1 DISS. ETH NO Modeling of orographic cirrus clouds A dissertation submitted to the ETH ZURICH for the degree of Doctor of Sciences presented by HANNA JOOS Dipl. Met., Hamburg University, Germany born 11 December 1979 citizen of Germany accepted on the recommendation of Prof. Dr. U. Lohmann, examiner Dr. P. Spichtinger, co-examiner Dr. M. Giorgetta, co-examiner 2009
3 Contents Abstract Zusammenfassung v vii 1 Introduction Cirrus clouds Orographic gravity waves Orographic cirrus clouds Overview over this dissertation Orographic cirrus in the future climate Introduction Model description Model verification: Simulation of the INCA-case Idealized Simulations Results Simulations with IPCC initial profiles Model setup South America: linear flow regime South America: hydraulic jump North America: linear flow regime North America: hydraulic jump Summary and Conclusions Orographic cirrus in the global climate model ECHAM Introduction Parameterization of cirrus clouds: homogeneous freezing Calculation of the vertical velocity Linear theory for gravity waves Orography in ECHAM Results Global simulation Comparison with measurements Summary and Discussion Influence of a future climate on the microphysical and optical properties of orographic cirrus clouds in ECHAM Introduction Model description Moist Brunt-Väisäla frequency in the calculation of the vertical velocity Parameterization of an orographic cirrus cloud cover Reduction of the vertical velocity Results The current climate A future climate Summary and Conclusions iii
4 5 Summary and Outlook Summary Cloud resolving simulations Global simulations Outlook Calculation of the vertical wave phase Coupling of gravity waves and liquid/ mixed-phase clouds Parameterization of gravity waves from additional sources Cirrus cloud cover A The subgrid-scale orography in ECHAM5 75 List of Symbols and Abbreviations 81 List of Figures 85 List of Tables 89 Bibliography 91 iv
5 Abstract Cirrus clouds play a crucial role in modulating the earth radiation budget. They cover approximately 30% of the Earth and consist purely of ice crystals. On the one hand, ice crystals scatter the incoming solar radiation back to space leading to a cooling (albedo effect of clouds). On the other hand, they effectively trap the outgoing long wave radiation leading to a warming (greenhouse effect of clouds). Which effect dominates depends on the cloud s macrophysical and microphysical properties like vertical extension of the cloud, optical thickness, ice crystal shape, ice water content and ice crystal number concentration. For optically thick clouds, the scattering of incoming solar radiation dominates the trapping of long wave radiation leading to a cooling, whereas for optically thin clouds the opposite effect dominates. Until now it is believed that, on a global scale, cirrus clouds warm the present climate. However, an estimate of the net radiative effect of cirrus clouds based on general circulation models (GCM) is difficult because GCMs simulate a fourfold difference in high cloud amount. This is caused by the fact that complex interactions of dynamical and thermodynamical processes that lead to cirrus formation, are not properly taken into account. Furthermore it has been shown that there is a lack of cirrus clouds over mountainous regions in GCMs as the dynamical processes leading to the formation of orographic cirrus clouds are not taken into account. The aim of this work is therefore to improve the understanding of orographic cirrus clouds, their microphysical and optical properties and their representation in global circulation models in the present-day and future climate. A cloud resolving model is used in order to determine the key parameters influencing the microphysical and optical properties of orographic cirrus clouds in the present and future climate. First, the capability of the model in realistically simulating orographic cirrus has been shown by comparing the simulated results to aircraft measurements of an orographic cirrus taken during the INCA-campaign. In order to investigate the influence of a warmer climate on the microphysical and optical properties, the model has been initialized with vertical temperature, moisture and wind profiles taken from an IPCC A1B scenario run representative for the present climate and for the period Two regions representative for North and South America for the corresponding summer and winter months have been investigated. Furthermore the behavior of orographic clouds in a linear and non-linear flow regime are shown. The additional moisture in a warmer climate leads to a slight dampening of the propagation of gravity waves and the associated vertical velocities. Together with the higher temperatures, fewer ice crystals nucleate homogeneously. Assuming that the relative humidity does not change in a warmer climate, the specific humidity in the model is increased, leading to an increased ice water content of the clouds. The net effect of an increased ice water content and a decreased ice crystal number concentration is an enhanced optical depth. This behavior is a robust feature in our simulations that appears in the summer and winter months, North and South America and in the linear and non-linear flow regime. In a next step, a parameterization of orographic cirrus clouds has been developed and implemented in the ECHAM5 GCM. With this new parameterization changes in orographic cirrus clouds in a future climate have been estimated. To improve the simulation of orographically excited cirrus in ECHAM5, a coupling of gravity wave dynamics and cloud microphysics has been implemented. The maximum vertical velocity induced by gravity waves is calculated and used directly in the calculation of the ice crystal number concentration. As the ice crystal number concentration strongly depends on the vertical velocity, the addition of a gravity wave induced vertical velocity leads to higher ice crystal number concentrations in the upper troposphere. A comparison of the new parameterization with measurements shows a better agreement. However, the simulated vertical velocities and ice crystal number concentrations are slightly overestimated. To investigate orographic cirrus clouds in a warmer climate, a newer model version of the ECHAM5 model has been used, in which the effective radius of ice crystals depends on the ice crystal number v
