Dissertation Universität Bremen Fachbereich 1 - Physik Institut für Umweltphysik
Studies on the Transport of CO and Related Biomass Burning Emissions Using Ground-based Fourier Transform Infrared Spectrometry
Vorgelegt von Voltaire Velazco zur Erlangung des Grades Doktor rerum naturalium (Dr. rer. nat) 1. Gutachter: Prof. Dr. Justus Notholt 2. Gutachter: Prof. Dr. Klaus Künzi
April 2006
TABLE OF CONTENTS
ABSTRACT ...... 3 ZUSSAMMENFASSUNG ...... 4 LIST OF PUBLICATIONS WITHIN THE LAST THREE YEARS ...... 5 1. INTRODUCTION...... 6 1.1. C OMPOSITION OF THE EARTH ’S ATMOSPHERE ...... 7 1.2. L AYERS OF THE ATMOSPHERE ...... 8 2. INSTRUMENTATION ...... 10 2.1. T HE FOURIER TRANSFORM INFRARED (FTIR) S PECTROMETER ...... 10 2.2. A DDITIONAL FEATURES AND ADVANTAGES OF A FOURIER TRANSFORM INFRARED (FTIR)12 SPECTROMETER ...... 12 3. RETRIEVAL THEORY ...... 14 3.1. I NTRODUCTION TO SOLVING INVERSE PROBLEMS ...... 14 3.1.1. The least Squares Solution to an Inverse Problem ...... 14 3.1.2. Estimating the Errors of the Solution ...... 15 3.1.3. Resolution Matrix of the Solution ...... 16 3.1.4. Optimal Estimation ...... 17 3.2. I MPROVEMENTS ON THE RETRIEVAL PROCESS ...... 18 3.2.1. Optimizing the Retrieval Strategy...... 18 3.2.2. A priori Profile Information of the Target Molecules ...... 18 3.2.3. Strategy for Specific Molecules ...... 21 3.2.3.1. Strategy for CO...... 21 3.2.3.2. CO Averaging Kernels...... 21 3.2.3.2. Strategy for C 2H6 ...... 24 3.2.2.3. O 3 Triplet Spectroscopy Applied to C 2H6 Retrievals ...... 24 3.2.2.3. O 3 Triplet Spectroscopy Applied to C 2H6 Retrievals ...... 25 4. RESULTS AND DISCUSSIONS...... 29 4.1. S ENSITIVITY STUDIES ON THE DEPENDENCE OF THE CO RETRIEVALS ON THE A PRIORI INFORMATION ...... 30 4.1.1. Introduction...... 30 4.1.2. Results...... 31 4.1.3. Summary ...... 35 4.2. L ATITUDE AND ALTITUDE VARIABILITY OF CARBON MONOXIDE IN THE ATLANTIC DETECTED FROM SHIP -BORNE FOURIER TRANSFORM SPECTROMETRY , M ODEL AND SATELLITE DATA ...... 36 4.2.1. Introduction...... 36 4.2.2. MOPITT data...... 36 4.2.3. Model data from MATCH-MPIC...... 37 4.2.4. Five Polarstern Cruises from 1996-2003 ...... 37 4.2.5. FTIR, Model and Satellite Data...... 41 4.2.6. Simulating Low- Resolution Retrievals from High Resolution Retrievals...... 41 4.2.7. Assessment of CO sources ...... 43 4.2.8. Summary ...... 46 4.3. T HE IMPACT OF FOREST FIRES ON CO CONCENTRATIONS IN THE NORTHERN HEMISPHERE 47
1 TABLE OF CONTENTS
4.3.1. Introduction: Fires in the Ecosystem...... 47 4.3.2 The El Niño Phenomenon ...... 47 4.3.2.1. The El Niño episode in 1997-1998...... 48 4.3.3 Long Term Measurements of Trace Gases in the Arctic and the effects of Fires in 1998 ...... 49 4.3.4. Measurements of CO, C 2H6 and HCN in Spitsbergen ...... 52 4.3.5. Summary ...... 56 4.4. A Q UANTITATIVE ASSESSMENT OF THE 1998 CO A NOMALY IN THE NORTHERN HEMISPHERE BASED ON TOTAL COLUMN MEASUREMENTS WITH FTIR S PECTROMETRY ...... 57 4.4.1. Introduction...... 57 4.4.2. Experimental Techniques...... 57 4.4.3. Error Estimates...... 57 4.4.4. Results...... 59 4.4.5. Quantifying the Effects of the 1998 Anomaly in the Free Troposphere ...... 60 4.4.6. Summary ...... 65 4.5. INVESTIGATIONS OF POLAR STRATO -MESOSPHERIC CARBON MONOXIDE MEASURED BY GROUND -BASED FOURIER TRANSFORM INFRARED SPECTROMETRY ...... 67 4.5.1. Introduction...... 67 4.5.2. Instruments...... 68 4.5.3. Retrieval of Strato-Mesospheric CO...... 68 4.5.4. The Chemical Transport Model...... 69 4.5.5 Results...... 72 4.5.5.1. FTIR data and model calculations ...... 72 4.5.5.2. Quantifying the CO above 18 km ...... 72 4.5.6. Summary ...... 76 4.6. M EASUREMENTS OF CO AND HCN IN PARAMARIBO , S URINAME ...... 77 4.6.1. Introduction...... 77 4.6.2 The Inter-tropical Convergence Zone (ITCZ)...... 78 4.6.3. Results...... 78 4.6.3.1. Peak in the signal caused by emissions from fires...... 78 4.6.3.2. Perturbations detected in the mid to upper troposphere...... 78 4.6.3.3. Backward trajectories confirm ITCZ migration ...... 79 4.6.5. Effects of the ITCZ migration on the total columns of HCN and CO...... 79 4.6.6. Summary ...... 85 5. SUMMARY AND CONCLUSIONS ...... 86 ACKNOWLEDGEMENTS ...... 88 6. REFERENCES...... 89
2
Abstract
This work deals with measurements and analysis of solar absorption spectra of atmospheric trace gases in the infrared spectral region, as recorded by a Fourier transform infrared (FTIR) spectrometer. The aim of this work is to study the transport of biomass burning emissions in the free troposphere, as well as its impact on the mid and high latitude atmospheres.
The first part of this work involves the analysis of spectra from measurements performed in the high Arctic at the NDSC (Network for the Detection of Stratospheric Change) station in Ny Ǻlesund, Spitsbergen (79° N, 12° E). After a short introduction to FTIR, it will be shown that the retrievals of several trace gases especially carbon monoxide (CO) have been improved and characterized, in order to yield usable profile information.
The second part of this work deals with measurements of CO profiles on board the German research vessel Polarstern. The measurements were taken during a ship cruise across the Atlantic, yielding altitude as well as latitudinal information on trace gases. The measurements of CO profiles have been combined with trajectory analyses and fire count data from satellites, in order to obtain a better understanding of the transport of biomass burning emissions and pollution in the troposphere. Furthermore, the measurements were validated using data from the MOPITT satellite and simulations from the Model of Atmospheric Transport and Chemistry of the Max Planck Institute in Mainz, Germany. The result is one of the first formal validations of real FTIR profile data with satellite and model, as well as a better illustration of the transport of pollution over the Atlantic.
The third part of this work is a collaboration with different institutes worldwide to achieve a quantitative assessment of the CO anomaly in the northern hemisphere brought about by widespread biomass burning events in 1998. The study was done using total columns of CO retrieved from FTIR and surface CO from in situ measurements . This is combined with a study on fire counts measured by the MODIS instrument and fire studies from government inventories. Measurements of HCN and C 2H6, also biomass burning products, were also analyzed to complement the CO measurements.
The fourth part of this work takes advantage of the increased altitude information that was achieved through the optimization of the retrieval methods. This part shows and establishes that FTIR spectrometry enables the detection of columns of strato-mesospheric CO, which consists of the columns in the layer between from 18 km and 100 km (stratosphere to upper mesosphere). With this information, an old data set from 1992 in Ny Ǻlesund turned into the longest ground- based measurements of strato-mesospheric CO ever published. As a result, new information on the nature of the transport and variability of strato-mesospheric CO have been uncovered.
The last part of this work discusses the results of a recent and on-going campaign in the tropics (Paramaribo, Suriname, 5.8° N, 55.2° W). The results shown here focus on the measurements of enhanced CO and HCN values and the effect of the migration of the ITCZ (Inter-tropical Convergence Zone) over the measurement site.
3
Zussammenfassung
Diese Arbeit beschäftigt sich mit der Messung und Analyse von Sonnenabsorptionsspektren atmosphärischer Spurengase im Infrarotspektralbereich mit Hilfe eines Fourier Transform Infrarot (FTIR) Spektrometer. Ziel dieser Arbeit ist es, den Transport der Emissionen aus tropischer Biomassenverbrennung in der freien Troposphäre sowie seine Auswirkung auf die Atmosphäre der mittleren und hohen Breitengrade zu untersuchen.
Der erste Teil dieser Arbeit behandelt die Analyse von Spektren, die in der hohen Arktis an der NDSC (Network for the Detection of Stratospheric Change) Station in Ny Ǻlesund, Spitzbergen (79° N, 12° E), gemessen wurden. Nach einer kurzen Einleitung über die Funktionsweise eines FTIR-Spektrometers wird gezeigt, daß die Herleitung der Konzentrationsprofile von den zu untersuchenden Gasen, besonders CO, verbessert und charakterisiert werden kann, um dadurch verwendbare Profilinformationen zu liefern.
Im zweiten Teil der Arbeit werden Messungen von CO Profilen, die an Bord des deutschen Forschungschiffes Polarstern aufgenommen wurden, vorgestellt und diskutiert. Die Messungen wurden während einer Kampagne auf dem Atlantik aufgenommen und lieferten Information über die Höhenverteilung sowie über die Breitengradverteilung der Spurengase. Die Messungen sind mit Trajektorienanalysen und Satellitendaten kombiniert worden, um den Transport der Emissionen von Waldbränden und die Verschmutzung der Troposphäre zu untersuchen. Weiterhin wurden die Messungen mit Daten des MOPITT Satelliten und Modellrechnungen des ’Model of Atmospheric Transport and Chemistry’ (MATCH) des Max Planck Instituts in Mainz verglichen. Dies ist mit die erste Validation der Breitengradabhängigkeit der FTIR- Spurengasprofile. Weiterhin ermöglicht die Analyse der Modellrechnungen eine bessere Abbildung des Transportes in der freien Troposphäre sowie eine Quantifizierung der CO- Quellen.
Der dritte Teil der Arbeit beruht auf einer Zusammenarbeit mit unterschiedlichen Instituten, um eine quantitative Einschätzung der CO-Anomalien in der Nordhemisphäre zu erzielen, die durch die extremen Waldbrände im Jahre 1998 verursacht wurden. Dies wird mit einer Analyse der Satellitendaten der Feuerstatistik des MODIS Instruments erreicht. Die Ergebnisse für CO werden durch Messungen von HCN und von C 2H6, auch Produkte der Biomassenverbrennung, ergänzt.
Der vierte Teil der Arbeit beruht auf den Informationen, die durch Verbesserung der Retrieval Methoden erzielt wurden. Dieses Kapitel zeigt , daß FTIR Messungen Säulen von strato- mesosphärische CO ermöglichen, die aus den Säulen in der Höhenschicht zwischen von 18 Kilometern und von 100 Kilometern beruhen . Die Messungen liefern mit den in Ny Ǻlesund seit 1992 vorhandenen Daten den längsten bodengebundenen Datensatz von strato-mesospherischem CO.
Im letzten Teil der Arbeit werden die Resultate einer neuen und „on-going“ Kampagne in den Tropen (Paramaribo, Suriname, 5.8° N, 55.2° W) kurz diskutiert. Die Ergebnisse konzentrieren sich auf die Messungen erhöhter CO- und HCN-Werte und den Einfluß der Migration der ITCZ (Inter-tropical Konvergenz-Zone) über dem Messungsungsort.
4
List of Publications Within the Last Three Years
Some parts of this thesis were published in:
Velazco V ., J. Notholt, T. Warneke, M. Lawrence, H. Bremer, J. Drummond, A. Schulz, J. Krieg, O. Schrems (2005), Latitude and Altitude Variability of Carbon Monoxide in the Atlantic Detected from Ship-borne Fourier Transform Spectrometry, Model and Satellite Data. J. Geophys. Res. 110, D09306, doi: 10.1029/2004JD005351.
Velazco, V ., S. W. Wood, M. Sinnhuber, I. Kramer, N. B. Jones, Y. Kasai, J. Notholt, T. Warneke, T. Blumenstock, F. Hase, F. J. Murcray and O. Schrems (2006), Trends of Polar Strato-mesospheric Carbon Monoxide Measured by Ground-based Fourier Transform Infrared Spectrometry, submitted to Atmos. Chem. Phys.
Yurganov, L. N. , T. Blumenstock, E. I. Grechko, F. Hase, E. J. Hyer, E. S. Kasischke, M. Koike, Y. Kondo, I. Kramer, F.-Y. Leung, E. Mahieu, J. Mellqvist, J. Notholt, P. C. Novelli, C. P. Rinsland, H.- E. Scheel, A. Schulz, A. Strandberg, R. Sussmann, H. Tanimoto, V. Velazco , R. Zander, and Y. Zhao, (2004), A quantitative assessment of the 1998 Carbon Monoxide emission anomaly in the northern hemisphere based on total column and surface concentration measurements, J. Geophys. Res ., Vol. 109, No. D15, D15305, doi:10.1029/2004JD004559.
Other Publications During the Course of this Work:
Warneke, T., J.F. Meirink, J-U. Grooß, J. Notholt, G.C. Toon, V. Velazco , A.P.H. Goede and O.Schrems (2006), Seasonal and latitudinal variation of methane: A ground-based and ship- borne solar IR spectroscopic study, submitted to Geophys. Res. Letters.
Warneke, T., Z. Yang, S. Olsen, S. Koerner, J. Notholt, G. C. Toon, V. Velazco , A. Schulz and O. Schrems (2005), Seasonal and latitudinal variations of column averaged volume-mixing ratios of atmospheric CO2, Geophys Res. Letters , 32, L03808, doi:10.1029/2004GL021597.
Palm, M., C. v. Savigny, T. Warneke, V. Velazco , J. Notholt and K. Kuenzi, Intercomparison of O3 Profiles observed by SCIAMACHY (2004), Ground-based Microwave and FTIR Instruments, Atmos. Chem. Phys., 5, 2091–2098, 2005.
Yurganov, L. N., P. Duchatelet, A. V. Dzhola, D. P. Edwards, F. Hase, I. Kramer, E. Mahieu, J. Mellqvist, J. Notholt, P. C. Novelli, A. Rockmann, H. E. Scheel, M. Schneider, A. Schulz, A. Strandberg, R. Sussmann, H. Tanimoto, V. Velazco , J. R. Drummond, and J. C. Gille, Increased Northern Hemispheric carbon monoxide burden in the troposphere in 2002 and 2003 detected from the ground and from space, Atmos. Chem. Phys ., 5, 563–573, 2005
De Maziere, M., C. Vigoroux, T. Gardiner, M. Coleman, P. Woods, K. Ellingsen, M. Gauss, I. Isaksen, T. Blumenstock, F. Hase, I. Kramer, C. Camy-Peyret, P. Chelin, E. Mahieu, P. Demoulin, P. Duchatelet, J. Mellqvist, A. Strandberg, V. Velazco, J. Notholt, R. Sussmann, W. Stremme, A. Rockmann (2005), The exploitation of ground-based Fourier transform infrared observations for the evaluation of tropospheric trends of greenhouse gases over Europe, Env. Sci , 2(2-3): 283-293.
5 CHAPTER I
1. Introduction
The Earth’s atmosphere is undergoing rapid changes, mainly due to human activities. The results of careless land use and industrialization have profoundly modified the composition of the atmosphere on a global scale which is leading to disturbing consequences. Long term depletion of stratospheric ozone, increase in greenhouse gas concentrations, acid rain, and changes in the self cleansing capacity of the atmosphere are just some of these consequences. Followed by the environmental and economic issues arising from these developments, international agreements such as the Montreal protocol and the Kyoto protocol have emerged, which were also partially pushed forward by global awareness and civic action of numerous non- governmental organizations (e.g. Greenpeace and WWF).
A major contributor in global atmospheric change and one of the main topics of this work is biomass burning. Emissions from biomass burning provide an abundant source of greenhouse gases. It is also a source of numerous chemically active gases, including carbon monoxide, non- methane hydrocarbons, and nitric oxide. These gases, along with methane, lead to the chemical production of tropospheric ozone (another greenhouse gas) they also control the concentration of the hydroxyl radical, which regulates the lifetime of almost every atmospheric gas. Biomass burning was once believed to be mostly concentrated in the tropics, however it was recently demonstrated that it is also a regular feature of the world's boreal forests, the effects of which will be shown in chapter IV of this work. Recent estimates indicate that almost all biomass burning is man-made and that this trend is increasing with time. With the formation of greenhouse and chemically active gases as direct combustion products and a longer-term enhancement of biogenic emissions of gases, biomass burning may be a significant driver for global change. 1
Understanding and quantifying atmospheric processes are essential for the improvement of climate predictions, for finding ways in mitigating detrimental impacts of climate change and for continuance of developments and observance of environmental treaties. In order to understand the physical and chemical processes of the atmosphere and to asses the human impacts on the atmosphere, a number of scientific efforts have been initiated worldwide. These efforts include laboratory studies of the properties of atmospheric trace gases, ground-based measurements of atmospheric constituents, continuous in situ samplings, satellite measurements and modeling of atmospheric processes.
For this work, I focused on ground-based FTIR measurements of atmospheric trace gases that are products of biomass burning. One of the most significant trace gases coming from biomass burning is carbon monoxide (CO). This is because 1) it is responsible for more than half of the total turnover of OH, which is a cleansing agent in the atmosphere, 2) CO is a very good tracer and indicator of pollution, because by looking at CO, it is possible to know how and where pollution is transported in the atmosphere. And so with this work, allow me to take you to the warm regions of the tropics, the temperamental waters of the Atlantic and to the pristine and secluded Arctic. Have a pleasant time reading, Voltaire Velazco.
1 Levine, J.S., “Biomass Burning and the Production of Greenhouse Gases” http://asd- www.larc.nasa.gov/biomass_burn/biomass.html.
6 CHAPTER I
1.1. Composition of the Earth’s atmosphere
The earth’s atmosphere started forming about 3.5 billion years ago from gases escaping from the interior of the earth. From about 3.3 billion years ago, the atmospheric composition evolved from an anoxic state to an oxic state, i.e. from a state without oxygen, to a state with oxygen. At present, the atmosphere is composed mainly of nitrogen (78.1 %), oxygen (20.9%) and noble gases (0.9%) (Figure 1.1. 1, left). The remainder consists of several trace gases and aerosols. Despite their negligible contribution, trace gases and aerosols are very important in atmospheric chemistry and physics. An example of these trace gases is Ozone and examples of aerosols are dust, smoke and haze.
THERMOSPHERE MESOPAUSE
MESOSPHERE
STRATOPAUSE
STRATOSPHERE
TROPOPAUSE
TROPOSPHERE
Figure 1.1. (left) Relative amounts of gases in the atmosphere . (right) The layers of the
Earth’s atmosphere determined by a temperature profile (black curve) taken from the tropics
(5.8° N)
1 http://en.wikipedia.org/wiki/Image:Atmosphere_gas_proportions.gif
7 CHAPTER I
1.2. Layers of the Atmosphere
The earth’s atmosphere changes from the ground up. Five layers of the atmosphere have been characterized on account of their thermal characteristics, chemical composition, density and dynamics. Each of the layers are bounded by “pauses” where the maximum changes occur (see Figure 1.1, right).
1. Troposphere: From the Greek word “tropos” meaning “to mix”. This is the lowest layer of the atmosphere starting at the surface to between 7 km at the poles and 17 km at the equator. The upper boundary is called the tropopause. In the troposphere, temperature decreases with height. Almost all weather occurs in this lower most layer of the atmosphere.
