• No results found

Transport of very short lived halogenated substances from the tropical East Pacific to the stratosphere and the influence of El Niño 2015/16

N/A
N/A
Protected

Academic year: 2022

Share "Transport of very short lived halogenated substances from the tropical East Pacific to the stratosphere and the influence of El Niño 2015/16"

Copied!
75
0
0

Laster.... (Se fulltekst nå)

Fulltekst

(1)

halogenated substances from the tropical East Pacific to the stratosphere

and the influence of El Niño 2015/16

Sarah G JERMO

Thesis submitted for the degree of Master in Meteorology and Oceanography (Ocean & Middle Atmosphere Interactions)

60 credits

Department of Geosciences

Faculty of Mathematics and Natural Sciences UNIVERSITY OF OSLO

November 14, 2017

(2)

Transport of very short lived halogenated substances from the tropical East Pacific to the stratosphere and the influence of El Niño 2015/16

http://www.duo.uio.no/

Printed: Reprosentralen, University of Oslo

(3)

OSLO UNIVERSITY

Abstract

MetOs

Department of Geosciences

Master in Meteorology and Oceanography

Transport of very short lived halogenated substances from the tropical East Pacific to the stratosphere and the influence of El Niño 2015/16

by Sarah GJERMO

Natural, oceanic sources of atmospheric very short-lived halogenated substances (VSLS) that are transported to the stratosphere contribute to catalytic ozone depletion. The aim of this thesis was to determine the contribution of the VSLS; methyl iodide, bromoform, and dibromomethane, emitted from the tropical East Pacific (EP) to the stratospheric halogen loading, and how this contribution was influence by El Niño 2015/16. The Lagrangian particle dispersion model FLEXPART with ERA-Interim meteorological re- analysis fields was used to simulate the path of the VSLS emitted from the tropical EP, for the two case studies; El Niño 2015/16 and neutral ENSO 2012/13. Two different groups of model experiments were conducted: In the first group the VSLS were released in the model simulation according to measurements taken on the ASTRA-OMZ cruise (Oct 2015) and on the M91 cruise (Dec 2012). The second group focused only on bro- moform, and bromoform was constantly released over a large area of the tropical and southern EP boreal Fall/Winter 2015/16 and 2012/13. The stratospheric contribution of the VSLS was found to be significantly affected by how close to maximum entrainment locations they were released. El Niño 2015/16 enhanced the vertical transport of the VSLS in the tropical EP by 37 % compared result from the neutral ENSO 2012/13, indicating that VSLS transport to the stratosphere from the tropical EP is considerably influenced by El Niño. The Peruvian upwelling appeared not to be an essential source region for methyl iodide contributing to the stratospheric iodide loading, whereas, it turned out to be a significant source region for bromoform and dibromomethane.

(4)
(5)

Contents

Abstract i

Contents iii

1 Introduction 1

2 Background 5

2.1 Atmospheric Transport and Circulation . . . 5

2.1.1 The Hadley Cell . . . 5

2.1.2 The Walker Circulation . . . 6

2.1.3 The El Niño Southern Oscillation . . . 7

2.1.4 Troposphere-to-Stratosphere Transport in the Tropics . . . 8

2.2 Methyl iodide, Bromoform and Dibromomethane . . . 10

2.2.1 Marine Sources . . . 10

2.2.2 Transport from the Ocean to the Stratosphere . . . 10

2.2.3 Atmospheric Removal and Ozone Depletion . . . 12

2.3 Meteorology during ASTRA-OMZ and M91 . . . 13

3 Data and Methods 19 3.1 The ASTRA-OMZ Cruise . . . 19

3.1.1 Meteorological Observations . . . 19

3.1.2 Surface Ocean and Atmospheric Halocarbon Measurements . . . . 19

3.1.3 Halocarbon Emissions . . . 19

3.2 The FLEXPART Model and ERA-Interim Data . . . 21

3.3 Cruise VSLS Emissions Experiment . . . 23

3.3.1 VSLS Lifetime Profiles . . . 23

3.3.2 FLEXPART Setup . . . 23

3.4 East Pacific Bromoform Emission Experiment . . . 25

3.4.1 Surface water concentrations . . . 25

3.4.2 Choosing surface atmospheric concentrations . . . 25

3.4.3 Final bromoform emission fields . . . 29

3.4.4 FLEXPART Setup . . . 30

4 Results and Discussions 31 4.1 ASTRA-OMZ and M91 Data . . . 31

4.2 Cruise VSLS Emissions Experiment . . . 39

4.3 East Pacific Bromoform Emission Experiment . . . 45

5 Summary and Outlook 53

List of Abbrevations 55

List of Figures 57

List of Tables 61

(6)

Bibliography 63

Acknowledgements 69

(7)

Chapter 1

Introduction

Oceanic halocarbons contribute to ozone depletion in the stratosphere (Carpenter et al., 2014b; Solomon et al., 1994), and thus their transport to the stratosphere is of interest. El- evated concentrations of oceanic halocarbons, especially over oceanic upwelling regions in the tropics and subtropics, are related to biological activity (Hepach et al., 2014, 2016;

Quack et al., 2007). Because of the pronounced coastal upwelling, the tropical East Pa- cific (EP) is one of the most productive regions worldwide (Mann and Lazier, 2013). El Niño–Southern Oscillation (ENSO) has a large impact on the equatorial Pacific and the Walker Circulation (WC), and is therefore also likely to affect oceanic halocarbon emis- sions and their transport to the stratosphere.

Halocarbons are carbon based compounds containing halogen atoms. They are po- tential carriers of halogens to the stratosphere, which contribute to catalytic ozone de- pletion. In this thesis three halocarbons prominent to ozone depletion, namely methyl iodide, bromoform and dibromomethane, are investigated. These halocarbons have life times of less than 6 months in the atmosphere, and are therefore termed very short-lived halogenated substances (VSLS) (Carpenter et al., 2014b). In this thesis the transport of the VSLS emitted from the Tropical EP, to the stratosphere is studied using the Lagrangian FLEXPART model together with ERA-Interim reanalysis data. Calculations of strato- spheric entrainment, over the tropical East Pacific will be compared to findings from other tropical oceans, and the impact of El Niño 2015 will be compared with ENSO neu- tral state in the EP in 2012. The El Niño 2015 was one of the four strongest El Niños since 1950 (Stramma et al., 2016).

Since the ozone hole over Antarctica in late austral winter and early spring first was discovered (Chubachi, 1984; Farman et al., 1985), a huge effort has been devoted to ozone research. Ozone has a unique role in the stratosphere, because it absorb ultraviolet solar radiation, and thus acts as a protective shield around our planet. The health of humans, animals and plants are affected by increased ultraviolet light transmitted through the ozone layer (Solomon, 1999). Before the ozone hole was discovered, anthropogenic ozone depleting substances (ODS), mainly chlorofluorocarbons (CFC’s), was widely used both in industry and in households. These halogenated substances had only been demon- strated to deplete ozone in laboratories, and environmental damage caused by strato- spheric ozone depletion had not yet been witnessed. In spite of the uncertainties, leaders from 36 nations joined together in 1989 and signed the Montreal Protocol to reduce or eliminate the use of these substances (DeSombre, 2000). The Montreal Protocol is now considered a major success in reducing the emission of anthropogenic ODS, and finally the ozone hole is showing signs of healing (Solomon et al., 2016). Research on the ozone depletion is however still of major interest, as the regular stratospheric WMO Ozone as- sessments highlight.

The principal cause of stratospheric ozone depletion is the long-lived anthropogenic substances (Carpenter et al., 2014a), but recent observations have shown that also the VSLS are important stratospheric halogen sources (Dorf et al., 2006; Laube et al., 2008;

(8)

Sturges et al., 2000). Dorf et al. (2006) found that the measured stratospheric ozone- depleting bromine burden (10 years of balloon-borne observations) was significantly higher than what long-lived gases could be accounted for. The discrepancy was among others attributed to natural originated VSLS (Dorf et al., 2006). Henceforth, researches have been striving for a better understanding of the VSLS sources, emissions to the at- mosphere, atmospheric chemistry, and their transport pathway to the stratosphere.