6 concentration. Furthermore a reduced vertical velocity for the freezing parameterization based on box model simulations has been calculated, as the comparison of the simulated and measured values shows an overestimation of the simulated vertical velocity and ice crystal number concentration. The box model results were further used to develop a parameterization of an orographic cirrus cloud cover dependent on the horizontal wave length of the gravity waves. The influence of additional moisture on the propagation of gravity waves is investigated by using the dry and moist Brunt-Väisäla frequency, respectively, in the calculation of the gravity wave induced vertical velocity in two different simulations. With these new parameterizations implemented, simulations of the present and future climate are performed. From the present to the future climate the vertical velocity increases, as a smaller Brunt-Väisäla frequency in the future climate leads to less flow blocking and higher effective mountain heights over most mountain ranges. The opposite effect can be seen over dry regions. The ice crystal number concentration decreases in the future climate despite the increased vertical velocities. Higher temperatures lead to a faster growth of ice crystals and the supersaturation is depleted faster such that no new crystals can nucleate. The ice water content increases as more water vapor is available in a warmer climate. The net effect of a decreased ice crystal number concentration and an increased ice water content is an increased optical depth in a future climate. This result agrees very well with the cloud resolving simulations. The effect of orographic cirrus clouds on the radiation budget is given by an increased short- and long wave cloud forcing whereas the latter dominates. The results of the cloud resolving as well as the global simulations point into the direction that an increased optical depth of orographic cirrus clouds might be expected in a future climate. Furthermore the microphysical properties of cirrus clouds in ECHAM5 could be improved as compared to satellite measurements by taking more realistic dynamical processes for cirrus formation into account. This result emphasizes the important role of subgrid-scale dynamical processes for the correct representation of cirrus clouds in GCMs. vi
7 Zusammenfassung Zirruswolken haben einen grossen Einfluss auf das Strahlungsbudget der Erde. Sie bedecken etwa 30% der Erde und bestehen ausschliesslich aus Eiskristallen. Einerseits streuen die Eiskristalle die einfallende Sonnenstrahlung zurück ins All, was zu einer Abkühlung führt und mit dem Albedo-Effekt der Wolken bezeichnet wird. Andererseits können sie durch Absorption und Rückstrahlung der von der Erde emittierten langwelligen Strahlung erwärmend wirken (Treibhauseffekt der Wolken). Welcher dieser beiden Effekte dominiert, hängt von den makro- und mikrophysikalischen Eigenschaften der Wolke ab. Dazu gehören beispielsweise die vertikale Ausdehnung der Wolke, die optische Dicke, die Form der Kristalle, der Eiswassergehalt oder die Eiskristallanzahl. Bei optisch dicken Zirren überwiegt der Albedo-Effekt der Wolken, während bei optisch dünnen Wolken der Treibhauseffekt dominiert. Auf einer globalen Skala überwiegt bei Zirren vermutlich der Treibhauseffekt. Solche Abschätzungen basierend auf Ergebnissen der Globalen Zirkulationsmodelle (GCM) sind jedoch schwierig, da diese bis heute nicht in der Lage sind, den Bedeckungsgrad der Zirren und deren mikrophysikalische und optische Eigenschaften richtig zu berechnen. Zusätzlich konnte gezeigt werden, dass GCMs über Gebirgen keine ausreichende Zirrusbewölkung simulieren, da die dynamischen Prozesse, die zur Bildung von orographischen Zirren führen, in den GCMs nicht enthalten sind. Deshalb ist das Ziel dieser Arbeit, unser Verständnis von orographischen Zirren, ihrer mikrophysikalischen und optischen Eigenschaften sowie deren Simulation in GCMs zu verbessern und den Einfluss eines wärmeren Klimas auf diese abzuschätzen. Um die wichtigsten Parameter, die die mikrophysikalischen und optischen Eigenschaften von orographischen Zirren im heutigen sowie im zukünftigen Klima bestimmen, zu untersuchen, wurden Simulationen mit einem wolkenauflösenden Modell durchgeführt. Die Fähigkeit des Modells realistische orographische Zirren zu simulieren wurde getestet, indem die simulierten Ergebnisse mit Messungen einer orographischen Zirruswolke verglichen wurden, die im Rahmen der INCA-Kampagne durchgeführt wurden. Um den Einfluss eines wärmeren Klimas auf die mikrophysikalischen und optischen Eigenschaften zu untersuchen, wurde das Modell mit vertikalen Temperatur-, Feuchte- und Windprofilen aus einem IPCC A1B Szenariolauf initialisiert, die repräsentativ für das heutige sowie das zukünftige Klima ( ) sind. Die Simulationen wurden für zwei Regionen, die Nord- und Südamerika representieren, sowie für die jeweiligen Sommer- und Wintermonate durchgeführt. Zusätzlich wurden alle Simulationen für ein lineares und ein nicht lineares Strömungsregime durchgeführt. Die zusätzliche Feuchte in einem wärmeren Klima führt zu einer schwachen Dämpfung der Schwerewellen und den mit ihnen verbundenen Vertikalgeschwindigkeiten. Zusammen mit einer Zunahme der Temperatur, bilden sich weniger Eiskristalle durch homogenes Gefrieren. Unter der Annahme, dass die relative Feuchte in einem zukünftigen Klima konstant bleibt, nimmt die spezifische Feucht im Modell zu, was zu einer Zunahme des Eiswassergehaltes der Wolken führt. Die Abnahme der Eiskristalle führt in Kombination mit der Zunahme des Eiswassergehaltes zu einer höheren optischen Dicke. Dieses Ergebniss zeigt sich in allen hier durchgeführten Simulationen. In einem nächsten Schritt wurde eine Parametrisierung für orographische Zirren entwickelt und in das ECHAM5 GCM implementiert. Mit Hilfe dieser neuen Parametrisierung ist es möglich, den Einfluss eines wärmeren Klimas auf orographische Zirren auf einer globalen Skala abzuschätzen. Um die Simulation von orographischen Zirren in ECHAM5 zu verbessern, wurde die Dynamik der Schwerewellen mit der Wolkenmikrophysik gekoppelt. Dazu wurde die maximale Vertikalgeschwindigkeit die in einer orographischen Schwerewelle auftritt berechnet, und direkt in der Berechnung der Eiskristall anzahl verwendet. Da die Eiskristallanzahl stark von der Vertikalgeschwindigkeit abhängt, führt dies zu einer beträchtlichen Zunahme der Eiskristallanzahl in der oberen Troposphäre. Der Vergleich dieser neuen Parameterisierung mit Messungen zeigt eine bessere Übereinstimmung. Allerdings sind sowohl die Vertikalgeschwindigkeiten als auch die Eiskristallanzahl leicht überschätzt. vii
