o 9 ? D 9 L

�� !�"��#��$%�"��&�'��( ��)��+*,��.-(-��/�0�1�2��.3�0�14�$��/3��(��5�67�980�&:;��5�<>=?-(��&=��"@A��-��B�2$%C��"��/$%��5<>=?-��D=?-�E)�F�/<�G��.H�$%�I�JLKNMPOFMPQBM;RSO�T9UVUVTXWBO�Y[Z�J]\_â`cb�êd,TfDg�hjiVk3lmg2nok?g)h�fDgqpsr�t+nvuolmwyxvnznzg2{|g2nv}j~jpmxN�D�Np

� � � � ��_��,��o� �j�D��m�'��o��9�?��D�9�L��,��"�o�/ P¡v¢�£�¤N�� V�+¥§¦a¨z©&ª¤�¡P«¬¤N�ov¤N�B¥�®c¯v�°¤�¡�±2²� v«m³z�o�

´ µB¶&· ¸7¹Aº¼»¾½B¿�ÀLÁÂ¹eÃ#ÀAÄc¿�¹#Å&½#ÆÈÇÉ¶�ÊBÅ�¶&¿BÆ�ÁÂ¶&µ#ÃËÃ#ÀÌ¹#µB¹LÍ�Î�Ïz¶ÄL¿B¶&¶&µ#½AÀ#Ðcº�¶ËÄL¹Aº�¶�º7Æ�µÒÑ�Í,¹Aº�ÓÔº�¹#ÁÂÅ�Í,¶�ºÕ¹#µBÖ×Æ�µØ¹LÁÚÙ&Æ�¶&µ#Ã¹#Æ�¿ÌÆ�µÌÃ§½B¶Û¿�¶+ÄLÆÈÀ#µÜÀLÑÞÝ�¹BÇ�Í,¹�Î�ßàeá�âäãåçævèmézê/ëoì áoí�î�ïPð èmñvò°ê/ë�ì á�ó�ô)ð õ éPåyò°ê/ë�ö áø÷§áúùvávûüõ2ýÈï òFþNÿ ô éPñ â�á�� è ô è��°ê

ê�� !�"$#%!'&��)(*�+!-,��.!/0��1�.�32-��4�5&��0�)�$&��62��879#:��!/;�<=2>�#@? AB�CD��.��E�)�F�)��1(HGI� J�KL��K�MNC��O�)(NPE�.!��$&��ÿ=A��CQ�1�.��F�)��R7S!/�$!'2'!T(VUW ��T�O"�X��)��&YAB��CQ�1�@��Z�5��CD� &��(\[$��@�/��0�T(^]B��0!�� (7S!��$!'2'!>_

`9abdc�a$e=fdg'a=hji+idk^l*mon.p�q.nQrVsQnQtvu5wxwzyQ{F|Y}jc/a\~��k��

��(��¬�/� �Ò�9�E�� 9�D��#�P�,��d�+�� Õ�E�I��,� �L��,�� 2��F�§�(��F��%�� D,�� O� ��z�>��.�-�/�z�-��D�j�/�d�¡ /¢j�-�-�/£/¢¤�L¥>¦x�¨§ ©�ªE«D¬@ ®°¯0±�²%®°³�´Dµ�©I§ ©0¶S³�²�·Lµ9¸O´Tµ0²%¹O«<º¤·L´T²»±�¼·Lµ�©

¶o«<ºZ½·L¹�³�µd¾5½·:¿zÀ±SÁQ±Â¹L«<µÃÅÄ ÆÈÇTÉ@Ê ËdÌ:É@Í�Ê Î�Ï-É@ÊQÏzÍÂÐÑ Ò�Ó)Ô�ÕÂÒ�Ò�Ò Ö ×ÅØ.ÙÂÚ�ÛDÜ0Ø@Ý�Ø¡Þ<ß0Ú+ÙÂÞ�àÈÝ�á<àDØ@âäãäÞTåBáÂæÙÂß0Ý�á�Ú�çTèDØ¡Ü0ãäéoç/á�âëê

âäÛQß0ã¼á�Þìá�í)Ø¡ÜÅß0èDØïîxÙÂÜ0ãäÚWÙ�Ü0Ø.ÙQðòñNÞ�áÂí�Ø¡Ü0í<ã¼ØLó*á�ædß0èTØô$õTö9÷Nø ù çQÜ0áÂú1Ø@é¡ß.Ö

×ÅØ@Þ<ÛDß�ûü ·L²E±�¼ý

Õ Ñ@Ò)Ô�ÕÂÒ�Ò�Ò Ö ×Wñ�þ9ÿ��oî^Ù�ÞTÙ�â��QÚ�ãäÚHáÂí�Ø¡Ü�é@á�Ý�çDâäØ��ìá�Ü0á�å�ÜOÙÂçTè��Vã¼Þß0èDØoé@á�Þ<ß0Ø��Qß+áÂæ�ß0èTØd×WñNîòØ��Qç/Ø¡Ü0ãäÝ�Ø@Þ�ß.Ö

� ÙTÙÂÜ��Åî�üTÙÂÞ�à ��ü õ é¡ã ÙÂâäá�Ý� Ñ�Ñ�Ô�ÕÂÒ�Ò�Ò Ö õ ß0Û�à��Ù�Þ�à � Ø@Ú ß0ÚváÂæ ø Ý�çQÜ0áÂí�Ø.à��9ÙÂã¼Þ ô Ú ß0ãäÝ¤Ù�ß0Ø@Ú

æ Ü0á�Ý ß0èDØ � �o×H× îxÜ0Ø@é@ãäçTãëßOÙÂß0ãäá�Þ �9Ù�àTÙÂÜ.Öù Ø¡Ü�Ü0Ø@ãëÜOÙ ù ü ·L²+±�¼ý

� Ñ.Õ�Ô�ÕÂÒ�Ò�Ò Ö ÿoã¼ÙÂå�ÞTá�Ú ß0ãäé@ÚWá�ædè��DàQÜOÙÂÛTâäãäé¤ú1ÛTÝ�çìÙÂÞ�àZå�Ù�ç��á.ó;ã¼ÞÚ ß�ÜOÙ�ß0ã��Ø.à��Tá.ó+Ú�á�í)ØLÜjß0á�ç/á�åÂÜOÙ�çDè��ïÖ

ÿ9Ü0á�DãäÞTÚ��<ã>îxü ·L²j±�¼ý� Ò<Ñ�Ô�ÕÂÒ�Ò<Ñ ø þ9ÿ�� ô�� ê5ñ��Ø¡Ü0á�Ú�á�âäÚ î�Ø@âäá�Þ��Dü ·L²o±�¼ý! #" ÙÂÛDß0Ø@ÞDØ¡ß$��ü ·L²o±�¼ý! ñ�âäæ ÙÂÜ0á

õ ü ·L²+±�¼ý% Ò�Õ�Ô�ÕÂÒ�Ò<Ñ ø þ9ÿ�� ô�� ê � ÜOÙ�ÞTÚ�ç/áÂÜ�ß &5á�ÞDÞ�Ù('@á�â¼Ù�×¨ü ·L²o±�¼ý! ô ß0è)�Ø " ü ·L²S±�¼ý! û*�Ø¡á�Þ

�Dü ù ü ·L²E±�¼ý+ ÿ �Ø@Ú0ÙÂâ¼Ý¤ÙÂÞ�à ù ü, Ò�)Ô�ÕÂÒ�Ò<Ñ ø þ9ÿ�� ô�� ê$þNÛ�ÙÂå�Ø@Ú+Ø¡ß�é@á�ÞQí�Ø@é¡ß0ãäá�Þ õ.-Ø/'¡Ø0��ü ·L²+±�¼ý+ ��á�é.Ù1�dü ·L²+±�¼ý+ 2" Ù�ÛQß0Ø@ÞTØLß õ ü

·L²j±�¼ý3 Ò � Ô�ÕÂÒ�Ò<Ñ �oÙÂç4�Tá�ó ãäÞ�Ù�Þ ñ�âäçTãäÞTØ í�Ù�âäâäØ�� àDÛQÜ0ãäÞTåòÙòÚ�è�ÙÂâäâ¼á.ó

Ú�á�ÛDß0è æ á�Ø@èDÞ Ø¡í)Ø@Þ�ß.ð��5DÚ�Ø¡Ü0í�ÙÂß0ãäá�ÞDÚ�Ù�Þ�à è��DàQÜOÙ�ÛDâäã¼éÙÂÞ�ÙÂâ¼á�å

ù â¼Ù�Ý¤Ù�Þ�ß " ü ·L²+±�¼ý

Ó Ò��Ô�ÕÂÒ�Ò<Ñ îã��QØ@â/å�Ø@á�Ý�Ø¡ß�Ü��Bæ:á�Ü�ÙSß0ã¼âëß0Ø.à76QÚ�ç�Ù�é¡Ø5/á�Ü0ÞDØ9ãäÞ�ß0Ø¡Ü�æ ØLÜ0á�êÝ�ØLß�Ü0ã¼éNÜOÙ�àDãäá�Ý�Ø¡ß0Ø¡Ü

8 ÙÂâ à<ß0Ø@ÛDæ:Ø@â-îxü ·L²j±�¼ýÑ.Ò Ò%)Ô�ÕÂÒ�Ò<Ñ ��Ø.ÙÂâäê°ß0ãäÝ�Ø5Ü0Ø¡ß�Ü0ãäØ@í�Ù�âQá�æ/ß0èDØ5ó+ãäÞ�à�æ Ü0á�Ý�Ù�âäã¼Ù�Ú�Ø@àBí)Ø@âäá�é¡ê