2. Stratosphere: The stratosphere extends from the tropopause up to about 50 km. This layer holds 19% of the atmosphere’s gases but the water vapor content is about 1000 times smaller than in the troposphere. Temperature profiles in the stratosphere increases with height due to the ozone layer that absorbs solar ultraviolet (UV) radiation. The top of the stratosphere (stratopause) has temperatures of about 270 K, which is almost the same as the ground level temperature. Above this, the temperature decreases with height. Another characteristic of the stratosphere is its dynamical stability due to vertical stratification with warmer layers above and colder layers below. This causes the vertical movements of the gases slow in this layer.
3. Mesosphere: This layer extends from the stratopause, the transition boundary which separates the stratosphere from the mesosphere up to about 80-85 km. The temperature profile increases with height.
4. Thermosphere: This layer extends from 80-85 km to about 640 km. The temperature here increases with height. In the lowest layer of the thermosphere CO 2 is photolyzed to yield CO, resulting in CO concentrations of more than 60 ppm (parts per million), which is 1000 times greater than the values in the troposphere.
An important characteristic of the atmosphere is the exponential drop of pressure with altitude. This characteristic can be illustrated by the hydrostatic equilibrium between gravity and the pressure gradient.
∂p = −ρg (2.1) ∂z where p is the pressure, z is the altitude, g is the gravitational acceleration and ρ is the density of air. The ideal gas law states
ρRT p = (2.2) mmol
8 CHAPTER I
with T as the thermodynamic temperature, R=8.31 J/mol-K the universal gas constant and 23 mmol is the mass on N A=6.022x10 molecules. Combining (2.1) and (2.2), we arrive at the barometric law (under the assumption that T is constant).
= − p(z) pO exp( z / H ) (2.3)
pO is the reference pressure at sea level (1013 hPa) and the pressure scale height H=RT/(m mol g) has a value of about 7 km.
9 CHAPTER II
2. Instrumentation
For this study, three different types of commercial spectrometers were used, the Bruker 120M, 120HR and 125HR. This chapter will provide the very basics of FTIR spectroscopy since thorough descriptions of the instrumentation techniques can already be found in numerous works and textbooks [ e.g. Davis et al. , 2003]
2.1. The Fourier Transform Infrared (FTIR) Spectrometer
The invention of the Fourier transform spectrometer (FTS) came about from A. Michleson’s classical attempt in 1880 to measure the “ether”, at the time, a medium believed to permeate space and allow the propagation of light through interplanetary space. It is probably the most famous experiment due to its failure.
In its simplest form, the FTS consists of two mirrors placed at right angle to each other. A beam splitter is placed at the vertex of the right angle and oriented at 45° angle relative to the two mirrors (Fig. 2.1.1)
Fixed Mirror M 1
Moving Mirror M Source 2
BS
Detector
Figure 2.1.1 . Schematic diagram of a simple Michelson interferometer
A parallel beam of radiation is directed from the source to the beam splitter (BS). The BS is a plate made of a partially reflecting and partially transmitting material e.g. potassium bromide (KBr). It is coated so that 50% of the incident radiation falling on it is reflected to mirror M 1 and 50% is transmitted to mirror M 2. The radiation from both mirrors return along the same path and is then recombined to a single beam at the beam splitter (of course, half of the total radiation is sent back to the source but this is unimportant).
If the source emits monochromatic radiation, the recombined beam leaving BS and going to the detector will show constructive or destructive interference, depending on the relative path lengths of BS to M 1 and B to M 2. This is the essence of the Michelson experiment [Banwell & McCash, 1994].
10 CHAPTER II
Hence, if the path lengths are equal or differ by an integral number of wavelengths, constructive interference will be observed from the recombined beam leaving BS. On the other hand, if the path length difference is a half integral number of wavelength, destructive interference will be observed. Moving the mirror M2 smoothly towards or away from BS will result in an alternating pattern of destructive and constructive interference that can be seen by the detector as incident radiation alternating in intensity.
Extending to Two or More Frequencies
If we have a source that emits 2 frequencies υ1 and υ2, the interference pattern (also considered as beat pattern) would be a more complicated intensity fluctuation as M 2 is moved. However, obtaining the original frequencies and intensities emitted by the source is very fast and easy by simply applying Fourier transformation to the resultant signal.
Figure 2.1.2 shows a typical interference pattern or interferogram for a white source (source: Tungsten, beam splitter: Kbr, detector: InSb). The wide range of frequencies causes the signal to diminish rapidly away from the position at which both mirrors are an equal distance from the beam splitter, the so- called zero-retardation peak or zero path difference (ZPD). In the real world, no source is truly white. Figure 2.1.3 shows the variation in total intensity as a result of varying source output and beam splitter efficiency across the infrared (IR) range (from 4100 cm -1 to 4400 cm -1 ). This background variation must be taken into account for each spectrum. If the beam from such a source passes through a sample before reaching the detector, the absorptions due to the sample will lead to gaps in the frequency distribution of the beam depending on the properties of the sample. Hence, the production of a spectrum is a two-stage process. In a first step, without a sample in the way of the beam, the movable mirror M 2 is moved smoothly over a period of time, covering a certain distance. During the whole traverse, the detector signal is recorded regularly and each piece of information is put serially onto different storage locations in the computer. After the whole path is covered by the moving mirror, the computer carries out the Fourier transform of the stored data. In the second step, an interferogram is recorded in exactly the same way, this time with a sample in the path of the beam. The interferogram then undergoes a Fourier transformation. The result is ratioed against the background spectrum in order to yield the transmittance spectrum.
tensity
Relative In
Scanner Position
Figure 2.1.2. A typical interference pattern for a white source (Tungsten lamp)
11 CHAPTER II
Relative Intensity
Wave Numbers (cm -1)
Figure 2.1.3. Tungsten spectra near 4000 cm -1 recorded on board the research vessel Polarstern on Feb. 7, 2003. beam splitter: KBr, detector: InSb.
2.2. Additional Features and Advantages of a Fourier Transform infrared (FTIR) Spectrometer
The FTIR spectrometer has a number of essential advantages compared to common grating or prism spectrometers. These advantages are more than enough to compensate for the computational expenses in calculating the Fourier transform of the recorded interferogram.
1. Jaquinot or throughput advantage: This arises from the fact that the circular aperture used in FTIR spectrometers have a larger area than the linear slits used in grating spectrometers. This property enables higher throughput of radiation.
2. Fellget or multiplex advantage: the total scanning time for an FTIR spectrometer is considerably less than that required for s dispersive spectrometer to produce a spectrum of the same sensitivity and resolution. For dispersive spectrometer, each element is scanned consecutively. For a total scan time T, the time t taken to record one resolution element is T/n; where n is the number of resolution elements. The entire frequency range is sampled for the total scan time T in FTIR. If t is the recording time, it is found that the signal (S) is proportional to t and the noise (N) is proportional to t or S S ∝ t or N ∝ t and so ∝ t N Therefore, we may write;
S = T (2.1) N FTIR
12 CHAPTER II
S T = (2.2) N DISP n
S N FTIR = n S (2.3) N DISP
The result is a gain of n in signal-to-noise ratio for an FTIR spectrum recorded over the same total time as a dispersive spectrum. In comparison, if it took 10 minutes to record a dispersive spectrum, an FTIR spectrum of the same S/N could be obtained in 10 seconds [ Banwell and Mc Cash , 1994].
3. The Connes advantage: In an FTIR spectrometer, the position of the moving mirror is measured by counting the interference fringes from a HeNe laser. Since the frequency of the laser is known to a high precision, it is possible to measure the mirror position and therefore the frequency very accurately.
4. Resolving Power: The resolving power of an FTIR spectrometer is constant over the entire spectrum. In contrast, a grating or a prism spectrometer’s resolving power depends on the angle of the incident radiation and hence varies with frequency. The resolving power is usually poor at the ends of the spectrum. The resolution of an FTIR spectrometer is related to the mirror travel or OPD;
Resolution = 1/OPD (2.4)
A limitation in FTIR spectrometry is the ability to make the moving mirror travel smoothly and perfectly straight over long distances. Difficulties from this limitation arise for high resolution work.
13 CHAPTER III
3. Retrieval Theory
3.1. Introduction to Solving Inverse Problems
3.1.1. The least Squares Solution to an Inverse Problem
This section gives an overview of the methods used in the retrieval of volume mixing ratio (VMR) profiles of atmospheric trace gases. I start with the simple case of solving systems of linear equations then proceed with a solution to underdetermined problems. Finally, I will expand this discussion to the optimal estimation method (OEM), which is the core of the retrieval algorithm used for this work (SFIT-2).
The following discussions were derived from the lecture “Inverse Methods and Data Analysis” of Prof. Dr. Schlitzer given to students of Environmental Physics at the University of Bremen.
We first consider a linear problem relating measurements y to a state vector x
y=K·x (3.1)
The vector x can represent for example, VMRs of a certain molecule at different discretized layers of the atmosphere. The vector y then represents the intensity of radiation at certain wavelengths detected by a spectrometer (a spectrum) and K would be a matrix relating the VMR to the measurements.
In an ideal case, where there is enough information, solving for x is straightforward. If we consider errors of the measurements σi, then the least squares solution can be obtained by minimizing the cost function F after the equation has been normalized by the errors, where;
F=e T·e = (K·x-y)T(K·x-y) (3.2)
Solving for the gradient of F leads to the normal equations;
KT·K·x’ =KT·y (3.3)
And from this, the optimal solution x’ is;
x’ =( KT·K )-1KT·y (3.4)
For underdetermined problems where the number of independent equations is less than the number of unknowns, this method fails to offer a solution.
There is still “hope” for a solution by taking advantage of the information that we can get out of the determined part of K and by identifying the undetermined (or null space) of K. This is where singular value decomposition comes in. Any matrix can be written as;
14 CHAPTER III
K= U·S·V T (3.5)
If K has dimensions (mxn) then U will have dimensions (mxm), S is (mxn) and VT is (nxn). S contains the so-called singular values which are arranged in descending order along the diagonal of S. The index of the last non-zero singular value determines the rank of K which also corresponds to the number of linearly independent equations. Since the remaining p+1 elements of S are zeroes, we could truncate S up to the p th row and p th column. Similarly, U can be truncated up to the p th column and V can be truncated up to the p th row, so that;
T K=U p·S p·V p (3.6) and K will still remain the same!
The columns of the full matrix V are orthogonal vectors of length n so that we can represent x as a linear combination of these vectors;
p n = α + β x ∑ i v i ∑ j v j i=1 j= p+1 (3.7)
I have separated the contributions coming from the first p columns of V and the remaining p+1 to p=n columns of V, which contains the null space of K so that we can write;
x=x part + x null (3.8)
With this, we could separate the part of x that we can determine (solve uniquely) which we call the particular solution xpart and the part that we do not know at all ( xnull ).
null Since A·x = 0, we could choose any arbitrary values for the coefficients ß i and they will not affect the cost function F. This means that there is an infinite number of possible solutions and they will all fulfill the least squares criteria. From all these infinite possible solutions, we can choose one solution such that xnull = 0. Hence, we look for one optimal solution which is the smallest (which gives the smallest ||x’|| ). This is given as;
part -1 T x = V p · S p ·U p · y (3.9)
This solution fulfills the normal equations in (3.3).
3.1.2. Estimating the Errors of the Solution
Now that we have a solution, the next task is to characterize that solution, i.e. to determine how good the solution is. We do this by examining its covariance matrix. The retrieved state vector can be written as;
xpart =C·y (3.10)
15 CHAPTER III
where y is a vector with a covariance matrix given by cov(d) and C is some matrix that maps y on x. The covariance of x is written as;
cov(x part )= C·cov·(d) ·C T (3.11)
-1 T Substituting C= V p · S p ·U p from (3.9) we arrive at;
part -1 T T -1 T cov(x )= V p ·S p ·U p · cov(d) · U p · S p ·V p (3.11)
2 2 For uncorrelated data y i, with variances all equal to σd , we can write cov(d)= σd · I , which is just an identity matrix ( I) multiplied by the variance of the data. With this, (3.11) can be simplified to;
part 2 -2 T cov(x )= σd Vp ·S p · Vp (3.12)
This tells us that, to calculate the errors of the solution vector, it is not necessary to know the data vector y. All we need is an information on the errors of y. This is a very important prognostic tool for the errors of a remote sensing instrument. For example, we can know the errors of the measurements that a satellite will give, even before launching the satellite into space.
3.1.3. Resolution Matrix of the Solution “The truth is rarely pure and never simple.” - Oscar Wilde
Because of insufficient information, the “true” state vector x can never be known. As illustrated in the previous section, we can only find an optimal solution xpart . How xpart relates to x is described by the resolution matrix (or later referred to as Averaging Kernel Matrix). Let T K=U p·S p·V p be the singular value decomposition of K with the value of p being less than the number of unknowns and p= rank( K). Substituting K to the normal equations from (3.3), and T noting that because of orthogonality, Up ·U p = I, we have;
2 T T Vp · S p ·V p ·x = V p · S p · U p · y (3.13)
-2 T T Multiplying with Vp · S p ·V p from the left and noting that Vp ·V p = I, we have;
T -1 T Vp · V p ·x = V p · S p · U p · y (3.14) or T part Vp ·V p ·x = x (3.15)
Equation (3.15) describes the relationship between our “smallest” optimal solution xpart , which T we obtained from (3.9) and the general optimal solution x. We can call Vp ·V p =R as the part part resolution matrix of the parameter x. Equation (3.15) states that a component of x , say xj is composed of a linear combination of the components of the “true” x, i.e. n part = x j ∑r jk xk (3.16) k=1
16 CHAPTER III
T T Note that: R is a square matrix (nxn). If K is regular, i.e. p=n, then Vp ·V p = V ·V = I and R=I, hence, xpart = x.
3.1.4. Optimal Estimation
We have seen in the previous section that solving inverse problems can lead to disaster if we do not have sufficient and useful information. Despite the lack of information, we have also seen that we could still come up with a best optimal solution by imposing criteria or constraints to the solution. The optimal estimation method (OEM) is a another method of solving for x’ especially if the problem is ill-posed or underdetermined. It employs the Bayesian approach in solving the inverse problem. One requirement is that there are some prior understanding or expectation of the quantities we want to determine. We then want to update this understanding based on new information (for a discussion on Bayes theorem and how it is applied to optimal estimation, please consult Rodgers [2000], chapter 2).
The measurement process is expressed as a forward model which maps the state space (such as x) into measurement space (say, y). Bayes theorem provides a formalism to invert this mapping and calculate a posterior probability density function (PDF) by updating the prior PDF with a measurement PDF.
Here, we introduce a term called “virtual measurement” or sometimes called a priori (xa), with a covariance matrix Sa. The a priori , in our case, can be previous measurements of the state vector x, it could also be a guess of what x may be. In optimal estimation, xa and Sa act as constraints to the solution.
Given the measurement error covariance matrix Sε (this could be the noise in the spectra), the a priori xa and its covariance Sa , and applying Bayes theorem, we arrive at a solution for x [Rodgers , 2000];
= + ⋅ T ⋅ ⋅ T + −1 − ⋅ x' x a Sa K (K S a K Sε ) (y K xa ) (3.17)
The covariance matrix of the solution given by (3.17) is;
= − ⋅ T ⋅ ⋅ T + −1 ⋅ (3.18) S x' S a S a K (K S a K S ε ) K S a
The equivalent of the solution resolution matrix R, described in the previous section is the averaging kernel matrix; A=GK, with;
= ⋅ T ⋅ ⋅ T + −1 G S a K (K Sa K S ε ) (3.19) So that;
= T ⋅ −1 ⋅ + −1 −1 ⋅ T ⋅ −1 ⋅ (3.20) A (K Sε K S a ) K Sε K
17 CHAPTER III
Another useful quantity is the degrees of freedom for signal (DOFs). It is calculated from the trace of A. If A represents the averaging kernels of the state vector subspace containing the VMR profiles of a certain gas in the atmosphere, then the DOFs represent approximately how many independent layers can be retrieved.
The SFIT-2 algorithm is an empirical approach to the OEM in the sense that Sa, Sε, and xa can be “tuned” in order to obtain stable profiles of a certain atmospheric trace gas.
3.2. Improvements on the Retrieval Process
As mentioned in the previous section, SFIT-2 uses an empirical approach to the OEM. Therefore, this leaves a lot of room for many possible ways in using SFIT-2, it may also leave a lot of possibilities for mistakes. In this work, improvements have been made on the retrieval process using SFIT-2, specifically on the tuning parameters used, on maximizing information content and on the characterization of the retrievals.
3.2.1. Optimizing the Retrieval Strategy
A new retrieval strategy has been established within the frame of this work specifically for the retrieval of CO, C 2H6, O 3, N 2O and CHF 2Cl. Optimizing the retrieval strategy involved;
• Choosing optimal micro-windows • Using an updated line list • Choosing an appropriate layering scheme. • Choosing the best a priori profiles • Optimizing the tuning parameters (SNR, a priori uncertainty, etc.) • Characterizing the errors and information content of the retrievals
The micro-windows were chosen and optimized according to the measurement site, the type of instrument and the line list database. The HITRAN-2000 database with 2001 official updates, called hereinafter as HITRAN 2000+ was used for all the retrievals. This database + contains specific additions for C 2H6 [ Pine and Rinsland, 1999] and O 3 provided by the LPMA , [De Maziere, & UFTIR Team, 2005a]. A 29-altitude layering scheme was used for the retrieval grid. Standard a priori profiles and covariance matrices were established and used for all the retrievals. In general, the retrieval strategy, which was developed, produced good results. In this section, a brief summary of the retrieval strategy for the species CO and C 2H6 will be discussed. A list of the target gases with the corresponding micro-windows, interfering species and the information content in terms of the degrees of freedom for signal is outlined in Table 3.1.
3.2.2. A priori Profile Information of the Target Molecules
A priori profile information of the target molecules may significantly influence the result of the retrieved profiles. Establishing the most appropriate a priori profile information was part of this work. Several factors were taken into account when choosing an a priori profile; 1) it
+ Laboratory for Molecular Physics and Applications, France (CNRS)
18 CHAPTER III should represent the true profile of the target molecule as closely as possible. 2) sudden changes and peaks in the a priori profile should be avoided, i.e. the profile should be smooth. 3) for the purpose of trend analysis in this work, one standard a priori profile for each molecule is used (see Figure 3.2.1).
For CO retrievals, the a priori profile that was used has been adapted from the a priori profile used in the MOPITT retrievals [ Bremer et al ., 2004]. The advantage of this is that it makes the comparison of FTIR and MOPITT CO profiles much easier, which will be shown in Chapter 4.2 A priori profiles for the retrieval of C 2H6 were taken from the average of balloon measurements done in the Arctic region.
Table 3.1. Summary of best retrieval micro-windows and performances. DoFs indicates the approximate number of independent elements of altitude-resolved information calculated for a solar zenith angle of around 60°; the precise value depends on the particular retrieval conditions like spectral signal-to-noise ratio, a priori information, station altitude, etc. Alt max indicates the approximate highest altitude for vertical inversion.
