ON THE EDGE OF A CLOUD On the edge of a cloud Proefschrift ter verkrijging van de graad van doctor aan de Technische Universiteit Delft, op gezag van de Rector Magnificus prof. dr. ir. J. T. Fokkema, voorzitter van het College voor Promoties, in het openbaar te verdedigen op dinsdag 9 december 2008 om 15.00 uur door Thijs HEUS doctorandus in de experimentele natuurkunde en in de meteorologie en fysische oceanografie geboren te Utrecht Dit proefschrift is goedgekeurd door de promotor: Prof. dr. ir. H. E. A. van den Akker en de copromotor: Dr. H. J. J. Jonker Samenstelling promotiecomissie: Rector Magnificus Voorzitter Prof. dr. ir. H. E. A. van den Akker Technische Universiteit Delft, promotor Dr. H. J. J. Jonker Technische Universiteit Delft, copromotor Prof. dr. P. H. Austin University of British Colombia Prof. dr. A. A. M. Holtslag Wageningen Universiteit Prof. dr. A. P. Siebesma Technische Universiteit Delft Prof. dr. B. Stevens Max-Planck-Institut fur¨ Meteorologie Prof. dr. P. P. Sullivan National Center for Atmospheric Research Dit werk is financieel gesteund door de Nederlandse Organisatie voor Wetenschap- pelijk Onderzoek (NWO). Voor het gebruik van supercomputer-faciliteiten bij dit onderzoek is subsidie verleend door de Stichting Nationale Computer Faciliteiten (NCF). Copyright c 2008 hoofdstuk 3 en 5, en figuur 1.2 American Meteorological Society ° Copyright c 2008 hoofdstuk 4 Royal Meteorological Society ° Copyright c 2008 voor het overige Thijs Heus ° Uitgegeven door Grafisch Bedrijf Ponsen & Looijen BV ISBN: 978-90-6464-301-9 PROPOSITIONS Belonging to the PhD thesis On the edge of a cloud by Thijs Heus 1 Parameterizing cumulus clouds without taking the role of the subsiding shell into account, may only be successful due to a prefactor or with luck. 2 The latent heat release of condensation, in combination with the dry, warm environment turns clouds into one of the clearest examples in nature of the unsteady puff convection as described by Hunt et al. (J. Fluid Mech., 2003). 3 The only spectral gap existing in the atmospheric sciences is between textbooks and the scientific state-of-the-art. 4 The h-index would be more valuable if it were calculated over papers published in the last 10 years only. 5 The average scientific age of a supervisor, roughly 20 years, sets a time scale for rediscovering scientific results from earlier days. 6 Clouds and washing machines show that virtual potential temperature and socks are non-conserved quantities. 7 The beauty of scientific studies cannot be explained but must be experienced. 8 Both climate research and evolution theory grew up in public, resulting in a persistent belief in their teething troubles. 9 Universities would find more and better PhD candidates if they would advertise the creativity and the diversity of methods involved in science, rather than focus on the limited scope of a PhD project. 10 Referenda imply a disqualification of politicians for their job, and justify salaries in the public sector that are higher than the salary of the prime minister. 11 The position of head of the multi-scale physics department at TU Delft is an underpaid job. These propositions are considered opposable and defendable and as such have been approved by the supervisor, prof. dr. ir. H. E. A. Van den Akker. STELLINGEN Behorende bij het proefschrift Over de rand van een wolk door Thijs Heus 1 Zonder rekening te houden met de neergaande ring kan een model van cumu- luswolken alleen maar slagen dankzij een voorfactor of dankzij mazzel. 2 De latente warmte die vrijkomt bij condensatie, in combinatie met de warme, droge omgeving, zorgt ervoor dat wolken e´en´ van de duidelijkste voorbeelden zijn van unstable puff convection zoals beschreven door Hunt et al. (J. Fluid Mech., 2003). 3 Het enige spectrale gat in de atmosferische wetenschappen ligt tussen studie- boeken en het front van de wetenschap. 4 De h-index zou zinniger worden door hem slechts te berekenen over de artikelen die de afgelopen 10 jaar gepubliceerd zijn. 5 De gemiddelde wetenschappelijke leeftijd van een begeleider, ruwweg 20 jaar, geeft een tijdschaal waarop ouder wetenschappelijk werk herondekt wordt. 6 Wolken en wasmachines laten zien dat de virtuele potentiele¨ temperatuur en sokken geen behouden grootheden zijn. 7 De schoonheid van wetenschappelijk onderzoek kan niet worden uitgelegd, maar moet worden ondergaan. 8 Het probleem van zowel de evolutieleer als klimaatonderzoek is dat beide en plein public zijn opgegroeid; dit resulteert in een hardnekkig geloof in hun kinderziektes. 9 Universiteiten zouden meer en betere promovendi vinden als ze de creativi- teit en verscheidenheid in de wetenschap zouden aanprijzen, in plaats van de beperkte breedte van een promotietraject. 10 Referenda impliceren een diskwalificatie van politici voor hun werk, en geven dus een rechtvaardiging voor salarissen van topbestuurders hoger dan dat van de minister-president. 