ãëß0ãäØ@Ú+Ý�Ø.ÙÂÚ�ÛDÜ0Ø.à9��Èÿoá�çTçDâ¼ØLÜjÜOÙ�àTÙ�Ü0Ú� ÙTÙÂÜ��Åî�ü ·L²j±�¼ý

Ñ�Ñ Ò�,�Ô�ÕÂÒ�Ò<Ñ " á�Ý�çDß0Ø¡ê°Ü0Ø@ÞTàDÛRàQØ�â¼Ù ø�ø�ø:-Ø@Ý�Øïú1á�ÛQÜ0Þ)�Ø@ØZÚ ßOÙ�ß0ãäÚ ß0ã<;�ÛDØø î õ û>=@?$Ù�âäØ@ÛQÜ0ÚjØ��<ß�Ü -Ø@Ý�Ø@Ú:A

þoÙ.í)Ø.ÙÂÛWî�ü ·L²j±�¼ýÑ�Õ Ò3)Ô�ÕÂÒ�Ò<Ñ õ Ý¤Ù�âäâÚ�é.Ù�âäØ¤å�Ü0á�ÛTÞTàQêB�ÙÂÚ�Ø.à ÙÂé¡ß0ãäí)Ø�Ý�ãäé¡Ü0á.ójÙ.í)ØBÝ�Ø.Ù�ê

Ú�ÛQÜ0Ø@Ý�Ø@Þ�ß0ÚEá�í)Ø¡Ü+æ:á�Ü0Ø@Ú ßNé.Ù�ÞDá�çDã¼Ø¡ÚÿoØ@éOè�ÙÂÝCDÜ0Øo×¨ü ·L²j±�¼ý

D�EzÐ@É.Ì:É.ÏzÉ9FSÌ:Ê<Í�Í�Ê�G�H-Ì@IKJ2EML�ÇON#P Ç2Q�Ê ê)R ²%²ëÁTS UVUXWYW�W5ý¼® ÁTµL¼ý ¿¡«<µOµ0®»·L«Tý ZO¹÷ ÞTãäí)ØLÜ0Ú�ãäß��Ø�îãäØ¡Ü�Ü0ØSØ¡ß9×WÙÂÜ0ãäØ " ÛDÜ0ãäØ ÷ ÞTãäí)ØLÜ0Ú�ãäß��Ø5?$Ø¡Ü0Ú0ÙÂãäâ¼âäØ@Ú õ ÙÂãäÞ<ß ö ÛDØ@Þ�ß0ã¼Þ&�Ñ.Ò)Õ�ê � Ñ[��ê ô � " á�âäâ -Ø@å�Ø0?5ÙÂÛOTÙ�Þ� 6�çDâ¼Ù�é@Ø��ÛDÚ�Ú�ãäØ@Û � ,�6O/á�ÛDâ¼Ø¡í�Ù�ÜOà�?$Ù�ÛOTÙ�Þ,��Õ��Õ�îxÙÂÜ0ãäÚ " Ø.àDØ��ÅÒ�� ,3�Ò � ,\�9Û��<Ù�ÞDé@á�ÛQÜ�ß " Ø@àDØ��]0^7_.`�aCbT`Xa$c5dO_2e+fge<_ih`�a�j�k7a�_.lOm7no`�k7_�pgq`Xa0bT` q�r jtsou2vw ÿoãëÜ0Ø@é¡ß0Ø@ÛQÜNàDØSâ¼ÙIçTÛXTâäãäé.ÙÂß0ãäá�Þ-ð ��Ø.Ù�Þx��á�ÛX'@Ø@âw " á�Þ�ßOÙ�éLß.ð ûzÙ�ÛQÜ0Ø@Þ�ßN×ÅØ@Þ<ÛDß�y Ý�Ø@Þ<ÛDß�z9âäãäÚ0ÙQü ÛDÞTãäí�ê�çTÙÂÜ0ãäÚ@Ñ�Õ�ü æ:Ü�{w " á�Ý�ãëß��Ø�à>ÖB�Ø.àQãëß0ã¼á�Þ-ð |x�Ø@â -Ø@ÞDØ " èTØ¡çDæ ØLÜ�=°û�×Åÿ�A�6~}9í)Ø@Ú ÿoÙ�Þ�àQá�ÞDÞTØ.ÙÂÛ�=°û��9ÿ5� " A�6

" �QÜ0ãäâäâäØ ù â¼ÙÂÝ¤Ù�Þ�ßC= õ ñiA�6 " �<Ü0ã¼âx×Åá�ÛTâäãäÞ�=°û õ " ô A�6�ñNâ¼Ù�ãäÞ�îxÜ0á�êßOÙÂß0= " ô�� î�A�62��Ø@Ý�ã.��á�é.Ù�=°ûz×Åÿ�A

A new Gas Chromatographic measurement system to analyze greenhouse gases.

1

A NEW GAS CHROMATOGRAPHIC EXPERIMENT TO ANALYZE GREENHOUSE GASES IN FLASK SAMPLES AND IN AMBIENT AIR IN THE

REGION OF SACLAY.

L. PÉPIN1, M. SCHMIDT1, M. RAMONET1, D.E.J. WORHTY2 AND P. CIAIS1 1Laboratoire des Sciences du Climat et de l’Environnement, Gif-sur-Yvette, France. 2Environment Canada, Atmospheric Environment Service, Toronto, Ontario, Canada.

Abstract In the prospect of accurate measurements of greenhouse gases, a new gas chromatographic (GC) measurement system has been developed at LSCE. It gives the opportunity to measure precisely CO2, CH4, N2O and SF6 at ambient atmospheric concentrations (i.e. ±0.045 ppm, ±0.65 ppb, ±0.40 ppb and ±0.05 ppt (expected) respectively) in air sample either in flasks or in ambient air from the Saclay region. The basic principle of gas chromatography are explained, followed by a detailed experiment descriptions. We have tested different points of the system likewise the measurements accuracies and explain the results of these investigations. Finally, typical records from flasks and ambient measurements illustrate the GC applications and enable us to provide first estimations of greenhouse gases emission rates in the region of Saclay based on our ambient measurements and co-located radon 222 measurements. Introduction In the context of climate change studies, some fundamental questions must be addressed: how are the composition and properties of the atmosphere changing as a result of human influences? How do these changes compare to the natural forces of change? How do we expect these changes to evolve in the future? The most significant atmospheric compounds that can be modified by both natural and human induced influences, external to the climate system, are greenhouse gases others than water vapor, particularly carbon dioxide (CO2), methane (CH4) and nitrous oxide (N2O). Since the beginning of the industrial revolution, some 200 years ago, the atmospheric concentration of CO2 has increased by about 30%. The background CO2 concentration was estimated to be 367 ppm (part per million, 10-6) in 1999, whereas it was about 280 ppm for the last millennium. Atmospheric methane concentrations show that pre- industrial levels (about 650 ppb (part per billion, 10-9)) have more than doubled during the past two centuries (around 1750 ppb in 1999). Other greenhouse gases, less concentrated than CO2 but much more radiatively active, such as the N2O (315 ppb in 1998) have been rising slowly but steadily (about 0.8 ppb/year during the 90’s). The SF6 sources are entirely anthropogenic, mainly coming from leaks into electrical installations, where this gas is used as insulator. In the late 90’s the mean concentration of SF6 was about 4 ppt (part per trillion, 10-12) and rose at a rate of about 0.2 ppt/year.


2

The state of the art concerning the knowledge of the geochemical cycles (sources and sinks) of greenhouse gases can be found in the IPCC report [2001]. The understanding of the geochemical cycles of the greenhouse species (CO2, CH4, N2O, SF6…) is fundamental to evaluate, for the next centuries, the radiative forcing due to the drastic change of the atmosphere’s composition that is predicted. Moreover, policy makers try now to control atmospheric growth rates of greenhouse gases (Kyoto Protocol) and to constrain countries to emission quotas. In this context, it becomes crucial to determine as accurately as possible the greenhouse gases emission rates on the scale of a country or a region. Greenhouse gases budgets are rather well estimated at global scale [IPCC, 2001]. Nevertheless, for poorly documented species or at regional scales, budgets are less accurately known. Because of the importance of the stakes, more and more scientific groups are working on this subject. Different methods may be used to estimate the greenhouse gases emission rates at regional scale. We generally distinguish between the “bottom-up” and the “top-down” approaches. The principle of “bottom-up” approaches is to aggregate some local information to deduce regional estimates. In the case of emission rates, field campaigns are done to evaluate the signature of every process that takes part of the geochemical cycle of the studied species. Then the measured fluxes are summed up to derive regional emission rate of greenhouse gas. The major drawback of such method is due to the difficulty of extrapolating point data. Moreover, unidentified processes are occulted. Consequently, such estimations yield to large uncertainties [IPCC, 2001]. The “top-down” approach is directly based on atmospheric observations of the greenhouse gases concentrations. The spatial distribution of a trace gas concentration in the atmosphere combines effects coming from the sources and sinks distribution and also from the atmospheric transport. The late parameter must thus be taken into account via an atmospheric transport model or, alternatively, via the use of a “reference” atmospheric tracer. An atmospheric tracer is a compound for which sources and sinks are pretty well known; radon, C2CL4 or SF6 are used as such tracers [Bakwin et al., 1997]. The experiment which has been developed since September 2000 at LSCE, and that we describe in this report, has two kind of applications. On one hand, the gas chromatographic (GC) system is designed to analyze flasks that are sampled at RAMCES (Réseau Atmosphérique de Mesures des Composés à Effet de Serre) monitoring stations. These results are then used to monitor the concentration changes of greenhouse gases in different part of the world and to be used in global 3D atmospheric tracer transport models. On the other hand, we have developed a second application for the GC system that is to monitor, quasi continuously, ambient concentration of CO2, CH4, N2O and SF6 in air sampled on the roof of our laboratory building in Saclay. The main goal of such study is to derive emission rates estimates of the greenhouse compounds. To do so, we will use the radon tracer method [Schmidt et al., 1996; Kuhlmann et al., 1998; Levin et al., 1999; Biraud et al., 2000; Schmidt et al., 2001].


