SOFF data enhances AI applications in weather and climate science
This article was written by Florian Pappenberger, Deputy-Director General & Director of Forecasts and Services at the European Centre for Medium-Range Weather Forecasts (ECMWF).
Data generated and internationally shared as a result of investments by the Systematic Observations Financing Facility (SOFF) is becoming increasingly vital for advancing machine learning (ML) and artificial intelligence (AI) applications in weather and climate science. SOFF supports countries to provide the good-quality data and insights necessary for developing robust ML models, which improve predictive capabilities, support effective early warning systems, and inform national climate change adaptation and mitigation plans.
The rise of machine learning in weather and climate science
Machine learning has played a role in weather and climate science for many years, with postprocessing being a prominent and valuable example. In the last few years, first in nowcasting and then in medium-range forecasting, machine learning has emerged as a tool for making forecasts. The technology is new and fast evolving, but meteorological organisations are embracing developments to ensure all avenues are explored in creating the best possible forecasts. In climate science, developments are ongoing but less mature, with early signs that machine learning could play a key role in uncertainty quantification.
More, better, and real-time data improve Machine Learning
Good-quality data is vital for making accurate ML predictions. SOFF plays a pivotal role in enhancing the quality and quantity of surface-based weather and climate data, essential for the effectiveness and added value of ML/AI applications. By providing financial support to implement the WMO mandated Global Basic Observing Network (GBON), SOFF ensures that critical weather and climate observations are made and shared globally, particularly addressing gaps in data-sparse areas such as Small Island Developing States as well as Least Developed Countries by installing new or improving existing surface and upper air stations. This expansion significantly increases the volume of data available for producing reanalyses of the past Earth system state. Reanalyses, such as the ERA(what is this, spell out) reanalysis (produced by ECMWF in the framework of the EU Copernicus Climate Change Service), provide an invaluable resource for training robust ML models.
According to GBON standards, SOFF promotes data reliability and standardization, for example by supporting regular inspection and calibration of instruments. This will ensure that the data will meet international requirements and standards, such as for continuity and timeliness. The wealth of good-quality data is crucial for detailed weather forecasting, allowing ML models to predict weather patterns as well as extreme events with greater precision. Advanced ML techniques benefit from the vast and diverse datasets provided by countries supported by SOFF.
Real-time data processing is another critical aspect of SOFF’s contribution, facilitating timely updates and ensuring that ML models receive the most current information. In this case, SOFF investments ensure that good-quality data from the countries will be shared internationally through the WMO Information System (WIS) 2.0.
Enhancing human capacity to better process data
SOFF plays a crucial role in promoting the use of advanced technologies and investing in capacity building worldwide. Recognizing the diverse challenges faced by different regions, including Small Island Developing States and Least Developed Countries, SOFF aims to support the development and maintenance of robust observing networks. These networks collect data vital for machine learning models. This data is used daily in numerical prediction systems around the world to create the most accurate initial conditions for the physics-based or machine learning predictions. Through collaborative training programmes with SOFF Peer Advisors, local expertise in meteorology and data science will be strengthened across various regions. These initiatives aim to equip scientists and technicians globally with the skills needed to operate and ensure the long-term sustainability of meteorological instruments, fostering a more inclusive and comprehensive global weather observation system.
Informing national climate change adaptation and mitigation plans
Countries implementing SOFF investments provide valuable data and insights that can be used by ML models to develop effective strategies for climate change adaptation and mitigation. By supporting the collection of long-term climate data, these investments will enable ML models to analyze historical climate patterns and trends, essential for understanding vulnerabilities and predicting future impacts. This data-driven approach allows for the development of localized adaptation plans that address specific regional challenges and enhance community resilience.
SOFF plays a crucial role in enhancing early warning systems by providing comprehensive and good-quality weather and climate data. ML algorithms can analyse the vast datasets provided by SOFF to detect anomalies that may indicate the onset of extreme weather events. By integrating ML models with conventional early warning systems, National Hydrometeorological Services can be empowered to automat alerts and disseminated them to to relevant authorities and the public, ensuring timely warnings, and reducing the time lag between detection and notification.
Partnerships for a global public good
SOFF fosters international partnerships and collaborations, facilitating the exchange of knowledge and best practices. These collaborations enhance the global capacity to collect and analyze weather and climate data. For example, ECMWF supports SOFF countries by offering free real-time access to ECMWF forecast data.
In conclusion, closing data gaps in most data sparse areas around the world is increasingly important for advancing ML/AI applications in weather and climate science. Particularly Small Island Developing States and Least Developed Countries require financial support and capacity-building to collect and internationally exchange basic weather and climate data. With SOFF support in collaboration with ECMWF, countries not only improve their national forecasts, but a sound basis for developing robust ML models and improving outcomes for communities worldwide can be achieved additionally, contributing to a global public good.
YOU MAY ALSO LIKE...