8 Um den Einfluss eines zukünftigen Klimas auf orographische Zirren zu untersuchen, wurde eine neuere Modellversion von ECHAM5 verwendet, in der die Berechnung des Effektivradius der Eiskristalle von der Eiskristallanzahl abhängt. Zusätzlich wurde basierend auf Boxmodellsimulationen eine reduzierte Vertikalgeschwindigkeit berechnet, die in die Berechnung des homogenen Gefrierens eingeht, da der Vergleich von simulierten und gemessenen Werte eine leichte Überschätzung der simulierten Werte gezeigt hat. Die Simulationen des Boxmodells wurden weiterhin für die Entwicklung einer einfachen Parametrisierung für den Bedeckungsgrad für orographische Zirren verwendet. Dieser hängt von der horizontalen Wellenlänge der Schwerewelle ab. Der Einfluss einer höheren Feuchte in einem zukünftigen Klima wird untersucht, indem zwei unterschiedliche Simulationen durchgeführt werden, wobei in einer die trockene und in der anderen die feuchte Brunt-Väisäla Frequenz in der Berechnung der schwerewelleninduzierten Vertikalgeschwindigkeit verwendet wird. Mit diesen neuen Parametrisierungen wurden Simulationen des heutigen und zukünftigen Klimas basierend auf dem IPCC A1B Szenario, durchgeführt. Die Vertikalgeschwindigkeit nimmt über den meisten Gebirgen im zukünftigen Klima zu. Verursacht wird dies durch eine Abnahme der Brunt-Väisäla Frequenz in einem zukünftigen Klima, was zu einer Verminderung des flow-blocking und somit zu einer erhöhten effektiven Berghöhe führt. Der gegenteilige Effekt zeigt sich über den sehr trockenen Regionen der Erde. Trotz der meist höheren Vertikalgeschwindigkeit in Zukunft, nimmt die Eiskristallanzahl ab, da höhere Temperaturen zu einem schnelleren Wachstum der Kristalle führen. Die Übersättigung wird schneller abgebaut, so dass keine neuen Kristalle mehr gebildet werden können. Der Eiswassergehalt nimmt in einem zukünftigen Klima zu, da mehr Wasserdampf vorhanden ist. Die Kombination dieser beiden Effekt führt zu einer Zunahme der optischen Dicke. Der Einfluss von orographischen Zirren auf den Strahlungshaushalt besteht in einem erhöhten kurzwelligen und langwelligen cloud forcing, wobei das langwellige clolud forcing dominiert. Die Ergebnisse der wolkenauflösenden sowie der globalen Simulationen deuten auf eine Zunahme der optischen Dicke in einem zukünftigen Klima hin. Ausserdem konnte die Simulation der mikrophysikalischen Eigenschaften von Zirren in ECHAM5 allein durch die Berechnung realistischerer dynamischer Prozesse sowie deren Kopplung an die Eismikrophysik, verbessert werden. viii
9 Chapter 1 Introduction 1.1 Cirrus clouds Cirrus clouds play an important role in the climate system. They consist purely of ice crystals and cover approximately 30 % of the Earth (Wylie and Menzel, 1999). As ice crystals interact with the short-wave and long-wave radiation, cirrus clouds have a great potential to modulate the Earth s radiative budget. They can either warm or cool the Earth-Atmosphere system depending on their microphysical and optical properties. On the one hand, ice crystals scatter the short-wave radiation back to space and thus lead to a cooling (albedo effect of clouds). On the other hand, the long-wave radiation can be trapped effectively which leads to a warming (greenhouse effect of clouds). For optically thin cirrus, the absorption of infrared radiation and re-emission at lower temperatures dominates the scattering of incoming solar radiation, leading to a warming. For a high optical depth the scattering of incoming solar radiation dominates, leading to a cooling. Which effect is more pronounced thus depends on macrophysical properties such as the optical thickness of the cloud, which in turn depends on the microphysical properties like ice crystal number concentration, ice water content, crystal size and shape (Wendisch et al., 2007). As these properties are mainly determined by the interaction of dynamical and thermodynamical processes, it is very difficult to estimate the overall effect of cirrus clouds on the radiative budget. Until now it is believed that cirrus clouds lead to a net global warming (Chen et al., 2000). However, Zhang et al. (2005) showed that different models simulate a fourfold difference in high cloud amount. This uncertainty arises from the complex interaction of dynamical and thermodynamical processes which are not implemented in the models or not resolved by the coarse model grid. The poor representation of cirrus thus exhibits a significant source of uncertainty in predicting future climate. Thus, one basic requirement for an improved prediction of climate change is the correct representation of cirrus clouds in general circulation models (GCM). For the formation of cirrus clouds, high supersaturations with respect to ice are needed, hence cirrus formation is associated with (cloud free) ice supersaturated regions (ISSR). This is obvious when comparing the spatial distributions of ISSRs and cirrus clouds (Spichtinger et al., 2003b). The supersaturation needed for cirrus formation depends on the freezing process. Cirrus can form via heterogeneous or homogeneous freezing. Heterogeneous freezing is initiated by solid aerosols, the so-called ice nuclei (IN). The properties of these ice nuclei determine the critical supersaturation required for the initiation of the freezing process. Homogeneous freezing denotes the freezing of supercooled solution droplets. The freezing of droplets of a certain size is initiated when the supersaturation exceeds a critical threshold. Koop et al. (2000) showed, that the threshold only depends on temperature and is mostly independent of the nature of the solute. In general, the supersaturation threshold is lower for heterogeneous freezing (DeMott et al., 2003). However, the lack of efficient ice nuclei in the upper troposphere, the frequent measurement of high ice crystal number concentrations and high relative humidities and the prevalence of mesoscale temperature fluctuations strongly suggests that the homogeneous freezing is the most important freezing mechanism (Sassen and Dodd, 1989; DeMott et al., 2003; Haag et al., 2003). The presence of IN can modify the microphysical properties of cirrus especially at low cooling rates and for optically thin clouds (Käercher, 2002; Spichtinger and Gierens, 2009b). ISSRs form due to different cooling mechanisms, including synoptic-scale vertical motion, turbulence, convection or gravity waves, leading to an increase in relative humidity beyond ice saturation or due to moisture advection. Aircrafts measure a typical horizontal extent of ISSRs of 150 km (sometimes a few 1000 km) in the mid-latitudes, however, very large extended ISSRs up to 4000 km were rarely 1
10 2 Chapter 1. Introduction observed (Gierens and Spichtinger, 2000). Typical vertical extents are 500 m at mid and high latitudes whereas the variability in layer thickness is large (Spichtinger et al., 2003a). The ice supersaturated regions define the overall thermodynamical conditions necessary for cirrus cloud formation. The microphysical properties, on the other hand, are determined by the mesoscale variability in cooling rates at the point of freezing (Haag and Kärcher, 2004; Hoyle et al., 2005). Various studies show, that the observed microphysical properties can only be calculated correctly by taking mesoscale vertical velocity/temperature fluctuations into account. Sources for these mesoscale velocity/temperature fluctuations are mesoscale gravity waves arising from different sources like convection, geostrophic adjustment, baroclinic instability and stratified flow over mountains (Fritts and Alexander, 2003). The mesoscale velocity fluctuations exhibit one key factor determining the ice crystal number concentration in a homogeneous freezing event. They induce adiabatic cooling and the relative humidity with respect to ice (RHi) increases. If the critical threshold for homogeneous freezing is exceeded, crystals start to form. The higher the vertical velocity and the associated cooling rate, the more crystals can form because