3

In the first part of this report we explain briefly the basic principles of the gas chromatography, and describe in details the experiment and the running procedure. We will also evaluate the accuracy of GC’s measurements, scrutinizing some test results. In the second part, we present several records obtained, with the GC experiment, in Saclay’s ambient air and in flask samples, and estimate some emission rates for CO2 and CH4 in the catchments area of the lab, using the radon tracer method.

PART I: DESCRIPTION AND PERFORMANCE OF THE MEASUREMENT SYSTEM Gas Chromatography: Principle The aim of this analytical technique is to separate, detect and quantify the different species contained in an air sample. The sample is injected into a packed column on which molecules are alternatively adsorbed and desorbed. This column is continuously flushed by the gas that transports the sample over the column (carrier gas). Depending on the typical characteristics (geometry, molecular weight…) of each compound and also on experimental parameters i.e. flow rate of the carrier gas; temperature, length and packing material of the column… the different species of the air sample are eluted out of the chromatographic column at different times (retention times). Following this step of individualization it is then possible to guide the different molecules toward an appropriate detector. Most of detectors react to the composition change of the effluent as a perturbation in regard to an initial state maintained by the continuous flow of carrier gas into the column and the detector. A good detector has thus to be stable, sensitive and to have a short response time to detect the composition change of the effluent. Here we use two different detectors.

• Flame Ionization Detector (FID): A hydrogen flame burns into the air, producing ions. When organic compounds are introduced into the flame from the column effluent, electrically charged species are formed. A polarizing voltage attracts these ions to a collector located near the flame. The current produced is proportional to the amount of sample being burned. This current is sensed by an electrometer, converted to digital form, and sent to an output device.

• Electron Capture Detector (ECD): The ECD contains a cell plated with 63Ni, a radioactive isotope. The 63Ni release β that collide with carrier gas molecules to produce low-energy electrons (each β particle produce approximately 100 electrons). The free electron bring about a small current (reference current). When some electro-acceptor compound enters the detector, it captures part of the free electrons, creating negatively charged ions. The voltage across the cell is pulsed to collect the remaining


4

free electrons while the heavier ions are relatively unaffected and swept out the vent with carrier gas flow. Cell current is measured and compared to the reference current. The pulse rate is adjusted to maintain a constant cell current. The more free electrons, the lower the pulse frequency required to match the reference current. When a component that capture electrons passes through the cell, the pulse rate rises. This pulse rate is converted to voltage and recorded. The quantification of the concentration of each species analyzed in the sample is performed comparing the signal produced by this sample on the detector with the one of at least one reference gas i.e. which composition is known. GC description Since September 2000, an automated gas chromatographic system (HP-6890) has been developed by Doug Worthy (Atmospheric Environment Service, Environment Canada [Worhty et al., 1998]) and optimized at LSCE for measurements of CO2, CH4, N2O and SF6 in flask samples as well as for semi-continuous measurements of ambient air in Saclay. This system, shown in Figure 1, is equipped with two detectors i.e. it allows to analyze simultaneously CH4 and CO2 on a flame ionization detector (FID) and N2O on an electron capture detector (ECD). The ECD is also designed to detect SF6 but, as the working standards we use now do not contain any SF6, we are not able to quantify yet the concentration of this compound in our samples. In the case of flask samples measurements, we prepare the analysis by evacuating the injection line between each flask and the 8-port selection valve. We reach a vacuum of around 7 10-3 mbar with a primary pump. The ambient air samples are first dried trough a cold trap (1 l) maintained around -50°C by a cryocooler. Then, the analytical procedure is common for every type of gas injected, i.e. flask samples than ambient air or standard gases. First, injected gases pass into a second, but smaller, cold trap (5 ml, -50°C). It has mainly been designed to dry effectively the flask samples, even if air is dried with a perchlorate magnesium cartridge during sampling. Moreover, it insures us that all types of gas are injected in the same temperature conditions. Following the flushing of the two sample loops (15 ml each) by about 125 ml of gas, the sample equilibrates with atmospheric pressure and temperature for 30 seconds. Then, the CO2/CH4 and N2O/SF6 sampling valves are switched to the inject position (as shown on the Figure 1), allowing the column carrier gas to transfer the contents of the sample loops to their respective separation columns. For the CO2 and CH4 analysis, Nitrogen (5.0) is used as carrier gas. The sample loop content is directly transferred onto the analytical column, packed with Haysep-Q (12’Î 3/16’’ OD, 80/100). CH4 and CO2 retention time are respectively 2.42 and 2.95 min. in our experimental conditions. First, the effluent is thus directly directed to the detector (position of the shunting valve depicted


5

on the Figure 1). At 2.62 minutes this va lve is switched, allowing the sample to go through a Ni catalyst where CO2 is reduced (in presence of H2) to CH4 prior to FID detection.

Table 1a: Working conditions for the detection of CH4 and CO2.

In the case of N2O/SF6 analysis, the carrier gas in a mixing of CH4 (4.5, 5%) in Argon (6.0). The sample is first passed onto a pre-column (Haysep-Q, 4’Î 3/16’’ OD, 80/100), from which N2O and SF6 elute after around 1 minute. They are then directed to the analytical column (Haysep-Q, 6’Î 3/16’’ OD, 80/100). At this point the pre-column is back-flushed, avoiding contaminants passing through the ECD and reducing analysis time. With this configuration, N2O and SF6 are respectively eluted 3.20 and 3.66 minutes after the sample’s injection.

Table 1b: Working conditions for the detection of N2O and SF6.

Sample loop Volume Flow rate

15 ml 125 ml/min (2.5 psi)

Chromatographic column Type Length External diameter Packing Temperature

Haysep-Q 12’ 3/16’’ 80/100 mesh 80°C

Detector Type Temperature

FID 250°C

Catalyst Type Temperature

Ni 390°C

Flow rates Carrier Gas: N2 FID supply: Air Catalyst supply: H2

53 ml/min (23-33 psi) 300 ml/min 100 ml/min

Sample loop Volume Flow rate

15 ml 125 ml/min (2.5 psi)

Chromatographic columns

Use Type Length External diameter Packing Temperature

Pre-column Haysep-Q 4’ 3/16’’ 80/100 mesh 80°C

Analytical column Haysep-Q 6’ 3/16’’ 80/100 mesh 80°C

Detector Type Temperature

ECD 395°C

Flow rates Carrier Gas: 95% Ar (6.0), 5% CH4 (4.5)

82 ml/min (23-23 psi)


6

Figure 1: Schematic of the experiment.

ECDFID

Cat.

FID Air

H2

N2

Sample

Ar/CH4

Oven (80°C)

Cold Trap(1 l, -50 C)

Cold Trap(5 ml, -50 C)

Working Standards

Flow Controllers

16 Ports Selection Valve

8 Ports Selection Valve

Pump(5 l/min)

CO2/CH4 Sampling (inj ect position)

N2O/SF6 Sampling (inj ect position)

N2O/SF6 Backflush(backf lush =off )

Shunting Valve(direct position)

Analytical Columns

PressureRelease Valve

(~1 bar)

6 Ports 2 Positions Valves

(Injection Pump)


7

The complete injection procedure takes 4.5 minutes. It is called a “method”; an example of it is described and commented in Table 2.

Table 2: Basic method description. After the detection step, the HP software (ChemStation) integrates the different peaks following prescribed instructions. Figures 2a and 2b show, for FID and ECD respectively, the entire chromatogram together with a zoom on which we can see how the peaks are integrated. The operator fixes the length of the time window and the slope threshold for which the integration is allowed. For each injection, integration results are saved into a file. The complete procedure takes about 5 minutes. To manage a particular analysis (flask, ambient air or any other gas), methods are arranged into tables named “sequences” which determines the succession of injections. Raw data associated to each sequence are saved in subdirectory containing the method’s files.

Time (min.)

Specifier Parameter and setpoint

Comments

0.00 0.00 0.00 0.00 0.70 0.70 0.75

1.25 1.75

2.62

2.70 2.70 3.45

4.45

4.50

Valve 1 Aux. 5 pressure


Valve 3 Aux. 4 pressure Aux. 5 pressure

Valve 1 Valve 1

Valve 4

Aux. 4 pressure


Valve 4

Aux. 3 pressure

off 2.5 psi

On 23.0 psi

Off 35.0 psi 0.0 psi

On Off

On

23.0 psi

On 0.0 psi

Off

33.0 psi

Sampling valves into the load position Flushing the sampling loops (125 ml/min) Backflush valve into Backflush position Initial flow rate of the ECD carrier gas Backflush valve into the direct position Adjustment of the Ar-CH4 flow rate Stop flushing the sample loops (pressure/temperature equilibration) Injection after 30 second equilibrium time Return into the load position for the sampling valves Shunting valve into the catalyst position (between CH4 and CO2 elution time) Reduction of the Ar-CH4 flow rate Backflush valve into Backflush position FID Nitrogen carrier gas flow rate reduced (prevent ing flame out) Shunting valve goes back into the direct position Recovering of the FID Nitrogen’s flow rate


8

Figure 2a: FID chromatogram and peaks integration. FID response is in pico-ampere (10-12 A).

