Newsletter-banner-No-153

OpenIFS users explore atmospheric predictability

Glenn Carver

 

The fourth OpenIFS user workshop, held in June 2017 at the International Centre for Theoretical Physics (ICTP) in Trieste, Italy, was devoted to ‘Atmospheric Variability: seasonal predictability and teleconnections’. It attracted scientists from institutes in Europe and further afield and was undoubtedly the most successful OpenIFS workshop to date.

OpenIFS experiments
%3Cstrong%3EOpenIFS%20experiments.%3C/strong%3E%20Participants%20presented%20and%20discussed%20their%20OpenIFS%20seasonal%20forecast%20experiments%20with%20different%20sea-surface%20temperatures.%20(%3Cem%3EPhoto:%20Filip%20V%C3%A1%C5%88a%3C/em%3E)
OpenIFS experiments. Participants presented and discussed their OpenIFS seasonal forecast experiments with different sea-surface temperatures. (Photo: Filip Váňa)

The OpenIFS activity at ECMWF provides a supported version of the operational Integrated Forecasting System (IFS) for research and education. User workshops are an opportunity for scientists to interact, present their work with OpenIFS and learn more about ECMWF. Each meeting focuses on an active research area at ECMWF. This was the first such event to focus on seasonal forecasting.

Italian connection

The choice of location reflected increasing interest in OpenIFS from Italian research groups. ICTP was chosen for its excellent facilities and international reputation, particularly for seasonal prediction. The workshop took place over five days instead of three for previous meetings, as feedback indicated participants wanted more time for hands-on activities with the model.

Fifty participants ranging from young to senior scientists took part, twice as many as in previous OpenIFS workshops. For the first time, the workshop was also significantly oversubscribed with over ninety applicants. The meeting was opened by two keynote presentations from ECMWF. Professor Erland Källén, then Director of Research, spoke about ‘Research and development at ECMWF’. He was followed by Dr Franco Molteni on ‘Experimentation on extended-range prediction and multi-year variability at ECMWF’. Both talks were well received with many questions.

Each morning was a combination of invited and contributed presentations. With eleven invited speakers, nine contributing speakers and twenty-one poster presenters, the scientific standard was high and engaging. A range of topics on the workshop theme were presented, some based on non-ECMWF models such as the European Earth system model EC-Earth and the SPEEDY global circulation model developed at ICTP. Some speakers gave examples of using OpenIFS for research on other topics and teaching at universities. The afternoons were spent first learning about OpenIFS and then running practical experiments in teams. The last day of the workshop was devoted to team presentations and discussion of the results.

El Niño 2015/16 experiments

El Niño is a term used to describe a significant warming of the sea-surface temperature (SST) mainly in the central and eastern Pacific, which occurs irregularly every few years. This interacts strongly with the atmosphere and has consequences for global weather patterns. The strength and area of each El Niño varies, but the event of winter 2015/2016 was one of the strongest on record.

Franco Molteni designed an experiment with OpenIFS based on the 2015/16 El Niño. The OpenIFS model was run in two experiments, each with a 10-member ensemble from November 2015 to the end of April 2016; one used observed SST values, the other climatological values, a 20-year mean SST from ERA-Interim. These forecasts were provided to participants to explore the impact of El Niño by comparing a seasonal forecast with SST anomalies to one without. Ten teams of participants were then asked to design experiments in which the observed SST anomaly over an oceanic area of their choice was changed to either the ERA-Interim climatology (removing the anomaly) or to some multiple of the difference (for example doubling the anomaly).

Using Metview tools developed by ECMWF’s Sándor Kertész, participants were able to alter the SST field used in these forecasts to create their own experiments and ask questions such as: what would the change in European weather be if we removed or increased the Pacific El Niño anomaly? By using ensembles of forecasts, some measure of significance could be determined for these changes.