Microwindow (s) -1 (cm ) used Interfering species; DoFs / Alt max (km) simultaneously 1000.0 – 1005.0 1110.0 – 1113.0 O H O, N O, CH 5 / 35 3 1117.3 – 1117.9 2 2 4 1120.1 – 1122.0
2057.70 – 2057.91 CO 2069.55 – 2069.72 H2O, N 2O, O 3, solar lines 4 / 18 2157.40 – 2159.35
C2H6 2976.50 – 2977.20 H2O, O 3, CH 4 2 / 30
N2O 2481.30 – 2482.60 2526.40 – 2528.20 Option 4 µµµm H O/HDO, CO , O , CH 4.5 / 30 2537.85 – 2538.80 2 2 3 4 2540.10 – 2540.70
1161.34 – 1161.66 Option 8.5 µµµm 1182.49 – 1182.83 O3, HDO, CH 4 4.5 / 30 1183.25 – 1183.74
H2O/HDO, CO 2, O 3, ClO, CHClF 2 (HCFC-22) 828.8 – 829.35 1 / 25 C2H6, solar lines
19 CHAPTER III
CO C2H6 90 90 a 80 b 80 70 70 60 60 50 50
40 40
Altitude(km)
Altitude(km) 30 30 20 20 10 10 0 0 0.0E+00 2.0E-07 4.0E-07 6.0E-07 8.0E-07 0.0E+00 1.0E-09 2.0E-09 3.0E-09 VMR VMR
O3 N2O 90 90 80 c 80 d 70 70
60 60
50 50 40 40 Altitude (km) Altitude 30 Altitude(km) 30 20 20 10 10 0 0 0.00E+00 2.00E-06 4.00E-06 6.00E-06 0.00E+00 1.00E-07 2.00E-07 3.00E-07 4.00E-07 VMR CHF2CL VMR 90 80 e 70 60
50
40
Altitude(km) 30 20 Figure 3.2.1 . A priori profiles used for the retrievals of CO (a), C H (b), O 10 2 6 3 (c), N 2O (d) and CHF 2Cl (e). 0 0.00E+00 5.00E-11 1.00E-10 1.50E-10 2.00E-10 VMR
20 CHAPTER III
3.2.3. Strategy for Specific Molecules
For the purpose of this study, the details of the retrieval strategies for CO and C 2H6 will be presented in the following sub-sections. The sub-sections contain discussions on the information content of the retrievals, the spectral fits and typical averaging kernels. Further details of the strategies for O 3, CH 4, CHF 2Cl, and N 2O are provided in De Maziere, & UFTIR Team [2005a].
3.2.3.1. Strategy for CO
Modifications and improvements were made for the retrieval of CO that were originally described in the studies of Pougatchev and Rinsland [1995]. The spectroscopic appearance of terrestrial CO lines has several important characteristics in the infrared region. First, there are several atmospheric CO absorption lines in the 2.3 µm and 4.7 µm regions. Second, there is a saturated line (absorption= 100%) in the center of the strong lines in the 4.7 µm region. Third, the spectral regions occupied by the CO lines are strongly contaminated by absorption lines of other atmospheric gases such as H 2O, CO 2, O 3, OCS and CH 4. Fourth, each terrestrial CO line overlaps with a corresponding absorption line in the solar atmosphere, which is Doppler shifted and formed at high temperatures.
In this work, CO is retrieved using three of the four micro-windows that were presented in the work of Pougatchev & Rinsland [1995] with modifications in the limits and the sizes of the micro-windows. The two micro-windows in the 2069 µm and 2057 µm regions contain 13 12 CO lines. The saturated line centered at 2158.3 µm contains a CO line. The H2O, N 2O, O 3, and Solar lines are taken into account in all of the micro-windows. Examples of the micro- windows recorded for Spitsbergen are shown in Fig 3.2.2 a-c. Additional weak lines also occur in the interval and each terrestrial CO line is overlapped by solar CO absorption from the same transition.
Figure 3.2.3 shows the K-Matrix or the Jacobians plotted against the spectral points. This plot provides a visualization of where the information is coming from. For example, the saturated line starting at 2157.3 cm -1 tells us that most of the information is coming from the wings of the line and no information from the line center (2158.3 cm -1) at all. The figure also shows that most of the information coming from this line is from CO in the troposphere. The two narrow micro- windows starting at 2057.7 cm -1 and 2069.5 cm -1 provide information on CO at high altitudes.
3.2.3.2. CO Averaging Kernels
The averaging kernels for the retrieval of CO are shown in Figure 3.2.4 (left). They were calculated for a retrieval with a solar zenith angle (SZA) of 51°, a spectrometer optical path difference of 90 cm and a signal to noise ratio of 500. The amplitudes of the kernels give an indication of the sensitivity of the retrieval in each layer. The widths of the averaging kernels represent the vertical resolution for that layer. In the case of CO, not all of the averaging kernels are independent, i.e. some kernels overlap with the others. This means that information on the volume mixing ratio of one layer may be composed of information coming from other layers. A very conservative estimate on the resolvable layers for CO retrievals would yield three to four
21 CHAPTER III layers. The averaging kernels for three merged layers are shown in Figure 3.2.4 (right) . This figure shows that, with the given settings and conditions, we can independently resolve at least two layers in the troposphere and one in the stratosphere.
a
ie sn n it y t
b
e lar iv e
c
Figure 3.2.2 (a-c). Example of the fits for the retrieval of CO profiles using three microwindows. The spectra were recorded with a resolution of 0.005 cm -1 with SZA=72.99°. CO: 04060404.0pb.
22 CHAPTER III
(km) Altitude
Wave numbers (cm -1)
Figure 3.2.3. A plot of the spectral points from the three micro-wind ows used in the retrieval of CO (black lines). The Jacobians (K-matrix) in the background (color coded z-axis, with altitude in the x-axis and the corresponding spectral points in the y-axis) indicate where the information is coming from, as a function of altitude and which spectral points contribute to that information.
Averaging Kernels 40 0.02 - 1 km 35 1 - 3 km 3 - 5 km 5 - 7 km 30 7 - 9 km 9 - 11 km 25 11 - 13 km 13 - 15 km 15 - 21 km 20 21 - 31 km
Altitude (km) Altitude 15
10
5
0 -0.2 0 0.2 0.4 0.6 0.8 1
Figure 3.2.4. (left) Averaging kernels for an optical path diff erence (OPD) of 90 cm, Solar Zenith Angle (SZA) of 51° and a signal to noise ratio of 500. The amplitudes of the kernels give an indication of the sensitivity of the retrieval in each layer. The widths of the averaging kernels can be regarded as an indicat or of the vertical resolution for that layer. (right) Averaging kernels for the merged layers 0.02-3 km, 3-11 km, 11-19 km.
23 CHAPTER III
3.2.3.2. Strategy for C 2H6
-1 -1 Ethane is retrieved in the range 2976.51 cm to 2977.21 cm . The total columns of the interfering gases H 2O, O 3, and CH 4 are also retrieved in this interval. The uncertainties in the water vapor mixing ratios become a problem in the retrieval for stations such as Jungfraujoch in the Swiss Alps and La Reunion Island in the South Atlantic ( E. Mahieu and C. Vigoroux, personal communication ). Therefore, the H 2O profile for these stations is usually retrieved in the first step from the same spectrum. As a second step, the resulting profile is scaled while fitting ethane. However, this method is not used for retrievals in Spitsbergen for the following reasons; First, this might result in a strong bias in the retrievals that could influence the retrieved ethane columns. Second, daily water vapor measurements are performed in Spitsbergen which can serve as a priori information for the water vapor profiles. The figure below shows the variability of water vapor taken from sonde measurements in Spitsbergen (with the exception of January and February due to very scarce measurements). The sondes are carried by Helium filled balloons that are launched daily at noon at the Koldewey station.
Figure 3.2.5. A plot of the variability of water vapor taken from sonde measurements in Spitsbergen (with the exception of January and Februa ry due to very scarce measurements). The sondes are carried by Helium filled balloons that are launched daily at noon at the Koldewey station.
24 CHAPTER III
3.2.2.3. O 3 Triplet Spectroscopy Applied to C 2H6 Retrievals
p The spectroscopy being used for the Q3 branch of C 2H6 is based on Pine and Rinsland [1999]. Improvements to the spectroscopic line information have been made within the UFTIR -1 project. This includes revising the spectroscopic parameters of the O 3 triplet at 2977 cm [Mikhailenko et al., 2002]. The updated parameters allow significant improvement on the fit of these weak Ozone lines. An example of spectral fits using the old parameters and the new parameters is shown in Figure 3.2.6. The fits and the residuals in the 2976.9 cm -1 – 2977 cm -1 are clearly better in contrast to the fits and residuals using the old parameters.
A narrower interval (2976-66 cm -1- 2976.95 cm -1) has also been tested. Using this interval could help reduce the impact of the strong water vapor line located near 2977.95 cm -1. However, the information content in this interval is lower compared to the wider window. The retrieved columns have also shown a higher scatter.
r Another option for retrieving Ethane is using a window in the Q0 branch at around -1 p 2986.6 cm . Retrievals of total columns using this window agree with the retrievals in the Q3 branch with some small differences that vary with season (Fig 3.2.7). However, there are no updated spectroscopic parameters available for these lines. Moreover, this branch is even closer to a strong water vapor line compared to the 2976 window and there is a risk of inconsistencies that may arise with the parameters provided by Pine and Rinsland [1999]. Considering these deficiencies, it is not yet recommended to use this domain either separately or in combination p with the interval in the Q3 branch.
Retrieving profiles for Ethane with the current parameters and techniques is very difficult, if not almost impossible. Most of the information from the current Ethane line comes from below 10 km. A plot of the Jacobians showing the contributions of the different altitude levels to the information can be seen in Figure 3.2.8. An averaging kernel for a typical Ethane retrieval is shown in figure 3.2.9. The averaging kernel shows that almost all of the profile information for the layer 0 km-15 km comes from below 5 km, as indicated by the peak of the kernel. Therefore any interpretation that can be said about Ethane retrievals pertain only to the atmosphere from 0 km to about 10 km
25 CHAPTER III
relativeintensity
relativeintensity
Figure 3.2.6. An example of spectral fits using the old parameters and the new parameters. -1 -1 The fits and the residuals in the 2976.9 cm – 2977 cm (red boxes) are clearly better compared to the fits and residuals using the old parameters.
26 CHAPTER III
Figure 3.2.7. Retrievals of total columns of Ethane from 2000-2002 using three different windows. The retrieved trends agree with some small differences that vary with season.
(km) Altitude
Wave numbers (cm -1)
Figure 3.2.8 . A plot of the spectral points from the micro-window used in the retrieval of Ethane (black lines). The Jacobians (K-matrix) in the background (color coded z-axis, with altitude in the x-axis and the corresponding spectral points in the y-axis) indicate where the information is coming from as a function of altitude and which spectral points contribute to that information.
27 CHAPTER III
.
Figure 3.2.9 . C2H6 Averaging Kernels calculated for a retrieval with a spectrometer optical path difference (OPD) of 180 cm, signal -to-noise ratio of 300 and a solar zen ith angle of 57.83°. The kernel has a peak below 5 km, indicating that most of the contribution to the volume mixing ratio profile for this layer mostly comes from this altitude.
28 CHAPTER IV
4. Results and Discussions
This chapter presents the results gathered using the method of FTIR spectrometry. First, I will show sensitivity studies on the effects of choosing different a priori information for the retrieval of volume mixing ratio profiles. After establishing the validity of the a priori information used and of the retrieval method, I will show measurements of CO in the Atlantic on board the research vessel Polarstern in chapter 4.1. This chapter will show the latitude and altitude variability of CO and its distribution measured from 5 ship cruises. The measurements are also validated by model and by satellite data. Chapter 4.3 deals with the effects of fires on the CO in the atmosphere detected from long term measurements in the Arctic. This chapter is supplemented by studies of other trace gases (HCN and C 2H6), which are also products of biomass burning. A quantification of the amounts of CO released during intense burning events in the northern hemisphere in 1998 is shown in Chapter 4.4. This chapter is based on a work which is published in Yurganov et al., (2004), done in cooperation with Dr. Leonid Yurganov of JAMSTEC. Chapter 4.5 is a product of the increased altitude information that was obtained by optimizing the retrieval methods. This part shows and establishes that the FTIR spectrometry enables the detection of columns of strato-mesospheric CO from 18 km up to the upper mesosphere (100 km). Chapter 4.6 discusses the results of a recent and on going campaign in the tropics (Paramaribo, Suriname, 5.8° N, 55.2° W). This focuses on the measurements of enhanced CO and HCN values and the effect of the migration of the ITCZ (Inter-tropical Convergence Zone) over the measurement site.
29 CHAPTER IV
4.1. Sensitivity Studies on the Dependence of the CO retrievals on the A priori Information
4.1.1. Introduction
In this section, an illustrative view of the dependence of the retrievals on the a priori information is presented. It is shown that the derived information mostly comes from the measurement, an indication that the retrieval is robust. A sensitivity study on the data from one of the cruises is used for this purpose. The spectra were measured from a ship cruise that covers the latitudes 79°N to 70°S. This provides us a large ensemble of different CO profiles from different latitudes.
The experiments were performed using a Bruker 120M interferometer. The instrument was installed in a climate controlled laboratory container placed on top of the helicopter deck/ upper deck of the German research vessel (RV) “Polarstern“. A solar tracker mounted on the container‘s roof was modified to react faster in order to cope with the ship‘s movements. It is driven to follow the sun using a quadrant diode controller [ Notholt et. al., 2000]. The spectra were recorded between 750 and 6000 cm -1 using the sun as the light source. Measurements were made with an optical path difference of up to 200 cm corresponding to an unapodized resolution of 0.005 cm -1. Strict visual inspection by the personnel made sure that only spectra during cloud-free measurements were evaluated.
CO spectra were analyzed in three micro windows: 2057.70 cm -1 – 2057.90 cm -1; 2069.55 cm -1 – 2069.80 cm -1; 2157.30 cm -1 – 2159.35 cm -1. The interfering gases H 2O, N 2O, and O 3 are taken into account in all of the three micro windows. Their total columns are retrieved at the same time.
In order to study the dependency of the results on the a priori profiles, the retrievals at all latitudes were repeated with the CO a priori profile scaled by different factors, assigned a constant value and shifted in altitude. This was done to the combined cruise data in Dec. 1999- Jan. 2000 and Jul. 2000. The cruise data cover the southern hemisphere summer and the following northern hemisphere summer. Only the sensitivity of the retrievals with respect to the a priori is shown. Sensitivity studies of the retrieved profile to other parameters such as temperature, instrumental noise, etc. are discussed in e.g. Pougatchev & Rinsland [1995].
30 CHAPTER IV
4.1.2. Results
A latitude transect of the retrieved volume mixing ratio for CO during the cruise of Dec. 10, 1999- Jan. 18, 2000 (southern hemisphere summer) combined with the cruise from Jul. 1-26, 2000 (northern hemisphere summer) is shown in Figure 4.1.2 C. Also shown is the a priori VMR used for the retrieval. The ad-hoc a priori uncertainty was set to 20% for all layers. The averaging kernels for a typical retrieval have been shown in chapter III Figure 3.2.4. The averaging kernels for the averaged layers indicate that the CO retrieval is sensitive until about 20-30 km.
To show the dependency of the results on the a priori profiles, the retrievals at all latitudes were repeated with the CO a priori profile multiplied by 0.5, 0.8, 1.2 and 1.3. Selected results are shown in Figure 4.1.1A-F. The studies were done for the measurements between 40° S and 40° N. It has been shown that there is no strong variation in CO in the region south of 40° S [ Notholt et al., 2003]. Therefore this region has been excluded. As can be seen, the retrievals are only slightly influenced by the a priori depending on the scaling factor used. Note that the features are well retained. The CO enhancement just below the tropical tropopause is still clearly visible. The figures also show the clean southern hemisphere and the presence of pollution below 4 km in the northern hemisphere.
Even when the whole a priori profile is shifted downwards and upwards by 4 km (Figure 4.1.1 D and Figure 4.1.1 E respectively), the CO enhancements near the equator between 10 km- 15 km along with other features remain, although they are changed in intensity.
Finally, a CO a priori profile having equal values of 80 ppbv at all layers was used (Figure 4.1.1 F). Even with this unrealistic a priori profile, the CO enhancements near the equatorial region could still be resolved and the clean southern hemispheric air can be reconstructed despite the retrievals having yielded unrealistic and negative values above 30 km.
The retrievals were compared to in-situ measurements during the cruise (Figure 4.1.2). The measurements were taken from an in situ monitor, measuring the ultraviolet resonance fluorescence of CO at 150 nm. The instrument has an accuracy of ±1.3 ppbv and a detection limit of 3 ppbv (2 σ) for an integration time of 1 s [ Notholt et al., 2003]. The FTIR surface layer data (0 – 4km) showed good correlations with the in-situ data with the exception of a few measurements between the equator and 20° S. In this region, the FTIR surface layer data showed greater values compared to the in-situ data. Despite the perturbations in the a priori , the FTIR surface layer data showed small deviations.
Total CO column measurements are shown in Figure 4.1.3 (top). Using different a priori profiles has proven to have very little effects on the retrieval of the total column. The percentage differences of the total columns with respect to the original CO column using the standard retrieval (“VMR 1.0”) are mostly within ±2% (Figure 4.1.3 bottom plot).
31 CHAPTER IV
Altitude(km)
0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 -7 x 10
Figure 4.1.1. Latitude transects of the retrieved volume mixing ratios of CO and the corresponding a priori VMR profiles used. The measurements were taken from 70° S to 79° N. The CO enhancement at around 10 km-15 km in the equatorial region can still be resolved using different values of the a priori . The a priori VMR profiles were multiplied by 0.5, 1.0, 1.2, shifted down by 4 km, shifted up by 4 km and assigne d a constant value of 80 ppb, (A, B, C, D, E, and F respectively)
32 CHAPTER IV
CO In-situ and FT-IR Surface Data (0-4 km) from the Cruise 99-2000 300 In-Situ vmr = 80 ppb vmr x 0.5 250 vmr x 0.8 vmr x 1 vmr x 1.2 vmr x 1.3 vmr + 4km 200 vmr - 4km
150
CO(ppbv)
100
50
0 -80 -60 -40 -20 0 20 40 60 80 Latitude (°) Figure 4.1.2. Comparison between in situ CO data and CO as measured with the FTIR spectrometer on board the RV Polarstern by using different values of the a priori (the standard VMR used is labeled “VMR x 1“). The FTIR surface data were averaged from 0 to 4 km. Spikes from the in-situ measurements are attributed to background CO from t he ship’s exhaust and other local sources (e.g. Harbor in Cape town at 33° S).
33 CHAPTER IV
Total Column Measurements 18 x 1 0 with respect to Changing A priori Columns 3 VM R X 1.0 VM R X 0.5 VM R X 1.2 2.5 VMR = 80 ppb VM R -4 km VM R + 4 km 2 2
1.5
Molecules/cm
1
0.5 -8 0 -6 0 -4 0 -2 0 0 2 0 4 0 6 0 8 0 A-priori Latitude (degrees)
Percentage Differences of Total Column Retrievals 6 A priori VMR x 0.5 A priori VMR x 1.2 A priori VMR = 80 ppb 4 A priori VMR - 4km A priori VMR + 4km
2
0 %
-2
-4
-6 -80 -60 -40 -20 0 20 40 60 80 Latitude (degrees)
Figure 4.1.3. Total CO column measurements with respect to different a priori VMR profiles (top). The corresponding percentage difference of the total column retrievals with respect to the standard retrieval is shown in the bottom figure
34 CHAPTER IV
4.1.3. Summary
CO can be retrieved up to 20 km-30 km with FTIR spectrometry. The sensitivity studies show that the basic vertical structure of the CO profile with regards to the location of enhanced layers of CO is well captured despite the differences in the choice of the a priori profile. The retrievals cover the northern hemisphere, where industrial sources influence the atmospheric CO concentration, the tropics where biomass burning contributes to enhanced CO and the relatively clean southern hemisphere. The averaging kernels from the retrievals give 4-5 degrees of freedom corresponding to 4-5 independent pieces of information. The shape of the kernels indicates that for a few layers, a maximum altitude resolution of 4 km to 5 km can be achieved for CO retrievals. It is calculated that a relative change of 20% in the a priori VMR (6- 20 ppb) leads to less than 8% (<10ppb) change in CO concentration for each layer below 16 km. The retrievals also show reasonable results despite assuming a ±4 km-shift in the a priori and even when a constant a priori was used. The retrievals of surface CO concentrations with the FTIR also show good agreement with in-situ data. To summarize, we have shown that a single CO a priori profile could be used for FTIR retrievals of CO at all latitudes from the ground up to 16 km.