Figure 2b: ECD chromatogram and peaks integration. ECD response is in 5 Hz scale.

min 0.5 1 1.5 2 2.5 3 3.5 4

pA

0 1000 2000 3000 4000 5000 6000

CH4

CO2

min2.4 2.6 2.8 3 3.2 3.4

pA

0 10 20 30 40 50 60 70 80 90

CH4 (2.40)

CO2 (2.94)

FID

min 0.5 1 1.5 2 2.5 3 3.5 4

5 Hz

0 5000

10000 15000 20000 25000 30000

min2.8 3 3.2 3.4 3.6 3.8 4

5 Hz

150 160 170 180 190 200 210 220 230 240

ECD O2

N2O SF6

N2O (3.20)

SF6 (3.66)


9

Concentration calculation To determine the concentration of samples, we inject alternatively reference and sample gas in the GC. We use 3 reference gases, which cover the range of atmospheric concentrations (from “clean” Southern Hemisphere where flasks can be sampled, to the semi-urban atmosphere of Saclay). These gases provided by Air Liquide (France) are called working high (WH), target (WT) and low (WL) (see Table 3 for the exact composition of each reference gas). The basic sequence of injections follows: WH - WL - X1- X2- X3- X4 where Xi represents sample. In the case of flasks analysis, we inject 2 times the same flask sample, and 2 different flasks consecutively. To calculate the sample concentrations we use the area of the CO2 peak and the height of the CH4 and the N2O peak. Indeed, height of peaks is less sensitive to baseline variations than area, but in the case of non-gaussian peaks (for CO2 for example, see Figure 2a) the use of peak height does not give proper values. Then, we compare the chosen parameter (area or height) of the peak that we want to quantify, to the corresponding WH and WL. In addition, temporal drifts of detector responses between 2 standards injections are taken into account, as illustrated in Figure 3.

Figure 3: Illustration of the samples calculation method. The sample concentration is calculated using the following equation:

[ ] [ ] [ ] [ ]( ) ( )**

*

WLWHWLXWLWH

WLX−

−×−+=

(1)

where: ( ) ( )

( ) ( )WHWHNi

WHWHWHtN

ttWHWH

WLWLN

iWLWLWL

tNtt

WLWL

W HX

WLX

−×

+=−×

∆−

+=

−×

−

+=−×

∆−

+=

''

'1

'

*

*

>

>

WL* X WH*

WL

X

WH

Concentration

Response

t= tX

t0 t0+∆t tX=t0+i∆t

t0+N∆t

Response

Time

WH'WH*

WH

X

WLWL*

WL'

>

>


10

To check the quality and control the stability of our measurements (see “reproducibility and long term stability” paragraph), we inject every 2 hours a so-called “target gas” (WT). Its concentration is nevertheless calculated exactly in the same way as real samples (i.e. X4 is replaced by WT in the basic sequence). Calibration strategy As 125 ml of gas are used for every injection and working standards are analyzed 2 times per hour, we calculate a lifetime of working standards (WH and WL) of about 2 years (and about 9 years for WT). To ensure the continuity of our measurements and to be able to point out possible drift of working standards, we use a two different levels calibration scale. Every 3 month, the working standards should be calibrated against 2 international standards, which are provided by NOAA CMDL (Boulder, US) that constitute our primary scale (see Table 3 for the exact compositions). A third NOAA’s standard is used as a target, in the same way than for the working standards. If we detect a drift in the composition of one of our working standard, it is thus always possible to correct the calculated concentrations of samples that have been analyzed between 2 calibrations, assuming that this drift is linear in time.

Table 3: Composition of primary and working standards used in this study. For the primary scale, the accuracy is given by CMDL, it takes instrumental plus experimental

uncertainties into account. Concerning the working standards, we give the measurement reproducibility.

Non linearity for FID and ECD and co-elution tests Is the relation between detector response (area or height of peak for each species detected) and gas concentrations linear? Is the preliminary question that we have to solve. We have tested the degree of linearity of detectors used in this study for each analyzed species, comparing the

NOAA1 (CA04648)

NOAA2 (CA04673)

NOAA3 (CA04687)

WH (1330339)

WT (1366740)

WL (1750765)

CO2 (ppm) 442.48 ±0.07

390.20 ±0.07

365.82 ±0.07

421.583 ±0.023

390.439 ±0.038

369.896 ±0.043

CH4 (ppb) 1978.3 ±1.3

1875.9 ±1.3

1731.8 ±1.3

2113.4 ±0.92

2020.66 ±0.68

1909.31 ±0.64

N2O (ppb) 338 ±3.4

328 ±3.3

321.1 ±3.1

361.77 ±1.11

357.66 ±0.94

346.09 ±0.67

SF6 (ppt) 9.7 ±0.2

14.9 ±0.2

7.6 ±0.2

CO (ppb) 523.0 ±10.8

305.9 ±3.1

53.9 ±1.0


11

prescribed values for WH (i.e. the values determined calculating WH in regard to NOAA’s primary standards) with the ones calculated using WL and WT or WL only. This test is an extreme configuration, as we calculate one sample (WH) that is out of the range of the standards (WL and WT).

Prescribed WH WH /WT and WL WH /WL CO2 (ppm) 421.583 ±0.023 421.609 ±0.069 421.740 ±0.036

CH4 (ppb) 2113.14 ±0.92 2113.21 ±0.81 2114.43 ±0.72

N2O (ppb) 361.77 ±1.11 367.33 ±0.78 358.32 ±0.48 Table 4: Linearity test.

The results of this test are summarized in Table 4 and confirm that the FID response is linear for CH4 measurements. Concerning CO2, the linearity of this detector is less clear (note that the final response for CO2 takes also into account the methanization process). Comparing the prescribed value to the one calculated against WL alone, shows that the difference between the two values is higher than the accuracy of each measurement. Nevertheless, we obtain results that are not significantly different, if we compare the calculation of WH using WT and WL as standards and the prescribed value. The non- linearity of the FID is weak (about 0.003 ppm/ppm difference between the sample and the standard); we will thus consider that the FID’s responses are linear for both CO2 and CH4. Comparing the different calculation for N2O gives obviously different results (cf. Table 4), which means that the ECD response for this compound is non linear. An evaluation of the non- linearity of the ECD response for N2O [Schmidt et al., 2001] is given by:

stdmeas

measpres

ONON

ONON

2.2

.2.2

−

−

(2)

where N2Opres. refers to the prescribed value, N2Omeas. is the value deduced using one standard (N2Ostd) for the calculation. This calculation gives a correction of about 0.28 ppb per ppb difference between the measured value and the standard used. This value is in agreement with ECD non- linearity of N2O estimated by Schmidt et al. [2001] and Bader [2001]. To determine more precisely the ECD non- linearity for N2O analysis, we have to perform a dilution test in which the WH would be mixed with synthetic air, free of N2O and CH4. The dilution rate would be exactly known using the linearity of the FID’s response for CH4 and comparing prescribed concentration of CH4 and the measured ones. This test is not feasible until now. Nevertheless, in practice, it is not necessary to correct N2O values if they lie into the range of the two standards used. To confirm this assumption, we can compare the N2O’s WT concentration (calculated against WH and WL): 357.67 ±0.67 ppb, with the prescribed value (357.66 ±0.94 ppb).


12

An additional source of uncertainty to N2O analysis is the co-elution phenomenon that can occur between N2O and CO2 because these two species have the same molecular weight. Even if CO2 is not a good electron acceptor, it can capture a few electrons in the ECD, artificially enhancing the N2O signal. We have tested this possible bias, removing all or part of the CO2 with ascarite (sodium hydroxide silica, 5/20 mesh) in WH standard. This test is presented in Figure 4 and does not show any increase of the N2O concentration when CO2 increases.

Figure 4: Influence of CO2-N2O co-elution on N2O concentration. We thus conclude that there is no co-elution between N2O and CO2 in the particular experimental conditions we use (length and temperatures of the columns, flow rates of carrier gases…). Reproducibility and long term stability Quality of measurements is estimated in different way. First, we calculate the short term (i.e. within a sequence) reproducibility of standards. If Wn is a working standard injected at t0, Wn-1 and Wn+1 are respectively injected at t0-N∆t and t0+N∆t (taking the same notation than in Figure 3), then at t0 we can write:

[ ] [ ] ( )2

11.

−+ +=

nnpresn WW

WnWW

(3)

12:00 18:00 00:00 06:00 12:00 18:00 00:00 06:00 12:00 18:00 00:00 06:00 12:00

418

418.5

419

419.5

420

ppm

150

200

250

300

350

400

ppm

354.5

355

355.5

356

356.5

357

357.5

ppb

ReferenceAscarite

CO2

N2O

9-nov.-01 10-nov.-01 11-nov.-01 12-nov.-01


13

We evaluate the reproducibility of every species and for each working standards. Typically, it is: ±0.045 ppm for CO2, ±0.65 ppb for CH4, ±0.40 ppb for N2O. We consider these standard deviations as lower limits for the accuracy of samples calculations. The long term control of the accuracy of measurements is performed by scrutinizing the concentration variations, for every species, of the WT that is injected every 2 hours (Figure 5). Other interest of having such a target gas (either for the working scale than for the primary one) is, moreover, that it is less often injected, its life time is thus longer and allows using the target calculation as a continuity tracer when we change of working standards.

Figure 5: Time series of CO2, CH4 and N2O of WT from 10/16/2001 to 01/16/2002 compared to the prescribed values (solid lines).

Circled points have not been taken into account for the mean and standard deviation calculations.

