ECMWF’s commitment to working with Member and Co‑operating States and other partners to improve the quality, range and accessibility of weather forecasts was strengthened by the common necessity to tackle the challenges brought about by the global pandemic with enduring solutions.
During this unprecedented year, ECMWF made more data openly available to all and made strong progress with the European Weather Cloud to create a cloud computing infrastructure for the meteorological community. Despite the lockdowns, ECMWF delivered its training and workshop programmes in earnest. In many ways, events achieved a much broader reach, whilst reducing the associated carbon footprint. All ECMWF’s governance meetings went virtual for the first time, in some instances allowing larger delegations to attend.
ECMWF also continued to serve its community through its contribution to the European Union’s Copernicus Services, publishing key reports such as the European State of the Climate, and maintaining its regular and critical user engagement programme.
Making deliverables and expertise available
ECMWF’s operations remained unaffected despite the lockdowns caused by COVID-19, with staff working remotely thanks to the Centre’s reliable IT infrastructure and online resources.
Events moved to virtual platforms during the year, which saw the number of attendees rise to 1,305, up 98% from 2019. There was also a significant reduction in the carbon footprint associated with events. Staff and participants did miss the face-to-face engagement of events and networking, a consideration that will help shape the Centre’s future approach to serving Member and Co‑operating States and will inform the use of a range of different video-conferencing software.
The flexibility enabled by digital tools made shorter events of just one-hour duration viable, such as the Machine Learning series of seminars. As a result, ECMWF ran 19 events in 2020, compared with 7 in 2019. Speakers and guests were able to attend shorter events without the need to travel.
A wide-ranging and comprehensive programme of workshops, seminars, training, and task groups covered topics including assimilating satellite cloud and precipitation for numerical weather prediction; aircraft weather observations and their use; warm conveyor belts; and guidelines for post-processing ensemble forecasts to enable WMO Members to make more effective use of ensemble forecasts.
The Using ECMWF Forecasts (UEF) event in June attracted 227 people from 43 countries for the theme ‘Keeping users at the heart of operations’. The Annual Seminar in September was on the topic ‘Recent developments in numerical methods for atmospheric and oceanic modelling’, exploring the latest developments in computational techniques for solving the equations that govern atmospheric, wave and sea-ice dynamics. Speakers and participants at both events embraced the digital format to interact and engage with other guests.
Liaison visits, which are a key part of our engagement with the Member and Co‑operating States on ECMWF’s plans and products, also evolved as the pandemic unfolded. Before lockdown, knowledge sharing, networking and training visits were made to Ireland, Montenegro, the Netherlands, and Serbia, while virtual visits were made to Belgium, Estonia, Luxembourg and Switzerland before the end of 2020.
Strong progress was made on a number of initiatives to improve ECMWF products. To monitor and improve the transparency and performance of the dissemination system, ECMWF is working on producing regular key performance indicator (KPI) reports for Member and Co‑operating States.
A new way to browse ECMWF’s website allows searches of all ECMWF datasets in one place. The data is also discoverable by Google’s Data Search portal, and the retrieval of archive data online has also greatly improved. A reorganisation of services means users get a much faster turn-around for their requests.
The European Weather Cloud being piloted in collaboration with EUMETSAT made good progress, with Member States testing applications and providing feedback for configuring the future operational cloud services.
Open data for all
Since October 2020, ECMWF has made hundreds of its forecast maps free and available to all to encourage innovation and to support a thriving, data-based digital economy.
The open data covers the whole world and includes medium-range, extended-range and long-range forecast charts of temperature, wind, precipitation, clouds and ocean waves as just some of the products available. With ECMWF’s focus on ensemble prediction, charts also cover probability-based information, which provides a guide to forecast confidence. The likelihood of extreme conditions, as well as tropical and extratropical cyclone activity, are also included.
Users can share, redistribute and adapt the information as they require, even for commercial applications, as long as they acknowledge ECMWF as the source.
Previously, full access to ECMWF’s forecast charts was restricted to the national meteorological and hydrological services of Member and Co‑operating States, World Meteorological Organization Members and commercial customers, and access was subject to a range of licences and often incurred charges.
The changes also mean a move to an open data policy for historical information in ECMWF’s huge data repository – the Meteorological Archival and Retrieval System (MARS). MARS contains hundreds of petabytes of data including recent and past forecasts, analyses, climatological data and research experiments; it represents the largest archive of such data in the world. Opening MARS data simplifies and expands their use and re-use, stimulating further research and the development of applications related to weather and beyond.
Warm conveyor belts – a challenge to forecasting
ECMWF’s first virtual workshop took place in March, bringing together observation, assimilation, model, forecast and research communities to explore all aspects of warm conveyor belts (WCBs).