of the steep increase in the nucleation rate (Koop et al., 2000). Thus the calculation of a realistic vertical velocity is crucial to simulate realistic microphysical properties. This has also been shown by Lin et al. (1998) who investigated the broadness of the crystal spectra which develops if an air parcel follows sine waves with different amplitudes. The observed broad crystal distribution could be reproduced by varying the initial position of the air parcel in the wave trajectory whereas the maximum ice crystal number concentration is determined by the amplitude of the wave. No additional processes are needed to explain the observed properties. Hoyle et al. (2005) also investigated the origin of high ice crystal number concentrations in cirrus clouds and found that the observed high ice crystal number densities are obtained only when small-scale temperature/vertical velocity fluctuations are taken into account. Kärcher and Ström (2003) investigated the role of dynamical variability and aerosols for cirrus cloud formation based on the INCA (Interhemispheric differences in Cirrus properties from Anthropogenic emissions) measurements (Gayet et al., 2004). They could show, that variability in mesoscale vertical velocities combined with pure homogeneous freezing is responsible for most of the observed variance in the ice crystal number concentration. However, it is possible that the presence of IN influences the properties of optically thin or even subvisible cirrus with low ice crystal number concentration formed in weak updrafts (Gierens, 2003; Spichtinger and Gierens, 2009b). With increasing updrafts this influence is diminished as the amount of IN is limited and the evolving ice crystal number concentration exceeds the available IN. All these studies emphasize the importance of mesoscale dynamical features in determining microphysical and optical properties of cirrus clouds. Recent studies indicate that the ice crystal number concentration plays an important role for the transition from a warming of cirrus clouds to a cooling (Fusina et al., 2007). In order to account for the influence of ice crystal number concentration on the radiative properties of cirrus clouds and to predict the global net effect of cirrus clouds on the radiative budget and changes in a future climate it is necessary to include the effect of mesoscale dynamics on cirrus cloud properties in global circulation models. As orographic gravity waves are one important and widespread source for mesoscale fluctuations this dissertation attempts to improve the understanding of orographic cirrus clouds and their representation in global climate models also with regard to the future climate. 1.2 Orographic gravity waves Orographic gravity waves lead to the formation of orographic clouds. They can form only when the atmosphere is stably stratified and when air is forced to flow over mountains. A fluid parcel which is displaced vertically will then undergo buoyancy oscillations. In a fluid that has no upper boundary, like the atmosphere, gravity waves may propagate vertically and horizontally. As the density of the atmosphere decreases exponentially with height, the amplitude of a gravity wave that propagates high into the atmosphere, grows with height. The associated vertical displacement of air induces adiabatic cooling and provides regions for cloud formation throughout the whole troposphere. Gravity waves are able to transport momentum from the Earth s surface to regions in the atmosphere
11 1.3. Orographic cirrus clouds 3 where they dissipate. The magnitude of the momentum flux may be large enough to strongly influence the large-scale mean flow (McFarlane, 1987). Large mountains or weak winds can additionally lead to non-linear effects like flow blocking. This means that only part of the flow is able to flow over the mountain, whereas the flow in the low levels is blocked. This effect also leads to a transfer of momentum between the atmosphere and the surface of the Earth. The influence of mountains on the atmosphere has been recognized and investigated intensely during the last century. There are several publications describing the linear theory of mountain waves (see e.g. (Queney et al., 1960; Smith, 1979). The development of nonlinear mountain wave regimes are described in e.g. Smith (1989); Wurtele et al. (1996). Due to the processes described above, the representation of orographic gravity waves in numerical weather prediction and climate models is very important. McFarlane (1987) and Palmer et al. (1986) showed that the inadequate representation of the unresolved orography and the missing influence of vertically propagating gravity waves on the large-scale flow lead to deficiencies in the simulated large scale circulation especially in the northern hemispheric winter. When the effect of subgrid-scale orography on the momentum transport was parameterized, the systematic westerly bias could be reduced. In the scheme described in Palmer et al. (1986) the orography is isotropic and it is assumed that most of the momentum is transported by long, hydrostatic waves. However, this scheme was not able to predict the drag exerted by low-level wave breaking (Gregory et al., 1998). It therefore was developed further based on the work of Baines and Palmer (1990) and Lott and Miller (1997). It includes an anisotropic subgridscale orography and takes into account the drag due to flow blocking and low-level wave breaking. This scheme is implemented in the ECHAM5 GCM used in this dissertation. The subgrid-scale orography, which provides the basis for orographic gravity waves and is used in this parameterization, is described in more detail in appendix A. 1.3 Orographic cirrus clouds The vertically propagating gravity waves induce vertical displacement of air in the whole troposphere. Adiabatic cooling is induced in the rising air parcel and the relative humidity with respect to ice (RHi) within the parcel starts to increase. If the critical threshold for freezing is exceeded, ice crystals start to nucleate. The supersaturation is then depleted by growing ice crystals. Depending on the path of the air parcel, further freezing events are possible if the RHi gets again high enough. In the downdraft of the gravity wave, ice crystals start to evaporate. As ice crystals can survive in subsaturated air, they can be advected downstream over several hundreds of kilometers. Thus, orographically induced ice crystals are a plausible formation mechanism of large-scale cirrus clouds downstream of big mountain ranges (Dean et al., 2007). On the other hand, if RHi is not high enough and/or the ice crystals are small, they can completely evaporate leading to a cloud with the typical lenticularis shape. Figure 1.1 shows two examples of orographic clouds. In the left panel, ice crystals form in the updraft of a gravity wave. The ice crystals survive the downdraft region and are advected over 200 km. In contrary, in the right panel, a lenticularis cloud can be seen where the ice