In Figure 5 we show, the calculated values of CO2, CH4 and N2O of WT since October 2001 and compare them to the prescribed ones (i.e. the values deduced from the calibration of working standards against NOAA primary scale). During this period, the mean CH4 value for the calculated WT was 2021.30 ±0.69 ppb (the excluded data mainly correspond to the beginning of sequences) and fits within the accuracy of the prescribed value (see Table 3). The mean N2O concentration of the calculated WT (357.72 ±0.94) is also reliable with the prescribed value but seems to oscillate around it. For the CO2 concentration calculated in the WT gas, we observe a drift over the period. Until mid November we calculate a mean CO2 level around 390.64 ±0.06

CO2

CH4

N2O

16-oct.-01 30-oct.-01 13-nov.-01 27-nov.-01 11-déc.-01 25-déc.-01 8-janv.-02

390

390.2

390.4

390.6

390.8

391

ppm

2020

2022

2024

ppb

356

358

360

ppb


14

ppm. Then, the value decreases, reaching about 390.40 ±0.05 ppm during the beginning of 2002. This effect may have several causes. It can be due to regulator effect, either on the WT itself than on the two standards used for the calculation, especially because the different kind of gases are not injected with the same regularity (1 inj./2h for WT and 1inj./1/2h for the working standards). As we will change the regulators to minimize the dead volume in them, we hope to prevent in the future this kind of pressure regulator effect (which can also explain the systematic shift between the mean calculated and prescribed CH4 values). Nevertheless, the drift of the calculated CO2 concentration in the WT can also be due to an aging effect of the mixing itself or of the others working standards. To address this point we have to calibrate again the set of working standards with the primary scale. Flask measurements Typical sampling conditions imply that flasks are pressurized with about 0.5 bars overpressure. Due to sampling problems or to the number of analysis performed before the GC measurement, it can happen that a flask does not have enough overpressure. The injection system has thus been designed to allow the use of an injection pump (into bracket on the Figure 1). We have tested how reliable are multiple injections on flasks (1 or 2 liters) filled with over or near atmospheric pressure. 1 and 2 liters flasks have been filled with about 0.5 bars over pressure of a reference gas (cf. Table 5). The critical point is nevertheless that we don’t know the exact pressure in the prepared samples. First of all, we have performed 4 or 3 injections (for 2 and 1 liters flasks respectively), considering that the flasks were over pressurized (“over-pressure procedure” hereafter). Then we have injected 4 (2 liters flasks) or 3 (1 liter flasks) more times the flask samples using the injection pump (“pump procedure”). Figure 6 and Table 5 present the results of this test. Looking at Figure 6, we have chosen to eliminate some measurements for the mean and standard deviation calculations. This choice is somehow subjective because, as already noticed, the injection and flushing pressures are not recorded. Indeed, we guess differences between flasks caused by the initial filling. For example, on F1 we consider that 4 over-pressure injections +2 using the injection pump are reliable. On the other hand, only the first 3 injections on F5 seem to be confident, even if both flasks are 2 liters ones.


15

Figure 6: Multi-injection experiment on 1 and 2 liters flasks.

To explain this difference, one hypothesis is that the two flasks have not been filled with the same pressure and that the injection pump may not be strong enough to flush all the line before injecting the sample in the case of F5. It appears more and more clearly that the result’s interpretation should be improved if we record the flasks initial pressure and the flushing pressure of the line. It will be the next improvement that we plan to provide to the GC experiment.

Prescribed Over-Pressure procedure (selection)

Pump Procedure (selection)

CO2 (ppb) 369.896 ±0.043 369.65 ±0.058 Except inj. 4

of F5

369.87 ±0.14 2 first inj. of F1

+1st inj. of F2,

F3, F4

CH4 (ppb) 1909.31 ±0.64 1909.10 ±0.67 All 1909.90 ±0.93 3 first inj. 2L

flasks +2 first inj.

of 1L flasks

N2O (ppb) 346.09 ±0.67 346.09 ±0.56 All 345.62 ±0.74 All Table 5: Multi injection experiment on flasks (1 and 2 liters). Over-Pressure and Pump

procedures.

CO2

CH4

N2O

09:00 12:00 15:00 18:00

12 december 2001

368.5

369

369.5

370

370.5

371pp

m

1890

1900

1910

1920

1930

ppb

344.5345

345.5346

346.5347

347.5

ppb

F1 (2L)F2 (2L)F3 (1L)F4 (1L)F5 (2L)

Over-Pressure Pump


16

It can be also noticed that data kept for the statistic calculations depend on the studied species. Indeed, the absolute accuracy expected for CO2 measurements is at least 4 and 10 times better than for CH4 and N2O respectively. The CO2 GC measurements are thus much more sensitive to every contaminations or instability of the system than those of the other species and the criteria to select “CO2 reliable” injections are much more restrictive. In spite of the limitation that we have pointed out, this multi- injection test can be interpreted. Results obtained with the pump procedure are more scattered than with the over-pressure one. Nevertheless, considering the accuracy of the prescribed values and the reproducibility on the flask measurements, we show that the use of the injection pump don’t alter significantly the analysis of CH4 and N2O. The different results observed on CO2 are more ambiguous. On one hand, the over-pressure procedure gives less scattered value than the pump procedure, which has already been noticed in the case of CH4 and N2O. But, on the other hand, the mean value is significantly lower than the prescribed one. Moreover, the CO2 mean deduced with the pump procedure is closer from this reference. A CO2 enrichment seams thus to be caused by the pump use. It may be explain by room air contamination through the pump membrane. The flasks filling procedure could also introduce a bias between the tank composition and what is measured into flasks. Any way, flasks are over-pressurized in most of cases; the pump is thus usually disconnected to avoid possible contamination of samples. Furthermore, we routinely, analyze two times each flask sample. As shown on the previous figure, there is obviously, no problem of representativity for these two injections, even on 1 liter flasks. Future developments of the experiment In the strictly experimental point of view, we plan to couple the existing GC with an other one (HP 6890) equipped with an FID. This additional detector should allow to detect CO in the same samples than the previous species (CO2, CH4, N2O and SF6). In practice, it means that we will design a third sample loop that will inject samples into the CO chromatographic column in parallel with the CO2/CH4 and N2O/SF6 ones. Comparing to the present situation, this modification of the GC experiment will provide a gain of time and of total used volume. Moreover, the main improvement of such CO detection would be the result's accuracy. Indeed, CO concentration in flasks and ambient air was formerly measured with a Reduction Gas Analyzer (RGA3, TraceAnalytical) with an accuracy of about 3%. With this new system we expect to reach an absolute accuracy of at least ±1 ppb. Moreover this development of the greenhouse gases monitoring station will allow to analyze also CO in Saclay's atmosphere and then to calculate easily the CO/CO2 ratio, which could provide important information on CO2 sources [Potosnak, 1999].


17

PART II: APPLICATIONS Flask samples analysis As already mentioned we routinely analyze two times every flask sample. The histograms shown in Figure 7 present the difference between replicate injections (∆inj.) performed on flask samples analyzed since Mai 2001.

Figure 7: Replicate differences for CO2, CH4 and N2O, depending on sampling site. The solid lines represent instrumental accuracy (sum of the standards reproducibility, see Table3).

For CO2, about 64% of the replicate injections are less different than ±0.075 ppm. Considering CH4 and N2O respectively, we notice that 62.6% and 69.3% of replicate injections lay within the range of ±0.75 ppb. Moreover, this figure shows that, in the case of CO2 analysis, the experimental source of uncertainty (sampling, aging process…) is the major contributor of the total uncertainty, whereas the instrumental accuracy can explain most of the ∆inj. for CH4 and N2O analysis.

0

50

100

150

200

250

Rep

licat

es

AMSBDXBGUBZHCGOCTRGPCGRI

0

50

100

150

200

250

Rep

licat

es

HLEHNGMDF

MHDNCAORLPDMTRMTVR

0

50

100

150

200

250

Rep

licat

es

-0.125-0.075

-0.0250.025

0.0750.125

-1.25-0.75

-0.250.25

0.751.25

-1.25-0.75

-0.250.25

0.751.25

3.9%6.2%

22.2%

34.3%

14.3%

6.5%

12.6%

8.1%11.3%

21.2%23.6%

20%

8.4%7.4%5.1%

8.7%

19.6%

27.3%23.5%

10.7%5.1%

CO2

CH4 N2O

∆CH4 (ppb) ∆N2O (ppb)

∆CO2 (ppm)


18

In many cases, flasks are sampled into pair, to control the quality of the sampling and measurement procedures. In most of them, the pair difference lies into the range of uncertainty due to the reproducibility of each standard used for the calculation. If not, two different cases can occur: one of the flasks exhibit obviously aberrant values, it is then rejected and we only consider the mean of the two injections on the other flask; if we are not able to discriminate which of the two flasks is correct, it can append that the entire pair is rejected. The Figure 8 is an example of the quality insurance on flasks analysis for the CO2. It concerns the Mace Head monitoring station (Ireland ), which is part of the RAMCES network. In situ measurements are continuous ly performed (Siemens Ultramat, accuracy: ±0.1 ppm, [Biraud et al., 2000]) and pair flasks are sampled in certain conditions of wind speed and direction (to catch only long distance influence).

Figure 8: Comparison of CO2 continuously recorded at Mace Head Station (Ireland) and Flasks analyzed on the GC experiment.

Studying the different measurements, we show that, concerning the period from September to November 2000, the reproducibility on flask analysis is better than ±0.070 ppm, the mean pair difference for the valid measurements is ±0.15 ppm and the discrepancy between flask (mean value of the two replicate injections) and continuous measurements is on average 0.095 ±0.40 ppm. Note that the transit time in the two different sampling lines is not taking into account.

Mace Head CO2

355

360

365

370

375

380

385

390

8-sept-2000 22-sept-2000 6-oct-2000 20-oct-2000 3-nov-2000 17-nov-2000

ContinuousMeas.