The issue of choosing an a priori profile is significant for studying combined measurements from different groups and might give rise to problems during comparisons. A study by Yurganov et al., [2004], also shown in chapter 4.4, involves several FTIR measurements of CO total columns from different locations in the northern hemisphere. A standard CO a priori profile has not been established among the groups. However, through this study, we have shown that the a priori profile could have different values with very small effect on the retrieved column. Moreover, this study provides complementary information to the total column observations that will be shown in chapter 4.4 due to the altitude information that is presented here.
In-situ measurements are helpful in verifying the surface measurements of the FTIR. Both have been shown to agree well. However, in-situ measurements will not be able to represent the whole troposphere due to the fact that CO shows a strong variability as a function of latitude. For instance, high CO VMRs are often present on the surface and in the upper troposphere in the equatorial regions as discussed in Notholt et al., [2003]. In the northern hemisphere high values of CO VMRs are typical at lower altitudes. In the southern hemisphere high CO VMRs are often only found in the upper troposphere.
35 CHAPTER IV
4.2. Latitude and Altitude Variability of Carbon Monoxide in the Atlantic Detected from Ship-borne Fourier Transform Spectrometry, Model and Satellite Data
4.2.1. Introduction
Measurements of the global variations of trace gases are important for the understanding of chemical and dynamical processes that control the distribution of these trace gases both in the free troposphere and in the stratosphere. Fourier Transform Infrared (FTIR) spectroscopy has been found to be one of the most suitable instruments for the measurements of atmospheric trace gases [ Rao and Weber , 1992].
Measurements of the trace gas CO over the Atlantic are discussed in this section. CO is primarily produced from biomass burning, oxidation of CH 4 and other biogenic hydrocarbons and fossil fuel combustion. It has a lifetime ranging from weeks to a few months and it is an effective indicator of how transport processes distribute atmospheric pollutants from biomass and fossil fuel burning on a global scale. The main sink of CO is oxidation by OH. Atmospheric CO is responsible for more than half of the total turnover of OH [ Crutzen and Zimmermann , 1991]. In the tropics, CO from biomass burning events can be effectively transported upwards by deep convection and can reach high altitudes in the tropical troposphere [ Krishnamurti et al., 1996, Sherwood et al., 2000, Notholt, et al., 2003].
Along with other pollutants, biomass burning releases a significant amount of CO. Estimates of this amount range from about 300 Tg of CO per year to more than 700 Tg of CO per year (e.g. Bergamaschi et al., 2000, Holloway et al., 2000, Inter-governmental Panel for Climate Change (IPCC), 1996, IPCC 2001). Biomass burning also controls the CO concentration in localized regions near the surface, primarily in the tropics, yielding a 15-30% contribution to CO concentrations throughout most of the troposphere [ Holloway et al., 2000].
Due to the localized intermittent sources and its short lifetime, CO is not homogeneously distributed in the atmosphere. In the Atlantic, backward trajectories starting at different altitude levels indicate that air parcels originate from the African, American and the European continents, depending on the altitude. Upwelling in the tropical Atlantic also plays an important role in transporting pollutants to the free troposphere [ Holton, et al., 1995, Talbot, et al., 1996]. All of these processes lead to very uneven distributions of CO in the troposphere. Ship-borne i n-situ and surface measurements of total columns are not able to provide information on the vertical distribution of CO. Therefore there is not enough information about the transport of CO based on these measurements alone. In order to achieve a better picture of the transport of CO it is crucial to measure VMR profiles with a sufficient altitude resolution. In our case, CO VMR profiles could be retrieved up to 30 km with a resolution of up to 4 km for some layers.
4.2.2. MOPITT data
Data from space borne measurements of CO were obtained from the MOPITT instrument on board the Terra spacecraft launched in December 1999. Terra has a sun-synchronous orbit at an altitude of 705 km. The MOPITT instrument retrieves CO mixing ratios using gas correlation radiometry at 4.7 µm and 2.3 µm [ Drummond, 1992]. The instrument measures in the nadir
36 CHAPTER IV direction with a 22 km x 22 km pixel resolution. Global coverage is achieved in about 4 days. The retrieval algorithm is also based on the optimal estimation technique, which requires an a priori CO profile [ Deeter, et al., 2003]. The retrieval algorithm is discussed in more detail in the works of Pan, et al, [1998] and Deeter, et al., [2003]. From the analysis of the MOPITT averaging kernels, the retrieved CO profile has essentially two independent pieces of information; at 500-800 hPa and 150-350 hPa [ Bremer et al., 2004]. The retrievals were done only for cloud-free daytime data. The pixels were selected within a radius of 200 km from the position of the RV Polarsten and averaged for ± 2 days.
4.2.3. Model data from MATCH-MPIC
MATCH-MPIC, the Model of Atmospheric Transport and Chemistry has been developed for the investigation of global tropospheric chemistry. It has been described in detail in several publications [ Rasch et al., 1997 , Lawrence et al, 1999, 2003, von Kuhlmann, 2003]. It is described as an “offline” model that reads in gridded time-dependent values for the basic meteorological parameters; e.g. temperature, surface pressure, and horizontal winds. It then employs these parameters to diagnose further meteorological parameters such as cloud fields and convective mass fluxes that are required for atmospheric chemistry simulations.
MATCH-MPIC has been used to calculate CO mixing ratios along the track of the research vessel Polarstern. The positions of the ship every three hours (the model output frequency) were used to yield interpolated CO VMR profiles at the location of the FTIR measurements. This allows for a direct comparison of the latitude transects both from the ship measurements and the model results. The contributions of regional CO tracers from various sources (e.g. biomass burning, fossil fuel combustion and oxidation of methane and NMHCs) were also calculated from the model; the regions are the same as defined in Lawrence et al., [2003], with the addition of a breakdown of the CO tracers into source categories.
4.2.4. Five Polarstern Cruises from 1996-2003
A total of five cruises with the FTIR have been carried out on board the RV Polarstern within 1996-2003. The retrieved volume mixing ratio profiles for the cruises ANT-XIV (Oct. 12 – Nov. 4, 1996), ANT XVII-1&2 + ARK XVI-1 ( Dec. 10, 1999- Jan. 18, 2000 and Jul. 1-26, 2000) , ANTXX-1 (Nov. 7-20, 2002), ANTXX-3 (Jan. 25- Feb. 14, 2003) and ANT XXI-1 (Oct. 21- Nov 13, 2003) are shown in Figures 4.2.1 A-E. In the lower troposphere, the typical features are; the relatively CO- rich northern hemisphere, high CO volume mixing ratio profiles in the equatorial regions and a relatively clean lower troposphere in the southern hemisphere. The CO enhancement in the upper troposphere (between 10 km-15 km) in the tropical regions is worth noting. This is evident in all the cruises. These enhancements can also be seen to extend to 20° S to 30° S also at the 10 km -15 km level.
A three-dimensional slice of the atmosphere along the track of the RV Polarstern showing the VMR of CO at different altitude levels viewed from the west and east is shown in figure 4.2.2. The measurements took place in Jan.-Feb. 2003. The red dots correspond to fires measured by the MODIS web fire mapper for the same months (http://maps.geog.umd.edu/default.asp ). The backward trajectories correspond to different pressure levels along the track; 700 hPa or
37 CHAPTER IV about 2.6 km (green), 300 hPa or about 8.7 km (blue) and 140 hPa or about 14.2 km (magenta). The trajectories were obtained from the Deutsche Wetter Dienst (German Weather Service http://www.dwd.de). They were calculated using 3-dimensional winds. Note the vertical circulation of air parcels in the southern Atlantic indicated by the trajectories at 300 hPa and 140 hPa as well as the CO enhancements in this region. Air parcels directly coming from areas with high concentration of fires could be seen to contribute to high CO mixing ratios (reddish color) near the equator.
38 CHAPTER IV
A B a
C D
-7 x 10 2
1.8
1.6
E 1.4 1.2
1
0.8
0.6
0.4
0.2 VMR
Figure 4.2.1. A-E. Variations of CO volume mixing ratios with latitude and altitude measured
from five different cruises in the Atlantic Ocean. The CO enhancement in the upper
troposphere (about 10 km-15 km) in the tropics and sometimes extending to 20° S-30° S is
evident in all of the cruises. A biomass-burning plume was directly encountered during the cruise in Jan.-Feb. 2003, resulting in high mixing ratios (red) near the equator (d).
39 CHAPTER IV
Figure 4.2.2. A slice of the atmosphere along the track of the RV Polarstern showing the CO VMRs at different altitude levels viewed from the west and the east. The red dots correspond to fires measured by the MODIS web fire mapper (http://maps.geog.umd.edu/default.asp). The backward trajectories correspond to different pressure levels along the track; 700 hPa or about 2.6 km (green), 300 hPa or about 8.7 km (blue) and 140 hPa or about 14.2 km (magenta). Note the vertical circulation of air in the southern Atlantic indicated by the trajectories at 300 hPa and 140 hPa near the African continent as well as the CO enhancements in this region. (Backward trajectories from the German Weather Service: http://www.dwd.de ). For an animated sequence, please refer to the auxiliary mpeg material in the CD.
40 CHAPTER IV
4.2.5. FTIR, Model and Satellite Data
CO volume mixing ratio profiles from the MATCH-MPIC model, the FTIR on RV Polarstern and MOPITT on the Terra satellite are shown in Figure 4.2.3. The data are shown for the three cruises in Nov. 2002 (Fig. 4.2.3 A-C), Jan.- Feb. 2003 (Fig. 4.2.3 G-I) and Oct-Nov. 2003 (Fig. 4.2.3 J-L). Data from MOPITT and from MATCH are also shown corresponding to the positions of the RV Polarstern on these dates. The color plots are latitude transects along the track of the RV Polarstern. All three data sets show similar structures regarding the CO profiles near the equator below 5 km and for most of the measurements in the lower troposphere in the northern hemisphere. For visual clarity some of the FTIR measurements are interpolated in this plot (true measurements are indicated by a marker on top of each plot and interpolation, if necessary is made without exceeding 15° of distance between the data points). Where data is absent, for instance near the equator during the cruise in Oct. ’03, some disagreement is expected. The MOPITT data reaches only up to 16 km. Due to MOPITT’s coarse resolution, some features may not be resolved such as the second maximum in the CO VMR profile near the equator during the cruise in Jan. ’03, which can be seen in the FTIR and the model data. CO enhancements at high altitudes near 20° S that could be seen from the model and the FTIR data are also visible in the MOPITT measurements but to a certain extent are difficult to distinguish due to the limited vertical resolution.
4.2.6. Simulating Low- Resolution Retrievals from High Resolution Retrievals
The profiles from MATCH have a higher resolution compared to the FTIR and MOPITT. Due to the nature of the retrieval problem being under determined, the averaging kernels of the FTIR and MOPITT tell us that the retrievals at each designated atmospheric layer has contributions from neighboring layers. To account for the different characteristics of the FTIR and MOPITT instruments, especially the averaging kernels, we have included a study using the method of retrieval simulation, which is discussed in Rodgers and Connor [2003] and Palm et al,. [ 2004]. The FTIR retrievals were done with the same a priori profile (until 24 km) as used in the MOPITT retrievals. The MOPITT retrievals were then simulated by the FTIR retrievals using the averaging kernels of MOPITT. A similar method has been applied to the profiles from MATCH using the FTIR and MOPITT averaging kernels with the assumption that MATCH is the true profile.
The results of the measurements from Nov. 2002 are shown in Figure 4.2.3 (A-C). The top plots (A-C) show the original retrievals from the 3 platforms. The plot of the profiles from MATCH smoothed by the averaging kernels of the FTIR is shown in Fig. 4.2.3D. Figure 4.2.3E shows the FTIR profiles smoothed with the averaging kernels of MOPITT. The MATCH profiles smoothed with the MOPITT averaging kernels are shown in Fig. 4.2.3F.
The true VMR at a layer can never be known. Using the method of retrieval simulation, it can be said that Fig 4.2.3 D & F would show what the FTIR and MOPITT would have seen if the MATCH profiles represented the true CO profile. On the other hand, the plot in Fig. 4.2.3 E would show what MOPITT would have seen if the FTIR profiles represented the true CO profile.
41 CHAPTER IV
MATCH-MPIC FTIR MOPITT
Nov 2002 -7 x 10
A B C 2
1.8
1.6
1.4 D E F 1.2
1
0.8
Jan 2003
Altitude (km) 0.6 G H I 0.4
0.2
VMR
Oct 2003
J K L
Latitude (degrees)
Figure 4.2.3. A-C, G-I, J-L. CO volume mixing ratio data from MATCH-MPIC (left column), the FTIR on R V Polarstern (middle column) and MOPITT on the Terra satellite (right column). The color plots are latitude transects along the track of the RV Polarstern. The MOPITT data reaches only up to 16 km. Figure 4.2.3D shows CO profiles from MATCH for Oct. 2002 s moothed by FTIR averaging kernels. CO profiles from the FTIR and from MATCH for Oct. 2002 smoothed by MOPITT averaging kernels are shown in E and F respectively.
42 CHAPTER IV
4.2.7. Assessment of CO sources
In order to assess the origins of the CO enhancements, we have calculated the contributions of regional tracer fields from the MATCH-MPIC model. A selection of the most significant sources for the three cruises and their percentage contribution is shown in Figure 4.2.4. The absolute contributions are shown in Figure 4.2.5. Most of the CO in the equatorial regions come from African biomass burning (AFRBB), with more than 40% contribution during the cruise in Jan. 2003. Oxidation of non-methane hydrocarbons (NMHC) gives about 20-30% contribution to the enhancements in the upper troposphere in the southern hemisphere with the largest contribution during the cruise in Jan. 2003, where air parcels seem to have been entrained in vertical circulations. South American biomass burning (SAMBB) also contributes to the enhancements in the upper troposphere in the southern hemisphere especially for the cruise in Nov. 2002 and Oct. 2003 but SAMBB for the same region is almost absent during the cruise in Jan. 2003. Fossil fuel combustion from the North American continent (NAMFF) has a relatively stable contribution (up to about 20-25%) to the northern hemisphere CO measured in all three cruises. The most dominant source of the background CO is the oxidation of methane (see Figure 4.2.5). It covers large regions of the northern and southern hemispheres including the stratosphere with contributions well above 30%.
43 CHAPTER IV
Nov 2002 Jan 2003 Oct 2003
NAMFF
AFRBB
SAMBB
Altitude(km)
METHANE
NMHC
Latitude (degrees)
%
Figure 4.2.4. Contributions of each regional tracer field to the CO budget for the three cr uises calculated from MATCH-MPIC (in percent). The CO tracer fields are; North American Fossil Fuel (NAMFF), South American Fossil Fuel (SAMFF), African Biomass Burning (AFRBB). There are also tracer fields of CO from methane oxidation (METHANE) and from N MHC oxidation (NMHC)
44 CHAPTER IV
Nov 2002 Jan 2003 Oct 2003
NAMFF
AFRBB
SAMBB
Altitude(km)
METHANE
NMHC
Latitude (degrees)
ppb
Figure 4.2.5. Absolute contributions of each regional tracer field to the CO budget for the three cruises calculated from MATCH-MPIC (in ppb). The CO tracer fields are the same as in Fig. 4.2.4.
45 CHAPTER IV
4.2.8. Summary
A recurring CO enhancement in the upper troposphere (at 10 km-15 km) in the Southern Atlantic has been detected with the FTIR spectrometer in 5 ship cruises spanning a period of 8 years. However, the measurements are limited to one period (during and close to the southern hemisphere summer). Backward trajectories calculated for the cruise in Jan. 2003 suggest different sources that account for the CO enhancements at different latitudes. In the equatorial regions down to about 25°S, air parcels mostly originate from the African continent. The scenario is different for regions south of 25°S, air parcels found at about 8 km above the cruise track could have originated from South America, as the trajectories suggest. Convective transport of polluted air from the continents from low to high altitude most likely explains the high CO enhancements in the upper troposphere (10 km-15 km). Recycling of air in the Atlantic coming from Africa was also observed from the trajectories. This may lead to persistence of pollutants in the southeastern tropical Atlantic as observed by previous studies of Talbot et al., [1996] and Garstang et al., [1996]. Backward trajectories suggest that some air parcels appear to be carried along for days in vertical circular flow patterns near the coast of the southern African continent. These flow patterns also vary with season because they were not observed in the other cruises.
Measurements from the MOPITT instrument complement the measurements of the FTIR. There is a very good agreement between the FTIR and MOPITT data in the lower troposphere of the northern hemisphere. However, some features of CO enhancements (e.g., a second maximum) in the upper troposphere seen in the FTIR data is difficult to detect in the MOPITT data due to the coarse resolution of MOPITT. Retrieval simulation wherein the averaging kernels are considered leads to a reasonable agreement between MATCH, MOPITT and the FTIR.
The detected enhancements in the upper troposphere of the southern hemisphere by the FTIR agree with results from MATCH-MPIC. It is also important to note that the lower troposphere in the southern hemisphere is relatively clean as seen from both data sets, therefore in-situ surface and total column measurements alone would not be sufficient to detect the enhancements in this region.
The calculation of the contribution of the major regional tracers shows that methane oxidation provides a large background source of CO while biomass burning and other sources account largely for the variability in CO. Biomass burning emissions from the African continent provide the most dominant source in the equatorial regions. In the 10 km-15 km region south of 20°S, oxidation of non-methane hydrocarbons seems to contribute more to the CO enhancements than methane oxidation and biomass burning, especially during the cruise in Jan. ‘03. South American emissions from biomass burning have a significant contribution in the southern hemisphere except for the case during the Jan. 2003 cruise, where air parcels in this region appear to be carried along in circular flow patterns. At the same time CO from oxidation of NMHC’s during the cruise in Jan. 2003 are the highest compared to the 2 other cruises. This highlights the importance of the transport of pollution on CO concentrations.
46 CHAPTER IV
4.3. The impact of Forest Fires on CO concentrations in the Northern Hemisphere
4.3.1. Introduction: Fires in the Ecosystem
Recurring fires are part of the natural environment--as natural as rain, snow, or wind [Heinselman , 1978]. Fire-adapted ecosystems exist in North America. In these ecosystems called “fire-dependent ecosystems”, fire is an essential part in maintaining the function and balance of the whole ecosystem. Fire affects the functioning of ecosystems in numerous ways according to Heinselman , by:
• Regulating plant succession • Regulating fuel accumulations. • Controlling age, structure and species composition of vegetation. • Affecting insect and disease populations. • Influencing nutrient cycles and energy flows. • Regulating biotic productivity, diversity and stability. • Determining habitats for wildlife.
However, anthropogenic activities disrupt the natural cycle of recurring fires. Coupled with extreme drought conditions (e.g. during El Niño), uncontrolled fires caused by humans could lead to widespread burning events that could significantly affect the composition of the Earth’s atmosphere.
4.3.2 The El Niño Phenomenon
Fishermen who lived in South America in the 1500’s began to wonder about a current of unusually warm water that came to their shore every few years near Christmastime. The fishermen, who already embraced the Spanish language and Christianity at the time, associated this phenomenon with the birth of the Christ child at Christmas and named the warm water “El Niño”, which means "the infant" in Spanish.
El Niño has come to be used as a term for abnormal warming events that recur at intervals of about 2-7 years and typically last for a few seasons. It is a vast swath of warm water with roughly the size of Europe and usually appears off the coast of Peru.
During major events, the warming brings nutrient-poor tropical water southward along the west coast of South America. El Niño is associated with atmospheric circulations that produce wide ranging effects on global weather and climate. This expanse of warm water, about 24 to 29 degrees Celsius, remains in the western Pacific ocean near Australia most of the time, pushed and held there by westbound trade winds. But once in a while, the winds slacken, and even move in the opposite direction.
Normally, winds blowing from east to west (the trade winds) -- push water from the South American coast so that the ocean's surface is lower off Peru than it is off Indonesia by a few feet. During El Niño, the warm water piled up near Indonesia flows back across the Pacific to the South American coast.
47 CHAPTER IV
The reorientation of this flow can send devastating storms to the coasts of south west United States, droughts to Australia and south east Asia and cool, wet winters as far away from the Pacific as portions of Texas and other Gulf coast states (www.cnn.com/SPECIALS/el.nino/).