crystals evaporate in the downdraft of the wave. In general, the vertical velocities and cooling rates induced by orographic gravity waves are very high such that it is assumed that homogeneous freezing is the dominant freezing mechanism for orographic cirrus clouds at temperatures below the homogeneous freezing threshold (Kärcher and Ström, 2003). All simulations presented in this work therefore only consider homogeneous freezing. Measurements in an orographic cirrus cloud at very cold temperatures (T< 60 C) (Jensen et al., 1998) also provide evidence that the homogeneous freezing might be the dominant freezing mechanism. The measured high ice crystal number concentration ( 10 cm 3 ) can only be explained if homogeneous freezing of solution droplets is assumed. The reproduction of the measured values with an ice cloud model which only takes the homogeneous freezing into account, confirms the assumption of homogeneous freezing as the most important process here. Furthermore Jensen et al. (1998) show that changes in the updraft velocity can substantially change the maximum ice crystal number concentration whereas changes in the sulfate aerosol number density and size distribution only have a weak impact on the ice
12 4 Chapter 1. Introduction Figure 1.1: Satellite picture of a mountain wave induced cirrus over the Sierra Nevada (left panel). The ice crystals are advected over 200 km (from durrand/). The right panel shows a lenticularis cloud over the Andes, Bolivia (from crystal number concentration. During the INCA campaign (Gayet et al., 2004), an orographic wave cloud at temperatures of 47 C was measured. They found very high ice crystal number concentrations ( cm 3 ) and very small particles which again can only be explained by homogeneous freezing. A comparison of this cloud with one which formed at similar ranges of temperature and with similar aerosol properties but not in an orographic wave, shows completely different microphysical properties. Heymsfield and Miloshevich (1993) show aircraft measurements of a cold lenticular wave cloud in a temperature range -31 C to -41 C. They found homogeneous freezing below about -33 C. Above that temperature only liquid droplets were observed. Due to a lack of heterogeneous IN and a very low homogeneous ice nucleation rate at this temperature, no ice was formed. All these measurements emphasize the importance of dynamics for the formation of cirrus clouds and their microphysical properties. A modeling study of an orographic cirrus cloud presented in Kay et al. (2007) showed that the broad distribution of optical depth in an orographically induced cirrus results because homogeneous freezing driven by variable vertical velocities leads to variable optical depth. The addition of IN with a realistic background concentration has little influence on the optical depth. Considering orographic cirrus on a global scale, Fowler and Randall (1999) and Dean et al. (2005) showed that in the Colorado State University GCM and the Met Office HadAM3 respectively, there is a lack of cirrus clouds especially over continents. This fact can be ascribed to the missing link between dynamics of the subgrid-scale processes leading to orographic gravity waves and cloud microphysics. Dean et al. (2007) implemented a coupling of gravity waves and cloud microphysics in the UK Metoffice HadAM3 GCM by calculating temperature perturbations which can lead to the formation of orographic clouds. However, as in this model only the ice mass mixing ratio and no ice crystal number concentration is simulated, the important influence of mesoscale velocities on the ice crystal number concentration cannot be taken into account. As ECHAM5 contains a two-moment ice microphysics, which allows the calculation of ice crystal number concentrations, the vertical velocity induced by gravity waves is directly used for the calculation of the ice crystal number concentration. Furthermore, as the occurrence of high supersaturation is required for the formation of homogeneously frozen ice crystals, the saturation adjustment schemes often used in GCMs are not suitable. In these schemes it is assumed that all water vapor above saturation with respect to ice is deposited on the existing ice crystals. In the ECHAM5 GCM used for this dissertation, supersaturated regions are allowed to form and the depositional growth equation is solved explicitely (Lohmann and Kärcher, 2002). This is an important improvement and provides the basis for a realistic treatment of cirrus clouds as they coexist with ice supersaturated regions. In general, the ECHAM5 GCM is suitable for considering orographic cirrus clouds as it allows ice supersaturated regions and explicitly couples the
13 1.4. Overview over this dissertation 5 ice crystal number concentration to the vertical velocity. All the studies presented here, show that taking the dynamical processes for cirrus formation into account is necessary to calculate realistic spatial distribution and microphysical properties of cirrus. As the effect of cirrus clouds on the radiation budget strongly depends on their microphysical properties, it is important to identify the sources of mesoscale velocity fluctuations leading to cirrus formation and to couple them to the ice microphysics. This is especially important for climate models that are designed to predict the changing climate. If possible changes in the dynamics and the coupling to the ice microphysics are not taken into account, no predictions concerning changes in microphysical and optical properties are possible. As mountains provide a widespread source for mesoscale gravity waves/velocity fluctuations and thus trigger the formation of cirrus, this dissertation presents an improved representation of orographically forced cirrus clouds in the ECHAM5 GCM in order to improve the representation of the optical and microphysical properties in the current and future climate. Furthermore the key parameters which determine these properties are investigated with the cloud resolving model EULAG (Smolarkiewicz and Margolin, 1997; Prusa et al., 2008). 