Mean PairFlask

IndividualFlask Inj.


19

Typical record of semi continuous measurement of ambient air in Saclay We have already shown a few results of flasks analysis, but the GC is also designed to sample ambient air. The air inlet is located on the roof of the institute in Saclay at about 12m high. On the map presented in Figure 9, the environment of the lab is depicted.

Figure 9: Map of the region of Saclay. Black lines : highways; green zones: forests and gray zones: urban areas.

Ambient air is continuously pumped through a decakon line at a flow rate of about 5 l/min. The surplus volume is released in the room, before drying the air into a large (1 liter) glass trap immersed in a cold alcohol bath maintained near –50°C by a cryocooler (see Figure 1). Our strategy is to mix ambient air and flask samples analysis during days and nights measurements. During the day, we usually analyze 10 flasks (i.e. 20 injections), which means that 7 ambient air injections are recorded every 2 hours. During night time measurements, 12 flasks are analyzed, i.e. 11 injections of Saclay’s air every 2 hours. Moreover, during the week end, only ambient samples are measured which provide quasi continuous record (15 injections per 2 hours) of Saclay’s concentration of greenhouse gases for these periods.

5 km

LSCE

Paris


20

Figure 10: CO2, CH4 and N2O recorded in Saclay’s ambient air from June to December 2001.

19-juin-01 19-juil.-01 18-août-01 17-sept.-01 17-oct.-01 16-nov.-01 16-déc.-01

360

400

440

ppm

2000

2500

3000

3500

ppb

300

320

340

360

ppb

CO2

CH4

N2O


21

Figure 10 reproduces CO2, CH4 and N2O signals recorded since Mai 2001 in Saclay. Gaps into data correspond to test periods on the GC or calibration sequences. In the case of N2O, we don’t have any measurements from the 4th September to the 29 of October 2001. During this period, the EDC has been stopped because of a lack of carrier gas. In the cases of CO2 and CH4, synoptic variations come out. We plot, in Figure 11, the monthly mean diurnal cycles of these two species.

Figure 11: Month by month mean diurnal cycle of CO2 and CH4, from June to December 2001. For both, we show the seasonality of the diurnal concentration variations. Indeed, we observe an increase of the mid afternoon (from noon to 5 PM) minimal level from summer to winter. It is even more obvious for CO2 concentration, the mean level for this species being 362.4 ±3.9 ppm in August and 390.5 ±10.9 ppm in December. Both sources and sinks of CO2 can vary seasonally. The heating systems release CO2 mainly during the winter, in the Ile de France region the contribution of private and services combustions contribute for about 42% of the global CO2 emissions [CITEPA, 2000]. Furthermore, the biospheric CO2 sink, via the photosynthesis process that stocks carbon in plants, is strongly reduced during winter. Moreover, the meteorological situation during this season (i.e. lowering of the mixing layer height) can also favour the accumulation of greenhouse gases, and other pollutants, in the lower atmosphere where they are

0 2 4 6 8 10 12 14 16 18 20 22 24360

365

370

375

380

385

390

395

400

405

ppm

JuneJulyAugustSeptemberOctoberNovemberDecember

0 2 4 6 8 10 12 14 16 18 20 22 24

1850

1900

1950

2000

2050

2100

ppb

CO2 CH4


22

emitted. These processes going all into the direction of an increase of the background concentration of CO2 during winter months. In the case of CH4, we also observe a seasonal change of the mid afternoon level, but that seams less significant than for CO2: 1843.2 ±32.1 ppb in August and 1932.5 ±83.9 ppb in December. Indeed, for an industrial region of the Northern Hemisphere, like Saclay’s area, the mains CH4 sources are anthropogenic and not related to season except the natural gas consumption. The seasonal increase of the CH4 daily minima can then be attributed to the mixing layer height decrease during winter. Standard deviations associated to monthly averaged values are around ±10 ppm for CO2 and ±70 ppb for CH4, but are highly variable. We will study more precisely these diurnal cycles in regard to air masses origin. Note that N2O diurnal cycle is not clear enough to be calculate monthly mean, we have thus to improve our N2O accuracy before studying in more details this greenhouse gas diurnal and seasonal cycles.

Figure 12: Typical CO2, CH4 and N2O diurnal variations in Saclay’s atmosphere during August. the violet solid line for the N2O, plots the running average on 11 points.

9-août 10-août 11-août 12-août 13-août 14-août 15-août 16-août

340

360

380

400

420

440

460

ppm

1800

2000

2200

2400

ppb

312

316

320

324

328

ppb

CO2

CH4

N2O


23

On the Figure 12, we zoom to the period between the 9th and 16th of August. It points out the diurnal cycles of each species. It appears that CO2, CH4 and, less clearly, N2O are accumulating in the atmosphere from the late afternoon (6 PM U.T., which corresponds to 8 P.M. local time during summer) until 6 AM (U.T.). The greenhouse gases concentrations increase are allowed by the nighttime temperature inversion that isolates the surface layer, where pollutants are emitted, from vertical mixing. Then, the inversion breakdown as the sun rises and greenhouse gases concentrations drop down rapidly. In the case of CO2, the assimilation sink takes also place during daytime and contributes also to the decrease of CO2 level within the surface layer. We notice also that CO2 and CH4 diurnal variations are highly correlated. Emission rates estimation As an illustration of the GC result's interpretation for ambient air, we present here an estimation of the CO2 and CH4 emission rates for the 4 to 7 September 2001. This calculations have been based on the radon tracer method, i.e. we have compared the concentration variations of greenhouse gases and the activity of the isotope 222 of the radon gas (222Rn) on an hourly basis. It supposes that 222Rn is spatially and temporally homogeneously emitted [Lambert et al., 1982] and that sources of 222Rn and trace gases are co-located. 222Rn activity (Bq/m3) is determined from the measurements of it’s short- lived daughters (218Po, 214Bi, 214Pb) [Polian et al., 1986] produced in the atmosphere by disintegrations and quickly adsorbed onto sub-micron sized aerosol particles. 222Rn is measured with the active deposit method i.e. aerosols are accumulated on a paper filter during 1or 2 hours by pumping ambient air at a flow rate of about 8 m3/h (in the case of radon experiment the air inlet height is only 5 m). The filter is then placed under an alpha spectrometer composed by a scintillator plus a photomultiplier, in order to measure the radioactive decay of the filter for 1or 2 hours depending on the pumping time. The statistical error associated with this method is inversely proportional to the square of the number of disintegrations recorded. Comparison between different sites and sampling procedures indicate that the absolute error made on 222Rn peaks is about 20% [Biraud et al., 2000]. For the chosen period, we dispose of good quality's records of CO2, CH4 and 222Rn but, unfortunately, of no N2O data at this time. Records used to estimate the CO2 and CH4 emission fluxes are shown on Figure 13.


24

Figure 13: 222Rn, CO2 and CH4 variations from the 3 to the 8 of September 2001. First, we determine the accumulation period of the greenhouse gases into the atmosphere. During summer it typically takes place from sunset, when convective mixing stops, to sunrise (i.e. 6 PM to 6 AM). Assuming that the mixed layer has a constant height ( H ) during nighttime accumulation, we can approximate the rate of change of the trace gas concentration ( )ttCx ∆∆ )( by:

Hj

ttC xx =

∆∆ )(

(4a)

xj being the constant emission rate of x at which the trace gas is released in the atmosphere. Applying this equation, assumes that there is no mixing of air from the free troposphere into this stable layer. In the case of 222Rn, which is a radioactive species, the radioactive decay has to be considered, leading to:

)()(

tCHj

ttC

ixx λ−=

∆∆

(4b)

According to Levin et al. [1999], we only consider events for which the correlation factor (r²) of the trace gas and 222Rn co-variations is higher than 0.5. For the selected events we eliminate the

0

1000

2000

3000

4000

mB

q/m

3

3-sept.-01 4-sept.-01 5-sept.-01 6-sept.-01 7-sept.-01 8-sept.-01

360

370

380

390

400

410

ppm

1800

1850

1900

1950

ppb

222Rn

CO2

CH4


25

unknown mixing layer height, combining equations 4a and 4b (considering that λC222<< ∆C222/∆t):

∆∆

−∆∆

=tC

CCC

jj xx

222

222222

222222 1

λ

(5)

222j is taken to be 76.5 Bq/m3, which is a mean value for France [Eckhardt, 1990]. The correction factor (into brackets) that takes into account the radioactive decay of 222Rn offset by fresh emissions from soils reduces the flux estimation by about 3% [Schmidt et al., 2001]. Emission rates of CO2 and CH4 estimated for the studied period are presented in Table 7.

CO2/222Rn correlation

jCO2 (103 kgC/km²/day)

CH4/222Rn correlation

jCH4 (kgCH4

/km²/day)

3-4 sept. x= 2.84 10-3 y +367.877

r²= 0.096 x= 12.4 10-3 y +1853.55

r²= 0.47

4-5 sept. x= 13.42 10-3 y +357.837

r²= 0.99 12.8 x= 19.15 10-3 y +1836.49

r²= 0.86 21.3

5-6 sept. x= 8.49 10-3 y +365.835

r²= 0.65 8.1 x= 24.58 10-3 y +1843.41

r²= 0.64 27.4

6-7 sept. x= 11.4 10-3 y +359.533

r²= 0.98 10.9 x= 26.96 10-3 y +1809.62

r²= 0.89 30

Table 7: Emission rates of CO2 and CH4, estimated with the radon tracer method from the 4th to the 7th of September 2001.