The main research method for estimating global oceanic halocarbon emissions that has been used is: the bottom-up approach based on surface-water and atmospheric halo- carbon measurements (Quack and Wallace, 2003). In contrast, the top-down approach is based on chemical transport modeling studies reproducing measurements of atmo- spheric VSLS concentrations (Warwick et al., 2006). There is a gap in the estimation of the oceanic halocarbon emission between the two methods, where the top-down approach leads to higher VSLS emission values (Hossaini et al., 2010; Liang et al., 2010). A possible explanation for that gap may be an under-representation of coastal emissions and small spatial and temporal extreme emission events in the bottom-up emission estimations (Ziska et al., 2013). Hossaini et al. (2010) claims that one uncertainty bottom-up approach was the assumption of a constant prescribed lifetime of Br_y in the troposphere, and Tegt- meier et al. (2013) suggested that the available upper air measurements of methyl iodide were not representative of global estimates due to strong variations in the geographical methyl iodide entrainment distribution. Hence, the top-down emission approach may therefore lead to overestimations. A problem with both of these approaches is that they don’t include seasonality for the surface oceanic concentrations. A newer model study has now been introduced; using a biogeochemical model where both the biological and photochemical production mechanisms for the VSLS are considered, with the production following the seasonal insolation cycle (Hense and Quack, 2009; Stemmler et al., 2015).

There are several uncertainties to the processes of VSLS contribution to stratospheric halogen loading and ozone depletion. A large part of the uncertainty is related to the concentration of VSLS in the marine atmospheric boundary layer (MABL) (Quack and Wallace, 2003). Significant amounts of VSLS have been observed over oceanic upwelling in the eastern tropical Atlantic, and may be solely explained by oceanic sources, without the need of additional continental sources (Hepach et al., 2014). Fuhlbrügge et al. (2013) found that the MABL height variations influences the volume mixing ratio (VMR) of halocarbons especially over oceanic upwelling systems, and it has been hypothesize that a common phenomenon over coastal upwelling systems is low MABL height, high halo- carbon emissions and high atmospheric mixing ratios (Fuhlbrügge et al., 2013; Hepach et al., 2014).

Another major uncertainty is the process of stratospheric VSLS transport and scav- enging in the tropical tropopause layer (TTL). The injection of stratospheric halogen from VSLS comprises both the VSLS source gas injection (SGI) and product gas injection (PGI) (Ko and Poulet, 2003). Liang et al. (2014) studied the convective transport of the very short lived bromocarbons using the NASA Goddard Earth Observing System (GEOS) Chemistry Climate Model (GEOSCCM), and found that the amount of bromocarbons reaching the stratosphere was actually weakened for very strong convection conditions, opposite to earlier presumptions. This weakening was attributed to the increased scav- enging of the soluble product gases, which lead to a high decrease in PGI exceeding the minor SGI increase (Liang et al., 2014). In this thesis I am only focusing on the direct transport of the VSLS source gases.

Recent papers on the effect of ENSO variations on the convection of VSLS to the stratosphere are Aschmann et al. (2011) and Ashfold et al. (2012). By long-term modeling

(9)

the impact of VSLS on stratospheric bromine loading using a three-dimensional chem- istry transport model, Aschmann et al. (2011) found that intensified atmospheric con- vection leads to higher amounts of VSLS in the upper troposphere/lower stratosphere, especially under extreme conditions like El Niño seasons. Ashfold et al. (2012) used a trajectory model (NAME) to investigate the timescales over which air parcels reach the TTL above Borneo (West Pacific), and they compared the ENSO neutral year 2008 to the moderate El Niño year 2006 and the moderate La Niña year 2007. They found that more parcels traveled from the boundary layer to the TTL during the La Niña year, and less during the EL Niño year, although one should bear in mind that the influence of ENSO is different in other parts of the Tropics (Ashfold et al., 2012).

Researchers have now started to get a better understanding of the impact of the halo- carbons to the stratospheric ozone depletion (Carpenter et al., 2014b). However, there is still a gap between modeled halogen loading and halogen measurements in the strato- sphere, and therefore more studies are necessary. The pronounced oceanic coastal up- welling and the tropical atmosphere suggests that halocarbon emissions from the EP may be important for the stratospheric halogen loading. This is however not necessarily true, because the tropical EP is also a region with climatologically sinking of air masses, as it is the eastern branch of the WC, suppressing the vertical transport of the VSLS. Fur- thermore, the Peruvian upwelling along the West Coast of South America leads to a pro- nounced stable MABL layer, acting as a transport barrier for the VSLS (Fuhlbrügge et al., 2016a). Hardly any studies have investigated the transport of VSLS from the Tropical EP to the stratosphere (Aschmann et al., 2011). Hence, it is not yet known how important the region is for the stratospheric halogen budget. Moreover, the EP is a key region when studying the effect of El Niño, as the main convection over the West Pacific ocean shifts eastward during an El Niño event, closer to the EP VSLS sources. Furthermore, the weak- ening of the WC during El Niño events decreases the airmass suppression over the EP (Wallace and Hobbs, 2006), favoring more convection. The regional setting of this thesis, makes it therefore very relevant for this field of study.

The aim of this thesis is to investigate the role of VSLS transport to the stratosphere above the tropical EP, and to contribute to the research on El Niño’s influence on this transport. In particular, this thesis will examine three main research questions: 1. To what extent is the tropical EP a source for VSLS to the atmosphere? 2. How much of these VSLS is transported to the stratosphere? 3. How does El Niño affect the VSLS transport from the tropical EP to the stratosphere?The reader should bear in mind that this study is based on two ship campaigns carried out in the tropical EP, namely ASTRA- OMZ (Oct 2015) and M91 (Dec 2012), and so the data used is limited in time and space.

The overall structure of the thesis takes the form of five chapters, including this intro- ductory chapter. Chapter 2 begins by laying out the background information necessary for the scientific research. The 3. Chapter is concerned with the data and method used for this thesis. Chapter 4. presents the results and discussions, which is divided into three sections: In the first section a meteorological overview is given, together with halocar- bon measurements from the ASTRA-OMZ cruise, and a comparison with the M91 cruise among others. In Section 2, a FLEXPART model study of the transport of the VSLS to the stratosphere, based on the ASTRA-OMZ and the M91 campaigns, is presented. In the 3.

Section a regional case study of bromoform is given. Two model experiments were setup with bromoform emissions over the tropical and southern EP, one for El Niño 2015 and one for ENSO Neutral 2012. Modeled VSLS VMR results from this section are compared with in-situ cruise measurements in the MABL and in-situ aircraft measurements. The final chapter concludes the results and highlights the implication of the findings to future research in this area.

(10)
(11)

Chapter 2

Background

A brief introduction to the atmospheric transport and circulation relevant for this the- sis is presented in Section 2.1, consisting of the four subsections: The Hadley Cell, The Walker Circulation, The El Niño Southern Oscillation, and Troposphere-to-Stratosphere transport in the tropics. The three halocarbons methyl iodide, bromoform, and dibro- momethane, which transport to the stratosphere is investigated in this thesis, are pre- sented in Section 2.2. Section 2.2 consists of the subsections: Marine Sources, transport from the Ocean to the Stratosphere, Atmospheric Removal and Ozone Depletion. Fi- nally an overview of the meteorological setting at the time and place of the two cruises ASTRA-OMZ and M91, which this thesis is based upon, is given in Section 2.3.