1.4 Overview over this dissertation In this dissertation several studies of orographic cirrus clouds are presented. In order to investigate the most important parameters influencing the microphysical and optical properties, and to improve the representation of orographic cirrus on a global scale, simulations with three different models were performed. The key parameters determining the microphysical properties like the ice water content and the ice crystal number concentration have been investigated with a cloud resolving model. Therefore, simulations are performed, where under idealized conditions, the influence of a changing vertical velocity, temperature and relative humidity with respect to ice have been investigated. In order to estimate the changes of the microphysical properties in a warmer climate, additional simulations initialized with IPCC profiles were carried out. The results of the cloud resolving simulations are presented in chapter 2. In a next step the representation of orographic cirrus clouds in the global climate model ECHAM5 has been improved. Therefore a coupling of gravity wave dynamics and cloud microphysics has been implemented in the model and the results are compared to measurements. This new orographic gravity wave scheme is shown in chapter 3. The implementation of orographic gravity waves in ECHAM5 provides the basis for further investigation of the microphysical and optical properties of orographic cirrus in a future climate. Based on boxmodel simulations the calculation of the gravity wave induced vertical velocity used in the freezing process is improved and a cloud cover for orographic cirrus clouds has been developed. In order to investigate orographic cirrus clouds in a warmer climate, simulations of the IPCC A1B scenario have been performed. The results of this simulations are presented in chapter 4. In Chapter 5 the results of this dissertation are summarized and possible extensions of the work presented here are discussed.
15 Chapter 2 Orographic cirrus in the future climate A cloud resolving model (CRM) is used to investigate the formation of orographic cirrus clouds in the current and future climate. The formation of cirrus clouds depends on a variety of dynamical and thermodynamical processes, which act on different scales. First, the capability of the CRM in realistically simulating orographic cirrus clouds has been tested by comparing the simulated results to aircraft measurements of an orographic cirrus cloud. The influence of a warmer climate on the microphysical and optical properties of cirrus clouds has been investigated by initializing the CRM with vertical profiles of horizontal wind, temperature and moisture from IPCC A1B simulations for the current climate and for the period for two regions representative for North and South America. In a future climate, the increase in moisture dampens the vertical propagation of gravity waves and the occurring vertical velocities. Together with higher temperatures fewer ice crystals nucleate homogeneously. Assuming that the relative humidity does not change in a warmer climate the specific humidity in the model is increased. This increase in specific humidity in a warmer climate results in a higher ice water content. The net effect of a reduced ice crystal number concentration and a higher ice water content is an increased optical depth. H.Joos, P. Spichtinger and U. Lohmann, Orographic cirrus in the future climate. Atmos. Chem. Phys. Diss., 9,
16 8 Chapter 2. Orographic cirrus in the future climate 2.1 Introduction Cirrus clouds have a strong influence on the radiative budget of the earth. They can either cool or warm the Earth-Atmosphere system depending on their microphysical properties like ice water content or ice crystal number concentration. Generally, optically thick cirrus clouds exert a cooling effect and optically thin clouds a warming effect. The global net effect of cirrus clouds tends to warm the Earth- Atmosphere-system (Chen et al., 2000). Our understanding of the formation of cirrus clouds and their resulting microphysical and optical properties is crucial to predict changes in the radiative budget in the future climate. However, as the formation of cirrus clouds depends on very complex multi-scale dynamical and microphysical processes, their formation and life cycle is not well known (Spichtinger et al., 2005a,b). A significant part of the uncertainties in the prediction of future climate in general circulation models (GCM) arises from the representation of cirrus cloud formation (Zhang et al., 2005). As in most GCMs only large scale vertical velocities are calculated and the manifold dynamical processes which strongly influence the formation of cirrus are not taken into account, the cirrus cloud amount is underestimated in many GCMs and no reliable prediction of a change in cirrus cloud cover is possible as the correct underlying physical process is not taken into account. There are only two parameterizations for GCMs (the HadAM3 and the ECHAM5 model) available which at least take the formation of orographic cirrus clouds into account (Dean et al., 2005; Joos et al., 2008). Additionally, anthropogenic aerosols can change the ice crystal number concentration and exert an anthropogenic forcing comparable to the forcing on warm clouds (Penner et al., 2009). An estimation of the influence of a changing climate on the formation of (orographic) cirrus and the radiative budget is still difficult. In order to estimate the changes in orographic cirrus cloud cover in a warmer climate, it is necessary to determine the key processes, which influence the formation of orographic cirrus. Here, changes in the flow regime due to a change in atmospheric stability caused by a change in the temperature and moisture profiles, an increase of horizontal wind speed especially in the upper troposphere and, of course, the changes due to warmer temperatures have to be taken into account. There are several papers describing the influence of moisture on the propagation of orographically excited gravity waves (see e.g. Durran and Klemp (1983); Jiang (2003)). The additional moisture weakens the atmospheric stability and amplitudes of the gravity waves and thus the vertical velocities which strongly influence the ice crystal number concentration. Additionally, the vertical wavelength is increased. On the other hand, the increase in horizontal wind speed leads to an increase of the amplitudes and vertical velocities of the waves. In order to investigate the impact of these two opposing effects, which we expect in a changing climate, we present simulations with the non-hydrostatic anelastic model EULAG (Prusa et al., 2008). As in this model a detailed ice microphysical scheme is implemented (Spichtinger and Gierens, 2009a) it can be used in order to assess the dynamical and thermodynamical changes in a future climate and its influence on the formation of orographic cirrus clouds. In order to determine the importance of the individual processes, idealized simulations were carried out where changes in the temperature profiles, the relative humidity with respect to ice (RHi) and the position of the ice supersaturated layer have been investigated. In order to determine the changes in the microphysical and optical properties of orographic cirrus in a future climate, simulations initialized with vertical profiles taken from IPCC simulations for the beginning and the end of the 21 st century were performed. In section 2.2 the model used for this study is introduced briefly. In section 2.3 we show the model s capability to simulate realistic orographic cirrus by comparing the results of a simulation with aircraft measurements. In section 2.4 we present results of idealized simulations and discuss them and in section 2.5 the results of the simulations initialized with IPCC profiles are presented. In section 2.6 we summarize our work and draw some conclusions.
