Tropospheric NO2 from satellites


AT2

Tropospheric NO2 measured by satellites
10-12 Sept 2007

KNMI, De Bilt, The Netherlands

Report
Workshop report pdf (2.2 MB)

Presentations and abstracts

Download all contributions
(~235 MB - zip format)

Below you will find all contributions of the workshop on 'Tropospheric NO2 measured by satellites' arranged by session in ppt and pdf format. Click on titles to view abstracts.

List of Participants

Session: Retrieval - Validation - Modelling - Applications - Posters

Session Retrieval - Chairs: James Gleason & Eric Bucsela

	INVITED TALK: Tropospheric NO2 from space : retrieval issues and perspectives for the future Michel Van Roozendael Tropospheric NO2 columns have been derived for more than 10 years based on spectroscopic measurements from GOME on ERS-2, SCIAMACHY on ENVISAT and OMI on EOS-AURA. Owing to its strong and highly structured absorption features in the UV and visible regions, NO2 can be measured with high signal to noise ratio so that localized emissions can be easily detected providing a wealth of information on atmospheric events, processes and their evolution with time. Key applications of the recorded data sets include assimilation in air quality models, attribution of source by inverse modeling techniques, identification of long-range transport events and global or regional trend studies. For such quantitative applications, the assessment of the accuracy, stability and cross-platform consistency of the NO2 data sets are of major importance. In this presentation, I will review some of the main issues related to NO2 retrieval from space, based on the experience from GOME, SCIAMACHY and OMI. This will cover spectral fitting using DOAS, elimination of stratospheric background, evaluation of tropospheric AMFs and how to deal with clouds. Approaches currently adopted by different retrieval groups will be contrasted and discussed in terms of their respective strengths and weaknesses. Finally challenges for future tropospheric NO2 retrievals will be addressed in relation with the need for improved spatial and temporal resolutions.
	Direct fitting of NO2 from GOME-1, GOME-2, SCIAMACHY, and OMI Kelly Chance, T.P. Kurosu, R.V. Martin, T. Beck, S. Kondragunta We have implemented direct fitting of radiance spectra from the GOME-1, GOME-2, SCIAMACHY, and OMI satellite spectrometers. We are using these instruments in a study to attempt to analyze the spectra in as identical a fashion as possible, with the most complete treatment of underlying algorithm physics, in order to separate instrumental, algorithmic, and temporal differences in satellite measurements. Initial efforts, presented here, concentrate on NO2.
	Quantitative retrievals of NO2 from GOME Lara Gunn, Martyn Chipperfield, Richard Siddans, Brian Kerridge The direct measurement of tropospheric trace gases from space is difficult and so it is advantangeous to combine models and data to fully exploit the potential of observations. We present a new procedure for the quantitative determination of tropospheric total column amounts of trace gases from nadir viewing satellites such as the Global Ozone Monitoring Experiment (GOME) on-board the European Research Satellite 2 (ERS-2). The procedure constrains the stratosphere by assimilating chemical observations from the Halogen Occultation Experiment (HALOE), which flew on the Upper Atmosphere Research Satellite (UARS), into the SLIMCAT/TOMCAT three-dimensional (3-D) chemical transport model (CTM). The chemical data assimilation method is performed using the sequential sub-optimal Kalman filter scheme and here we assimilate HALOE CH4, H2O, O3 and HCl. The assimilation scheme preserves tracer correlations and the overall effect is to produce a more realistic stratosphere in the CTM. Using the stratospheric constraints from the CTM we then calculate tropospheric residuals. In the first instance we have applied this method to GOME observations of NO2. Retrieved slant columns are usually converted to vertical column by an air mass factor. The calculated air mass factor is sensitive to surface albedo, cloud fraction and height and aerosol properties. Our method uses data on cloud and aerosol from the Global Retrieval of ATSR cloud Parameters and Evaluation (GRAPE) project and surface albedo is retrieved from GOME. This data in conjunction with a multiple-scattering radiative transfer model, derived from GOMETRAN, will enable a more accurate air mass factor calculation. We will show quantitative results from the GOME period of derived tropospheric abundance of NO2. We will discuss the advantages of this scheme, and the improvements obtained by including the model stratosphere and retrieved aerosol and cloud properties.
	Impact of clouds on tropospheric NO2 retrieval Ping Wang, P. Stammes, R. van der A In general clouds have shielding, albedo and in-cloud absorption effects on trace gas retrievals. To correct for these effects the cloud fraction and cloud pressure are two important parameters, which have to be known.Thereto we have developed in recent years the O2 A-band cloud algorithms FRESCO and FRESCO+. The effective cloud fraction and cloud pressure retrieved from the O2 A-band using FRESCO(+) have been used in cloud corrections of ozone and NO2 retrievals from GOME, SCIAMACHY and GOME-2, by many users, e.g. in the TEMIS project. According to retrieval simulations using the radiative transfer model DAK we found that a Lambertian cloud with high albedo is a good assumption for cloud corrections of ozone and NO2. In FRESCO(+) the effective cloud fraction is the cloud fraction of a Lambertan surface with albedo 0.8 yielding the same radiance at the top of atmosphere (TOA) as the clouds in the scene. The tropospheric NO2 air mass factor differences between a scattering (Mie) cloud and a Lambertian cloud of albedo 0.8 are within 10% if the geometric cloud fraction is smaller than 0.2. If geometric cloud fraction is 0.5, the tropospheric NO2 air mass factor difference is within 20% except for the SZA larger than 70 degree and SZA equal to 0. Recently, FRESCO was improved by adding the single Rayleigh scattering. This new version is called FRESCO+. The difference in cloud pressure between FRESCO and FRESCO+ is significant for the less cloudy scenes, which are especially important for tropospheric NO2 retrieval. In this presentation we will show some improvements of using the FRESCO+ product in tropospheric NO2 retrievals.