2.1 Atmospheric Transport and Circulation

2.1.1 The Hadley Cell

The Hadley cell is named after George Hadley (1685-1768), who was an English meteorol- ogist (Wallace and Hobbs, 2006). When seeking for the origin of the trade winds, Hadley realized that they must be caused by the uneven distribution of solar insolation between the equator and the poles. He visualized one great thermally directly driven cell on each hemisphere. Heated air is convected over the equator and is transported towards the poles where it cools, sinks, and flows back towards equator (Holton, 2004, pp. 314–316).

Although Hadley’s idea of one major cell in each hemisphere did not prevail, his concepts of that differences in heating give rise to persistent large-scale atmospheric overturning circulation and of that zonal winds can be attributed deflection of meridional winds have (Aguado and Burt, 2010, pp. 201–202). A more realistic model is the three-cell model dividing the circulation of each hemisphere into three major transport cells, namely; the Hadley cell which circulates air between the tropics and sub-tropics, the mid-latitudinal Ferrel cell and the Polar cell (Aguado and Burt, 2010, p. 203).

The Hadley cell is a thermally direct cell, and it can be described as following: Con- vergence of warm and moist air near the equator, results in major convection, heavy precipitation, and release of latent heat. This zone of rising air, varies gradually over the seasons, and is called the Intertropical Convergence Zone (ITCZ). The rising air reaches the tropical tropopause layer (TTL), and is then forced poleward. The rotation of the earth leads to an eastward deflection of the winds, and the resulting subtropical jet streams. At about 30N/S, the winds start to subside. The now cool and dry air warms adiabatically, while sinking and traps the underlying moist and cold maritime air below. This results in an inversion layer called the trade wind inversion. The air then moves back towards the equator, absorbs moisture on its way, and completes the loop. Again because of the Coriolis effect, the winds are deflected westward, creating the southeasterly trade winds (Aguado and Burt, 2010, pp. 203–206). In Figure 2.1 a conceptual model of the Hadley cell is depicted.

(12)

FIGURE 2.1: Conceptual model of the Hadley Cells, together with the tropical tropopause layer (TTL) and the Inter-tropical convergence zone

(ITCZ). Adapted from Fiehn (2017, p. 6).

FIGURE2.2: Generalized Walker Circulation during ENSO Neutral condi- tions. NOAA Climate.gov drawing by Fiona Martin (Liberto, 2014)

2.1.2 The Walker Circulation

Due to the southeasterly trade winds over the equatorial Pacific, the surface water is pushed westward. This results in warm water piling up at the west Pacific, and a con- tinuous upwelling of cold water at the eastern boundary. The warm surface water in the west, heats the air above, leading to low surface pressure and a major convection center over the western Pacific. The rising air travels east over the ocean, resulting in high sur- face pressure over the eastern Pacific where dry air subsides. This east-west circulation of air over the equatorial Pacific ocean, is called the Walker Circulation (WC), and it is named after G. T. Walker who was the first to document the surface pressure pattern as- sociated with it (Holton, 2004, pp. 382–382). The WC is dependent on El Niño–Southern Oscillation (ENSO) conditions. The above description is valid for the ENSO neutral con- dition (see Figure 2.2).

(13)

FIGURE2.3: Conceptual model of a La Niña event, with the generalized Walker Circulation over a map of anomalous sea surface temperatures.

Blue indicates anomalous ocean cooling, and orange anomalous ocean warming. NOAA Climate.gov drawing by Fiona Martin (Liberto, 2014)

FIGURE2.4: Conceptual model of an El Niño event. NOAA Climate.gov drawing by Fiona Martin (Liberto, 2014)

2.1.3 The El Niño Southern Oscillation

ENSO describes an ocean atmospheric coupled oscillation, where EN stands for El Niño which is a recurrent pattern of positive temperature anomalies in the equatorial Pacific sea surface, and SO is the Southern Oscillation, an interannual fluctuation of atmospheric pressure between Darwin and Tahiti. ENSO is an irregularly interannual variation of the surface temperatures and winds over the Pacific Ocean, affecting the global climate (Wal- lace and Hobbs, 2006, pp. 431-438). ENSO has a warm phase and a cold phase, which are called El Niño and La Niña respectively, when the surface temperature anomaly is above 5 for several months in a row. During La Niña, the Walker Circulation is strengthened.

Even more warm surface water piles up in the west, leading to enhanced convection over the West Pacific, stronger subsidence of air over the East Pacific, intensified upwelling in the coastal East Pacific, and large Darwin-Tahiti pressure differences (Figure 2.3). While during an El Niño event (Figure 2.4), the Pacific WC is weakened, or even reversed, and the Darwin-Tahiti pressure difference is small. The convection center over the West Pa- cific propagates eastward, leading to less subsidence of air over the East Pacific, or even

(14)

FIGURE 2.5: Schematic of tropical deep convection, with the convective boundary layer (CBL), and the tropical tropopause level (TTL). The level of the cold point tropopause (CPT) and the level of zero radiative heat- ing (LZRH) is shown. Pink lines indicate typical tracer routes, red arrows mass redistribution, and a typical temperature profile is shown in green

(Carpenter et al., 2014a, p. 1.34).

convection i.e. precipitation over the Atacama desert (Holton, 2004, pp. 384–385). Cur- rently, many classification methods are used for identifying the type of an El Niño event.

One such method is to analyze the type according to where the surface ocean tempera- ture anomalies are situated (Pascolini-Campbell et al., 2015). There are two types of El Niño; the Eastern Pacific El Niño and the Central Pacific El Niño.

2.1.4 Troposphere-to-Stratosphere Transport in the Tropics

Over the past decades, it has become clear that the transition from the troposphere to the stratosphere is gradually rather than abrupt. In the tropics, a transitional regime between the troposphere and stratosphere extends over several kilometers. This transi- tional regime in the tropics has been named the Tropical Tropopause Layer (TTL). The TTL spans the distance between the connectively dominated overturning circulation of the Hadley cell, to the slow upwelling region of the lower stratospheric Brewer-Dobson circulation (Fueglistaler et al., 2009). Air enters primarily the stratosphere in the Trop- ics. Thus the TTL acts as a gateway for atmospheric tracers to the stratosphere (Holton et al., 1995), such as for the very short-lived halogenated substances (VSLS). The base of the TTL is defined as the height of the temperature lapse rate minimum. The cold point tropopause (CPT) is used to define the top, and , the level of zero radiative heat- ing (LZRH) is in the middle (Carpenter et al., 2014a) (Figure 2.5).

Transport and chemical processes in the TTL are important for the VSLS source gas injection (SGI) and product gas injection (PGI) (Carpenter et al., 2014a). Significant trans- port to the stratosphere of particularly short-lived substances, with atmospheric lifetimes of several days or less, is unlikely unless emitted close to deep convection (Hossaini et al., 2012). Very deep overshooting convection may transport air masses directly through the TTL, allowing also the shortest-lived VSLS to reach the stratosphere, although these

(15)

events are relatively rare (Carpenter et al., 2014a). See Figure 2.5 for an illustration of deep convection and overshooting convection, together with the TTL.

Trajectory calculations from several studies found that troposphere-to-stratosphere transport (TST) trajectories mostly entered the TTL over the West Pacific (Bonazzola and Haynes, 2004; Fueglistaler and Haynes, 2005; Fueglistaler et al., 2004; Hatsushika and Yamazaki, 2003). The tropical West Pacific is under neutral ENSO conditions, a region with major vertical transport of air, due to the rising branch of the WC. The largest mod- ifications of the TST occur due to major ENSO events (Fueglistaler et al., 2004). Krüger et al. (2008) found that the TTL becomes colder and drier during La Niña over the western Pacific, and warmer and less dry during El Niño.