17 2.2. Model description Model description For this study we use the anelastic, non-hydrostatic model EULAG (Smolarkiewicz and Margolin, 1997; Prusa et al., 2008; Grabowski and Smolarkiewicz, 2002). As cloud physics is implemented in this model, moist dynamics and a coupling of dynamics and thermodynamics is performed. Additionally, a twomoment bulk ice microphysics scheme is implemented. In this scheme the ice classes correspond to different nucleation processes (homogeneous and heterogeneous freezing) and include the processes of ice crystal nucleation, depositional growth/evaporation and sedimentation. In our simulations only homogeneous freezing is considered as it can be assumed that the formation of orographic cirrus is dominated by high vertical velocities and supersaturations and that homogeneous freezing is the dominant freezing mechanism (Kärcher and Ström, 2003). The homogeneous nucleation rate is parameterized according to Koop et al. (2000). The background aerosol (sulfuric acid) is distributed log-normally with a modal radius r m =25 nm for aqueous solution droplets and geometric standard deviation σ =1.4. In Spichtinger and Gierens (2009a) it was found that this setup gives reliable results for homogeneous nucleation events. For a more detailed description of the ice microphysics scheme see Spichtinger and Gierens (2009a). For all simulations shown in this study, a 2-dimensional (x-z plane) model domain is used. The detailed model setup for each simulation is described at the beginning of every chapter. 2.3 Model verification: Simulation of the INCA-case In order to show the model s capability to represent the formation of orographic cirrus clouds, measurements from the INCA (Interhemispheric differences in cirrus properties from anthropogenic emissions) campaign (Gayet et al., 2004) are used as comparison. The INCA campaign took place in April 2000 over Punta Arenas, Chile and in October 2000 over Prestwick, Scotland, respectively. The measurements used for this comparison were taken at 5 April 2000 between 18 and 19 UTC on a flight track at 53 S from 69.2 W to 76 W. During this flight, the vertical velocity, ice crystal number concentration and ice water content in an orographic cloud were measured. Vertical velocities were measured with a five-hole probe only during constant altitude flight sections. The accuracy of the vertical velocity is estimated to be on the order of 0.1 ms 1 (Bögel and Baumann, 1991). Ice particle concentrations were measured with a combination of two instruments, the FSSP-300 and 2DC-C optical probe inboard the DLR Falcon (Gayet et al., 2002, 2004). The particle concentrations used for this comparison refer to the particle size range micrometer in diameter. Furthermore, residual particle measurement with the Counterflow Virtual Impactor, CVI (Noone et al., 1993) have been carried out. In order to test the model s capability to represent orographic cirrus clouds in a correct way, these measurements are compared to a 2-dimensional simulation with the EULAG model. As initial profiles we used the temperature and wind data from the ECMWF (European Centre for Medium-Range Weather Forecasts) Reanalyse data for the 5 th of April 2000 at 18 UTC when the measurements were taken. Additionally, a realistic topography from the National Geographical Data Centre (NGDC, (Hastings et al., 1999)) is implemented. Figure 2.1 shows the initial profiles of temperature T (z), potential temperature θ(z), horizontal wind speed u(z) and pressure p(z), respectively. The wind direction is approximately 260. For simplicity we assumed a pure west wind here. In this simulation we used a horizontal model domain of 1000 km and 20 km in the vertical with a horizontal resolution dx = 1000 m and a vertical resolution dz = 50 m. In all simulations the dynamical time step is dt = 2.5 s and the microphysical time step is dt m = dt/10 = 0.25 s. The model is run for 5 hours. In all simulations the Coriolis force is neglected. Figure 2.2 shows the result of the flow over the realistic topography initialized with the ECMWF profiles after t = 5 h. The topography of the Andes induces gravity waves that propagate through the whole troposphere. The maximum and minimum vertical velocities amount to +8/-8 ms 1, respectively. They occur in a height between 4 and 6 km. According to the height of the flight, an ice supersaturated region (ISSR) has been implemented in a height of m with an initial supersaturation of RHi=130%. In order to compare the simulation with the measurements, histograms for the
18 10 Chapter 2. Orographic cirrus in the future climate Figure 2.1: Initial vertical profiles of temperature T, potential temperature θ, horizontal wind speed u and pressure p taken from the ECMWF Reanalyse data for the 5 April 2000 at 18 UTC at 53 S and 78 W. Figure 2.2: Flow regime for the INCA case at 5 April 2000, 18 UTC initialized with the ECMWF profiles. Grey lines denote lines of constant potential temperature and the colorbar indicates the vertical velocity in m s 1. The black box shows the initial position of the supersaturated layer. vertical velocity, ice crystal number concentration (ICNC) and ice water content (IWC) are shown. The temperatures measured during the flight lie between 230 K and 226 K. Therefore, we selected the simulated values at T=226K and T=230K for comparison. Additionally, the distribution sampled over the whole ISSR is shown. The simulated values sampled over the 5 simulated hours are shown in figure 2.3. It can be seen that the model reproduces the distribution of the measured variables remarkably well. The measured vertical velocity is in the range between +1.8 and -1.8 ms 1. In general, this can be reproduced although the model seems to overestimate the vertical velocities. This is due to the 2-dimensional theory that leads to an overestimation of the vertical velocity (Dörnbrack, 1998) and the effect of moisture on the propagation of gravity waves, which could lead to a decrease in vertical velocities. This is neglected here. On the other hand, as the airplane flies with 170 ms 1 it did not necessarily reach to measure