4.3.2.1. The El Niño episode in 1997-1998
The El Niño episode in 1997-1998 was the most important climatic oscillations affecting area burned and fire impacts in the 90’s according to the Food and Agricultural Organization of the United Nations [2001]. In these years, most of tropical Asia, Africa, the Americas and Oceania regions experienced extremely extended wildfire situations. During 1997-1998, the amount of land-clearing fires and other escaped fire situations have increased in the equatorial forest regions of Southeast Asia and South America. The northern temperate/boreal forest zone also experienced extremely dry years in the 90’s. The Far East of Russia was severely affected by wildfires during the 1998 drought.
Some ecosystems like the rain forests of Indonesia and Brazil and the cloud forests of Mexico, which are areas that are usually not seriously affected by forest fires, also sustained considerable damage in 1998 3.
The next sections of this work deal with measurements of enhanced levels of trace gases that are products of biomass burning. It will be shown that these enhanced levels were caused by anomalous fire events due to dry spells associated with the 1997-1998 El Niño episode.
3 Public Policies Affecting Forest Fires in the Asia-Pacific Region, James Schweithelm; and Public Policies Affecting Forest Fires in the Americas and the Caribbean, Robert W. Mutch et al; In FAO Meeting on Public Policies Affecting Forest Fires, FAO Forestry Paper 138, Rome 1999.
48 CHAPTER IV
4.3.3 Long Term Measurements of Trace Gases in the Arctic and the effects of Fires in 1998
Long-term measurements of trace gases in the northern hemisphere reveal significant enhancements of CO concentrations in 1998. This increase is attributed to widespread fires that occurred that year. Figure 4.3.1 shows fire counts in the northern hemisphere from the Along Track Scanning Radiometer (ATSR), an instrument on board the European Remote Sensing Satellite-2 (ERS-2). ( http://dup.esrin.esa.it/ionia/wfa/index.asp ). More information on the ATSR project can also be found on http://www.atsr.rl.ac.uk .
In the northern hemisphere, the year 1998 has the largest amount of fire counts recorded from 1997-2003 with the highest counts in August of that year (Fig. 4.3.1 A). However, an increase in the number of fires can already be seen in 1997 in the southern hemisphere (Fig. 4.3.1 B). This increase corresponds to the time of the widespread fires that occurred in Indonesia and other parts of South East Asia. Also, the year 1997 was an El Nino year. The fires in the southern hemisphere were mostly within the latitude belt 0° - 30° S and the abnormalities also occurred mostly in this region. On the other hand, increased fire counts in the northern hemisphere mostly took place within the latitude belt 30°N to 60° N, especially in 1998.
The map in Fig. 4.3.2 shows the locations of the fires that occurred in Aug. 1998. The red squares represent hot spots detected by the ATSR instrument. It can be seen from the figure that fires from South America and the African countries dominate the fire counts in the southern hemisphere.
In the northern hemisphere, 1998 was unusual for both North America and the former USSR countries in terms of fire events. The total area burned amounted to 5.3 to 5.6 million hectares for each of the regions. In the former USSR states, the area burned in 1998 was more than 4 times larger compared to the areas burned in other years (Fig. 4.3.3). This shows that the contribution to the increased fire counts were mostly from the former USSR states.
49 CHAPTER IV
A
Fire Counts Fire B
C
D
Fire Counts Fire
E
Figure 4.3.1 . Total Fire counts in the northern and southern hemisphere from 1997-2003 (A & B). A large percentage of the fires that occur red during the summer of 1998 were from the latitude belt of 30°-60° (D). In August 1998, the contribution of the fires between 30° and 60° comprises about 80% of the total fires in the northern hemisphere for that month. Data were taken from ERS-2 ATSR-2 night-time data (1995 - 2002), ENVISAT AATSR night-time data (2003 - present) at http://dup.esrin.esa.it/ionia/wfa/index.asp
50 CHAPTER IV
Figure 4.3.2. Fire counts (red squares) from ATSR for the month of August 1998. 25881 hot
spots. http://dup.esrin.esa.it/map/wms_docs/wms_index.php?context=fireatlas.xml
Total Area of Fires 8.0E+06 Europe 7.0E+06 former USSR
6.0E+06 North America 5.0E+06
4.0E+06
3.0E+06 Area (Hectares) Area 2.0E+06
1.0E+06
0.0E+00 1996 1997 1998 1999 2000 Year Figure 4.3.3. Total Area of fires. source: Economic Commission for Europe, Timber Bulletin 54/4, Volume LIV, No.4, 2001. http://www.unece.org/trade/timber/ff-stats.html
51 CHAPTER IV
4.3.4. Measurements of CO, C 2H6 and HCN in Spitsbergen
Total columns of CO, C 2H6 and HCN averaged over one day (blue circles) and over the whole month (red dots) are shown in Fig. 4.3.4. CO has a maximum column concentration during the Arctic wintertime. CO then starts to decrease from early spring to summer as sunlight comes back to the Arctic producing the OH radicals that clean up the CO accumulated during the polar night.
In order to estimate the total column anomaly in 1998, the monthly means for that year were normalized by the monthly means from the “normal” years, i.e. the years from 1996-2001 while excluding the months in 1998.
An enhancement of CO columns (up to 50% increase) is evident during late summer to early fall of 1998 as shown in Figure 4.3.5 (top). This anomaly is attributed to widespread biomass burning events that occurred during that year. In addition to this, enhancements can also be seen in HCN and C 2H6 (Fig. 4.3.5 (center & bottom plots).
A 300-hr backward trajectory run starting on September 20, 1998 (see Fig. 4.3.6) clearly suggests that air parcels coming into Ny Ǻlesund could have been transported from the main zones where intense boreal fires developed as identified by the fire map (Fig. 4.3.2). These zones are in Eastern Siberia, Northern Japan and the Yukon and Northwestern territories in Canada.
HCN is a sensitive tracer for large scale biomass burning [ Li, et al. , 2000]. At ground level, it is produced by agriculture, biomass burning and production of coke. Although typical seasonal cycles of CO and HCN differ, Fig. 4.3.5 (bottom) reveals an enhancement in HCN columns coinciding with the CO enhancement. This implies the similarity of their sources. C 2H6 exhibits the same seasonal trend as CO (Fig. 4.3.4 (center)). The enhanced C 2H6 in 1998 (fig. 4.3.5, center) also developed simultaneously with the enhancement of CO and HCN.
52 CHAPTER IV
18 x 10 CO Daily and Monthly Average 4 Monthly Average Daily Average 3.5
3
2.5
2
1.5
1
0.5 1996 1997 1998 1999 2000 2001 2002
) C H Daily and Monthly Average 16 2 6
2 x 10 6 Monthly Average Daily Average 5.5
5 ) 4.5 4
3.5
3 2.5 Total Column (molecules/ cm (molecules/ Column Total 2
1.5 1 1996 1997 1998 1999 2000 2001 2002 Years
15 x 10 HCN Daily and Monthly Averages Total columns (molecules/cm 14 Daily Average Monthly Average 12
10
8
6
4
2
0 1996 1997 1998 1999 2000 2001 2002 Years
Figure 4.3.4 . Daily and Monthly Averages of CO (top), C 2H6 (center) and HCN (lower) total columns showing the seasonal trends from 1996-2001.
53 CHAPTER IV
CO Related to Normal Monthly Means for Total Column, Spitsbergen 1.8
1.7 000, 2001) 000, 1.6 1.5 1.4 1.3 1.2 (Monthly Means)/ 1.1 1 (Ave. for 96,97,99,2000,2001 0.9 ((Monthly Means)/( averaged for 1996, 1997, 1999, 2 1999, 1997, 1996, for averaged Means)/( ((Monthly 0.8 1996 1997 1998 1999 2000 2001 2002 Years
HCN Related to Normal Monthly Means for Total Column, Spitsbergen 3
00, 2001) 00, 2.5
2
1.5
(Monthly Means)/
1 for(Ave. 96,97,99,2000,2001 (Monthly Means)/( averaged for 1996, 1997, 1999, 20 1999, 1997, 1996, for averaged Means)/( (Monthly 0.5 1996 1997 1998 1999 2000 2001 2002 Years C H Related to Normal Monthly Means for Total Column, Spitsbergen 2 6 1.5
1.4 2001) 00,
1.3
1.2
1.1 (Monthly Means)/ 1
(Ave. for 96,97,99,2000,2001 0.9 (Monthly Means)/( averaged for 1996, 1997, 1999, 20 1999, 1997, 1996, for averaged Means)/( (Monthly 0.8 1996 1997 1998 1999 2000 2001 2002 Years
Figure 4.3.5. Anomalies of CO, C 2H6 and HCN related to the normal monthly means from 1996-2001 with the exception of 1998.
54 CHAPTER IV
Figure 4.3.6 . Backward trajectories indicating the origins of the air parcels coming into Ny Alesund (obtained from the Hybrid Single-Particle Lagrangian Integrated Trajectory (HYSPLIT) Model at the National Oceanic and Atmospheric Administration we bsite: http://www.arl.noaa.gov/ready/hysplit4.html).
55 CHAPTER IV
4.3.5. Summary
Fires are an important part of the earth’s ecosystem. However, anthropogenic activities disrupt the natural cycle of recurring fires. Man-made forest fires and other biomass burning activities have changed natural ecosystems and now serve as the major sources of trace gases such as CO, C 2H6 and HCN. The El Nino episode in 1997-1998 amplified the amount of fires by causing droughts and dry spells in many areas especially in south-east Asia. The far eastern regions of Russia were also severely affected by wildfires during the 1998 drought. Due to this, statistics showed a large increase in the number of areas burned in the former Soviet Union in 1998. The areas burned are almost double compared to the other years. This leads to a significant increase in atmospheric pollution that could extend up to the Arctic region as indicated by backward trajectory analysis. Simultaneous enhanced columns above Spitsbergen in mid 1998 were observed for CO, HCN and C 2H6 using Fourier Transform Infrared Spectroscopy (FTIR). Prevalent biomass burning events in boreal areas are the most likely cause of the enhancements. Further investigations on the affected trace gases could provide more information and improve the quantification of the effects of biomass burning on the earth’s atmosphere .
56 CHAPTER IV
4.4. A Quantitative Assessment of the 1998 CO Anomaly in the Northern Hemisphere Based on Total Column Measurements with FTIR Spectrometry.
4.4.1. Introduction
This section focuses on the measurements of CO abundance in the high northern hemisphere (30°N to 90°N) between January 1996 and December 2001. This work was performed in order to provide information on the amount of CO that can be produced due to abnormal burning events. Parts of this section were results of a collaborative work published in Yurganov et al., [2004]. The measurements were obtained using two approaches; CO total column amounts retrieved using FTIR spectrometry and CO mixing ratios measured by ground- based in-situ stations.
Anomalies of the high northern hemisphere (HNH) CO burden were analyzed by averaging the data from different measuring stations in the northern hemisphere. “Anomalies” is defined here as the deviations of the total tropospheric CO mass between 30°N and 90°N from the mean seasonal profile. The mean seasonal profile is determined from a 5-year average (1996- 2001, excluding 1998).
4.4.2. Experimental Techniques
Data were gathered from FTIR spectrometers operated at nine FTIR sites and in-situ surface monitoring stations. Table 4.4.1 & Table 4.4.2 summarize the locations and further details of each site. Data from two neighboring stations (Rikubetsu and Moshiri in northern Japan, 155 km apart) were merged and is designated here as “Hokkaido”. All sites except Zvenigorod are operated within the Network for the Detection of Stratospheric Change (data are also available at the NDSC website: http://www.ndsc.ws ). The measurement frequency depended on weather conditions and duration of clear sky. Data were absent for the Arctic and sub Arctic stations during polar night and during very low solar zenith angle in early and late winter. The measurements at the Kitt Peak observatory were lower in number compared to the other sites. All column data are given as number of molecules per unit surface area above the station. Atmospheric pressure correction was not applied.
4.4.3. Error Estimates
The random error of a single measurement is around 2% - 3% and the systematic error is 5% as estimated by Rinsland et al., [1998]. Typical standard deviations of the daily mean are estimated at ±15% [ Zhao et al., 2002]; ±10% - ±12% [ Yurganov et al., 1999]. Changes in air masses with different amounts of CO concentration mainly account for this variability.
Another factor that might lead to errors is the choice of retrieval parameters, such as the a priori CO profile which serves as an input in the retrieval codes used by most groups. A standard CO profile has not been established among the groups, however it was shown in Velazco et al., [2005] and in chapter 4.1 of this work that there is a lot of freedom in choosing the a priori profile and that the effects of the differences in the a priori information on the total column retrievals are very small.
57 CHAPTER IV
Table 4.4.1. Characteristics of sites, spectrometers, and codes of retrieval procedures. Site Coordinates Altitude, Type of Typical Typical m asl spectrometer resolution, cm -1 number of spectra per day Ny Ǻlesund, 78.92º N 20 Bruker IFS 0.005 2-3 Spitsbergen 11.94º E 120HR Kiruna, Sweden 67.84º N 419 Bruker IFS 0.005 2-3 20.41º E 120HR Harestua, 60.22º N 596 Bruker IFS 0.005 12 ± 6 Norway 10.75º E 120M Zvenigorod, 55.70º N, 200 Grating, 0.18-0.23 17 ± 6 Russia 36.80º E home-made Zugspitze, 47.42º N 2964 Bruker IFS 0.0045 4 ± 3 German Alps 10.98º E 120HR Jungfraujoch, 46.55º N 3580 Bruker IFS 0.0028 and 6 Swiss Alps 8.00º E 120HR 0.0044 Moshiri, 44.37º N 280 Bruker IFS 0.0028 and 2-8 Hokkaido, Japan 142.27º E 120HR 0.0035 Rikubetsu, 43.46º N 370 Bruker IFS 0.0035 2-8 Hokkaido, Japan 143.77º E 120M Kitt Peak, U.S.A. 31.9º N 2090 McMath- 0.01-0.02 2-3 111.6º W Pierce FTS
Table 4.4.2. Surface CO monitoring locations Name Agency Latitude Longitude Altitude, m
Alert, Nunavut, Canada CMDL 82.45 -62.50 210 Ny-Ǻlesund, Spitsbergen, Norway CMDL 78.90 11.88 475 Barrow, Alaska, U.S.A. CMDL 71.32 -156.60 11 “M”, Ocean Station, Norway CMDL 66.00 2.00 7 Heimaey, Vestmannaeyjar, Iceland CMDL 63.25 -20.15 100 Shetland Island, U.K. CSIRO 60.17 -1.17 30 Cold Bay, Alaska, U.S.A CMDL 55.20 -162.72 25 Mace Head, Galway, Ireland CMDL 53.33 -9.90 25 Shemya Isl. Alaska, U.S.A CMDL 52.72 174.10 40 Vancouver, Estavan Pt, B.C., Canada CSIRO 49.38 -126.54 39 Zugspitze, Germany IMK-IFU 47.42 10.98 2964 Jungfraujoch, Switzerland EMPA 46.55 7.98 3578 Park Falls, Wisconsin, U.S.A CMDL 45.93 -90.27 868 Rishiri Isl, Japan NIES 45.07 141.12 35 Ulaan Uul, Mongolia CMDL 44.45 111.10 914 Sary Taukum, Kazakhstan CMDL 44.45 77.57 412 Kaz_mount, Plateau Assy, CMDL 43.25 77.88 2519 Kazakhstan Niwot Ridge, Colorado, U.S.A CMDL 40.05 -105.58 3475 Wendover, Utah, U.S.A CMDL 39.90 -113.43 1320 Ryori, Japan JMA 39.03 141.83 230 Azores, Terceira Island, Portugal CMDL 38.77 -27.38 40 St. Davids, Bermuda, U.K. CMDL 32.37 -64.65 30
58 CHAPTER IV
Monitoring agencies: CMDL = Climate monitoring and Diagnostics Laboratory, Boulder, Colorado, U.S.A. CSIRO = Commonwealth Science and Industry Research Organization, Canberra, Austaralia JMA = Japan Meteorological Agency, Tokyo, Japan NIES = National Institute of Environmental Studies, Tsukuba, Japan IMK-IFU = IMK-IFU, Forschungszentrum Karlsruhe, Garmisch-Partenkirchen, Germany EMPA = Swiss Federal Laboratories for Materials Testing and Research, St. Gallen, Switzerland The measurements from stations at Rishiri Isl., Ryori, and Zugspitze were carried out continuously. Measurements from weekly samples followed by laboratory analysis were used at other sites.
4.4.4. Results
The anomalies of monthly mean CO total column amounts for five low-altitude stations are plotted in Figure 4.4.1 (a). The averaged data (black curve) shows two peaks in August 1996 and Sep.-Oct. 1998 and a minimum in Sept. 1997. The 1998 anomaly was positive throughout the year, yielding about 2x10 17 molecules/cm 2 of CO during the first half and going up to 7x10 17 molecules/cm 2 of CO in October 1998. This is almost 40% of the normal total column. It is discussed in Yurganov et al., [2004] that the spring shoulder in January-April 1998 is a result of the Indonesian fires in Sep.-Nov. 1997 (El Niño year), where 130 Tg of CO were released into the atmosphere [ Duncan et al., 2003b]. Transport of CO from southern Mexico and central America may also be a possible minor source of the excess CO.
The strong anomaly over Hokkaido in Aug. 1998 is attributed to the influence of Siberian forest fires [ Zhao et al., 2002], with an estimated anomaly of up to 1.2x10 18 molecules/cm 2 of CO. This is an increase of 60% in comparison to the average total column amount during the reference period.
Observations of the CO mixing ratio anomalies from stations at the boundary layer show the same temporal shapes and absolute deviations as that of the total column anomalies (Fig. 4.4.1 b). The CO anomaly measured on Rishiri Island (150 km northwest of Moshiri) was 93 ppb higher than the hemispheric average anomaly in Aug. 1998. On the other hand, the average August anomaly at Ryori station, which is 560 km south of Moshiri (blue filled triangles), was only 28 ppb higher than the average HNH anomaly. It was established in Yurganov et al., [2004] that this difference is caused by the direct effects of CO transport to Rishiri Is. from the Siberian fires.
59 CHAPTER IV
4.4.5. Quantifying the Effects of the 1998 Anomaly in the Free Troposphere
Observations from mountain stations add to the information regarding the CO burden increase in the free troposphere. The column data for two alpine stations Jungfraujoch and Zugspitze (240 km apart), with the corresponding in-situ measurements are shown in figure 4.4.2a. The in-situ measurements tend to be more scattered than the column data. This could be explained by boundary layer processes that could lead to high but abrupt CO variability which are easily registered by in-situ samplers but are undetectable through FTIR column measurements (for example, see Barret et al., [2003]). Despite the scatter, the anomalies detected from the in-situ data agree well with the total column data. The main features such as the high values in 1998 are similar to the other sites in the northern hemisphere. Both data sets indicate an enhancement ranging from 20 ppb to 40 ppb in 1998.
Six other in-situ stations located above 900 m in altitude also show anomalies between 20 ppb and 40 ppb (Fig. 4.4.2b). These stations are considered as free tropospheric sites [ Yurganov et al. , 2004].
Both the alpine column and the in-situ measurements from the mountain sites were used to characterize the HNH free troposphere by taking the average of their anomalies. The anomalies were averaged over all stations comprising the boundary layer (BL), free troposphere (FT), and total columns (TC). The in-situ BL measurements comprise of 16 stations. The FT measurements comprise of 6 mountain in-situ stations and two alpine column stations. The TC measurements comprise of the five low-altitude total column stations. For a summary of the average anomalies in the three atmospheric domains, see Figure 4.4.3.