(16)

2.2 Methyl iodide, Bromoform and Dibromomethane

In this thesis, the troposphere-to-stratosphere transport (TST) of the three halocarbons methyl iodide, bromoform, and dibromomethane, is studied. Halocarbons are carbon based compounds containing halogen atoms, where halogen is the name for group seven of the periodic table of the elements, including e.g. chlorine, bromine and iodine. The three halocarbons studied in this project, are recognized as being contributers to the stratospheric halogen loading (Carpenter et al., 2014b), where they are involved in ozone depletion (Carpenter et al., 2014b; Salawitch et al., 2005). The knowledge about the halo- carbons spatial and temporal variability in emission, loss, and transport processes, are largely based on limited data (Hepach et al., 2016; Quack et al., 2007; Ziska et al., 2013).

It is known that the main atmospheric source for these halocarbons is oceanic (Carpenter et al., 2014b). Although, oceanic measurements of halocarbons are few, high concentra- tions especially over upwelling regions in the tropics and subtropics have been found.

Thus, they are considered important source regions (Hepach et al., 2016; Quack et al., 2007). The intense Peruvian Upwelling is considered one of the most productive oceanic regions in the world (Mann and Lazier, 2013). Therefore, studying halogen transport from this region is of great interest.

2.2.1 Marine Sources

Bromoform and dibromomethane are the primary natural contributors to atmospheric organic bromine (Quack et al., 2007). The source for the bromine-containing source gases is mainly natural and oceanic. For methyl iodide, the oceanic contribution to the atmo- spheric loading is more than 80% (Carpenter et al., 2014b).

The dominant producer of bromocarbons in the open ocean is phytoplankton (Moore et al., 1995b). Other known sources are macro algae and cyanobacteria (Nightingale et al., 1995; Wever and Horst, 2013). Hill and Manley (2009) suggested that a considerable formation pathway for the production of polyhalogenated compounds may be indirect through algal release of hypoiodous acid (HOI) and hypobromous acid HOBr, reacting with dissolved organic matter (DOM). The seaweed and phytoplankton produces H2O2 during photosynthesis and photorespiration (and stress for macro algae), which reacts with bromine in the water, to form HOBr, by enzymatic activity of haloperoxidases (Hep- ach et al., 2016). The HOBr then reacts with DOM, and forms polybromomethanes like bromoform and dibromomethane (Wever and Horst, 2013).

For methyl iodide is produces by non-biological, photochemical degradation of io- dide containing DOM, and biologically by algae and phytoplankton (Carpenter et al., 2014b; Tegtmeier et al., 2013). Methyl iodide may possibly also be formed via bacteria (Hepach et al., 2016)

Hepach et al. (2016) found the Peruvian upwelling region to be only a moderate source region for bromocarbons, but significant source region for iodocarbons, Decem- ber 2012. Previously high concentration of iodocarbons in the tropical oceans have been connected with mainly the photochemical source (Hepach et al., 2016). Stemmler et al.

(2015) used a global three-dimensional ocean biogeochemistry model to simulate bromo- form cycling in the ocean, and it was found to match observations well.

2.2.2 Transport from the Ocean to the Stratosphere

When the ocean is supersaturated with halocarbons, the halocarbons are emitted from the ocean, transported horizontally and vertically mixed in the marine atmospheric bound- ary layer (MABL). The rate at which the halocarbon ocean-to-atmosphere exchange oc- curs depends on the air-sea halocarbon concentration gradient. The air-sea concentration

(17)

FIGURE 2.6: Schematic of the oceanic sources and the atmospheric pro- cesses relevant for methyl iodide (CH3I), bromoform (CHBr3) and dibro-

momethane (CH2Br2).

gradient is significantly affected by oceanic upwelling (Fuhlbrügge et al., 2013, 2016a).

When there is coastal oceanic upwelling of cold water to the surface, the air over the ocean cools, resulting in a stable and isolated MABL and high atmospheric halocarbon mixing ratios. The high atmospheric mixing ratios decreases the halocarbon sea-air con- centration gradient, hence the emission to the atmosphere is reduced. Fuhlbrügge et al. (2013, 2016a) found that a strong trade inversion acts as a transport barrier, leading to a near-surface accumulation of halocarbons in the atmosphere. The trade inversion is a temperature inversion that occurs due to large scale subsidence in the descending branches of the Hadley Cell and the Walker Cell, and is found where the cold dry sub- sided air meets the underlaying warm moist air. The coastal emission of oceanic halo- carbons also vary due to the change in amount and types of algae, and with the diurnal and tidal cycles (Carpenter et al., 2014b; Wever and Horst, 2013). Local emission maxima linked to upwelling areas, over the tropical oceans, have been observed (Quack et al., 2007; Wever and Horst, 2013). Stemmler et al. (2015) simulated emissions of bromoform into the atmosphere, using observational-based estimates from Ziska et al. (2013) of near- surface atmospheric bromoform volume mixing ratio (VMR) as upper boundary condi- tion. These were found to be lower than previous estimates by Ziska et al. (2013). This is because seasonality is considered and less bromoform is produced in non-blooming seasons reversing the sea-air flux of bromoform, and also because coastal emissions of bromoform is not represented in this model (Stemmler et al., 2015).

VSLS are defined as substances with atmospheric lifetimes of less than six months (WMO, 2006). The current estimated atmospheric lifetimes at 10 km altitude are 17 days for bromoform, 150 days for dibromomethane, and 3.5 days for methyl iodide (Carpen- ter et al., 2014b). Hence, all three compounds belong to the VSLS category. Especially short-lived VSLS, like methyl iodide and bromoform, need to be emitted close to deep, convective systems to be able to reach the Stratosphere (Tegtmeier et al., 2013). Their impact on stratospheric halogen loading is still uncertain due to limited observations (Carpenter et al., 2014b). Several studies have found indications that methyl iodide and bromoform reaches the stratosphere over tropical regions, i.e., Fiehn et al. (2017), Hos- saini et al. (2015), Saiz-Lopez et al. (2015), and Tegtmeier et al. (2013) in despite of the stable MABL and the trade inversion.

The VSLS sources to the stratosphere halogen loading is typically separated between

(18)

SGI and PGI (Ko and Poulet, 2003), where the SGI is the direct injection of the VSLS sources, and the PGI is the injection of halogens from the atmospheric degraded VSLS products. In this thesis, only the SGI of VSLS has been focused on.

2.2.3 Atmospheric Removal and Ozone Depletion

The main sink of methyl iodide in the troposphere is photolysis (Carpenter et al., 2014b).

The tropospheric sinks of bromocarbons are OH oxidation and photolysis (Carpenter et al., 2014b). Photolysis is the most important removal for bromoform, and a major sink process is oxidation by OH radicals for dibromomethane (Carpenter et al., 2014b).

Other sinks of atmospheric VSLS are uptake by the oceans and soil microbial degradation (Carpenter et al., 2014b).

When bromine and iodine containing halocarbons are degraded in the atmosphere, they form reactive halogen radicals. This takes place both in the troposphere and strato- sphere, and they therefore differ from chlorofluorocarbons (CFC’s) which can only be broken down in the Stratosphere by ultra-violet radiation (Wever and Horst, 2013). One of the halogen radical’s most important reaction is ozone depletion. Ozone in the tropo- sphere is an active greenhouse gas, and it is also toxic for humans. The halocarbon emis- sions from the sea can lower tropospheric ozone, which contribute to reducing global warming, and inproving air quality. The VSLS source gases and product gases which reaches the stratosphere will take part in catalytic ozone depletion there (Wever and Horst, 2013). Thus, VSLS contribute to increased transmission of harmful ultraviolet light through the ozone layer.

PGI of bromine containing VSLS makes a non-negligible contribution to the strato- spheric bromine loading, with bromoform and dibromomethane as the most important sources. The PGI of brominated VSLS can range from 1.1 to 4.3 ppt. By including bromine containing VSLS in modeling studies, the modeled O3trends have been closer to obser- vations. The PGI of idodine containing VSLS to the stratosphere is still uncertain. It has been estimated to be less than 0.15 ppt, and is suggested to be a minor sink for O3 (Hossaini et al., 2012).

