 Clouds are important for the climate of our planet because they reflect short-wave radiation and they emit long-wave radiation into space and so it is really important to understand where clouds form and what types of clouds form. Now if we average over a long period, let's say a week and look at the earth then some features stand out. One is mean rising motion at the equator which is associated with deep convection and another one is slow subsidence of the equator which is associated with shallow convection. It is a common assumption that it is this slowly varying large-scale circulation also known as the Hadley cell that controls cloudiness. However, recent measurements show that reality is far more complex than the simple picture. These measurements were taken by my colleagues over the Atlantic and what they did was to fly an aircraft in circles of roughly 180 kilometers diameter while they were dropping instruments that measured the large-scale divergence and this divergence by mass conservation is directly related to the vertical velocity. And what they found was a surprisingly large variability both in the horizontal direction as well as in the vertical direction. Also the amplitudes were four times larger than what one would expect based on the budget analysis and at first they thought that this was noise, but they flew each circle twice and so one hour later they obtained almost the same result. This clearly showed that the measurements were not picking up noise. Rather they persisted these patterns and divergence persisted for more than one hour and given that one hour is quite long compared to the typical lifetime of clouds we think that these patterns and divergence are important. So in our study we are attempting two things. One is to confirm these patterns and divergence for a different location in a different data set and the second is to find a physical explanation for what drives this divergence. So when clouds form the condensational heating generates internal waves which are also known as gravity waves. These gravity waves are associated with vertical motion across a wide range of spatial scales anywhere from a couple of hundreds of meters to over a thousand kilometers. Our hypothesis is that these waves might explain the observed patterns in vertical velocity and divergence. In order to test this hypothesis we extract from observational data the divergence as well as information on the waves. These data were obtained during a field campaign many years back which took place near Darwin that's in the northwest of Australia and what was done during this campaign was that sands were released from the ground every three hours from five locations and these five locations were situated along the perimeter of a pentagon which had 200 kilometers diameter. So in this sense the experimental setup was similar to the circle flights because the region that was sampled was of similar size. This data set is really suitable for extracting information on waves because of the high frequency of the radius on launches and the long period of consecutive measurements which was over a week. In general it is really difficult to extract information on waves inside the troposphere because as the word says the troposphere is very noisy. In our case the analysis is successful because we combine the raw data which is profiles of horizontal winds with an additional data set and that is a variational analysis. What a variational analysis does is to apply small perturbations to the measurements of humidity, wind and temperature in order to conserve mass momentum energy and moisture within a certain domain. From this variational analysis we take the large-scale divergence and we take the background wind. This background wind profile is then subtracted from the individual soundings to get perturbations. The perturbation wind associated with low frequency gravity waves has a unique property and that is that it describes a spiral with height and you can think of the spiral very much as the DNA of gravity waves. So what we do is to determine in a statistical way the mean properties of the spiral and this in combination with a standard spectral analysis tells us the vertical wavelength, the horizontal wavelength and the periods of these waves. So the first question we asked was whether similar variability is found at Darwin as was reported over the Atlantic and the answer to that is yes. We find very similar variability in the data set that we analyzed. The second question is can we find a plausible physical explanation for this variability and we looked at whether or not gravity waves could serve as a plausible explanation. Now gravity waves are not a trivial solution to this problem because if they are an explanation this would mean that their periods would have to be consistent with the temporal autocorrelations in the divergence. The horizontal wavelength would have to match the spatial autocorrelations and the vertical wavelength would have to be the same as the typical scale of the variability in divergence in the vertical direction. At the same time however gravity waves obey an equation that is known as a dispersion relation and this equation relates the vertical wavelength, the horizontal wavelength and the period of the waves. The wave we extracted the typical wave that passed the domain has a horizontal wavelength of 600 kilometers, a vertical wavelength of 4 kilometers and a period of 12 hours. These match up really nicely with the scales of divergence and so we have now found a plausible explanation for what drives mesoscale variability. So clouds and their coupling to circulation are still a major uncertainty in climate projections. The waves we considered are large-scale and they're able to propagate far distances inside the troposphere. Now given that their sources, convection, is present most of the time somewhere in the tropics, we can expect these waves to be present in most locations most of the time and so if our hypothesis is true that it is these waves that force strong mesoscale vertical motion then this has some implications because it means that clouds are sources of waves that may in turn determine where clouds like to form and so in this way clouds are able to communicate with one another. This intertwinement between the cloud field and the wave field has some implications for the modeling of weather and climate. Course models with grid scales of let's say 100 kilometers are not able to resolve the sources of the waves, the waves themselves or the potential impact of the waves on convection. While I cannot say at this point how important this process really is or whether it would massively improve our models, I think it is worthwhile to find out. Most analysis techniques that deal with gravity waves have been designed for the middle atmosphere where there is much less noise than in the troposphere. So in order to continue our line of research, we need to constantly keep improving our methods. A promising aspect is that we have just entered the era of global cloud resolving simulations. These open up a new horizon for modeling weather and climate because the high resolution means that more processes are simulated explicitly and fewer need to be parametrized. This in turn means that fewer assumptions need to enter the model and the hope is that realism will increase. These high resolution simulations are starting to resolve many of the interactions that we are interested in and so on the one hand it means that we can use them to better understand these processes, but on the other hand we need to critically assess the models. Are they getting these processes right? And the only way to assess models is by confronting them with observations. Here we are very lucky to have a large field campaign in early 2020. Measuring area average vertical motion as well as extracting information on waves is a cornerstone of this campaign and so I hope that this will really help us make progress soon.