5th Workshop on Waves and Wave-Coupled Processes
What: | More than 70 experts from research and operational centers came together at the meeting organized by and at the European Centre for Medium-Range Weather Forecasts (ECMWF) to discuss the physics and the numerical solutions of coupled atmosphere–wave–ocean modeling at both short- and long-term scales. |
When: | 10–12 April 2024 |
Where: | European Centre for Medium-Range Weather Forecasts, Reading, United Kingdom |
1. Introduction
ECMWF (https://ecmwfevents.com/i/5th-workshop-on-waves-and-wave-coupled-processes/public/agend) held the fifth workshop on ocean waves and wave-coupled processes in Reading (United Kingdom) from 10 to 12 April 2024. It was attended by 80 scientists from across the world and followed workshops in Melbourne (2016), Qingdao (2017), Hangzhou (2018), and Uppsala (2023).
Ocean surface waves play a critical role in the Earth system, modulating many surface exchanges and acting in both atmospheric and oceanic boundary layers.
Accounting for their impact in weather and climate systems has recently attracted renewed interest, but the modeling, analysis, and observation of ocean waves and wave-coupled processes still face many challenges.
Forty-one presentations were given at the event, including eight keynote ones. Oral and poster contributions supplemented the talks, all of which are available on the ECMWF website.
2. Framing the situation
Time and location were well chosen as ECMWF is looking forward to framing their new developments. On a wider scale, the general wave community is also assessing the present situation and deciding where to focus their efforts over the coming years. Indeed, after the substantial developments of the 1980s and 1990s, the following decades have been mainly a period of refinement on different aspects with direct application of the available phase-averaged models, as the wave action models WAM, WAVEWATCH III (WW3), and SWAN. It is also fair to say that part of the improvements in these last 20–30 years was and is due to advances in the underlying meteorological models, i.e., the input wind fields to the wave models.
About the wave model physics, in particular about the three fundamental processes: (i) wind input, (ii) nonlinear interactions (NL), and (iii) dissipation due to wave breaking, we have an accepted theory for (i). We know perfectly (ii), but we are forced to use a simplified, less computer demanding, version of it (with substantial approximations). In practice, have no accepted theory of (iii), and we rely on parameterizations of the process. This last point will turn out to be a crucial limitation for the deeply coupled models we are aiming for.
During the last three decades, the counterpart of the operational wave models has been the numerical weather prediction (NWP) models providing the required wind input information. A partial, but significant, coupling was achieved with an atmospheric–sea surface momentum exchange depending on the stage of development of the local wave field (Janssen 2004). NWP systems have also moved toward a more Earth system approach adding an active coupling with an ocean circulation component. Thereto, ocean waves can play an active role in the exchanges between atmosphere and ocean. With the rapidly growing effort at climate modeling, either seasonal or long term, the time scale of the forecasts increases by one (seasonal) or two or three orders of magnitude. The physics of the wave modulated atmosphere–ocean exchanges needs to be improved accordingly or modified. These two alternative possibilities are explored in the next two sections.
3. The need for better physics
Granted the improved quality of the input wind fields, the general feeling is that the present errors in wave modeling are associated to a not sufficiently accurate description of the processes at the air–sea interface. Although not explicitly stated, but emerging from side discussions, the present concept of a clean logarithmic wind profile feeding energy into each spectral component, fully independently of the whole situation, although working, appears as an abstract idealization. For whoever has witnessed a stormy sea, with a foaming surface, lots of breakers, and consequent boundary layer detachments, it is obvious that the situation is much more complicated. This is connected to the spray production, hence to the energy, mass, and momentum transfers between ocean and atmosphere, whose parameterization is presently based mostly on wind speed only. A similar argument holds for the extremely large number of, great and small, bubbles that characterize the first few meters below the surface of a stormy sea. Moreover, Langmuir circulation can take the bubbles to greater depths.
Indeed, much of the push toward an improved physical quantification of the atmosphere–ocean exchanges at a stormy sea surface derives from the major effects these exchanges have when working especially on long time scales, up to climate ones (but see in this respect the next section). Whichever the time scale, the key point is that surface exchanges increase by one or two orders of magnitude in rough seas. This implies that the 15%–20% of the ocean that is on average in stormy conditions at any one time is actually controlling the bulk of the long-term air–sea exchanges. In the relatively short scale of weather forecasting, a better evaluation of these exchanges is also crucial for the correct evaluation of tropical cyclones.
Still on the physical side, and concerning strictly wave modeling, an open question is the correct evaluation of nonlinear interactions. Contrarily to wind input and white-capping dissipation, here the theory is well known, but the computer power is not available. This is also because whenever this turns out to be the case, the extra computing power is used to increase resolution, as it continues to be a much more effective use of it. It must also be said that no extensive test has been done showing better final results using the full nonlinear correct expression, yet studies have shown that aspects of the full spectral distribution of the wave field can potentially be improved with a more exact SNL, opening the door for the inclusion of waves related physics in many more processes (Alday and Ardhuin 2023).
A further, rather substantial, problem is the transfer of the eventually improved physics, particularly about atmosphere–ocean exchanges, to the climatological community. This takes us directly to the subject of the next session.