60 CHAPTER IV
1x10 18 a) Kiruna 18 2 1x10 Ny Alesund Zvenigorod 8x10 17 Hokkaido Harestua 6x10 17 Total mean
4x10 17
17 2x10
0
17 -2x10 Total column anomaly, molecules/cm 17 -4x10 1996 1997 1998 1999 2000 2001 2002
cold bay 140 barr b) Vancouv W iscon 120 Bermuda Alert 100 Azores Iceland """M""" 80 Ny Alesund Kazakhstan Rishiri 60 Ryori Shem ya Shetland 40 Mace Head 1 m ean 20
0 CO mixingratio anomaly, ppb
-20
1996 1997 1998 1999 2000 2001 2002
Figure 4.4.1. a) Total column CO monthly mean anomalies (actual monthly means subtracted by the “normal” values). b) Anomalies of the surface CO mixing ratio. The solid curve corresponds to arithmetic mean values for all sites for each month. adapted from Yurganov et al., [2004].
61 CHAPTER IV
Jung in situ a) 40 Zugsp in situ Jung total Zugsp total 20
0
Anomaly of mixing ratio, ppb ratio, mixing of Anomaly -20
-40 1996 1997 1998 1999 2000 2001 2002 Utah 60 Niwot Ridge b) 50 Jung in situ
Zugsp in situ 40 Kaz_mount Mongolia 30 mean in situ 20
10
0 Anomaly of mixing ratio, ppb ratio, mixing of Anomaly -10
-20
-30 1996 1997 1998 1999 2000 2001 2002
Figure 4.4.2. Monthly mean CO anomalies for high-altitude FTIR stations (with altitudes higher than 900 m asl). a) Jungfraujoch and Zugspitze; in situ and column amount anoma lies are compared. Total column amounts were divided by the total numbers of air molecules for the layer between the altitude of the station and 10 km asl (according to the 1976 US standard atmosphere). b) Monthly mean anomalies of CO mixing ratio, measured in situ at the mountain stations [from Yurganov et al., 2004].
62 CHAPTER IV
80 FT
70 BL
60 TC 50
40
30
20
10
0
-10
CO mixingAnomaly ratio, of ppb -20
1996 1997 1998 1999 2000 2001 2002
Figure 4.4.3. Avera ge anomalies for three atmospheric domains expressed in ppb. FT (free troposphere) includes data for all the mountain stations, both in situ and total column. BL (boundary layer) comprise the low level in situ data. TC (total column) is the average of the data for five low-altitude total column stations (total column amounts of CO were divided by the total numbers of air molecules) Yurganov et al., [2004].
The TC anomalies from the low-altitude stations (Fig. 4.4.1a) were converted into ppb (Fig. 4.4.3) by dividing the TC anomalies by the number of air molecules above the sites (from 0 km to 10 km), i.e.
TC Concentrat ion _[ ppb ] = Air _ column 0( − 5.1 km ) + Air _ column 5.1( −10 km )
(4.4.1) where; TC= total column measurements Air_column(0-1.5km)= 0.355x10 25 molecules/cm 2 Air_column(1.5-10km)= 1.23x10 25 molecules/cm 2 The anomalies for Zugspitze and Jungfraujoch in mol/cm2 were divided by the numbers of air molecules over these stations, setting 10 km as the upper boundary (0.945x10 25 mol/cm 2 and 0.834x10 25 mol/cm 2)
It is interesting to note that during the first half of 1998, the increase of the CO mixing ratio was higher in the FT than in the BL (Fig.4.4.3). However, the amount of CO increase in the
63 CHAPTER IV
BL overtakes that of the FT after the first half of 1998. Long-range transport between the hemispheres occurs mainly through the FT [ e.g. in Notholt et al., 2000] and inter-hemispheric transport takes about 15 months [ Finlayson-Pitts & Pitts, 2000]. This implies that the Indonesian fires in 1997 may have left the manifestations of its effects in the FT of the high northern hemisphere first. Furthermore, a model by Duncan et al., [2003b] shows that the Indonesian fires affected the NH, extending further to Europe where most of the stations are located.
The mass burdens in the three reservoirs were calculated with the following properties taken into account: BL: 0 km – 1.5 km, air column: 0.355x10 25 air molecules/cm 2; FT: 1.5 km – 10 km, 1.23x10 25 air molecules/cm 2; TC: 0 km – 10 km, sum of the FT and BL burdens (values taken using the 1976 U.S. standard atmosphere, [ Minzner, 1977]).
To calculate the tropospheric burden anomaly in Teragrams (Tg) coming from the TC anomaly, the surface area between 30°N and 90°N (1.275x10 18 cm 2) was multiplied by the TC anomalies measured from the five low-altitude stations (Fig. 4.4.1), then converted to ppb. The results are shown in Figure 4.4.4 (green triangles). The BL anomaly averaged over in-situ low altitude stations (in ppb) was multiplied by 0.355x10 25 air molecules (for 0 km – 1.5 km) then multiplied by the surface area to get the CO anomaly in Tg. The FT anomaly in Tg was obtained from the average anomaly in ppb from in-situ mountain stations and FTIR mountain stations (Fig.4.4.3) multiplied by 1.23x10 25 air molecules then multiplied by the surface area. The results are shown in Figure 4.4.4.
The monthly mean anomalies of the tropospheric CO burden in Tg are shown in Figure 4.4.4. The calculations from the BL and FT were combined (red squares) and then compared with the TC (green triangles). These two curves agree well. The annual mean for 1998 estimated from the “BL+FT” curve is 20.6 Tg and the annual mean for the same year estimated from the “TC” curve is 21.4 Tg. Both estimates differ by only 3.4%. This is a very good result considering the fact that both curves were obtained using independent techniques and that the measurements were performed on different locations in the northern hemisphere. The average of the two curves, taken as the simple arithmetic mean (shown by the “ALL DATA” curve, black with error bars) represents the best estimate for the anomaly of the 1998 trpospheric CO burden between 30°N and 90°N.
64 CHAPTER IV
50 Mean for 1998 BL+FT 20.6 Tg (BL+FT) 40 TC 21.4 Tg (TC) ALL DATA 30
20
10
0
anomaly, CO HNH,Tg
-10
-20 1996 1997 1998 1999 2000 2001 2002
Figure 4.4.4. Monthly mean anomalies of the tropospheric CO burden (total mass in Tg from sea level up to 10 km altitude) for the area covering 30º N to 90º N . The BL+FT curve is a combination of data from in-situ CO samplers at low altitudes and in the mountains, as well as the column data from two mountain stations. The TC curve is the average from five low altitude total colu mn stations. The ALL DATA curve is a simple arithmetic mean of these two curves (regardless of the number of stations). The error bars are STD * (N-1) -1/2 , where STD is standard deviation for individual locations, and N is the number of measurement sites. Adapted from Yurganov et al., [2004].
4.4.6. Summary
A significant positive enhancement of the CO during the northern hemisphere summer in 1998 was observed from the total column measurements as well as from the in-situ data. A CO build up was also evident during the summer months of 1996. These anomalies were observed for the entire mid to high latitude belt of the northern hemisphere (30°N to 90°N). The maximum CO enhancement ranging from 40% to 70% took place in Sep. – Oct. 1998 coinciding with the widespread fires that occurred during that year.
In August 1998, the CO anomaly in the high northern hemisphere was estimated to be more than 40 Tg/month from FTIR TC measurements. The annual 1998 emission positive anomaly was approximately 21 Tg/yr. Almost all the excess CO may be attributed to the emissions from boreal forest fires.
65 CHAPTER IV
In northern Japan, the increase of the monthly mean CO abundance from both total column and in-situ measurements significantly exceeded the hemispheric average. This is an indication of the direct effects of the westerly transport of emissions from Siberian fires.
66 CHAPTER IV
4.5. Investigations of Polar Strato-mesospheric Carbon Monoxide Measured by Ground- based Fourier Transform Infrared Spectrometry
4.5.1. Introduction
The high latitude regions have a larger seasonal atmospheric variation than anywhere else on earth [ Notholt et al., 1997]. The absence of sunlight during polar night, the varying meridional circulation and the presence of the polar vortices are just some of the precursors to important atmospheric chemical reactions in the polar regions. An excellent tracer for global transport and air mass descent rates in the polar stratosphere and lower mesosphere is carbon monoxide (CO). Its importance as an indicator of vertical transport in the mesosphere was highlighted in results from ground-based measurements e.g. Künzi and Carlson [1982]. At altitudes of about 50 km, the photochemical lifetime of CO is about 7 days, which is comparable to the vertical and horizontal advection time scales at these altitudes [ Dupuy et al. , 2004, Solomon et al., 1985].
The primary source of Carbon monoxide in the mesosphere and lower thermosphere is the photolysis of carbon dioxide:
CO 2 + h υ ==> CO + O
In the stratosphere, CO is produced through the oxidation of atmospheric methane, but OH rapidly destroys it through oxidation. This reaction acts as the main sink of CO. This process does not take place during the polar night, since OH is produced by reactions involving photolysis, e.g. of H 2O, and since the concentration of OH diminishes rapidly during polar night. Carbon monoxide follows the meridional circulation towards the winter hemisphere polar night region. The consecutive downward motion induces a sharp gradient in the CO concentrations down to the stratosphere [ Solomon, et al., 1985]. Meanwhile, uplifting of air masses with low CO content takes place in the summer. Solomon et al., [1985] predicted that very large abundances should accumulate in the polar night mesosphere because of the absence of photochemical destruction processes.
Previous ground-based measurements of mesospheric CO were shown in pioneering works such as that of Künzi and Carlson [1982], Clancy, et al ., [1982] Farmer, et al., [1980], Zander et al. , [1981], Goldsmith et al ., [1979], etc. However, there is a lack of long term ground- based observations of strato-mesospheric CO in the literature. Recently, satellites have been able to measure high-altitude CO [eg. Lopez-Valverde et al., 2003, Dupuy et al. , 2004, Clerbaux et al., 2005]. But until now, the longest reported time series of CO in the upper atmosphere was done by Forkmann et al ., [2003] over the Onsala Space laboratory, Sweden (57.4°N, 12°E). The time series spanned from Sept. 2000 to Sept. 2002.
Kasai et al., [2005] published the first reported ground based measurements of strato- mesospheric CO by FTIR spectrometry over Poker Flat, Alaska (65°N, 147°W). They have established the capability of the FTIR technique to detect the seasonal variability of strato- mesospheric CO.
67 CHAPTER IV
This chapter reports more than seven years of time series of strato-mesospheric CO measured from stations located in the high latitude regions of the Arctic (79°N) and the Antarctic (78°S). We augment this data set from two high-latitude stations and two mid-latitude stations. We also show comparisons with a global two-dimensional photolysis, chemistry and transport model in the stratosphere and mesosphere.
4.5.2. Instruments
Measurements were taken from three Arctic stations (Ny Ǻlesund 79° N, Kiruna 68° N and Poker Flat 65° N), one Antarctic station (Arrival Heights 78° S) and two mid-latitude stations (Bremen, Germany 53° N and Lauder, New Zealand 45° S). All stations are equipped with Bruker 120HR spectrometers except for Arrival Heights (120M) and Bremen (125HR). For the polar stations, measurements are limited by the polar night. Solar spectra in Ny-Ǻlesund can be recorded between March and early October. In Kiruna, the polar night is between early December and mid January. At the Arrival Heights station in Antarctica, the polar night is between late April and mid August. Measurements in Poker Flat are possible between early February and mid October.
4.5.3. Retrieval of Strato-Mesospheric CO
The program used for the retrievals of CO profiles for the 5 sites: Ny Ǻlesund, Poker Flat, Bremen, Lauder and Arrival Heights is SFIT-2 (Spectral Least Squares Fitting Program) version 3.8. The pressure and temperature profiles necessary for the forward model were obtained from balloon sondes that were launched daily from the stations. Above the altitude limits of the sondes (approximately 30 km), standard pressure and temperature profiles were taken. In Kiruna, pressure and temperature profiles were taken form NCEP [ Kanamitsu, 1989].
68 CHAPTER IV
Table 4.5.1. Summary of the site characteristics.
Station, Location Instrument Typical Windows used Interfering gases Resolution (cm -1) Ny Ǻlesund, Bruker, 120HR 0.005 2057.70-2057.91 H2O, N 2O, O 3, Spitsbergen, 79°N, 12°E 2069.55-2069.72 OCS Solar CO 2157.40-2159.20 Kiruna, Sweden, 68°N, Bruker 120HR 0.005 2057.5 - 2058.2 H2O, N 2O, O 3, 20°E 2069.4 - 2069.9 OCS Solar CO 2140.4 - 2141.4 2153.2 - 2160.0 Poker Flat, Alaska, Bruker 120HR 0.0019 2057.68-2058.00 H2O, N 2O, O 3, 65°N, 148°W 2069.56-2069.76 OCS Solar CO 2157.40-2159.2 Bremen, Germany 53°N, Bruker 120HR, 0.005 2057.70-2057.91 H2O, N 2O, O 3, 9°E Bruker 125HR 2069.55-2069.72 OCS Solar CO 2157.40-2159.20 Lauder, New Zealand, Bruker 120HR 0.005 2057.68-2058.0 H2O, N 2O, O 3, 45°S 170 °E 2069.55-2069.76 Solar CO 2157.40-2159.20 Arrival Heights, Bruker 120M 0.005 2057.68-2058.00 H2O, N 2O, O 3, Antarctica 78°S, 167 °E 2069.55-2069.76 Solar CO 2157.40-2159.20
CO spectra were analyzed in the CO micro-windows based on the previous work of Rinsland et al., [1998]. The Poker Flat data were retrieved using an additional window described in Kasai et al., [2005]. A summary of the micro-windows used and the interfering gases that are taken into account are given in Table 1. For all stations, the HITRAN2k line list [ Rothman et al., 2003] was used plus its recent updates to 2001. The program used to retrieve the Kiruna data is PROFFIT 9 developed by F. Hase [2000]. A detailed description and comparison of both retrieval are shown in Hase et al., [2004]. Typical averaging kernels for the retrievals in Ny Ǻlesund and Arrival heights are shown in Figure 4.5.1. The figure shows that the 2 partial columns from 0.2- 18 km and from 18 km to 85 km can be separated. For some cases, the kernel for the 18 km to 85 km is not perfect, i.e. it does not have a maximum of 1. Furthermore, the stratospheric columns and the mesospheric columns are not separable, thus measurements are provided as strato-mesospheric columns.
4.5.4. The Chemical Transport Model
The model used is a global two-dimensional photolysis, chemistry and transport model of the stratosphere and mesosphere. It is a coupled chemistry-dynamics model which combines the THIN AIR meteorological code [ Kinnersley, 1996] and the SLIMCAT chemistry code [Chipperfield, 1999]. Temperature, pressure and wind fields are calculated by the THIN AIR
69 CHAPTER IV code on isentropic surfaces from the bottom up to ~ 100 km with a vertical spacing of about 3 km.
The model has a horizontal resolution of about 9.5° extending from -85.3°S to 85.3°N in 19 evenly spaced latitude bins. The chemistry module uses with JPL-2003 photochemistry data [Sander et al., 2003] . Though the SLIMCAT model is not appropriate for the troposphere, it is applied to the entire vertical range of the model. Heating rates are calculated in the THIN AIR module, using ozone, NO 2 and CH 4 values provided by the chemistry module. CO 2 is also used to calculate heating rates, however, as this is very long-lived in the stratosphere and mesosphere, it is not accounted for in the chemisty code. In the past, the model has been used for a number of studies concerning the composition of the middle atmosphere [ Sinnhuber et al, 2003; Rohen et al, 2005; Chipperfield and Feng , 2003]. Inorganic chlorine, bromine and fluorine compounds as well as greenhouse gases like CH 4, N 2O and CO 2 are read-in daily into the lowermost model box; the values are derived from monthly mean values of the WMO A1B scenario [ WMO 2002 , 2003].
Model runs are started in 1988, and run to 2005. For this study, two model runs were carried out. The 'base' scenario uses the original SLIMCAT chemistry which does not contain CO 2. This means that CO is produced solely from CH 4 oxidation. In a second model run, called 'thermospheric', CO in the uppermost model box is fixed to the CO 2 value. This means that CO 2 is transported unchanged into the thermosphere, where it is transferred into CO immediately. CO will then be transported down into the mesosphere and stratosphere during polar winter; after polar sunrise, CO will react with OH, re-forming CO2 in a couple of days.
70 CHAPTER IV
Figure 4.5.1 . Th e typical averaging kernels for the retrieval of CO from 18 km to the top most layer of the retrieval are shown by the green curve. The blue curve is the averaging kernel for the retrieval of the CO column from the ground to 18 km. (Left: Ny Alesund; Cente r: Arrival Heights, Right: Poker Flat)
2 molecules/cm 16 x10
Figure. 4.5.2. Comparison of strato-mesospheric CO columns between the model and FTIR measurements from all stations. Cyan line: base model run without mesospheric CO. Green line: Mode l run with mesospheric CO. Blue dots: Retrieved columns from the FTIR. Magenta line (dashed): model run smoothed with the averaging kernels of the FTIR according to Rodgers and Connor, [2003]. The shaded area represents polar night.
71 CHAPTER IV
4.5.5 Results
4.5.5.1. FTIR data and model calculations
A comparison of the FTIR data and the model results for 2003-2004 are shown in Figure 4.5.2. The partial column densities measured by the FTIRs in molecules/cm 2 from 18 km to the top of the atmosphere are shown by the blue dots. The cyan curve represents the base run, where thermospheric CO is neglected. The green curve represents the complete run with the thermospheric CO. The model run smoothed by the typical averaging kernels of the FTIR is represented by magenta curve. The smoothing was done according to the formalism described in Rodgers and Connor [2003] and shown in Velazco et al., [2005] for FTIR, model and satellite data comparison of CO profiles. The smoothed curves represent what the FTIR should “see” if the model were to represent the true CO.
Long term FTIR measurements from the three polar stations compared with the model are shown in Figure 4.5.3. The gray shaded areas represent the polar night where solar absorption measurements are not possible. Unlike the seasonal cycles of CO in the troposphere, the seasonal cycles in the strato-mesosphere show very steep gradients, with maximum values occurring in January in the Arctic and in June-July in the Antarctic. As shown by the model (and partially by the measurements), the CO column above 18 km increases from about 4.0x10 16 molecules/cm 2 in summer to about 14x10 16 molecules/cm 2 in winter (an increase of about 3.5 times). This rapid increase is followed by a rapid decrease as soon as the sun re-appears in spring.
A small enhancement of CO columns can be observed in late summer from the measurements at the high-latitude stations. This could be seen as a small “bulge” from the average curves from Ny Ǻlesund and Poker Flat in Figure 4.5.4. The Kiruna and Arrival Heights data exhibit this “summer bulge” for some years (eg. 1998, 2002 & 2004 in Arrival Heights) but it tends to be averaged out due to the changing patterns in the data. Note also that Kiruna is often at the edge of the polar vortex. This summer enhancement is also seen in the model, it is much clearer in the “base” model run. In the thermospheric model run, it is superimposed by the much sharper signal of mesospheric CO. This summer enhancement is produced by methane oxidation in the stratosphere, which occurs faster in summer. CO is an intermediate product of methane, it is then slowly transformed into CO 2.
4.5.5.2. Quantifying the CO above 18 km
To quantify the columns detected by the FTIR in Ny Ǻlesund, we calculated the partial columns from different layers using the model (Fig. 4.5.5). The relative contributions are shown. Generally, about 20%-80% of the column above 18 km comes from the stratosphere (18-26 km). In winter, a significant portion of the column comes from above 56 km, i.e., the mesosphere and lower thermosphere. This is also the time in the Arctic when the downward vertical transport in the mesosphere due to the mean meridional circulation is strongest [ Murgatroyd and Singleton, 1961, Solomon et al. , 1985]. The summer maximum can be seen to originate in the stratosphere, where CH 4 oxidation plays a significant role in the production of CO. It cannot be seen above 56 km. It is most pronounced in 18 - 26 km. This means that the “summer bulge” detected by the FTIR really originates from the stratosphere.
72 CHAPTER IV
The Lauder data (45° S) do not show the very high values of strato-mesospheric CO (Fig. 4.5.6). Although there is a strong variability in the columns below 18 km [ Jones et al., 2001], the strato-mesospheric columns show almost no variability. This shows that the columns below 18 km do not influence the columns above 18 km and that the retrieval could clearly separate both columns. The only exceptions to the monotonous strato-mesospheric CO trend in the mid- latitudes are the values measured in the winters of 2002-03 and 2004-05 over Bremen (53°N). The high strato-mesospheric CO values are also correlated with high HF columns, confirming the intrusion of Arctic vortex air or high-altitude air down into the latitude of Bremen (Fig. 4.5.6).