WCBs are ascending, poleward-moving warm moist airstreams in the warm sector of extratropical cyclones. They can have major repercussions on downstream weather, but their evolution is often associated with great uncertainty. The aim of this workshop was to bring communities together to discuss how the latest research can improve understanding of WCBs. Extended knowledge of this type of air flow can help to inform improvements in operational ensemble forecasting – a goal which would be very difficult for a single such community to achieve on its own.
In recent years, close work with ECMWF Fellow Heini Wernli and his team at ETH Zurich has helped to improve our understanding of WCBs and how they can be represented well in numerical weather prediction. Heini co‑organised the workshop, and helped to set the scene with an overview of the history and relevance of warm conveyor belts and by posing some opening questions.
Questions addressed in the 31 talks and four working group sessions covered aspects such as predictability, observations, models and impacts.
Predictability is an issue because WCBs are associated with great atmospheric instability. This leads to a rapid growth in forecast uncertainty. It is therefore important to identify the key aspects of WCBs which lead to such uncertainty.
Another topic was how well we can estimate the features of WCBs at the start of weather forecasts and whether new observation sources would help. Talks also addressed the question of how well models represent the complex physical processes in WCBs and what aspects deserve particular attention.
In terms of impacts, the workshop looked into the role WCBs play in weather extremes, regime transitions, and global climate.
Strong progress with European Weather Cloud
The joint project by ECMWF and EUMETSAT to create a cloud computing infrastructure for the meteorological community entered an exciting new phase. Following the procurement and deployment of the clusters for development and production, as well as storage, networks, and monitoring, the focus in 2020 was on working with users. This involved trialling applications on the pilot infrastructures to evaluate the platform and to shape the future operational services.
The cloud will lay the foundations for a strong, domain-specific European IT infrastructure enabling direct access to and effective processing of data. This means that observation data, forecast data and meteorological products can be accessed together as if they were collocated. Federation with relevant cloud infrastructures from other partners will make it possible to further widen the range of accessible data.
The project has consistently attracted more use cases, showing a good range of the possibilities offered by the cloud. Users were able to port existing services or develop new ones, benefiting from the fast access to archive and forecast data.
The pilot also demonstrated how users can bring their processing to the data and apply new forms of data analytics, like machine learning, on the large datasets they require.
In May and November, joint workshops with EUMETSAT hosted over 325 participants, and 120 participants joined a virtual ‘Cloud IT technical workshop’, providing a technical overview and sharing updates on recent developments, experience gained, and next steps.
In real-world practice, the European Weather Cloud was used to support Croatia in the aftermath of an earthquake that shook Zagreb. The quake and aftershocks damaged the 17th century building hosting the Croatian meteorological service (DHMZ). Fortunately, no staff were hurt so they were able to keep their operations running, and ECMWF’s User Services were pleased to be able to support them using the European Weather Cloud to continue to fulfil their official duties.
The European Weather Cloud has benefited from technical expertise accumulated with cloud developments of the Copernicus Climate Data store and WEkEO Data and Information Access Service (DIAS). It is expected to become operational in 2022 after the pilot phase was extended by another year.
ECMWF’s strength relies on a variety of collaborations and partnerships with individuals, institutions and international organisations. Seven highly respected scientists from Europe and the United States were appointed as ECMWF Fellows in December 2020: Prof. Sándor Baran (University of Debrecen); Prof. Hannah Cloke (University of Reading); Prof. Dr Daniela Domeisen (ETH Zurich); Prof. Patrick Eriksson (Chalmers University of Technology); Dr Christian Grams (Karlsruhe Institute of Technology (KIT), Institute of Meteorology and Climate Research (IMK-TRO)); Prof. Dr Daniela Jacob (Climate Service Center Germany (GERICS)); and Dr Gabriele Pfister (Atmospheric Chemistry Observations and Modeling Lab (ACOM)).
During their initial three-year terms, the Fellows will work closely with ECMWF on scientific and technical research in a range of areas related to Earth system modelling, ensemble forecasting, and applications.
The Centre extended its thanks to Prof.Tilmann Gneiting (Heidelberg Institute for Theoretical Studies (HITS) and Karlsruhe Institute of Technology (KIT)), Prof. Rupert Klein (Freie Universität Berlin) and Prof. Tim Palmer (University of Oxford), who were appointed in July 2014 when the Fellowship scheme launched and completed their terms in July 2020.
The programme has helped strengthen links between ECMWF, the Fellows and their wider research groups in several areas, including the calibration and post-processing of forecasts; novel numerical methods; the use of ‘single precision’ in ECMWF’s Integrated Forecasting System (IFS); predictability; and machine learning.
Delivering environmental information
The Copernicus Atmosphere Monitoring Service (CAMS) and Copernicus Climate Change Service (C3S), both implemented by ECMWF on behalf of the EU, made strong progress throughout the year. ECMWF also contributed to scientific improvements for the Copernicus Emergency Management Service (CEMS), in particular the early warning systems for flood and fire danger.