(19)

2.3 Meteorology during ASTRA-OMZ and M91

The meteorological setting at the time and place of the two cruises ASTRA-OMZ and M91, is presented in this section. The cruises took place along the west coast of South America, during October 2015 (ASTRA-OMZ) and during December 2012 (M91). More information about ASTRA-OMZ is given in Chapter 3, and the M91 cruise is described in more detail in Fuhlbrügge et al. (2016a) and Hepach et al. (2016).

The ASTRA-OMZ cruise in took place during the development of a very strong El Nino. The development of the 2015/16 El Niño was anticipated by researchers at the beginning of 2015 (Hu and Fedorov, 2017). The year before, in 2014, the same projection was made, but no El Niño developed. This time, however, researchers were right. An extreme El Niño event developed in 2015/16 (Wang and Hendon, 2017). According to Stramma et al. (2016), the 2015 El Niño was a clear EP El Niño in October 2015. However, the 2015/2016 event became dominated by the CP El Niño dynamics after October 2015, as reported by Paek et al. (2017). The El Niño started early in 2015, but the shift to El Niño water mass distribution in the equatorial East Pacific (EP) was surprisingly slow (Stramma et al., 2016). In October 2015 the El Niño signal was found to be strongest at the equator, along the ASTRA-OMZ cruise track (Stramma et al., 2016). Large positive temperature anomalies over the equatorial EP in October 2015, can be seen in Figure 2.7a, highlighting the ongoing El Niño event (CDB, October 2015). At the time of the M91 cruise in December 2012, there was an ENSO neutral state, as seen in Figure 2.7b (CDB, December 2012). The negative outgoing longwave radiation (OLR) anomaly along the Peruvian coast indicates enhancement of oceanic upwelling (Figure 2.7b).

A discrepancy in the convective situation over the tropical Pacific between ASTRA- OMZ and M91 is expected because of the different ENSO states. A measure of deep con- vection is the anomaly of OLR. Negative OLR anomalies indicates enhanced convection, more cloud coverage, and higher and colder cloud tops which emits less infrared radia- tion into space. The opposite occurs for positive OLR anomalies. As expected, stronger convection in the tropical EP prevail during October 2015 than in December 2012 (Figure 2.8). There are no significant anomalies in the OLR in December 2012, which is typical for an ENSO neutral year, although there is some enhanced convection over the Andes Mountains.

The evaluation of the convective situation during El Niño 2015 and ENSO neutral 2012, is of interest since this thesis investigating the transport of VSLS to the strato- sphere. Hence, a time-longitude section of the anomalous equatorial OLR for 2015/16 and 2012/13 is shown in Figure 2.9. The shift of the convection centre from the West Pacific to the East and Central Pacific can be seen in Figure 2.9a.

Under El Niño conditions, more accentuated tropical convection has previously been found to drive a stronger and narrower Hadley Cell (HC) (Chang, 1995; Seager et al., 2003). The width of the Hadley Cell (HC) is limited by the subtropical jet. Hence, the 200 hPa winds are presented for October 2015 and December 2012 in Figure 2.10. The subtropical jets were situated farther north over the EP in October 2015 than in November 2012 (Figure 2.10). The position and stenght of the subtropical jets, from November to December in 2012 and 2015 appeared similar, although the southern subtropical jet was slightly enhanced for 2015 Center (CDB).

In Figure 2.11, radiosonde measurements of relative humidity (RH) from the two cruises ASTRA-OMZ and M91 are shown. It should be noted that two different types of radiosondes were used; ASTRA-OMZ used Graw, and M91 used Vaisala sondes. The av- erage CPT is located at about 17 km for both cruises (Alina Fiehn and Steffen Fuhlbrügge, personal communication, Nov 2017).

(20)

(A)

(B)

FIGURE2.7: Monthly sea surface temperature and temperature anomaly average for (A) ASTRA-OMZ , (B) M91 (CDB, Oct 2015 and Dec 2012).

A pronounced humid layer at around 1 km is found for both cruises (Figure 2.12), which is characteristic of the trade inversion layer. The trade inversion layer was higher for ASTRA-OMZ, especially when the cruise crossed the equator (Figure 2.12). In the free troposphere, less humidity was detected during ASTRA-OMZ than during the M91 cruise. This may indicate weaker vertical transport of humid air masses from the MABL

(21)

(A)

(B)

FIGURE2.8: Monthly outgoing longwave radiation and radiation anomaly average for (A) Oct 2015, ASTRA-OMZ , (B) Dec 2012, M91 (CDB, Oct 2015

and Dec 2012).

into the free troposphere during ASTRA-OMZ than during M91. It may also be due to

(22)

(A) (B)

FIGURE2.9: Anomalous outgoing longwave radiation averaged between 5N-5S for (A) 2015/16 and (B) 2012/13 (CDB, Mar 2016 and Mar 2013).

radiosonde differences (Kirstin Krüger, personal communication, Nov 2017). The mean height of the MABL was 307 m for M91 (Fuhlbrügge et al., 2016a) and 470 m for ASTRA- OMZ (Alina Fiehn, personal communication, Nov 2017).

(23)

(A)

(B)

FIGURE2.10: Monthly wind speed and wind speed anomaly average for (A) Oct 2015, ASTRA-OMZ , (B) Dec 2012, M91 (CDB, Oct 2015 and Dec

2012).

(24)

FIGURE 2.11: Relative humidity in the troposphere and lower strato- sphere, for the M91 cruise (left) and ASTRA-OMZ (right), measured with Vaisala and Graw radiosondes respectively. CPT = cold point tropopause, LRT = lapse-rate tropopause (Alina Fiehn and Steffen Fuhlbrügge, per-

sonal communication, Nov 2017).

FIGURE 2.12: Relative humidity in the lower troposphere, for the M91 cruise (left) and ASTRA-OMZ (right) (Alina Fiehn and Steffen Fuhlbrügge,

personal communication, Nov 2017).

(25)

Chapter 3

Data and Methods

3.1 The ASTRA-OMZ Cruise

Data from the two cruises ASTRA-OMZ and M91 is used in this thesis. The ASTRA-OMZ cruise was conducted on the R/VSONNE(SO243; 5 to 22 October 2015) from Guayaquil in Ecuador to Antofagasta in Chile, and the M91 cruise was carried on R/VMETEOR(1 to 26 December 2012) starting and ending in Lima, Peru. Descriptions of the meteorological observations, and the oceanic and atmospheric halocarbons, for ASTRA-OMZ, are given in the subsequent sections. The M91 cruise have been described earlier by Fuhlbrügge et al. (2016a) and Hepach et al. (2016).

3.1.1 Meteorological Observations

Meteorological observations of sea surface temperature (SST), surface air temperature (SAT), wind speed, wind direction, relative humidity, and air pressure were done every second. The wind were measured at about 30 m height above sea level. The data were averaged to 10 min intervals. GRAW DFM-09 radiosondes were regularly launched every six hours, with a total of 64 launches (Alina Fiehn, personal communication, Nov. 2017 and Marandino, 2016).

3.1.2 Surface Ocean and Atmospheric Halocarbon Measurements

Both surface oceanic and atmospheric halocarbon measurements were collected every 3 hours. Water samples were taken at a depth of about 5 m from a continuously working water pump in the hydrographic shaft of the ship. The water samples were then analyzed for halogenated trace gases with a gas chromatographer attached to a mass spectrome- ter (GC/MS) onboard the ship (Alina Fiehn, personal communication, Nov. 2017 and Marandino, 2016). The precision of the analysis is of 10% (1α). More detailed description of the measurements is given by Hepach et al. (2014).

