energy and risk management sectors.
APPLICATION SECTORS
- Electricity Companies and Renewable Energy Operators.
- Energy-Intensive Companies (Steel, Chemical and Paper Industries).
- Large-Scale Agricultural Holdings.
- Insurance and Reinsurance Companies.
- Emergency Management and Civil Protection Agencies.
PRINCIPAL INVESTIGATOR
José Antonio García Orza – Department of Applied Physics