11 De afdelingsvoorzitter van multi-scale physics aan de TU Delft wordt onderbe- taald. Deze stellingen worden opponeerbaar en verdedigbaar geacht en zijn als zodanig goedge- keurd door de promotor, prof. dr. ir. H. E. A. van den Akker. Summary in English Driving home, the sky accelerates And the clouds all form a geometric shape THE FLAMING LIPS Cumulus clouds have since long been one of the greatest challenges in the atmo- spheric sciences. For a correct representation of clouds in weather and climate models, where they are the biggest unknowns, a good understanding of the inter- action between cloud and environment is of prime importance. In this thesis, this problem is attacked with a combination of detailed large-eddy simulations (LES) and air-plane observations. While the traditional view states that air inside the cloud goes up, and all the air outside goes down in compensation, it is found here that on average, the air far away from the cloud hardly moves. There, air only moves in oscillating gravity waves or in small scale turbulence. Most of the compensating downward motion happens in the direct vicinity of the cloud, in a subsiding shell. The subsiding shell can be clearly observed in various case studies, in LES as well as in airplane observations. To appreciate the subsiding shell as more than just an area with slightly negative velocity, it is important not to look at the vertical velocity (as is intuitive in air-plane observations), but to look at the vertical mass flux. This way, the relatively large area of the shell compared to the cloud core can be taken into account. The origin of this subsiding shell is examined by analyzing the individual terms of the vertical momentum equation. Buoyancy is found to be the driving force for this shell, and it is counteracted by the pressure gradient force. This shows that evaporative cooling at the cloud edge, induced by lateral mixing of cloudy and environmental air, is the responsible mechanism behind the descending shell. The role of the shell in cloud-environment interaction is further explored and described in a conceptual three-layer model of the cloud. In the end, the shell creates a buffer layer that interferes with the interaction between cloud and environment, although vertical shear is able to skew the cloud, hence creating a preferred location for the subsiding shell. A related classical problem in cumulus cloud research is the determination of the source of in-cloud air. Here, this problem is studied by two distinct methods: 1) by analyzing conserved variable mixing diagrams (Paluch diagrams), and 2) by tracing back cloud-air parcels represented by massless Lagrangian particles that follow the LES flow. The obtained Paluch diagrams are found to be similar to many results found in literature, but the source of entrained air found by particle tracking deviates from the source inferred from the Paluch analysis, even more than can be explained by buoyancy sorting. Whereas the classical Paluch analysis x Summary seems to provide some evidence for cloud-top mixing, backwards particle tracking shows that virtually all mixing occurs laterally. Particle trajectories averaged over the entire cloud ensemble also clearly indicate the absence of significant cloud-top mixing in shallow cumulus clouds. To study the evolution of clouds in time as well as in space, a statistical life-cycle analysis has been conducted. Although trained observers have no problem in dis- tinguishing between the various lifestages of a cloud, this process proves difficult to automate, since cloud-splitting and cloud-merging events complicate the distinction between a single system consisting of several cloudy parts and two independent systems that collided. Because human perception is well-equipped to capture and to make sense of these time-dependent three-dimensional features, a combination of automated constraints and human inspection in a 3D virtual reality environment is used to select clouds that are exemplary in their behavior throughout their entire lifespan. The considerable number of selected clouds warrant reliable statistics of cloud properties conditioned on the phase in their life cycle. The most dominant feature in this statistical life-cycle analysis is the pulsating growth that is present throughout the entire life time of the cloud. The pulses are a self-sustained phenomenon, driven by a balance between buoyancy and horizontal convergence of dry air. The convective inhibition just above cloud base plays a crucial role as a barrier for the cloud to overcome during its infancy stage, and as a buffer region later on, ensuring a steady supply of buoyancy into the cloud. Samenvatting in het Nederlands In de auto, van de week zag ik een wolk die op Arafat leek SPINVIS Mooiweerwolken behoren sinds jaar en dag tot de grootste uitdagingen in de at- mosferische wetenschappen. Voor een goede representatie van wolken weer- en klimaatmodellen, waar wolken een grote onzekerheid zijn, is een goed begrip van de interactie tussen wolk en omgeving van belang.