Copernicus Atmosphere Monitoring Service
Two significant system improvements were realised in CAMS in 2020.
The CAMS Atmosphere Data Store launched in late spring, designed to eventually replace the Service’s data catalogue and allow CAMS data to be downloaded in one place and in a few easy-to-use formats. In addition, the CAMS global forecasting system was successfully upgraded on 6 October. The upgrade combined four significant scientific changes, making the CAMS global forecasts even more accurate and robust.
Throughout the year CAMS continued to provide near-real-time monitoring and global and regional forecasting of events such as desert dust storm emission, transport and deposition, and wildfire activity; 2020 was a year of extremes for the latter in particular. The Arctic and western US saw record high levels of activity during the summer, whilst Canada and tropical Africa saw below-average activity. In fact, the reduced activity, especially in tropical Africa (which typically accounts for about a third of total global wildfire emissions), helped to make 2020 one of the least active years in the CAMS Global Fire Assimilation System dataset, going back to 2003.
CAMS continued to monitor the ozone layer as it has since launch, combining information from its detailed numerical models of the atmosphere with satellite and ground-based (in situ) observations through data assimilation. Currently, CAMS uses ozone satellite observations from the SBUV-2, OMI, MLS, GOME-2 and Sentinel-5P/TROPOMI instruments. In addition to its ongoing monitoring and reporting of the Antarctic ozone hole, which regularly attracts press attention for CAMS, the Service also recorded that ozone levels over the Arctic reached record-breaking low values during 2020’s northern hemisphere spring, generating broad interest.
Copernicus Climate Change Service
In April the third annual European State of the Climate (ESOTC) report looking at key climate indicators in 2019 was published. A key focus was that 2019 was the warmest year on record for Europe. All seasons of the year were warmer than average, but some parts of Europe reached daily average temperatures up to 9°C higher than normal during the summer months. The ESOTC also explored the record-breaking levels of rainfall in autumn 2019 that resulted in the wettest November on record for some countries, and flooding in parts of western and southern Europe.
The launch of the ESOTC 2019 coincided with an expansion of the media partnership with Euronews to include Climate Now Social Live – an hour-long live social media programme featuring specialists discussing a topic relevant to climate change, with the opportunity for the audience to have their questions answered. The programme went from strength to strength through the year, covering topics such as biodiversity, health and use of artificial intelligence (AI). By the end of 2020, the C3S-sponsored Euronews online content had received over 1.1 million page views in total.
Later in the year the Climate Data Store (CDS) hit a major milestone with the registration of its 60,000th user in October. Nearly 36,000 TB of data have been delivered since the service launched in June 2018, and by the end of 2020 total data requests were rapidly approaching 200 million. At the end of the year, a significant back extension to the ERA5 reanalysis dataset was released, going all the way back to 1950. The dataset provides information on a wide range of climate variables; initially preliminary, the ERA5 Back Extension was separated from the main ERA5 dataset, with plans to relaunch at the end of 2021, and, in the future, extend it back even further.
Last but not least, C3S supported the response to the COVID pandemic by quickly developing an application allowing scientists to study how temperature and humidity conditions affect the spread of COVID-19.
Copernicus Emergency Management Service
ECMWF operates the computational centre for both the European Flood Awareness System (EFAS) and the Global Flood Awareness System (GloFAS). Highlights in 2020 included a major upgrade of the European system, multiple resolution global land surface maps, new forecast skill layers for both EFAS and GloFAS, and a thorough investigation of global hydrological simulations and work towards improving hydrological representation in ECMWF’s Integrated Forecasting System (IFS).
In the summer, GloFAS forecasts helped to mitigate the impacts of monsoon floods in Bangladesh. Over five million people were affected, there were 41 casualties and tens of thousands of people were evacuated with their livestock. The flood forecasts delivered with a ten-day lead-time gave local authorities the opportunity to warn vulnerable communities.
ECMWF continues to contribute to scientific improvements for the CEMS Early Warning Systems Fire component.
A new AI product estimating the probability of ignition from lightning, the first release of a global fire weather index reanalysis dataset based on ERA5 in the CDS, and a first attempt to perform calculation for the fire weather index at different times of the day were particular highlights in 2020.
CEMS-Fire has helped the collaboration between ECMWF and some Member States (Météo-France, IPMA, and AEMET) in activities related to fire danger control in Europe under the umbrella of the ARISTOTLE-ENHSP partnership.
A kick-off meeting was hosted at ECMWF in February 2020 to design a 24/7 European multi-hazard virtual advice service for natural disasters, to support the work of the EU’s Emergency Response Coordination Centre (ERCC).
As a result, in addition to providing fire-related meteorological data such as drought indices, CEMS-Fire made the ecCharts interactive web chart platform available to display new products developed by Member States to support fire management in their own countries.
*Figures based on average distances and modes of transportation, not taking into account that some of those travelling from outside Europe would likely combine several meetings in the same trip.