4. The need for simplified physics and models
While there is a general consensus that improved physics, possibly also added complexity, in the ocean wave component of Earth system models used in NWP will also benefit the short-term meteorological forecasts, caution is felt about their application in climate models. Given the time scale, number, and complexity of the long-term processes that must be captured by coupled general circulation models, it is natural to wonder about the use of current third-generation wave models in this mix. True, the correct, or also only improved fluxes at the interface are crucial for long-term climate modeling. However, the key question is whether a full 2D spectrum, of the order of 103 frequency-direction components at each grid point, is required for a correct evaluation of the fluxes. Indeed, there is a general feeling that to move toward climate applications, we need to simplify our present models, although how and how much are not at all clear. Feeling and reasoning suggest that a suitable solution could be somewhere in the middle, between second and third generation models, although this is still to be fully defined.
If our focus is the coupling with climate models, our main concern is a correct estimate of the fluxes at the interface. As for computer resources, at ECMWF, the ecWAM presently requires 7% of the overall daily forecast IFS time. This is evenly split among physics, nonlinear interactions and advection. Please remember that the 2.3% of NL is for the highly simplified solution, while the correct one would be heavier by at least one to two orders of magnitude. This is an area where machine learning could potentially offer much faster and sufficiently accurate solutions for the purpose of climate applications. This does not exclude, mutatis mutandis, similar applications also for the daily forecasts, particularly, we expect, for the NL.
5. Independent physical knowledge
Remarkably improved physics in forecasts and climate modeling cannot be expected to derive only from theoretical contributions. Obviously, field data are required, but more advanced and possibly sophisticated than the routine data we daily receive from buoys and satellites. Rather than putting an instrument into the sea or in the above air, and collecting data, we need to slow down for a moment to frame clearly what we need and would like to have. Then, suitable experiments (better if multiple for later cross comparison of the data and derived physics) are specified to be carried out in the sea. This will not be easy, also because the right conditions are to be met in time and at the measurement locations. In any case, this will not be a quick task. Apart from the difficulties of open-sea measurements, new data, especially if taken in not previously similarly explored conditions, typically bring with them new questions and the need for further action. A key point is that it is not at all clear which is the smallest scale we need to deal with to achieve the required physical knowledge and practical formulations. This is an open question, and no definite answer is in sight.
6. Data
The volume of data available and continuously collected is growing at an almost exponential rate. Satellites keep beaming down terabytes of data. Hundreds of instruments into the sea offer a steady flow of information. Although not only about physical parameters, the oceanographic tower of the CNR Institute of Marine Sciences, in the Northern Adriatic Sea, has about 50 instruments actively recording 24 h a day. Other similar platforms with multiple instruments exist around the globe, e.g., the Ekofisk platforms in the central North Sea. To derive meaningful conclusions for our present purposes from this enormous availability a strong, organized selection is required. As already mentioned, also the deeper look into the very details of the processes of interest will provide large sets of specialize data. Rather than a distributed collection of possibly different data from sparse different sources, it would probably be better, if not necessary, to focus on a small number of “hot spots” where ideally all the relevant parameters are continuously measured for later, also distributed, analysis. Obviously, this not only cannot be the work of the single researchers but also not the one of n independent researchers. A collective organized effort is required.
7. Implementation
In discussing and framing what to do and where to aim to, we cannot forget that at the ECMWF meeting, we were mainly a wave community, although hosted at a very favorable place. Indeed, besides being at the forefront of the meteorological forecast, it was here that already 30+ years ago the first coupling (Janssen 2004) between NWP and wave models took place. With different quantifications and improved theory, we can still move in this direction. However, a climatological community is probably harder to convince. With focus on a much wider variety of physical processes, the ones at the atmosphere–ocean interface have been parameterized to the best of the present knowledge, but not yet with a deep look at the single local processes. However, we now know that such a deeper look at the interface may provide insights about fluxes that can lead to new parameterizations, different from the ones presently used in the climate community. The question is how and with which compromises we can bring the two communities together. One action has been alluded to in section 3: to produce a simplified, but still fit for the purpose, wave model. This is something to be done interacting with the climate community to see from the start how best to act for a possible merging of the two disciplines.
8. A final general comment
The task ahead is not an easy one. Of course, we talk about joining efforts with the climate community. For the daily weather forecasts, the direction is clear, although still with physics to be defined.
About climate and contribution from the wave community, given the scale of the effort the question concerns the best way to approach the problem. On the one hand, we could proceed as most of us usually do, i.e., working on our own or in small groups, possibly with external cooperation within a large project. However, the scale of the efforts required may suggest that a large-scale organized effort is the way to go. On a smaller scale, the development of the WAM (Komen et al. 1994) was the product of a small and dedicated community with a strong leadership. At the opposite end of the scale, the relevance of climate for the human population and the Earth environment is such that a parallel with the Manhattan project (United States, 1942) may not be fully out of scale. Ending with a more humanistic parallel, in this respect, the research community should act as a well-conducted orchestra. Each group of instruments knows their part, and it is only through the collective effort that the best performance is achieved. However, this can be only under an enlightened and well-determined guide.
References
Alday, M., and F. Ardhuin, 2023: On consistent parameterizations for both dominant wind‐waves and spectral tail directionality. J. Geophys. Res. Oceans, 128, e2022JC019581, https://doi.org/10.1029/2022JC019581.
Janssen, P., 2004: The Interaction of Ocean Waves and Wind. Cambridge University Press, 300 pp., https://doi.org/10.1017/CBO9780511525018.
Komen, G. J., L. Cavaleri, M. Donelan, K. Hasselmann, S. Hasselmann, and P. A. E. M. Janssen, 1994: Dynamics and Modelling of Ocean Waves. Cambridge University Press, 532 pp.