The atmospheric air samples were taken at about 10 m height above sea level at the bow of the ship, using a jib of 4 meters. The samples were collected in stainless steel canisters, pressurized to 2 atm, and later analyzed at the Rosenstiel School for Marine and Atmospheric Sciences, University of Miami (Alina Fiehn, personal communication, Nov.

2017 and Marandino, 2016). Details about the atmospheric very short-lived halogenated substances (VSLS) samplings can be read in Fuhlbrügge et al. (2013).

3.1.3 Halocarbon Emissions

For calculating sea-to-air VSLS emission, the level of halocarbon saturation in the ocean surface layer is taken and converted to a flux by multiplying with an average transfer ve- locity (kw) (Moore et al., 1995a). The level of saturation is given as the difference between

(26)

the actual water concentration (cw) and the water concentration at which the concen- tration is at equilibrium with the above air. The theoretical equilibrium concentration is given as catmH , wherecatm is the atmospheric concentration of the halocarbon, and His Henry’s law constant. Henry’s law constant is defined as the concentration in air devided by the equilibrium water concentration (Moore et al., 1995a).

F= kw·(cwcatm

H ) (3.1)

The transfer velocity coefficient by Nightingale et al. (2000) was used for the cruise VSLS emissions experiment.kwvaries with sea level pressure, sea surface temperature, sea sur- face salinity, and the wind speed at 10 m height. Air pressure and sea surface temperature are taken from the ERA-Interim monthly means, and the sea surface salinity is taken from the World Ocean Atlas 2005. The 10 m wind speeds are parameterized from the observed wind speeds during ASTRA-OMZ and M91, using a logarithmic wind profile:

u10=u(z) κ

√CD κ

√CD+log(10z) (3.2)

whereκ = 0.41 is the von Kármán constant, CD is the neutral drag coefficient (Garratt, 1977), andz is the height of the observed wind speed u. For more information on the method of calculating the halocarbon emission flux see Hepach et al. (2014).

(27)

3.2 The FLEXPART Model and ERA-Interim Data

FLEXPART ("FLEXible Particle dispersion model") is a Lagrangian particle dispersion model originally designed for forecasting mesoscale point source pollutant dispersion, such as radionuclides released in a nuclear power plant accident (Stohl et al., 1998).

FLEXPART has since been applied to studies of intercontinental pollutant transport, global pollution transport on climatic time scales, stratosphere–troposphere exchange, and more.

Lagrangian dispersion models have proven useful for gaining a better understanding of the atmospheric flow properties, such as mixing of tracers, transport, and dispersion (Bowman et al., 2013). For this thesis the FLEXPART versions 9.2 has been used.

The FLEXPART model was chosen for this thesis because it is a widely used and tested Lagrangian particle dispersion model (Hegarty et al., 2013). In Lagrangian models individual infinitesimally small air parcels are simulated, forward or backward in time, in the atmosphere. Hence the trajectory information for each parcel is provided, which is favorable when having point sources (e.g. ship measurements as in our case). In nature, the atmosphere is Lagrangian in the sense that air constitutes of tiny molecules, flowing with the winds. Thus Lagrangian models are ideal for modeling atmospheric transport, and flow phenomena like turbulent eddies, transport barriers, and mixing. Additionally, Lagrangian models have minimal numerical diffusion (because sharp gradients are well simulated) and they are always numerically stable. Moreover, they conserve mass, en- ergy, and momentum, and they are computationally cheap (Lin, 2013). Another great advantage is that Lagrangian models are independent of a computational grid, unlike Eulerian models. In Eulerian models, the concentration of tracer released from a point source is immediately mixed within a grid box, while in Lagrangian models subgrid- scale information are carried by the air parcels, providing the best possible resolution (Hegarty et al., 2013). A drawback of the parcel information not being bound to any grid, is that in order for parcels to represent, e.g., dynamic volume, additional procedures are necessary (Lin, 2013).

The meteorological input I have used for the FLEXPART experiments is global 3 hourly ERA-Interim atmospheric reanalysis data produced by the European Center for Medium-Range Weather Forecast (ECMWF) a numerical weather prediction model, pro- viding horizontal and vertical wind components, temperature, specific humidity, surface pressure, total cloud cover, dew-point temperature, large scale and convective precipita- tion, sensible heat flux, surface stress, and topography (Stohl et al., 2005). ERA-Interim data has a 1°×1° resolution, and 60 vertical model levels from the surface to 1 hPa. It includes a 4-dimensional variational analysis (4D-Var) with an analysis window of 12 hours(Dee et al., 2011). 4D-Var is a four-dimensional data assimilation method, which purpose is to determine a best possible initial state based on available observations. An evolving forecast error covariance is calculated, where observations are used at the ob- servation time, or as close as possible. The atmosphere is integrated forward and then backward in time, so that the initial state is optimized to fit the observations, from the beginning to the end of the 12 hourly window (Kalnay, 2003).

Weaknesses in the Era-Interim meteorological reanalysis data are affecting the FLEX- PART results, i.e., representing the convective overturning in the troposphere, since that is not resolved sufficiently by the spatial model resolution of the Era-Interim reanalysis.

FLEXPART therefore have the option to use a moist convection scheme developed by Emanuel and Živkovi´c-Rothman (1999). The following brief explanation of the scheme is based on Forster et al. (2007). The parameterization is called every synchronization time step, and it uses time-interpolated specific humidity and temperature profiles from the Era-Interim reanalysis to redistribute particles within a column. Convection is activated

(28)

when:

TvpLCL+1 ≥TvLCL+1+Tt. (3.3) HereTvpLCL+1is the virtual temperature of a surface air parcel lifted to the level above the lifting condensation level (LCL) andTvLCL+1is the virtual temperature of the environment at the same level.Tt= 0.9 K is the threshold temperature value. The virtual temperature of a moist air parcel is the temperature at which a dry air parcel would have the same pressure and density as the moist air parcel. The mass fraction displaced from one ver- tical level to another is calculated based on the buoyancy sorting principle, and whether an individual particle is displaced. The position of the particles in their respective des- tination layers, are determined by selecting a random number between [0,1]. After the random displacement of particles by the convection, a compensating subsidence velocity acts on the remaining particles in the grid box.

Certain boundary layer parameters are parameterized in FLEXPART. Friction veloci- ties and heat fluxes in the boundary layer are parameterized using available accumulated surface sensible heat fluxes and surface stresses from the ECMWF reanalysis. The atmo- spheric boundary layer (ABL) heights are parameterized by using the critical Richardson number concept, according to Vogelezang and Holtslag (1996). Spatial and temporal variations in the ABL heights, that are not resolved by the ECMWF, are accounted for by using a subgrid terrain effect parameterization. This subgrid terrain effect parameteriza- tion have been switched on for the model experiments in this thesis.

Forster et al. (2007) evaluated the subgrid-scale convection scheme in FLEXPART, and found that the convection was greatly dependent on whether the parcel was located over sea or land. This is a challenge since the cruise measurements that were used as tracer sources for the FLEXPART runs were taken very close to land and the steep Andes moun- tains. However, Forster et al. (2007) also found that the total precipitation, combining the convective precipitation and the large-scale precipitation from the ERA-40 data (the pre- vious ECMWF reanalysis), was closer to observations than without including the convec- tion precipitation. Furthermore, FLEXPART model simulations and aircraft observations of VSLS volume mixing ratio (VMR) in the upper tropical tropopause layer (TTL) and the free troposphere have shown good agreement when including the boundary layer scheme and the convective scheme (Fuhlbrügge et al., 2016b; Tegtmeier et al., 2013).

In this thesis, the two FLEXPART model versions 9.2.2 and 9.2.3 have been used.

Tegtmeier et al. (2012) upgraded FLEXPART version 9.2 by supplying with a chemical lifetime height profile, as an option to the constant lifetime for the substances. In FLEX- PART version 9.2.3, an online calculation of the cold point tropopause is included (Alina Fiehn, personal communication, Nov. 2017). This calculation is helpful for estimating the amount of VSLS entering the bottom of the Stratosphere. Other than adding extra information about the height of the cold point tropopause (CPT), the FLEXPART version 9.2.3 is the same as version 9.2.2.