	INVITED TALK: Reducing errors in using tropospheric NO2 columns observed from space Folkert Boersma In this presentation I will try to give a balanced discussion of the state-of-science with respect to the retrieval and use of tropospheric NO2 columns from nadir UV/VIS sensors. First I will give an overview of the basic retrieval method that is common amongst retrieval groups. Then I will discuss the different choices and assumptions made by various groups and show how these lead to considerable differences between the retrieved columns. Subsequently I will focus on a number of important issues for the future: (1) How can we minimize the errors in the various retrievals? Is it simply a question of optimizing the forward model parameters, or (2) Should we be moving towards a common retrieval approach? Such a convergence requires --amongst others-- an extensive validation dataset that is representative for a wide range of situations (spatially and temporally) thereby serving as an arbiter to find the best possible retrieval approach. (3) With the new generation of high-resolution sensors SCIAMACHY, OMI, and GOME-2, new types of retrieval and interpretation errors are becoming more important. These are resolution-related; for instance albedo maps and a priori profile shapes that are currently used in retrievals are available at spatial resolutions much coarser than the retrieval resolution. I will conclude the presentation by showing how tropospheric NO2 data sets from different instruments with different horizontal resolutions and overpass times can be combined.
	A new approach to eliminate the broadband absorption in DOAS spectra Milagros Ródenas, E. Soria, J.D. Martín DOAS data analysis is based on the separation of the broadband structure overlapped with the high frequency absorptions of the compounds to study. This way, the wavelength-varying contribution of Mie and Rayleigh scattering, as well as the low frequency of the compound itself, are removed. It also accounts for the effect of the detector etalon, inaccuracies in the instrument calibration or deficiencies in the knowledge of broadband absorption spectra (eg. O4). Software for analysis of DOAS data carry out a polynomial to model and remove these undesired broadband features. The process is mathematically equivalent to a high-pass filtering. In particular, a global polynomial models the low frequencies within all the evaluation range. In practice, the order of the polynomial is chosen depending on the width of the fitting window and on the width of the widest trace gas absorption feature to be fitted. Nevertheless, it doesn’t exist a general solution for choosing the right polynomial degree, and even though it posses a small effect in data retrieval, mean errors of 8% have been found (Aliwell et al., 2002). This study proposes a generalization of the classic approach mentioned above. Polynomial modelings are used to eliminate local broadband components. In other words, the proposal is basically a polynomial fit based on windows with a certain length. In the particular case of considering the whole length of interest, our algorithm becomes the usual procedure for DOAS analysis. We have benchmarked our approach with the classic one. A case study on a real problem, such as, the measurement of NO2 is presented. References: Aliwell, S.R. et. al. (2002). Analysis for BrO in zenith-sky spectra: An intercomparison exercise for analysis improvement J. Geophys. Research, V-107, NO-D14.
	Tropospheric NO2 derived by Limb-Nadir-Matching Andreas Heckel, A. Richter, A. Rozanov, J.P. Burrows One step in the retrieval of tropospheric NO2 from satellite measurements is to remove the stratospheric part of the total column. Good estimations of the stratospheric contribution are an important and critical point especially for conditions with lower tropospheric columns. Current algorithms use Nadir measurements plus either a priori assumptions on the horizontal stratospheric NO2 distribution or assimilation into stratospheric models. Depending on atmospheric conditions the uncertainties of the tropospheric vertical column introduced by the removal of the stratospheric column are in the order of 0.5 to 1e15 molec/cm2. In polluted regions with high tropospheric columns this error is negligible compared to the uncertainties introduced by the tropospheric AMF. However in regions with lower tropospheric NO2 values the accuracy of the removal of the stratospheric column becomes the dominating factor in the error budget. The SCIAMACHY instrument with the combined Limb and Nadir measurements offers the possibility to derive the stratospheric component independently. The Limb measurements are analyzed to derive a stratospheric profile of NO2. Integrating this profile gives the stratospheric column. The tropospheric column is then derived from the difference between the total column (Nadir) and the stratospheric column (Limb). This study will concentrate on the tropospheric NO2 columns derived from the SCIAMACHY data of 2005. The introduction of the method and comparisons to currently available tropospheric NO2 products and aircraft measurements will be presented. Improvements and limitations of this Limb-Nadir-Matching approach will be discussed based on different time scales such as single orbits, daily, monthly or annual averages.