19 2.3. Model verification: Simulation of the INCA-case 11 Figure 2.3: Comparison of the simulated and measured vertical velocity (upper panel), ICNC (middle panel) and IWC (lower panel). For the ICNC the black line denotes the combined measurements taken with the FSSP-300 and 2DC-C optical probe. Additionally, the measurements with the CVI are shown in dark blue. The simulated results for the temperature range 226 K and 230 K are shown in red and light blue, respectively. The purple line shows the simulated results sampled over the whole ISSR. All simulated values contain data sampled over all time steps. the highest occuring velocities. If all these restrictions are taken into account, one can say that the model is quite able to represent the measured values (see fig. 2.3). The results for the ICNC and IWC also agree very well with the observation. Gayet et al. (2006) stated that shattering of ice crystals leading to an overestimation of measured ICNC was unlikely as different techniques were used. The comparison of the results obtained with the different techniques showed little difference such that shattering can be ruled out for this case. Based on these results, we assume that the model is able to simulate realistic vertical velocities and microphysical properties of orographic cirrus clouds and can thus be used for further investigations of orographic cirrus clouds with idealized simulations as described in the next chapter or
20 12 Chapter 2. Orographic cirrus in the future climate to investigate the changes in microphysical properties in a changing climate as discussed in chapter Idealized Simulations To investigate the key parameters, which determine the microphysical and optical properties of orographic cirrus clouds in a future climate, idealized simulations have been carried out. In a future climate, there are two main processes which influence the properties of an orographic cirrus cloud. On the one hand, an increase in moisture could lead to a damping of the gravity waves amplitude and reduce the vertical velocities. Furthermore, the vertical wavelength could increase such that the ISSR shifts in a different vertical phase of the wave. On the other hand, the temperature increase changes the available water vapor under the assumption of a constant relative humidity and has an influence on the depositional growth. In order to assess the importance of these individual thermodynamical and dynamical processes and its influence on the formation of cirrus clouds, the temperature inside the ISSR has been changed and the height of the ISSR is shifted to a lower/higher position such that a change in the wave phase is simulated. The model is initialized with the ambient potential temperature and pressure profiles θ(z) and p(z) according to Clark and Farley (1984), using a constant Brunt-Väisäla frequency N over the whole troposphere. From θ(z) and p(z) the physical temperature T (z) and the density ρ(z) can be calculated. Additionally, a wind profile u(z) is prescribed: between 0 and 2 km height, u(z) increases from 4 ms 1 to 9 ms 1. From 2 km to 12 km the horizontal velocity is constant (u(z) = 9 ms 1 ). Above that level it decreases linearly until u(z = 15 km) = -10 ms 1. For these simulations we use a 2D domain (x-z-plane) with a horizontal extension of 320 km and a vertical extension of 20 km with a bell shaped mountain in the middle of the domain. The mountain shape can be described as H(x) = h x2 a 2 (2.1) where a = m is the half-width of the mountain and h 0 = 600 m the maximum height. The non-dimensional mountain height ĥ = Nh 0 /u = 0.6 which leads to a hydrostatic mountain wave. The horizontal and vertical resolutions are dx = 250 m and dz = 50 m, respectively. The simulations have been carried out for 5 hours. An ice supersaturated layer (ISSR) with a depth of 1 km has been implemented in the model additionally. First, a reference case has been defined: The ISSR is situated in the vertical range between 8500 m and 9500 m. This corresponds to the height where the highest vertical velocity in the developing hydrostatic wave occurs. The initial temperature profile has been chosen in a way that the temperature in the middle of the reference ISSR is 220 K and the reference initial supersaturation with respect to ice (RHi) is 120%. Then two different effects were investigated. 1. A shift of the initial θ profiles such that the temperature in the middle of the reference ISSR is 210 K and 230 K, respectively (see Fig.2.4). 2. A change of the height of the ISSR, which causes a different temperature and a different position in the wave phase. The height of the ISSR has been shifted so that the temperature in the middle of the lowest layer is 230 K and in the highest layer 210 K (see figure 2.5). Figure 2.4 shows the initial T (z), θ(z) and u(z) profiles for the reference case (black line) and the two shifted θ-profiles. In order to obtain the same flow regime for all cases, the θ-profile has been shifted by adding a constant. Thus, the Brunt-Väisäla-frequency is the same in all simulations. The developing flow regime and the position of the different ISSR are shown in figure 2.5 (left panel). The results shown here are after t = 5h. A nearly hydrostatic wave develops, which propagates vertically through the whole troposphere and is absorbed at the tropopause. The highest vertical velocity occurs at 9 km height and amounts to 0.8 m s 1.