By Bob Plant
I’ve been asked to write a blog post to go live on 17 June, the opening day of the 2019 OpenIFS user workshop. As I’m involved in the organisation, it would almost seem strange not to talk a little about that.
The IFS (Integrated Forecasting System), is the modelling system developed and used at the ECMWF, and which underlies all of their forecasting, data assimilation and reanalysis activity. Brief outlines can be found here for the dynamics and here for the physics. The OpenIFS version is designed to be used outside of the centre. This allows universities to collaborate more easily with the ECMWF on research projects and supports more teaching-focussed activities.
Students hear a great deal about weather and climate modelling during their studies but have traditionally had little or no opportunity to work directly with the models. Even those whose main interests do not lie in numerical modelling will inevitably rely on modelling results, or will want to analyse model data. So some hands-on modelling experience is valuable, just as those of us who take a more theoretical or model-based perspective nonetheless benefit from being exposed to real experimental data. It’s important that the models should not be looked upon as black boxes that magically generate data, but that students get the opportunity to take out a torch and at least have a bit of a look around in the murky interior.
At the same time, there are obvious practical issues with using full-scale operational-type models in a classroom context. We often look for substantial high-performance computing for model-based research projects and expect to submit jobs that return results after some hours, or perhaps days. Also, while a model might be very nicely designed for the operational or expert research context, it may not be easy for a non-expert to pick up and get started with quickly. The OpenIFS provides a pretty good balance: it is relatively easy to use, but not so easy as to encourage a black-box syndrome.
I was keen to try out OpenIFS for teaching applications in the department, starting with an MSc dissertation project in summer 2015. While not totally plain-sailing, it was sufficiently encouraging to offer something for the MSc team project week in the following year, with Sue Gray and I each supervising a team so that we could help each other out with any teething issues. That worked well, and further team projects and dissertation projects have followed. There is more about those experiences in a short article in the ECMWF newsletter .
Getting back to this week’s workshop, it is a bi-annual event to introduce researchers from across Europe (and occasionally further afield) to the OpenIFS. We also have a scientific theme concerning the impact of moist processes on storm evolution, and there will be various talks and posters on this, alongside others relating to techniques and examples in using the model for research projects.
The key link between the modelling and the theme is our choice of case study. Storm Karl occurred in September 2016. It started out as a tropical system before undergoing an extratropical transition and ultimately produced much rain over Norway. It was observed as part of the NAWDEX (North Atlantic Waveguide and Downstream Impact Experiment) field campaign and there is an overview in this BAMS article. Apparently, it is the first system to undergo an extratropical transition to have been observed with research aircraft at each stage of its evolution, and so I would imagine that it will continue to be a focus of research over the next few years. The article highlights the importance of mid-level moisture, especially for the behaviour of the “warm conveyor belt” in the extratropical regime. Below are example plots from a preliminary OpenIFS simulation. There are also some very nice loops of the satellite imagery, and Met Office global model forecasts at this page, courtesy of Ben Harvey. We plan to perform a variety of modelling experiments and to interpret and understand our results by drawing on ideas from the talks and posters, and of course, plenty of discussions amongst the participants.
Example plots from a preliminary model run, for which thanks to Marcus Koehler. Left: 10m winds at T+42, 18UTC on 26 September. Karl is to the south-east of Greenland. Right: precipitation at the same time.
Numbers are limited for the hands-on computing part of the workshop, but if you are around in Reading and would like to come along to some interesting talks then feel free to join us in GU01 any morning from Tuesday to Friday. Or if you would like to talk about storms or modelling with 50-odd researchers also interested in such things, then again feel free – we’ll be in 1L61 for Tuesday to Friday morning coffee and over the lunch break. Our programme can be seen here.
I mustn’t forget to give credit where it is due. Under the small assumption that all is going to go wonderfully well, that will have been due to Glenn Carver, Gabi Szepszo and Marcus Koehler from ECMWF, and from the Reading side to Sue Gray and myself, Kathryn Boyd, Maria Broadbridge, Ben Harvey and Jake Bland. And finally thanks to our sponsors: we are funded by bringing together contributions from EGU, ESiWACE, the university environment theme, the department visitor fund and ECMWF.