	A combined retrieval, modelling and assimilation approach to estimate tropospheric NO2 from OMI measurements Henk Eskes, Ruud Dirksen, Ronald van der A, Folkert Boersma Spaceborne imaging spectrometers like OMI, SCIAMACHY, and GOME, provide on a regular basis high quality reflectance spectra of the Earth\'s atmosphere which allows for the detection of various species of atmospheric trace gases. This paper presents a method to retrieve troposheric nitrogen dioxide (NO2) concentrations from space. The retrieval consists of two steps; the first step uses the DOAS (Differential Optical Absorption Spectroscopy) approach to compute the total absorption optical thickness along the light path (the slant column). For OMI the DOAS was implemented in a KNMI/NASA collaboration. The second retrieval step, which was developed at KNMI, estimates the stratospheric column based on data assimilation and vertical profile estimates from the TM chemistry-transport model. In combination with the retrieved slant column, cloud fraction and cloud top height the tropospheric vertical column of NO2 is estimated. Because of its wide swath (115 degrees FOV) and high spectral resolution (13x24km^2 ground pixel size at nadir), the OMI instrument observes NO2 concentrations on an urban scale with global daily coverage. Our presentation will focus on aspects of the OMI retrieval, comparison with the operational NO2 product of OMI, and on the results of the latest OMI reprocessing of summer 2007.
	Synergistic use of multiple sensors for tropospheric NO2 measurements Andreas Richter, A. K. Heckel, J. Leitao, J. P. Burrows Tropospheric columns of NO2 can be retrieved from measurements of backscattered UV/vis light taken from satellite, e.g. from the GOME, SCIAMACHY and OMI instruments. These data have already been used to investigate NOx emissions and their spatial and temporal variability in a number of studies. In the retrieval process, a priori information on profile shape, stratospheric NO2, aerosol loading and other parameters is used that results in a relatively large uncertainty of the product derived. Therefore, for the use of data from multiple sensors e.g. for long-term time series one has to make sure that consistent retrieval assumptions are being used to avoid artificial changes in the time series. At the same time, measurements from different platforms taken at different times also enable a check on the consistency of the different time series and to a lesser extent also on the assumptions made. In this study, tropospheric NO2 data from SCIAMACHY, OMI, and GOME-2 is compared and analysed for their consistency and possible systematic differences.
	Development of Operational GOME-2 Tropospheric NO2 Product for Air Quality Applications Trevor Beck, Shobha Kondragunta, Kelly Chance, Thomas Kurosu, Lawrence E. Flynn NOAA/NESDIS is developing an operational tropospheric NO2 product from Metop-A GOME-2 instrument, flying in the morning orbit with a 9:30 AM equator crossing time, to meet air quality monitoring and forecasting requirements in the United States (U.S.). Initial focus is on the development and characterization of GOME-2 NO2 product with the end goal being assimilation of the data into the National Weather Service (NWS) air quality model to constrain NOx emissions and improve forecasting. Algorithm development work will be consistent with Aura/Ozone Mapping Instrument (OMI) flying in the afternoon orbit with 1:38 PM equator crossing time. Methodologies to convert slant column amounts to vertical column densities and to extract tropospheric amount from total column amount will be similar to OMI, so diurnal changes in NO2 can be tracked. NO2 product status and preliminary comparisons between GOME-2/OMI NO2 and NWS operational Community Multiscale Air Quality (CMAQ) predictions of NO2 on different spatial and temporal scales will be presented.
	First results on tropospheric NO2 from the GOME-2 instrument on MetOp Pieter Valks, N. Hao, D. Loyola, W. Zimmer The Global Ozone Monitoring Experiment-2 (GOME-2) is one of the new-generation European instruments carried on MetOp, which has been jointly established by ESA and EUMETSAT. GOME-2 will continue the long-term monitoring of atmospheric ozone and minor trace gases, started by GOME on ERS-2 and SCIAMACHY on Envisat. GOME-2 is a scanning spectrometer that measures the Earth’s backscattered radiance and extraterrestrial solar irradiance in the ultraviolet and visible part of the spectrum (240-790 nm), and contains two Polarisation Measurement Devices (PMDs). The advanced GOME-2 observes four times smaller ground pixels (80 x 40 km) than GOME on ERS-2, and provides a global coverage within about one day. The ozone and minor trace gas column retrieval algorithms for GOME-2 have been developed by DLR, in the framework of EUMETSAT´s Satellite Application Facility on Ozone and Atmospheric Chemistry Monitoring (O3M-SAF). In this contribution, we present the first results of the GOME-2 NO2 products, derived with the Differential Optical Absorption Spectroscopy (DOAS) method. The operational NO2 retrieval algorithm for GOME-2 is based on the GOME Data Processor (GDP) version 4.0, and includes a new algorithm for the retrieval of the tropospheric column density of NO2 for polluted conditions. For the calculation of the tropospheric NO2 column, an Air Mass Factor based on an assumed tropospheric NO2 profile is used, which has been derived from a MOZART-2 NO2 climatology. We present maps of total and tropospheric NO2 column densities. In addition, comparisons with other satellite data products of NO2, such as derived from GOME-1, SCIAMACHY and OMI will be shown.