The results are comparable to previous European stud ies [Biraud et al., 2000; Levin et al., 1999], but the uncertainty on the fluxes derived from this method remains important: about 30% according to Biraud et al. [2000]. The main source of uncertainty comes from the hypothesis on the homogeneity of radon emission. Indeed, j222 vary, depending on the soil texture, its water content…[Eckhardt, 1990; Kataoka et al., 2001 and references herein]. To improve our fluxes estimation we plan to make field measurements of direct 222Rn gas with flux chambers in the catchment area of the Saclay station. Nevertheless, as the radon tracer approach is independent of inventories, we can compare fluxes derived by both methods. For example, the CITEPA (Centre Interprofessionel Technique d’Etudes de la Pollution Atmosphérique, France) [2000] gave for the Département de l’Essonne in 1995 net emissions of CO2 and CH4 of about 6.06 109 kgCO2 and 19.31 106 kgCH4 respectively. Considering the surface of this region, we can approximate emission rates of 2.5 103 kgC/day/km² for CO2 and 29.31 kgCH4/day/km² for CH4. For the CH4 emission rates, our calculation fits very well with the statistic estimation of the CITEPA for 1995. The emission flux of CO2 that we deduce from the radon tracer method is nevertheless approximately 5 times higher than the one estimated from the CITEPA mean emission for Essonne in 1995. Part of this difference is probably due to the respiration flux that is not taken into account in the CITEPA


26

inventories. Indeed, at night, plant plus soil respiration add up CO2 to the nocturnal boundary layer. An other way to improve our greenhouse gases flux estimates is to take an other species than radon, for which the mean emission rate is more accurately known, as atmospheric transport tracer. SF6 can be considered as such a species, if there are not too much punctual sources of it near Saclay. According to CITEPA [2000], the emission rate of SF6 for France in 1999 was 101 tons, it represent a mean emission rate of about 0.18 gSF6/km² . With our new working standards, containing SF6, we would then measure the transport tracer and the greenhouse gases, for which we want to estimate the emission rates, in the same samples. This method has, of course, to be validate by comparison either with statistical inventories than with estimations deduced by the radon tracer method. References Bader, Gaschromatographische Messungen von atmosphärischem CH4, CO2, N2O und SF6 am

Schauinsland und an der Zugspitze, Univ. Heidelberg, Heidelberg, Germany; 2001. Bakwin, P.S, D.F. Hurst, P.P. Tans and J.W. Elkins, Anthropogenic sources of halocarbons,

sulfur hexafluoride, carbon monoxide, and methane in the southeastern United States, J. Geophys. Res., 102 (D13), 15915-15925, 1997.

Biraud, S. P. Ciais, M. Ramonet, P. Simmonds, V. Kazan, P. Monfray, T.G. Spain and S.G. Jennings, European greenhouse gas emission estimates from continuous atmospheric measurement at Mace Head, Ireland, J. Geophys. Res., 105 (D1), 1351-1366, 2000.

Centre Interprofessionnel Technique d’Etude de la Pollution Atmosphérique (CITEPA), Inventaire es émissions des gaz à effet de serre en France au cours de la période 1990-1999, CITEPA 430, Paris, 2000.

Eckhardt, K., Messung des Radonflusses und seiner Abhängigkeit von der Bodenbedchaffenheit, Univ. Heidelberg, Heidelberg, Germany; 1990.

Intergovernmental Panel on Climate Change (IPCC), Climate Change 2001: The Scientific Basis, Cambridge Univ. Press, New York, 2001.

Kataoka, T. et al., A study of the atmospheric boundary layer using radon and air pollutants as tracers, Boundary-Layer Meteorology, 101, 131-155, 2001.

Kuhlmann, A.J., D.E.J. Worthy, N.B.A. Trivett and I. Levin, Methane emissions from wetland region within the Hudson Bay Lowland: An atmospheric approach, J. Geophys. Res., 103 (D13), 16009-16016, 1998.

Lambert, G. G. Polian, J. Sanak, B. Ardouin, A. Buisson, A. Jegou and J.C. Le Roulley, Cycle du radon et ses descendants : application à l’étude des échanges troposphère-stratosphèere, Ann. Geophys., 38, 497-531, 1982.


27

Levin, I., H. Glatzel-Mattheir, T. Marik, M. Cuntz, M. Schmidt and D.E.J. Worthy, Verification of methane emission inventories and their recent changes based on atmospheric observations, J. Geophys. Res., 104 (D3), 3447-3456, 1999.

Polian, G., G. Lambert, B. Ardouin and A. Jegou, Long range transport of continental radon in subantarctic and Antarctic areas, Tellus, Ser. B, 38, 178-189, 1986.

Potosnak, M.J., S.C. Wofsy, A.S. denning, T.J. Conway, J.W. Munger and D.H. Barnes, Influence of biotic exchange and combustion sources on atmospheric CO2 concentrations in New England from observations at a forest flux tower, J. Geophys. Res., 104 (D8), 9561-9569, 1999.

Schmidt M., H. Glatzel-Mattheir, H. Sartorius, D.E. Worthy and I. Levin, Western European N2O emissions: A top-down approach based on atmospheric observations, J. Geophys. Res.106 (D6), 5507-5516, 2001.

Schmidt M., R. Graul, H. Sartorius and I. Levin, Carbon dioxide and methane in continental Europe: a climatology, and 222Radon-based emission estimates, Tellus Ser. B, 48, 457-473, 1996.

Worthy D.E.J., I. Levin, N.B.A. Trivett, A.J. Kuhlmann, J.F. Hopper and M.K. Ernst, Seven years of continuous methane observations at a remote boreal site in Ontario, Canada, J. Geophys. Res.103 (D13), 15995-16007, 1998.

�� !��"$#��

%'&)(+*-,/.+*102( 34*6587:9202;25)&8,<7=.>&@?BA�C D�EGF2HID�JLKNMOEGPRQ2% &)(+*-,/.>*<02( 32*S5UT VXWZY4[

\^]6_`bac`bdZe4fNe4gRhidj�`lk�mon \qpsrut v>wZx4y{z=|=}�~/|-��}��i�� [�T V�A��X.>&�.>04.qW�&)*<(>(+*<��YG&)�:?OA�[b7O92587O,�*{��VXWZY4[N��*-�X.�02A2*��*13 �*1(>7�.+&)?OA�34*'(+*-,��2*<(�,��2*��c02&�(+*1�O(>?O029�* �+&��:5)7O;L?B(>7�.+?B&@(>*1�*1A�( �*1�O&)?OA � (�7�AL,�&)5@&)*<A2A�*:��i�sW�Q4[b��qQL[�Y�� Q2[b��% Q[�¡S%£¢��'Q�Y4¤'��¥

�^¦¨§ }£v/©�ª�«U�¬«U4«¨w��}<~/~+}<®�xL�¬¯�z=©uyb°�±�²@°B³>}!óv>wZx4y¨µ6�8ª'°Z¶�}�·=®}<~+°�«¸�¬|=}¹~+}�ª<}>°�~+³ § �¬©�ª/«¸�¬«U4«�}^« § °�« ºB°�« § }�~�ª»ª��¼½²@°O¾�z=~>°=«�z=®~/��}�ª'�¬©�« § } w�°=~/�8ª6°�~+}�°¿´�À�Á ¦ w�Âby�Ã�ÀNÄ�Â�yNxNÀ�Á�Âby�ÄÆÅ�ÂysÇ�Å�ÈiÀNÂ�x4É£µ�Ê� [iT VXW�Y4[�*1�R.s�+?O0L�Z587 .+04.>*<5)5@*$,<?OA=ËX?B&@Ac.>*S340»�Z*1AB.>(+*�Ì'7=�.>&@?BA�7�5£34*¿5)7^Ís*1,��*<(�,��2*�Y4,<&@*1AB.>&�Î��B0�*OQi34*-� Ï A2&)ÐO*<(��+&�. �*1�Wi&@*1(+(>*£*�.i��7O(+&)*'�Z02(>&@*s*<.�ÑÒ*<(��+7O&@5)5)*1��Y47O&@Ac.R�RÓ'02*1AB.>&@AbQB340�Z?B�:�!&8�>�+7O(+&87=.$Ô7:5¸T ��A�*<(>�O&)*S¤�.>?O�:&)�c02*BQ�34*65UT V�A��R.+&@.+02.'34*Í�*1,��2*1(>,��2*�9�?O02(�5)*�% �*<ÐO*15@?B929�*<�:*<Ac.1Q�34*�5UT �Z,�?O5)*�Ì�?O(+��7O5@*ÕYG0�9 �*<(>&)*<02(>*'*<.�34*�5¸T �i,<?O5)*�W�?O5)ÖB.>*1,��2A�&)�c02*B¥

� v>wZx4y×�8ª 4©�·O}<~Ø« § }�Ù<z=�¬©�«Ú«¸4«�z=~�ª § � ±ÛzR¶ÜÀÞÝ'ßSx´�à�~+°=©¨³�}�á ª�¯�°�Ù<z�~�¾�°�ª/��³<®�~+}/ª<}�°=~>³ § z�~¸ºB°�©��@â-°=«U��z=©Oµ�Â»ÀbÁ�É´�à�~+°=©¨³�}�á ªã°=«�z=¯!��³ä}<©�}<~¸ºOåæ~+}/ª<}�°=~>³ § ³>}<©�«�}<~�µ-Â�v+ß�Å´�à�~+°=©¨³�}�á ª�³�z1z�±2}<~+°=«U�¬|=}ç~>}/ª<}�°=~+³ § °�©�·×·O}<|�}<²@z>±¨¯:}<©�«°1ºB}<©�³<å�µè°�©�·�àÞ~>°=©�³�}Lá ª»¶�z=4~é²@}�°O·=�¬©Gºê�¬©2ª/«U�¬«¸4«U��z=©�ªézX¶§ ��º § }�~q²)}>°�~/©��¬©Gº{�¬©ë« § }ìª<³��}<©�³�}�ª<�îíN©��¬|=}�~�ª/�¬«Uåìw£��}�~/~+}}<«�Ä»°�~/��}$À�4~/��}�Â í�©L�¬|�}<~�ª/�¬«¸å�ïL}<~�ª<°=�¬²¬²@}�ªsx¨°��¬©�«U®/ð£�}�©L«¸�¬©�ÂÁ�³>z�²@}SÝÕz=~/¯�°=²)}$x�-± ñ}<~/��}�4~>} °�©�· Á�³>z�²@}�wsz=²�å=«�}�³ § ©��ò<�}-Ê