2
molecules/cm
Years
Figure 4.5.3. FTIR measurements from different stations (blue dots) compared to the model (dashed magenta curves).
73 CHAPTER IV
Ny Alesund 9.E+16 Alaska Arrival Hts. + 6 mos.
Kiruna
7.E+16
2
5.E+16
molecules/cm 3.E+16
1.E+16 2003-10 2004-01 2004-04 2004-08 2004-11 2005-02 Years
Figure 4.5.4. Average Curves calculated from the time series of the 4 Polar stations: Ny - Alesund, Poker Flat, Kiruna and Arrival Heights. The summer bulge in July-August can be clearly distinguished from the Poker Flat and Ny Alesund data. The Arrival heights data (shifted by six months) only shows small “lumps” in June and Augus t. The variability in the data averages out the summer bulge. (The average curves are interpolated every 15 days)
Figure 4.5.5. Relative contribution of each layer to the column above 18km calculated from the model These curves indicate where the signal should come from. In winter, most of the CO part ial column above 18 km is dominated by the CO coming from above 56 km. And from the 26-36 km layer. This is also the time when the downward transport from the mesosphere is strongest. In summer, the column above 18 km is dominated by CO at 18-26 km where CH oxidation plays a significant role in the production of CO. 4
74 CHAPTER IV
2
molecules/cm
Years
Figure 4.5.6. CO columns above and below 18 km from the two mid-latitude stations: Bremen 53° N (red squares) and Lauder 45° S (blue) circles.
75 CHAPTER IV
4.5.6. Summary
The downward transport of strato-mesospheric CO above 18 km in the winter polar regions, which is strongly influenced by the meridional circulation can be seen in the data. The strong gradient showing a maximum in winter and minimum in summer are well captured by the measurements and verified by the model. It was shown that this feature is generally not observed in mid-latitude stations, with the exception of values during two anomalous winters in Bremen.
CO in the mesosphere is influenced by the competition between downward transport from the thermosphere and OH oxidation. The measurements show that the pattern of the strato- mesospheric CO columns for all years in one station are almost similar because CO has a shorter lifetime compared to the downward vertical transport. On the other hand, there is a difference in column amounts of the strato-mesospheric CO between the two poles, i.e. in spring, there is relatively more CO above 18 km in Ny Ǻlesund than in Arrival Heights based on measurements. We assign this to the presence of more subsidence in the Arctic compared to the Antarctic.
Comparisons with a global two-dimensional photolysis, chemistry and transport model in the stratosphere and mesosphere were shown. The assumption in the model that all the CO 2 in the thermosphere gets converted into CO via photolysis is very reasonable. There is a good agreement between the model results and the measurements. Furthermore, the model calculations gives us information on what the FTIR measures. According to the model, the production of CO from CH 4 oxidation in the mid to upper stratosphere results to a signal indicated by the “summer bulge” in the FTIR data seen in Ny Ǻlesund, Poker Flat and occasionally in Kiruna and Arrival Heights. The comparisons also confirm that the downward transport of mesospheric CO produced in the thermosphere could be detected through FTIR spectrometry.
This study concludes that long term observations of the seasonal variations in the strato- mesospheric CO through FTIR spectrometry gives us more information to further characterize the nature and the intensity of lower mesospheric and stratospheric vertical transport processes.
76 CHAPTER IV
4.6. Measurements of CO and HCN in Paramaribo, Suriname
4.6.1. Introduction
The tropical atmosphere comprises about half the volume of the global atmosphere. It is a very significant region, which determines the chemical composition of the global atmosphere and climate of the earth. The tropical regions also provide a stage where transport of water vapor and trace gases from the troposphere to the stratosphere take place; a process that is not yet well understood. Therefore, rapid environmental changes in the tropics as a consequence of deforestation, urbanization and industrialization may affect the atmosphere on a global scale.
A measurement campaign in Suriname took place between the months of Sept. 2004 and March 2005. The campaign under the STAR (Support for Tropical Atmospheric Research) project, was motivated by the increasing scientific significance of tropical atmospheric research. The measurement site was located in Paramaribo (5.8°N 55.2°W), the capital of Suriname at the northern coast of South America (Fig. 4.6.1). The site is located in a relatively clean environment due to its close proximity to the ocean and to the Amazon forest, which extends up to the southern borders of Suriname.
As air is transported by the easterly trade winds over this location, the relatively flat topography on this part of the South American continent leaves the background atmosphere relatively unperturbed. This is useful for tracing the atmospheric composition, measured at the site, back to its regions of origin using backward trajectory calculations.
Figure 4.6.1. The measurement site in Paramaribo, Surin ame is located close to the Amazon rainforest with French Guiana to the East, Brazil to the south and the Atlantic Ocean to the north.
77 CHAPTER IV
4.6.2 The Inter-tropical Convergence Zone (ITCZ)
Tropospheric air masses between the northern and southern hemispheres mix at a relatively slow rate. For example, a molecule released in one of the hemispheres typically takes about 15 months to reach the other hemisphere, with mixing primarily occurring in the mid to upper troposphere [ Finlayson-Pitts & Pitts, 2000].
Near the equator, from about 5° north and 5° south, the northeast trade winds and southeast trade winds converge in a low-pressure zone known as the Intertropical Convergence Zone or ITCZ *. The ITCZ acts like a barrier that prevents the exchange of air masses between the northern and southern hemispheres. In this chapter, we will sometimes refer to the northern hemisphere atmosphere as the part of the atmosphere north of the ITCZ and the southern hemisphere atmosphere as the atmosphere south of the ITCZ.
The ITCZ varies with season, typically lying north of the equator in July and south of the equator in January. During this seasonal displacement of the ITCZ, exchange of air between the hemispheres may take place. Inter-hemispheric exchange also occurs due to turbulent mixing in the upper troposphere near the equator [ Finlayson-Pitts & Pitts , 2000].
The Inter-tropical convergence zone migrates twice a year over Suriname, which is very interesting from a scientific point of view. The ITCZ here is well discernable by looking at rainfall statistics over Paramaribo. The records show a distinct increase in rainfall during the months when the ITCZ lies overhead (December-January, April-July), with dry seasons in between [STAR website: www.knmi.nl/samenw/star]. Hence, this station can provide data that are representative of atmospheric features associated with the Northern as well as the Southern Hemisphere, and in addition, the features which are unique to the ITCZ.
4.6.3. Results
4.6.3.1. Peak in the signal caused by emissions from fires
Total column measurements of HCN and CO are shown in Figure 4.6.3 and 4.6.4 respectively (plots on top). Both total column measurements show a distinct peak measured on October 18 and 20, 2004. This peak can also be observed in the MOPITT data (Fig. 4.6.4). The majority of the trajectories arriving in Suriname at the 850 hPa and 500 hPa levels originate from the Atlantic ocean (see examples in Figure 4.6.5). However, the trajectories on October 18 and 20 are the ones that have traveled a longer path along the South American continent, passing over areas with a large amount of fire counts. This suggests that biomass burning from neighboring countries such as French Guiana, Brazil and Suriname itself are the source of the pollution-rich air parcels (see Figure 4.6.6).
4.6.3.2. Perturbations detected in the mid to upper troposphere
* Due to the lack of horizontal air movement, the Intertropical Convergence Zone has also been called the doldrums by sailors. In this zone, the air simply rises with convection. The ITCZ is also known as the Equatorial Convergence Zone or Intertropical Front
78 CHAPTER IV
The volume mixing ratios of CO and HCN are also plotted in Fig. 4.6.2 and Fig. 4.6.3 (bottom plots). The VMR profiles were divided and averaged in three layers according to the sensitivity of the retrievals. The plots of the VMR profiles show that in both CO and HCN total column measurements, the peak in the signal that was observed on October 18 and 20, 2004 mostly came from the atmospheric layer above 4 km (mid troposphere to upper troposphere). There were only small perturbations of the concentrations of CO and HCN in the layer below 4 km.
4.6.3.3. Backward trajectories confirm ITCZ migration
It can be seen from the trajectories shown in figure 4.6.5 (a, b, c) that most of the air parcels at 850 hPa and 500 hPa originated from the northern hemisphere. The patterns of the trajectories indicate the westward blowing trade winds. This confirms that the atmosphere above Paramaribo was part of the sourthern hemisphere in Sept.- Nov. 2004.
The trajectories take on a different pattern as the months of February and March came in. This time, the air parcels at the 850 hPa and 500 hPa levels came from the northeast, confirming that the atmosphere above Paramaribo belonged to the Northern Hemisphere in Feb.-Mar. 2005.
4.6.5. Effects of the ITCZ migration on the total columns of HCN and CO
The shift in the ITCZ did not influence the measured total columns of CO very much. The trend seems to be constant as shown in Figure 4.6.2. The same observation can be seen from the CO data measured by the MOPITT satellite using from could-free measurements within a 500-km radius from Paramaribo (Figure 4.6.4). Note that MOPITT also captured the high CO peaks on Oct. 18 and 20, 2004. In contrast to the almost constant CO trend, a slight but noticeable decrease can be seen in the total column measurements of HCN as the ITCZ moves south of Paramaribo. This clearly indicates a difference in the HCN concentration in both hemispheres. This may be due to the fact that HCN is known to come mainly from biomass burning while CO has different sources (biomass burning, fossil fuel combustion and oxidation of methane and non-methane hydrocarbons). These CO sources are also abundant in the northern hemisphere while the main source of HCN, which is biomass burning, may not be as large in the northern hemisphere as in the southern hemisphere during the time of the measurement.
79 CHAPTER IV
18 CO Total Columns (0.5 km -85 km) Daily Averages x 10 3
2.8
2.6 2.4
2 2.2 2
1.8
1.6
molecules/cm 1.4
1.2
1
Sep04 Oct04 Nov04 Dec04 Jan05 Feb05 Mar05 Apr05 May05
-7 CO Volume Mixing Ratio x 10 2 12 km - 36 km 1.5
1
0.5
-7 x 10 2 4 km - 12 km 1.5
1
0.5 -7 x 10 2 0 km - 4 km Volume Mixing Volume Ratio 1.5
1
0.5 Sep04 Oct04 Nov04 Dec04 Jan05 Feb05 Mar05 Apr05 May05 Years Figure 4.6.2. Measurements of Total Columns of CO (top) and VMR profiles (below) of CO above Paramaribo. The peak in the total columns on Oct. 18 & 20 is also seen in the VMR measurements in the 4 km – 12 km layer.
80 CHAPTER IV
16 HCN Total Columns (0.5 km -85 km) Daily Averages x 10 2
1.8
1.6
1.4
2 1.2 1
0.8
0.6 molecules/cm 0.4
0.2
0
Sep04 Oct04 Nov04 Dec04 Jan05 Feb05 Mar05 Apr05 May05
-10 HCN Volume Mixing Ratio x 10 10 16 km - 36 km 8 6 4 2 -10 x 10
10 4 km - 16 km 8 6 4
2
-10 x 10 10 0 km - 4 km 8 Volume Volume Ratio Mixing 6 4 2 Sep04 Oct04 Nov04 Dec04 Jan05 Feb05 Mar05 Apr05 May05 Years Figure 4.6.3. Measurements of Total Columns of HCN (top) and VMR profiles of HCN (below) over Paramaribo. The peak in the total columns on Oct. 18 & 20 is als o seen in the VMR measurements in the 4 km – 12 km layer. A slight decrease in the HCN total columns was observed when the ITCZ moved south of Paramaribo.
81 CHAPTER IV
2
molecules/cm
Date
Figure 4.6.4. Measurements of the total columns of CO from MOPITT (blue dots) and from the FTIR (red circles) agree well. The MOPITT data shown here were taken using cloud-free measurements within a 500-km radius from the FTIR. MOPITT Data courtesy of Holger Bremer Institute of Environmental Physics, University of Bremen.
82 CHAPTER IV
Figure 4.6.5. Ten-day backward trajectories from the Web Trajectory service of the British Atmospheric Data Center show that the ITCZ was north of Paramaribo in Oct.-Nov. (A-C) and south of Paramaribo in Jan.-Feb. (D-F). Figures adapted from Warneke et al., [2005]. Fire counts and maps from MODIS
83 CHAPTER IV
Figure 4.6.6. The trajectories on October 18 and 20, 2004 have traveled a path along areas where high amounts of biomass burning took place, suggesting that bi omass burning from the nearby countries such as French Guiana, Brazil and Suriname itself are the source of the CO and HCN-rich air parcels. Figures adapted from Warneke et al., [2005].
84 CHAPTER IV 4.6.6. Summary
Measurements and analysis of the trace gases CO and HCN were successfully carried out in the tropics during the migration of the ITCZ and within the ITCZ. A distinct peak in the total column concentrations of CO and HCN, which occurred concurrently on Oct. 18 & 20, 2004 was measured. This peak is clearly caused by emissions from fires in the neighboring areas as confirmed by backward trajectory analysis and fire counts from satellite data. Further analysis of this peak shows that the enhancement mostly occurred in the mid to upper troposphere above 4 km. The concentrations of CO and HCN in the layer below 4 km were almost undisturbed during this period.
The migration of the ITCZ had little or no effects on the trend of the total columns of CO. This could be accounted for by the different sources of CO (biomass burning, fossil fuel combustion, and oxidation of both methane and non-methane hydrocarbons), which are present in both hemispheres. However, HCN total columns showed a slight decrease after the migration of the ITCZ, showing that the HCN source in the northern hemisphere is not as abundant as in the southern hemisphere. It has been established that the source of HCN in the atmosphere is mostly biomass burning [ Li et al, 2000], however several sources that are still unaccounted for may contribute to the uncertainty of the atmospheric HCN budget.
85 CHAPTER V 5. Summary and Conclusions
This work presented analyses of spectra from the high Arctic station in Ny Ǻlesund, Spitsbergen (1992-2005), from five cruises of the RV Polarstern , from the recent campaign in Paramaribo, Suriname and the latest spectra from Bremen, Germany, a newly established NDSC complementary site. A few improvements on the retrieval methods have been established, thereby achieving more altitude information on the retrieved vertical profiles of some trace gases.
One of the Polarstern cruises coincidentally took place during the occurrence of interesting atmospheric disturbances wherein we directly measured biomass burning plumes, enhanced trace gas concentrations and aerosol optical depths and high atmospheric methane values. The synergistic combination of this cruise with data from four other cruises, fire count data, trajectory analyses, MOPITT data and a model for atmospheric transport and chemistry led to a better understanding of the origins and transport processes of CO over the Atlantic ocean. The data set from the three different sources were also formally validated with very good results. This kind of validation has been shown for the first time.
Biomass burning is a very significant source of CO. A quantification of the CO anomaly during abnormal biomass burning events in 1998 was shown through a collaborative work with other stations worldwide. It has been found that in Sept.-Oct. 1998 alone, abnormal biomass burning accounted to an excess of 40%-70% in CO amounts within the latitude belt of 30° N-90° N. This CO contribution from biomass burning is twice as large as that from fossil fuel combustion.
In this work, the capability of the FTIR technique to measure strato-mesospheric CO columns consisting of the stratosphere up to the upper mesosphere (18 km – 100 km) was established. The strong gradient in the seasonal cycles of the strato-mesospheric CO from both poles was well captured by the measurements and verified by a chemical transport model. The presence of more subsidence in the Arctic compared to the Antarctic shows a slight influence on the CO column amounts above 18 km. The polar data were augmented with data from 4 other stations. With this study, the columns of the strato-mesospheric CO were also quantified. It was demonstrated that 20%-80% of the measured columns in summer originate from the stratosphere but a significant portion of the column comes from above 56 km in winter. Another interesting feature was observed; a small enhancement in strato-mesospheric CO columns during summer in the polar regions called the “summer bulge” in this study. This comes from CH 4 oxidation in the stratosphere. As an outlook, the summer bulge and its characteristics will be further studied in the future.
Because of the altitude information gained in the CO retrievals during the course of this study, the Ny Ǻlesund data set was turned into the longest time series of ground-based measurements of strato-mesospheric CO, giving new insights on the nature of the transport and variability high-altitude CO and paving the way for more possibilities to study the middle atmosphere with ground-based FTIR.
Part of the objectives of the on-going campaign in the tropical station in Paramaribo, Suriname was to study the tropical atmosphere and the inter-tropical convergence zone (ITCZ). Due to the ITCZ, which acts like a barrier between the atmospheric hemispheres, data that are
86 CHAPTER V representative of the features of the two hemispheres were gathered and analyzed. The migration of the ITCZ had little or no effects on the trend of the CO total columns. This is due to the different sources of CO that are present and abundant in both hemispheres. However, HCN columns showed a slight decrease after the migration of the ITCZ towards the south of the measurement site. This demonstrates that the HCN source in the northern hemisphere is not as abundant as in the southern hemisphere. Up to now, several HCN sources remain unaccounted for, contributing to the uncertainty in the atmospheric HCN budget. It can be concluded that the difference in both hemispheres is mostly due to the strong biomass burning source of HCN and it can only be speculated that there are other local sources that are still unaccounted for. This will be a topic of future investigations.
87 Acknowledgements Acknowledgements
This work will not be completed without the support given by the following people.
I would like to thank Prof. Dr. Justus Notholt for supervising my studies, for his patience in being a mentor. His optimism, enthusiasm in science and brilliant ideas made my studies both fun and fulfilling. Thanks for showing the real meaning of the word “Doktorvater”. I am grateful to Thorsten Warneke for his support and advice that helped a lot during my studies. He also acted as an adviser for my studies and for my thesis. Thanks also to Christine Weinzierl for all the support in technical aspects, for the support as a colleague and especially for her patience when I was seasick during the Polarstern cruise.
I am especially indebted to Prof. Jörn Bleck-Neuhaus and Dr. Jürgen Sültenfuß for founding and organizing the Environmental Physics program and for their support through its completion. I also acknowledge all my Professors in this course, especially Prof. Dr. Reiner Schlitzer Prof. Dr. Klaus Künzi and Prof. Dr. Otto Schrems. Many thanks to Stefanie Bühler, Anja Gatzka and Barbara Kozak for their assistance. Thanks to Birgit Teuchert and Sabine Packeiser-Pohl for their assistance in official and administrative matters.
I thank all my colleagues and members of the Institute of Environmental Physics, particularly Prof. John Burrows for his motivating and constructive comments, Mathias Palm for all the helpful discussions, for his ideas and for his patience in sharing his mathematical talents. Thanks to Björn Martin Sinnhuber for all his helpful advice. Thanks to Miriam Sinnhuber and Holger Bremer for their suggestions and for the collaborative work with CO.
Many thanks to Mashrab for sharing his technical talents. Thanks to Max, Kwon Ho, Manami, Yuren, Katha, Koshi, Tim, Waldemar, Janina, Joel, Nina, Cécile, Johannes, Wang Yi and Jihyun. I am very grateful to the family of Dr. Hinrich and Feliza Elmenhorst for showing me true hospitality during my stay in Bremen and for showing me the beauty of German and European culture, you have been like a family to me.
I wish to express my sincere thanks to my parents for their teachings, encouragements and support in many ways , I couldn’t ask for anything more and I can’t ever thank you enough. I am grateful to my grandparents; Nanay for introducing me to science and Tatay who influenced my appreciation for nature. Thanks to my brothers Jan Vincent and Malcolm.
I would like to acknowledge the financial support provided by the European Union for the UFTIR project (EVK2-CT-2002-00159). I am also grateful to the German Ministry for Research and Education (BMBF), which funded the Polarstern work via the DLR-Bonn (Grants 50EE0013 and 50EE0014) and to the Helmholtz Association within the virtual institute PEP (VH-VI-100) for financial support. I’m also grateful to the city of Bremen, Germany. Thanks also to the involved crew in the Polarstern cruises
Special thanks to the Taranaki Alpine Cliff Rescue Team (Rob, Arain, Alec and others), all the staff of ward three at Taranaki Base Hospital in New Zealand (Carolyn, Rose, Kathy, Robyn and others) and to Buzz, I will not be here if it weren’t for you.