Seasonal forecast experiment with OpenIFS
%3Cstrong%3ESeasonal%20forecast%20experiment%20with%20OpenIFS%3C/strong%3E.%20The%20charts%20show%20the%20effect%20of%20combining%20modifications%20to%20the%20sea-surface%20temperatures%20(SST)%20in%20the%20tropical%20Pacific%20and%20the%20Barents/Kara%20Sea.%20Three%20separate%2010-member%20seasonal%20forecasts%20were%20made%20in%20which%20the%20warm%20anomaly%20in%20the%20SST%20was%20replaced%20by%20a%20cold%20anomaly%20throughout%20the%20forecast.%20The%20top%20left%20panel%20shows%20the%20SST%20changes%20that%20were%20applied%20to%20the%20observed%20SST%20field.%20The%20top%20right%20panel%20shows%20the%202-metre%20temperature%20monthly%20mean%20difference%20between%20the%20experiment%20with%20the%20modified%20SST%20applied%20just%20over%20the%20tropical%20Pacific%20and%20the%20unmodified,%20observed%20SST;%20the%20bottom%20left%20panel%20shows%20the%20same%20but%20with%20the%20modified%20SST%20applied%20just%20over%20the%20Barents/Kara%20Sea%20region;%20and%20the%20bottom%20right%20panel%20shows%20the%20same%20but%20with%20the%20modified%20SST%20applied%20in%20both%20the%20Pacific%20and%20the%20Barents/Kara%20Sea.%20It%20can%20be%20seen%20that%20the%20colder%20SST%20combines%20to%20give%20increased%20warming%20over%20Eurasia.%20All%20figures%20are%20monthly%20means%20for%20January%202016%20from%20forecasts%20from%20the%20beginning%20of%20November%202015.%20(%3Cem%3ECharts:%20Tido%20Semmler%20and%20Team%203%3C/em%3E)
Seasonal forecast experiment with OpenIFS. The charts show the effect of combining modifications to the sea-surface temperatures (SST) in the tropical Pacific and the Barents/Kara Sea. Three separate 10-member seasonal forecasts were made in which the warm anomaly in the SST was replaced by a cold anomaly throughout the forecast. The top left panel shows the SST changes that were applied to the observed SST field. The top right panel shows the 2-metre temperature monthly mean difference between the experiment with the modified SST applied just over the tropical Pacific and the unmodified, observed SST; the bottom left panel shows the same but with the modified SST applied just over the Barents/Kara Sea region; and the bottom right panel shows the same but with the modified SST applied in both the Pacific and the Barents/Kara Sea. It can be seen that the colder SST combines to give increased warming over Eurasia. All figures are monthly means for January 2016 from forecasts from the beginning of November 2015. (Charts: Tido Semmler and Team 3)

These ensemble seasonal experiments were run from ICTP on the ECMWF high-performance computing facility (HPCF), the first time an OpenIFS workshop has made use of the HPCF remotely. Metview was used to compute monthly means from the model results, which were then transferred to the ICTP classroom for plotting using Metview. Participants were able to look at the statistical significance of the changes to the SST used in their experiments. Some teams developed their own diagnostic tools and applied them to the results, often working late into the evening!

There was a range of interesting experiments. Some teams chose to alter the tropical Pacific SST anomaly either over large areas, to look at teleconnections over Europe, or over smaller Pacific coastal areas for local impacts. Others focused on the role of the Indian Ocean, whilst two teams looked at the impact of changes to the Arctic sea-ice cover. The tools provided allowed the SST values to be altered in multiple areas, allowing for some interesting experiments looking at the interaction of different anomalies. All teams gave excellent presentations on their findings, resulting in scientifically interesting discussions on the interpretation of the results, with perhaps a few questions remaining too.

Plans for the future

The sensitivity experiments using the El Niño of 2015/16 yielded some interesting results and a follow-up publication is planned. Participants said they found the workshop informative and interesting and the opportunity to meet ECMWF scientists was much appreciated. Some participants would have liked to learn more about the model itself. However, fitting additional lectures into a very full programme would have been difficult given the large number of high-quality invited and contributed presentations. This is something to consider for future workshops.

OpenIFS workshops are important to the growing user community and will continue to be hosted at different European institutes. They also contribute to raising awareness of ECMWF and its ongoing research programme. The presentations highlighted the substantial research and development now under way using the OpenIFS model.

The workshop was organised jointly by: ICTP, ECMWF, CETEMPS University of L’Aquila, and ISAC-CNR Bologna. We gratefully acknowledge financial support from the EU ESiWACE programme, CETEMPS, ICTP and a Young Scientist Travel Award from the European Meteorological Society, which was won by Lenka Novakova from the University of Reading. The workshop would not have been such a success without the contribution and support from co-organisers Fred Kucharski (ICTP), Paolo Ruggieri (CETEMPS L’Aquila) and Susanna Corti (ISAC-CNR Bologna) as well as a number of ECMWF colleagues, particularly: Sándor Kertész for preparing the Metview tools; Filip Váňa for OpenIFS support; Franco Molteni, who suggested the practical experiment; Tim Stockdale, who provided the IFS experiment; Erland Källén; Sarah Keeley and Iain Russell. We look forward to the next user meeting in 2019.

Group photo. Fifty scientists took part in the workshop. (Photo: Filip Váňa)
%3Cstrong%3EGroup%20photo.%3C/strong%3E%20Fifty%20scientists%20took%20part%20in%20the%20workshop.%20(%3Cem%3EPhoto:%20Filip%20V%C3%A1%C5%88a%3C/em%3E)
Group photo. Fifty scientists took part in the workshop. (Photo: Filip Váňa)