(29)

3.3 Cruise VSLS Emissions Experiment

The first model experiment that was carried out for this thesis was the Cruise VSLS Emis- sion Experiment. The aim of this experiment was to model how much of the VSLS, mea- sured during ASTRA-OMZ and M91, entrained the stratosphere. Similar studies have been conducted for other tropical oceans before (Fiehn et al., 2017; Fuhlbrügge et al., 2016a; Hepach et al., 2016; Tegtmeier et al., 2012, 2013). The method is based on run- ning FLEXPART ERA-Interim point emissions, releasing VSLS from the location where the cruise measurements were taken. From surface ocean and atmospheric VSLS mea- surements, an emission is calculated and used as the respective release in the model run.

The general setup of the experiment is shown as a flowchart in Figure 3.1. FLEXPART version 9.2.2 was used for this experiment. This version does not include a calculation of the CPT, hence the 17 km height, which was found to be the height of the CPT above the tropical East Pacific derived from the radiosonde measurements (Chapter 2.3), was used for both cruise experiments. A description of the VSLS lifetime profiles, the calculation of the VSLS emissions, and the FLEXPART model setup are given in the subsections below.

3.3.1 VSLS Lifetime Profiles

Figure 3.2 shows the VSLS lifetime profiles used for this experiment. For methyl iodide a constant lifetime of 3.5 days was used (Carpenter et al., 2014b), since no profile was available. The lifetime of the bromocarbons is taken from Hossaini et al. (2010), who used a chemical transport model (CTM) with a detailed chemical scheme for the degradation of the two bromocarbons with height. The lifetime average is 17 days and 150 days for bromoform and dibromomethane respectively.

3.3.2 FLEXPART Setup

Six forward FLEXPART runs were executed for the cruise VSLS emissions experiment, one for each of the VSLS (methyl iodide, bromoform and dibromomethane) and for both cruises (ASTRA-OMZ and M91). Releases of the VSLS were carried out whenever mea- surements of the compounds were successfully taken. The compounds were released from a 0.002×0.002 (0.55 km×0.55 km) grid box at each measurement location at the surface during 1 hour, and carried by 10,000 particles. Further information about the FLEXPART runs are given in Table 3.1. The setup was the same for the ASTRA-OMZ and the M91 experiment for each of the VSLS, except for the runtime of dibromomethane.

This was because the ERA-Interim data was limited to the 31. December 2016. Thus, the dibromomethane experiment ran for 437 days (75 days less than M91). The runtimes was chosen to be long enough for the VSLS mass, represented in the FLEXPART particles, to have decayed to a unsignificant amount.

TABLE 3.1: FLEXPART model setup for the three VSLS; methyl idode (CH3I), bromoform (CHBr3), and dibromomehtane (CH2Br2). Two exper- iments are carried out for each compound. The runtime for the ASTRA- OMZ cruise experiment is shorter (see number in brackets), because me- teorological input data were not available at the time of the thesis calcula-

tions.

Compund Runtime [days]

Input interval [hours]

Sync time [seconds]

Output interval [hours]

Averaging time [seconds]

Sampling rate [seconds]

CH3I 10 3 900 3 1800 900

CHBr3 92 3 1800 6 3600 1800

CH2Br2 548 (437) 3 1800 24 7200 1800

(30)

Measured atm.

and oceanic VSLS

VSLS emission

Input

VSLS lifetime data EI fields

FLEXPART version

9.2.2

Post-Processing and Visualization

FIGURE3.1: Setup of the Cruise VSLS Emission Experiment. Oceanic and atmpospheric measurements of the VSLS’s methyl iodide, bromoform and dibromomethane were taken on the cruises ASTRA-OMZ (October 2015)

and M91 (December 2012).

0 5 10 15 20

Height[km]

0 5 10 15 20 25 30 35

Lifetime [days]

0 5 10 15 20

Height[km]

0 5 10 15 20 25 30 35

Lifetime [days]

Methyl iodide Bromoform Dibromomehtane/10

FIGURE3.2: Lifetime profiles used in FLEXPART for methyl iodide (Car- penter et al., 2014b), bromoform, and dibromomethane (scaled 1/10) Hos-

saini et al. (2010).

(31)

3.4 East Pacific Bromoform Emission Experiment

In order to further investigate the transport and dispersion of bromoform, a second ex- periment was designed. The approach of the East Pacific bromoform emission experi- ment was to release bromoform over a large area, so that the VMR of bromoform could be derived from the FLEXPART simulations. The results would then be comparable to, e.g., VMR measurements in the marine atmospheric boundary layer (MABL) and in the TTL.

Several aircraft campaigns have been conducted in the TTL over the East Pacific (Tegt- meier et al., 2013). Consequently, approximately 800 000 particles, containing bromoform mass information, were released over a large area, covering the Peruvian upwelling and the cruise track airmass sources. From the backward trajectories shown in Figure 3.3, it can be seen that both the ASTRA-OMZ and the M91 cruise mainly passed through air masses coming from the Peru Basin and the South Pacific Basin (Steffen Fuhlbrügge, per- sonal communication, Nov. 2017). Hence, the release field was chosen to be between 100W to 70W, and 20N to 50S 3.9. From this area oceanic bromoform emissions were released continuously, once per day for about 6 months, during a forward FLEXPART run. To get a realistic measure of the atmospheric bromoform VMR at the time of the cruises, the simulation was started three months ahead, so that the amount of bromo- form in the atmosphere could build up. A setup period of three months was chosen because it is reasonable to assume that the bromoform mass carried by the particles older than three months, would be perished by its e-folding lifetime of 16 days.

For this experiment the newly updated FLEXPART version 9.2.3 was used to calcu- late an estimate of the atmospheric bromoform VMR. Two model experiments were con- ducted; theEl Niño Expand theENSO Neutral Exp. The meteorological input fields and bromoform lifetime profile were the same as for the cruise VSLS emissions experiments.

This time, however, I used a bromoform emissionfieldfor a large area as input. To cal- culate the bromoform emissions, both an ocean and an atmospheric bromoform concen- tration field is needed, which is described in the following sections: The chosen surface water concentrations are described in the first section. Next, the process of choosing sur- face atmospheric concentrations are explained. In the last section the final bromoform emission fields are presented.

3.4.1 Surface water concentrations

For the surface water concentrations, the Stemmler et al. (2015) field was used. Stemmler et al. (2015) used a global coupled ocean biogeochemistry model to represent large scale variations of bromoform surface concentration in the open ocean. Coastal bromoform concentrations, which are generally higher than in the open ocean, were not included.

The Stemmler field was therefore scaled for each experiment with cruise measurements, i.e., the El Niño Exp with the ASTRA-OMZ measurements and the ENSO Neutral Exp with the M91 measurements. This was done by comparing each measurement of bro- moform in the surface ocean with the corresponding regional and temporal value in the Stemmlers field, using the average difference as the scaling factor. The scaling factor was found to be 2.28 for the El Niño Exp, and 0.61 for ENSO Neutral Exp.

3.4.2 Choosing surface atmospheric concentrations

Three different atmospheric bromoform VMR fields were considered for calculation of the ocean-atm bromoform flux. The first option was to use a constant atmospheric bro- moform VMR of 1 ppt. This option does not reflect the increased atmospheric concentra- tion of bromoform close to the coast, but it is relevant for the open ocean, and close to the global average of 1.2 ppt (Carpenter et al., 2014b, Table 1-7). The second option was

(32)

FIGURE 3.3: 10 days backward trajectories for ASTRA-OMZ and M91 (Steffen Fuhlbrügge, personal communication, Nov. 2017).