88 CHAPTER VII 6. References
Banwell, C. and E. McCash (1994), Fundamentals of molecular spectroscopy, McGraw-Hill Internationsl Ltd., U.K.
Barret, B., M. De Maziere, E. Mahieu, Ground based FTIR measurements of CO from the Jungfraujoch: characterization and comparison with in situ surface and MOPITT data, Atmos. Chem. and Phys., 3, 2217-2223 , 2003.
Bergamaschi, P., R. Hein, M. Heimann, P. Crutzen (2000), Inverse modeling of the global CO cycle 1. Inversion of CO mixing ratios, J. Geophys. Res., 105, pp. 1909-1927,
Bremer, H., J. Kar, J. Drummond, F. Nichitu, J. Zou, J. Liu, G.C. Gille, M. N. Deeter, G. Francis, D. Ziskin, J. Warner (2004), Spatial and temporal variation of MOPITT CO in Africa and South America: A comparison with SHADOZ ozone and MODIS aerosol, J. Geophys. Res. , 109, D12304.
Chipperfield, M. P. (1999), Multiannual simulations with a three-dimensional chemical transport model, J. Geophys. Res. , 104, 1781-1805.
Chipperfield, M.P., and W. Feng (2003), Comment on: Stratospheric Ozone Depletion at northern mid-latitudes in the 21st century: The importance of future concentrations of greenhouse gases nitrous oxide and methane, Geophys. Res. Letters , Vol 30, 10.1029/2002GL016353.
Clancy, R. T., D. O. Muhleman and M. Allen (1984), Seasonal variability of CO in the terrestrial mesosophere, Journal of Geophys. Res., vol. 89, No. D6, pages 9673-9676.
Clerbaux, C., P. F. Coheur, D. Hurtmans, B. Barret, M. Carleer, R. Colin, K. Semeniuk, J. C. McConnell, C. Boone and P. Bernath (2005), Carbon Monoxide distribution from ACE-FTS solar occultation measurements, Geophys. Res. Letters, Vol. 32, doi: 10.1029/2005GL022394.
Crutzen, P. and P.H. Zimmermann (1991), The changing photochemistry in the troposphere, Tellus , 43, 136-151.
Davis, S. P., M. C. Abrams, J. W. Brault (2003) , Fourier Transform Spectrometry , Academic Press, U.S.A.
De Maziere, M. and the UFTIR Team (2005a), Time series of Upper Free Troposphere observations from a European ground-based FTIR network (UFTIR) 2 nd Management Report.
De Maziere, M., C. Vigoroux, T. Gardiner, M. Coleman, P. Woods, K. Ellingsen, M. Gauss, I. Isaksen, T. Blumenstock, F. Hase, I. Kramer, C. Camy-Peyret, P. Chelin, E. Mahieu, P. Demoulin, P. Duchatelet, J. Mellqvist, A. Strandberg, V. Velazco, J. Notholt, R. Sussmann, W. Stremme, A. Rockmann (2005b) , The exploitation of ground-based Fourier transform infrared observations for the evaluation of tropospheric trends of greenhouse gases over Europe, Env. Sci, 2(2-3): 283-293.
89 CHAPTER VII Deeter, M. N., et al. (2003), Operational Carbon Monoxide retrieval algorithm and selected results for the MOPITT instrument, J. Geophys. Res. , 108 (D14), 4399, doi: 10.1029/2002JD003186.
Drummond, J. R. (1992), Measurements of Pollution in the Troposphere (MOPITT), in The use of EOS for Studies of Atmospheric Physics . edited by J. C. Gille and G. Visconti, pp. 77-101, North-Holland, New York.
Duncan, B. N., I. Bey, M. Chin, L. J. Mickley, T. D. Fairlie, R. V. Martin and H. Matsueda (2003b), Indonesian wildfires of 1997; impact on tropospheric chemistry, J. Geophys. Res., 108 (D15), 4458, doi: 10.1029/2002JD003195.
Dupuy, E., et al. (2004), Strato.mesospheric measurements of carbon monoxide with the Odin sub-millimeter radiometer: retrieval and first results, Geophys. Res. Letters, Vol. 31, doi: 10.1029/2004GL020558.
Farmer C. B., O. F. Raper, B. D. Robbins, R. A. Toth and C. Muller, 1980: Simultaneous spectroscopic measurements of stratospheric species: O 3 , CH 4, CO, CO 2, N 2O, H 2O, HCl and HF at northern and southern mid-latitudes. J. Geophys. Res., 86, 5179-5184, 1980.
Forkman, P., P. Eriksson and A. Winnberg, Longest ground-based measurements of mesospheric CO, Geophys. Res. Letters , 30, No. 10, 1532, doi: 10.1029/2003GL016931, 2003.
Finlayson-Pitts, B., & J. Pitts, Chemistry of the upper and lower atmosphere, pp. 660, Academic Press, USA, 2000.
Fire Data from: http://dup.esrin.esa.it/ionia/wfa/index.asp
Garstang, M., P. D. Tyson, R. Swap, M. Edwards, P. Kallberg and J. A. Lindesay (1996), Horizontal and vertical transport of air over southern Africa. J. Geophys. Res., 101, 23,721- 23,736.
Goldsmith, P. F., M. M. Litvak, R. L. Plambeck and D. R. Williams, 1979, Carbon monoxide mixing ratio in the mesosphere derived from gound based microwave measurements. J. Geophys. Res., 84, 416-418.
Hase F., Inversion von Spurengasprofilen aus hochaufgeloesten bodengebundenen FTIR- Messungen in absorption (2002), Wissenschaftliche Berichte Forschungszentrum Karlsruhe , FZKA 6512; ISSN 0947-8620.
Hase, F., J. W. Hannigan, M. T. Coffey, A. Goldman, M. Höpfner, N. B. Jones, C. P. Rinsland and S. W. Wood (2004), Intercomparison of retrieval codes used for the analysis of high- resolution, ground-based FTIR measurements , J. Quant. Spectrosc. Ra. , 87, 25-52.
Heinselman, M.L. 1978. Fire in wilderness ecosystems. In J.C.Hendee, G.H. Stankey & R.C. Lucas, eds. Wilderness Management . USDA Forest Service, Misc. Pub. 1365.
90 CHAPTER VII Holloway, T., H. Levy II, and P. Kasibhatla (2000), Global distribution of Carbon Monoxide (2000), J. Geophys. Res. , 105, 12,123 – 12,147.
Holton, J., P. Haynes, M. McIntyre, A. Douglas, R. Rood, L. Pfister (1995), Stratosphere- Troposphere Exchange, Rev. Geophys. , 33 , 4, pp. 403-439.
Houghton, J.T., L. J. Meira, B. A. Callander, N. Harris, A. Kattenberg and K. Maskell (Eds.) (1996), Intergovernmental Panel on Climate Change (IPCC) 1996 , Climate Change 1995: The Science of Climate Change, Contribution of Working Group I to the Second Assessment Report of the Intergovernmental Panel on Climate Change , 572 pp., Cambridge University Press, Cambridge, U.K. and New York, NY, USA.
Houghton, J.T., Y. Ding, D. J. Griggs, M. Noguer, P. J. van der Linden, X. Dai, K. Maskell, C. A. Johnson, (Eds.) (2001), Intergovernmental Panel on Climate Change (IPCC) 1996 , Climate Change 2001: The Scientific Basis, Contribution of Working Group I to the Third Assessment Report of the Intergovernmental Panel on Climate Change , 881 pp., Cambridge University Press, Cambridge, U.K. and New York, NY, USA.
Jones, N. B., C. P. Rinsland, J. Ben Liley, J. Rosen (2001), Correlation of aerosol and carbon monoxide at 45°S: Evidence of biomass burning emissions, Geophys. Res. Letters , 28, No. 4, 709-712.
Kanamitsu, M., Description of the NMC global data assimilation and forecast system, Weather Forecasting , 4, 335-342, 1989.
Kasai Y., T. Koshiro, M. Endo, N. B. Jones, Y. Murayama (2005), Ground based measurement of strato-mesospheric CO by a FTIR spectrometer over Poker Flat, Alaska, Adv. in Space Res., Adv. Space Res., 10.1016/j.asr.2005.04.099.
Kinnersley, J.S. (1996), The climatology of the stratospheric THIN AIR model, Quart. J. Roy. Meteorol. Soc., 122, 219-252.
Krishnamurti, T. N., M. C. Sinha, M. Kanamitsu, D. Oosterhof, H. Fuelberg, R. Chatfield, D. J. Jacob and J. Logan (1996), Passive tracer transport relative to the TRACE A experiment. J. Geophys. Res. , 101 , 23,889-23,907.
Künzi, K. F., and E. R. Carlson (1982), Atmospheric CO mixing ratio profiles determined from ground based measurements of the J=1 → 0 and J=2 → 1 emission lines, J. Geophys. Res., 87, 7235-7241. von Kuhlmann, R., M. G. Lawrence, P. J. Crutzen, and P. J. Rasch (2003), A model for studies of tropospheric ozone and non-methane hydrocarbons: Model description and ozone results, J. Geophys. Res. , 108 (D9), 4294, doi: 10.1029/2002JD002893.
Lawrence, M. G., P. J. Crutzen, P. J. Rasch, B. E. Eaton, and N. M. Mahowald (1999), A model for studies of tropospheric photochemistry: description, global distributions, and evaluation, J. Geophys. Res ., 104 , 26,245-26,278.
91 CHAPTER VII Lawrence, M. G., P. J. Rasch, R. von Kuhlmann, J. Williams, H. Fischer, M. de Reus, J. Lelieveld, P. J. Crutzen, M. Schultz, P. Stier, H. Huntrieser, J. Heland, A. Stohl, C. Forster, H. Elbern, H. Jakobs, and R. R. Dickerson (2003), Global chemical weather forecasts for field campaign planning: predictions and observations of large-scale features during MINOS, CONTRACE, and INDOEX, Atmos. Chem. Phys. , 3, 267-289.
Li, Q., D. J. Jacob, I. Bey, R. Yantosca, Y. Zhao, Y. Kondo, J. Notholt, Atmospheric Hydrogen Cyanide: Biomass Burning Source, Ocean Sink?, Geophys. Res. Lett. 27, 357-360, 2000. Mutlow, C.T., D.L, Smith & M.J. Murray, The Along Track Scanning Radiometer (ATSR) Instruments on ERS-1 and –2, Earth Observation Quarterly , March 2000.
Mikhailenko, S., A. Barbe, and Vl. G. Tyuterev , (2002) Extended Analysis of Line Positions and Intensities of Ozone Bands in the 2900–3400 cm −1 Region, J. Molec. Spectrosc., 215(1) , 29 41.
Minzner, R.A. The 1976 standard atmosphere and its relationship to earlier standards. Rev. Geophys. Space Phys ., 15 , 375-384, 1977.
Murgatroyd, R. J., & F. Singleton (1961), Possible meridional circulation in the stratosphere and mesosphere, Quart. J. Roy. Meteorol. Soc., Vol. 87, No. 372.
Norton R.H. and C.P. Rinsland (1991), ATMOS data processing and science analysis methods, Appl. Opt. , 30 , 389-400.
Notholt, J., G. Toon, F. Stordal, S. Solberg, N. Schmidbauer, E. Becker, A. Meier, B. Sen (1997), Seasonal variations of trace gases in the high Arctic at 79°N, J. Geophys. Res., 102, D11, 12-855-12861.
Notholt, J., G.C. Toon, C.P. Rinsland, N.S. Pougatchev, N.B. Jones, B.J. Connor, R. Weller, M. Gautrois and O. Schrems (2000), Latitudinal variations of trace gas concentrations in the free troposphere measured by solar absorption spectroscopy during a ship cruise, J. Geophys. Res., 105, No. D1, 1337-1349.
Notholt, J., Z. Kuang, C.P. Rinsland, G.C. Toon, M. Rex, N. Jones, T. Albrecht, H. Deckelmann, J. Krieg, C. Weinzierl, H. Bingemer, R. Weller, O. Schrems (2003), Enhanced upper tropical tropospheric COS: Impact on the stratospheric aerosol layer, Science , 300, pp 307-310 .
Palm, M., C. v. Savigny, T. Warneke, V. Velazco, J. Notholt and K. Kuenzi, Intercomparison of O3 Profiles observed by SCIAMACHY (2004), Ground-based Microwave and FTIR Instruments, Atmos. Chem. Phys., 5, 2091–2098, 2005.
Pan, L., J.C. Gille, D. P. Edwards, P.L. Bailey, C. D. Rodgers, Retrieval of tropospheric Carbon Monoxide for the MOPITT experiment (1998), J. Geophys. Res., Vol 103, D24, pp. 32,277- 32,290.
Pine, A. S. and C. P. Rinsland, The role of Torsional hot bands in modelling atmospheric ethane, J. Quant. Spectrosc. Ra. ,, 62 , 445-458, 1999.
92 CHAPTER VII Pougatchev, N. S. and C. P. Rinsland, Spectroscopic study of the seasonal variation of Carbon Monoxide vertical distribution above Kitt Peak (1995), J. Geophys. , Res., 100, 1409-1416.
Pougatchev, N S., B. J. Connor, and C.P. Rinsland (1995), Infrared measurements of Ozone vertical distribution above Kitt Peak, J. Geophys. Res. , 100, 16,689-16,697.
Rasch, P. J., N. M. Mahowald, and B. E. Eaton (1997), Representations of transport, convection and the hydrologic cycle in chemical transport models: Implications for the modelling of short- lived and soluble species, J. Geophys., Res ., 102, 28,127-138.
Rao, K. N. and A. Weber (editors) (1992), Spectroscopy of the Earth’s Atmosphere and Interstellar Medium , 2 nd ed., Academic Press Ltd., U.K.
Rinsland, C. P., N.B. Jones, B.J. Connor, J. A. Logan, N. S. Pougatchev, A. Goldman, F. J. Mucray, T. M. Stephen, A.S. Pine, R. Zander, E. Mahieu and P. Demoulin (1998), Northern and Southern hemispheric ground-based infrared spectroscopic measurements of tropospheric Carbon Monoxide and Ethane, J. Geophys. Res ., 103, D21, 28197-28217.
Rinsland, C.P., N.B. Jones, B.J. Connor, J.A. Logan, N.S. Pougatchev, A. Goldman, F.J.Murcray, T.M. Stephen, A.S. Pine, R. Zander, E. Mahieu, and P. Demoulin (1998), Northern and southern hemisphere ground-based infrared spectroscopic measurements of tropospheric carbon monoxide and ethane, J. Geophys. Res ., 103, 28197-28217.
Rinsland, C. P., E. Mahieu, R. Zander, P. Demoulin, J. Forrer and B. Buchmann (2000), Free tropospheric CO, C 2H6, and HCN above Central Europe: Recent measurements from the Junfraujoch including the detection of elevated columns during 1998, J. Geophys. Res ., 105, 24, 235-24,249.
Rodgers, C. (2000), Inverse Methods for Atmospheric Sounding: Theory and Practice vol 2 , Series on Atmospheric, Oceanic and Planetary Physics, World Sci., River Edge N.J.
Rodgers, C. D. and B. Connor (2003), Intercomparison of remote sounding instruments. J. Geophys. Res., 108, (D3), 4116, doi: 10.1029/2002JD002299.
Rohen, G., von Savigny, C., Sinnhuber, M., Eichmann, K.-U., Llewellyn, E., Kaiser, J., Jackman, C.H., Kallenrode, M.-B., Schroeter, J., Bovensmann, H., and Burrows, J. (2005), Ozone depletion during the solar proton events of Oct./Nov. 2003 as seen by SCIAMACHY, J. Geophys. Res-Space, 110(A9), doi:10.1029/2004JA010984.
Rothman, L. et al. The HITRAN molecular spectroscopic database: edition of 2000 including updates through 2001 HITRAN Special Issue 2003, J. Quant. Spectr. Ra,. 82, 1.
Sander, S.P., R.R. Friedl, D.M. Golden, M.J. Kurylo, R.E. Huie, V.L. Orkin, G.K. Moortgat, A.R. Ravishankara, C.E. Kolb, M.J. Molina, and B.J. Finlayson-Pitts (2003), Chemical Kinetics and Photochemical Data for Use in Atmospheric Studies, Evaluation No. 14, JPL Publication 02- 25, Jet Propulsion Lab, Pasadena, CA.
93 CHAPTER VII Schlitzer, R., Lectures on Inverse Methods and Data Analysis , Winter semester 2002-2003, University of Bremen, Germany
Sherwood, S., A. Dessler (2000), On the control of stratospheric humidity, Geophys. Res. Lett., 27, 2513-2516. Sinnhuber, M., Burrows, J., Künzi, K.F., Chipperfield, M.P., Jackman, C.H., Kallenrode, M.-B., and Quack, M. (2003), A model study of the impact of magnetic field structure on atmospheric composition during solar proton events, Geophys. Res. Lett ., 30, No. 15, 1818, doi:10.1029/2003GL017265.
Solomon S., R. R. Garcia, J. J. Olivero, R. M. Bevilacqua (1985), P. R: Schwartz, R. T. Clancy and D. O. Muhleman, Photochemistry and transport of Carbon Monoxide in the middle atmosphere, J. Atmos. Sci., vol. 42, No. 10.
Talbot, R. W., et al. (1996), Chemical characteristics of continental outflow over the tropical South Atlantic ocean from Brazil and Africa, J. Geophys. Res ., 101, 24, 187-24,202.
Velazco V., J. Notholt, T. Warneke, M. Lawrence, H. Bremer, J. Drummond, A. Schulz, J. Krieg, O. Schrems (2005), Latitude and Altitude Variability of Carbon Monoxide in the Atlantic Detected from Ship-borne Fourier Transform Spectrometry, Model and Satellite Data. J. Geophys. Res. 110, D09306, doi: 10.1029/2004JD005351.
Warneke, T., J. Notholt, V. Velazco, M. De Maziere, O. Schrems, Measurements of atmospheric trace gases in the tropics by FTIR spectrometry, Poster Presentation AS3.15-1WE1P-0220, European Geophysical Union GA, Vienna, Austria, 2005.
WMO (World Meteorological Organization), Scientific Assessment of Ozone Depletion: 2002, Global Ozone Research and Monitoring Project—Report No. 47 (2003), 498 pp., Geneva.
Yurganov, L. N., E. I. Grechko and A. V. Dzola (2002), Long-term measurements of Carbon Monoxide over Russia using a spectrometer of medium resolution , Recent Res. Devel. Geophysics , 4, 249-265.
Yurganov, L. N., E. I. Grechko, and A. V. Dzola, Zvenigorod carbon monoxide total column time series: 27 yr of measurements, Chemosphere Global Change Sci., 1 , 127– 136, 1999.
Yurganov, L. N. , T. Blumenstock, E. I. Grechko, F. Hase, E. J. Hyer, E. S. Kasischke, M. Koike, Y. Kondo, I. Kramer, F.-Y. Leung, E. Mahieu, J. Mellqvist, J. Notholt, P. C. Novelli, C. P. Rinsland, H.-E. Scheel, A. Schulz, A. Strandberg, R. Sussmann, H. Tanimoto, V. Velazco, R. Zander, and Y. Zhao, (2004), A quantitative assessment of the 1998 Carbon Monoxide emission anomaly in the northern hemisphere based on total column and surface concentration measurements, J. Geophys. Res ., Vol. 109, No. D15, D15305, doi:10.1029/2004JD004559.
Zhao, Y. Strong, K., Kondo, Y., Koike, M., Matsumi, Y. Irie H., Rinsland, C.P. Jones, N.B. Suzuki, K. Nakajima, H. Nakane, H., and Murata I., Spectroscopic Measurements of Tropospheric CO, C 2H6, C 2H2, and HCN in Northern Japan 2002. J. Geophys. Res. , 107(D18) , 4343, doi:10.1029/2001JD000748, 2002.
94