The 2015 edition of the ECMWF Annual Seminar, entitled 'Physical processes in present and future large-scale models', took place from 1 to 4 September. More than 80 scientists from around the world attended it.
Both early-career scientists and experts benefited from a wide range of presentations on the modelling of physical processes from leading scientists at ECMWF and internationally.
They also enjoyed the opportunity to discuss recent science developments in the field and how Earth system models can benefit from them. Special emphasis was put on the representation of physical processes in a modelling environment where resolution keeps increasing and where processes related to convection and orography are partially resolved.
The seminar was organized into sessions on Radiation, Convection, Clouds, Gravity waves, Boundary layer, Land surface, Physics in data assimilation, and Verification/Uncertainty. The sessions typically included three to four 45-minute talks covering fundamental aspects, process representation in large-scale models, and evaluation.
In the session on radiation it was shown that current codes are very accurate, but too expensive to run at full temporal and spatial resolution. The uncertainty is dominated by the input from clouds, aerosols and precipitation. The development of efficient codes to deal with high resolution on massively parallel computer architectures and accelerator hardware is an active area of research.
Convection is one of the processes that is already partially resolved in limited-area models, and will be increasingly resolved in future global models. The presentations covered a new way of modelling convection with so-called super-parametrization, the interaction of convection with atmospheric waves and the handling of convection at very high resolution.
Although a lot of progress has been made, the representation of clouds in large-scale models still comes with large uncertainty and remains an area of active research. In the presentations it was emphasized that both the representation of sub-grid heterogeneity and microphysics require a unified approach involving dynamical, boundary layer and convective processes.
Scientists from 17 countries attended the event at ECMWF in Reading, UK.
Gravity waves are a prominent feature of the atmosphere and affect both the troposphere and the stratosphere. Their impact was demonstrated from an observational and modelling perspective. Although the representation of gravity waves in large-scale models is still surrounded by many questions, it is expected that, in future, schemes can be better constrained by observations and results from high-resolution models.
The sessions on the boundary layer and land surface illustrated the interactive nature of the processes. There are strong feedbacks between precipitation and evaporation, and between turbulent heat transport in the atmosphere and heat transfer in the soil/snow. The drag exerted by the surface on the atmosphere also has a strong impact on the atmosphere. Research on extremes has shown that a dry anomaly in the soil is a necessary (but not sufficient) condition for a heat wave, and that there is predictability on the sub-seasonal timescale which is probably not fully exploited yet by state-of-the-art models. Another aspect of the boundary layer is the simulation of low clouds, which has strong implications for forecasting at all time ranges, and particularly for climate change.
The seminar concluded with sessions on data assimilation, uncertainty and verification. There is an obvious synergy between verification and assimilation, because model errors can often be isolated through the systematic confrontation of the model with observations in the data assimilation system. This is increasingly the case with modern satellite observations that are sensitive to hydrometeors. Examples were shown of systematic errors that have an impact on the feedbacks relevant for El Niño forecasts. Talks were given on the estimation and representation of model error in ensemble systems and the modelling of meteorological processes in the transport of tracers. Finally, the application of linear models in 4-dimensional variational (4DVAR) data assimilation was presented. The physical processes were shown to be important, but the limitations were illustrated, too. For very high resolution it may be necessary to consider alternative data assimilation techniques.
ECMWF Director of Research Erland Källén closed the Seminar, noting that remarkable progress had been made in the modelling of physical processes in the atmosphere, and that exciting prospects were offered at the meeting for the coming years.
Andreas Dörnbrack from the Institute of Atmospheric Physics of the German Aerospace Centre and Gunilla Svensson from Stockholm University are two of the lecturers who gave presentations at the seminar.
Andreas Dörnbrack uses ECMWF products when planning and conducting aircraft field campaigns to collect observations. His presentation on orographic gravity waves included lessons learnt from a recent field campaign in New Zealand and the relevance of the observations for NWP and ECMWF in particular.
Most speakers talked about recent work - Andreas Dörnbrack
He saw the Seminar as a valuable opportunity to get an overview of different topics. “Most speakers are talking about recent work,” he said. “Atmospheric physics is an interdisciplinary field. The Seminar would appeal to anyone working in the field and it gives people pointers of where to look if they want to know more.”
Gunilla Svensson, who presented work on Arctic boundary layers, echoed this sentiment. For her, the most important part of the Seminar is learning about developments in a broad range of fields. “If you’re working on model development, on one part of it, it’s important to know what’s going on in the others because everything is related.”
The Annual Seminar has always been popular with early-career scientists. Both Gunilla and Andreas felt it was important to pitch information at the right level for younger scientists to help them develop their careers. They praised the pedagogic nature of several presentations, adding there was plenty of useful information for more senior scientists as well.
The Seminar is a good event for bringing the forecasting and climate communities together - Gunilla Svensson
Gunilla’s presentation was on the challenges of describing the processes that occur in the Arctic, specifically in the Arctic boundary layers, and their importance for the interaction with the surface. The aim of her work is to improve models for both climate and weather forecasting. “It’s important that the weather forecasting and climate communities become closer,” she said. “The Seminar is a good event for bringing these communities together. They can both learn from each other and it’s important to have good collaboration on both sides.”
Presentations
Download the Seminar 2015 presentations.