� [�T VXWZY2[Ú(>*<�:925)&�.�02A�*^.>(+&)925)*^�:&8�+�+&)?OAé32*�(>*1,��*<(�,��2*OQ3�T *1A��R*1&@�BA2*<�:*1AB.S*�.634* �+*<(>ÐG&),<*Õ3ÞT ?O;��+*<(>Ð=7=.+&)?OAb¥ [�T �*�.>0�34*32*1�634&@ó �*1(+*1Ac.+*1�S,�?B�:9L?c�+7OAB.>*1�'34*:5¸T *<AGÐG&@(>?OA�A2*<�:*<Ac.'.+*<(+�(>*1�R.+(>*Ú��?G, �*17OA�QÆ7�.+�:?B�+92��Ô*1(+*BQu;2&)?B�+92��Ô*1(+*BQÆ,�(>ÖO?c�R9��ÞÔ*<(>*OQ�+02( � 7O,<*1�I,�?BAB.>&@A�*<Ac.>7O5@*-�>�ô,�?BA��R.+&@.+02*l5UT ?B;4ËX*1,�.+& � ,�*1Ac.+(�7�532*�(>*1,��2*1(>,��*�34*¿5¸T VXWZY2[�¥i�Z*�.+.+* �*�.>0�34*ÆÐ=7�34*¿5¸T �*1,��*<5)5@*5)?4,<7O5@*éÔ7{5¸T �*1,��*<5)5@*q�B5@?B;�7�5)*OQ�*15@5)*õ,<?OA�,<*<(>A2*q5¸T �*1ÐO?O5)04.>&@?BA9L7O�>� �*1*ö*<. � 02.+02(>*�34*l5)7è92587�A�Ô*�.+*l��*<(>(+*BQ�5¸T �*�.>0�34*l34*5UT *<AGÐG&@(>?OA2A�*<�:*<Ac.Æ&)?OA2&8� �*�34*�587ì��*<(>(+*�*<.¿,�*15@5)*»34*-�u*1A4�ÐG&)(+?BA2A2*1�!*1Ac.>�'925)7OA �*<.>7�&)(>*1�1¥ ��5)5@*!�R* � ?OA�32* �+02('0�A2*!7O94�9�(+?4,��2*�&)A�,�5)0�7OAB. 3 �*<ÐB*<5)?O929�*<�:*<Ac.��Z*<�49 �*<(>&)�!*1Ac.>7�02�¨Q4?B;4��+*<(>Ð=7=.>&@?BA�*<.��!?43 �*<5)&)�>7=.>&@?BA�¥

��¦¨§ }�¯!�8ª�ª��z�©2ª»zX¶uv>wZx4yê�¬©�³<²÷�·O}»~+}�ª�}�°�~+³ § Â «�}�°O³ § �¬©cº°�©�·!ª<³<��}�©L«¸� ø�³�¯�z=©L�¬«�z�~/�¬©cºcÊ ¦¨§ }$~+}�ª<}�°=~+³ § ±¨~+z+ºO~+°�¯ ¯�}/ª³�z�©�·=�³<«�}>· �Ò�¬« § �¬©�« § }Òv/©�ª/«¸�¬«U4«�}i�¬©�³<²÷�·B}�« § }Zª�«U�·=å6zX¶i« § }¯�°=�¬©ù³>z�¯�±�z=©¨}�©L«�ª�zX¶�« § }�Ás°�~/« § á ª�}<©�|-�¬~+z�©�¯�}�©L«Z¶�~+z�¯« § }!²)z1³�°=²N«�zÆ« § }�ºO²@zB¾>°�²Nª�³�°�²@}�´Xz1³�}�°=©�Âs°�«¸¯�z�ªU± § }<~+}/Â ¾<��z=®ªU± § }<~+}/Â ³<~/åOz�ªU± § }�~>}/Â'³�z=©L«¸�¬©¨}�©L«�°�²bª/4~o¶�°B³>}�ª�µOÊ ¦¨§ }�ª�}�~+}�®ª<}�°=~>³ § ³�z=©¨³>}<~/©I« § }Õ±2°=ª/«Õ°=©�·:¶�4«U4~+}�}<|�z�²÷4«¸��z�©{zX¶�« § }±¨²@°=©¨}�«�Ás°�~/« § ÂN« § }�ª/«¸�·�åÕzX¶s« § } ��z=©L�8ª�}�·:}<©�|-�¬~+z=©L¯:}<©�«ÒzX¶« § }�Ás°�~/« § °=©¨·�zX¶�±¨²@°=©¨}�«�°=~/å�°=«U¯:z=ª¸± § }<~+}�ªÕ�¬©^« § }Õª<z=²)°=~ª/å�ª/«�}�¯�Ê ¦¨§ }�ª<}¿ª<³��}<©�«U� øZ³�°B³�«U�¬|1�¬«U��}�ª�°=~>}¿¾�°=ª�}�·¹z=©½}>¼O®±�}<~/�¬¯:}<©�«�°=²�·B}�|=}�²)z>±�¯�}<©�«¸ª>Â�zO¾�ª�}<~/|=°=«U��z=©¿°�©�·6¯�z-·O}<²¬²÷�¬©cºcÊ

� [iT VXW�Y4[�*<.��R*-�Þ587�;�?O(�7=.+?B&@(>*1�¨�+?OAc.�(�7=.R.�7O,�� *1�¨7�04�6�i,<?O5)*1�%'?4,/.>?O(�7�5)*1�NúBY4,<&@*1A�,�*-��34*�5UT �iAcÐG&)(+?BA2A2*1�!*1Ac.>ús*<.ÒúB¤ �R.+(>?��9��cÖ4�+&)�c02*-ú:3ÞT V�5)*��34*<��û�(>7OA�,�*B¥

��¦¨§ }Zv/©�ª�«U�¬«U4«2w£��}�~/~+}<®�xL�¬¯�z=©:yb°�±�²@°B³>}'°=©¨·6�¬«�ªZ²@°O¾�z=~>°=«�z=®~/��}�ª£°�~+}�±�°=~/«bzR¶Z« § }Sü£~+°B·=�°�«�}sx¨³ § z-z=²ýª$þ+Á�©L|1�¬~>z=©�¯�}<©�«�°=²xÞ³��}<©�³�}�ªBþ$°=©¨·�þ>ÉSª/«U~+z>± § å�ª/��³�ªOþ zX¶�v�²@}<®R·O}<®¸àÞ~+°�©�³�}-ÊVXW�Y4[^�ZÏ WZ�^�ö��7B�R*:ÿ�� VXW�Y4[^�iÏ�Ñ6Y�Ó�LQ�W�587O,<*��B0��>�R&)*<0 ��2QG(>02*$340�Ís* � 0��O*��$WÒ7O(+&8��Z*134*<��Zû�(>7OA�,�* � ��ÑÒ*1(>�>7�&)5)5@*-�i��û2(�7�A�,<*

§ «¸«�±Þ� ��L�Ò�Ò�iÊ)� ±Lª/²@Ê Ù/Gª�ª/��}<�Ê ¶�~

�ZF��iJ��=F��1J��=E�� Z*1AB.>(+*63ÞT �Ò.>0�34*1�s32*1�s��AGÐG&)(+?BA2A2*<�:*1AB.��Z��*<(>(>*1�R.+(>* *�. Wi5)7OA �*<.>7�&)(>*1��ý��i�sW�� c.R.>9�C! � #"$"$"�¥ ,�*<.+9�¥ &@9L�R5U¥ � (� [�7�;�?O(�7=.+?B&@(>*634*��Z&)?4,��2&)�!&)*�*�.S�Z��&@�:&)*$��7�(>&@A�*1�%� [��&�� [�7�;�?O(�7=.+?B&@(>*634*1��Y4,<&@*1A�,�*-��340��Z5@&)��7=.�*<. 34*�5¸T ��AGÐG&)(+?BA2A2*<�:*1AB.'� [�Y�� c.R.>9�C! � #"$"$"£��5)�>,�*B¥ ,<*17�¥ � (� [�7�;�?O(�7=.+?B&@(>*634*�� *�. �*<?B(+?B5@?B�O&)* %'ÖcAL7��:&)�c02*(� [b�»%)� �c.R.>9�C! � #"$"$"�¥ 5@��3Þ¥ ËX0��>�R&)*<0b¥ � (� [�7�;�?O(�7=.>?O&)(+*�3ÞTý¡S, �*17OA2?O�B(>7O92�2&)*i%'ÖcAL7��:&)�c02*s*<.i34*'�Z5)&@��7=.>?O5)?O�B&@*'� [N¡S%£¢$�*� �c.R.>9�C! � #"$"$"�¥ 5@?434Ö4,O¥ ËX0��>�+&@*10�¥ � (� YG*1(+ÐG&8,�*�3ÞT ¤ �*1(+?BA2?O�:&)*+� Y4¤)� �c.R.>9�C! � #"$"$"�¥ 7�*1(+?L¥ ËX0��>�+&@*10�¥ � (

Documents

o 9 ? D 9 L