(A) (B)

FIGURE3.4: Surface concentrations of bromoform Stemmler et al. (2015) which are scaled with the (A) ASTRA-OMZ and (B) M91 cruise measure- ments. This was done by comparing each surface concentration measure- ment of bromoform with the corresponding regional and temporal value in the Stemmlers field, using the average difference as the scaling factor,

which is 2.28 for ASTRA-OMZ and 0.61 for M91.

to use an average bromoform VMR measured during the two cruises, using the ASTRA- OMZ average of 3.2 ppt for the El Niño Exp, and the M91 average of 2.9 ppt for the ENSO Neutral Exp. A problem with using these numbers is that it overestimates the VMR over the open ocean, because both cruises were close to the Peruvian coast, where the con- centrations are generally higher. The third option was to use the Ziska updated fields (Figure 3.5), were the original Ziska et al., 2013 field has been updated with the M91 measurements among others, but does not include the ASTRA-OMZ data yet. Thus, it includes the difference in VMR close to land and the open ocean, although not the ENSO variations.

(33)

FIGURE3.5: The Ziska updated bromoform VMR at the surface (Ziska et al., 2013) .

FIGURE3.6: Calculated bromoform emission fields using oceanic surface concentrations by Stemmler et al. (2015) and three optional atmospheric bromoform volume mixing ratios; 1 ppt (a and c), 3.2 ppt (b and d), and the field by Ziska et al. (2013) (c and f). The bromoform emission estimations from the ASTRA-OMZ (a, b, and c) and the M91 (d, e, and f) cruises are

plotted on top.

(34)

(A)

(B)

FIGURE3.7: Bromoform emission comparison between the ship emission estimations from a) ASTRA-OMZ and b) M91, and three different esti- mated emission fields. The emission field are estimated using the Stemm- ler et al. (2015) surface water bromoform concentration field and a marine bromoform VMR field; either a constant field value of 1 ppt (blue), the av- eraged measured volume mixing ratio; a 3.2 ppt field for ASTRA-OMZ and a 2.9 ppt field for M91 (red), or the updated Ziska et al., 2013 field (yellow).

In Figure 3.6 the resulting bromoform emission fields for the three options, stated above, with the estimated ASTRA-OMZ and M91 emissions plotted on top is presented.

Negative values are set to zero in this plots. The three options give quite different re- sulting emission fields. The Peruvian coast is a region with oceanic upwelling, but the upwelling is not continuous (Stemmler et al., 2015), thus the three fields shows emissions according to an upwelling average. I therefore found it the best to compare the average field values with the average cruise values, and an overview of the averages are given in Table 3.2 below:

(35)

TABLE3.2: Mean oceanic bromoform emissions for two experiments; the El Niño Exp and the ENSO Neutral Exp. In the first column the mean of the corresponding cruise emission estimated are shown (ASTRA-OMZ for the El Niño Exp and M91 for the ENSO Neutral Exp). For the other columns a mean over the stated field is shown. The field averages were taken over a big enough lat-lon box to include the respective cruise track. The values in

the brackets includes negative emissions.

BROMOFORM EMISSIONS [p mol m2h1]

Experiment Cruise 1 ppt 2.9 ppt 3.2 ppt Ziska update field El Niño Exp 1639 (1588) 1225 (1225) – 821 (818) 1079 (1079) ENSO Neutral Exp 232 (117) 227 (227) 15 (-97) – 36 (-32)

FIGURE 3.8: Schematic of method for calculating emission fields for ASTRA-OMZ and M91.

The average oceanic bromoform emissions of the fields were calculated for a large area for the East Pacific (EP) including the respective cruise track. By taking the mean over this area, open ocean values are included. However, open ocean emissions are gen- erally lower than the coastal emissions (Quack and Wallace, 2003). As the two cruises followed the Peruvian coast, measuring mostly coastal emissions, hence higher mean emissions from the cruises are expected than from the generated EP. It is apparent that cruise mean for both cruises is closest to the mean when using a constant 1 ppt VMR for the overlaying atmosphere (Table 3.2). To check further in detail the in situ cruise emis- sions are compared with the three optional field emissions at that same location (Figure 3.7). It can be seen that the ASTRA-OMZ emissions corresponded best with the emission field using the 1 pp for the atmospheric VMR, with a correlation coefficientR2 = 0.11.

Thus, using this atmospheric field for the El Niño Exp seemed the best. It is also no- ticeable that all emissions for the "1 ppt" emission field of the El Niño Exp, are positive.

However, this is not the case for the ENSO Neutral emission fields. The best field cor- relation with M91 data (Figure 3.7b) is R2 = 0.05 for the "2.9 ppt" emission field, but including quite a lot of negative emissions. The next best correlation is with the "1 ppt"

emission field where R2 = 0.04. Since this field included far less negative emissions, and the overall calculated averages corresponded best with this field (3.2), the "1 ppt"

emission field is used for both cruise experiments.

3.4.3 Final bromoform emission fields

The final calculation of the bromoform emission fields is summed up in Figure 3.8, and the resulting emission fields for the two model experiments; the El Niño Exp and the

(36)

(A) (B)

FIGURE3.9: Monthly averaged total released bromoform (Oct-Dec), per 1×1grid, in the FLEXPART simulations of (A) the El Niño Exp and (B)

the ENSO Neutral Exp.

ENSO Neutral Exp, is shown in Figure 3.9. Negative emissions in the ENSO Neutral Exp is omitted in the following. There is no negative emissions in the El Niño Exp. The monthly mean 10 m wind speed from ERA-Interim was used to calculate the bromoform emission calculations for the EP bromoform emission experiments.

3.4.4 FLEXPART Setup

Two model runs were exhibited, one for the El Niño Exp and one for the ENSO Neutral Exp. The model setup was the same for both runs (Table 3.3), except for the runtime. The simulation of the El Niño experiment started 05.07.2015 and ended 31.03.2016. The simu- lation of the ENSO Neutral Exp started 12.09.2012 and ended 31.03.2013. The simulation time was longer for the El Niño Exp since the ASTRA-OMZ cruise was conducted two months earlier than the M91 cruise, and I wanted to have an estimate of the atmospheric bromoform VMR at the time of the two cruises. 5 particles were released from each 1×1 release box, at the surface. The total number of released particles were 797,180.

The total released bromoform mass for the two experiments was 9.2 million kg for AS- TRA and 0.9 million kg for M91.

TABLE3.3: FLEXPART model setup for the East Pacific bromoform emis- sion experiment.

Compund Input interval [hours]

Sync time [seconds]

Output interval [hours]

Averaging time [seconds]

Sampling rate [seconds]

CHBr3 3 1800 12 7200 1800

Referanser

RELATERTE DOKUMENTER

The perpetrator’s type of leadership (e.g. the degree of support from the armed forces and previous record of violence against civilians) and existing ethnic or sectarian fault

The system can be implemented as follows: A web-service client runs on the user device, collecting sensor data from the device and input data from the user. The client compiles

As part of enhancing the EU’s role in both civilian and military crisis management operations, the EU therefore elaborated on the CMCO concept as an internal measure for

The dense gas atmospheric dispersion model SLAB predicts a higher initial chlorine concentration using the instantaneous or short duration pool option, compared to evaporation from

Based on the above-mentioned tensions, a recommendation for further research is to examine whether young people who have participated in the TP influence their parents and peers in

Overall, the SAB considered 60 chemicals that included: (a) 14 declared as RCAs since entry into force of the Convention; (b) chemicals identied as potential RCAs from a list of

Azzam’s own involvement in the Afghan cause illustrates the role of the in- ternational Muslim Brotherhood and the Muslim World League in the early mobilization. Azzam was a West

The ideas launched by the Beveridge Commission in 1942 set the pace for major reforms in post-war Britain, and inspired Norwegian welfare programmes as well, with gradual