Session validation - Chair: Pieternel Levelt

	A comparison of OMI with vertical columns constructed from in-situ measurements in Switzerland Dominik Brunner, Brigitte Buchmann The quality of tropospheric NO2 columns retrieved by OMI over Switzerland is assessed by comparison with in-situ observations of the Swiss air quality monitoring network NABEL. For each OMI pixel sampled above one of the regional NABEL stations Tänikon and Payerne a concurrent in-situ NO2 column is constructed by combining the measurements of stations located at altitudes between about 400 and 3500 m above sea level and taking into account actual PBL heights. Because NO2 is (mostly) measured with Molybdenum converters, known interferences with PAN and HNO3 are taken into account based on results of extended parallel measurements with both photolytic and Molybdenum converters. Furthermore, the ability of OMI to resolve regional air pollution structures in Switzerland is assessed by comparison with a high-resolution NOx emissions inventory and NO2 immission maps generated by a simple air pollution dispersion model.
	Relationship between ground-level NO2 concentrations and OMI NO2 column Lok Lamsal, Randall V. Martin, Edward Dunlea, Martin Steinbacher, Edward A. Celarier Recent studies have demonstrated that tropospheric NO2 columns are highly sensitive to lower tropospheric pollution. We present an approach to infer ground-level NO2 concentrations from tropospheric NO2 columns retrieved from OMI. The derived surface NO2 concentrations are compared with in situ NO2 data from 279 monitoring sites from the AQS/NAPS networks in the United States and Canada. The in situ NO2 measurements from commercial chemiluminescent NO2 analyzers equipped with molybdenum converter are corrected for interference from other reactive nitrogen species. The OMI derived ground-level NO2 data exhibit significant agreement with the in situ measurements after accounting for the interference. This comparison serves as an indirect validation of the tropospheric NO2 columns, and demonstrates the capability of monitoring surface air quality from space.
	NO2 lidar profiles measured during the DANDELIONS validation campaign 2006 Hester Volten, Ellen Brinksma, Stijn Berkhout, Daan Swart, René van der Hoff, Hans Bergwerff, Pieternel Levelt, Gaia Pinardi, Michel Van Roozendael DANDELIONS (Dutch Aerosol and Nitrogen Dioxide Experiments for validation of OMI and SCIAMACHY) is a project that encompasses validation of NO2 measurements by the Ozone Monitoring Instrument (OMI) and SCIAMACHY (Scanning Imaging Absorption SpectroMeter for Atmospheric CartographY), and of aerosol measurements by OMI and the Advanced Along-Track Scanning Radiometer (AATSR), using an extensive set of ground-based and balloon measurements over the polluted area of the Netherlands. Validation measurements were performed during two campaigns at the Cabauw Experimental Site for Atmospheric Research (CESAR, located in the centre of the Netherlands) which ran from May 8-July 14, 2005, and September 1-September 30, 2006, respectively. The campaign measurements were aimed at validation of the satellite products total and tropospheric nitrogen dioxide and ozone, and aerosol optical thickness. An extensive dataset of the obtained ground-based, balloon and satellite data on NO2, aerosols, and ozone, will be available for validation of Aura instruments through the Aura Validation Data Center (AVDC). Here we focus on the NO2 lidar data. For thorough validation in polluted areas, measurements of the NO2 profiles are required, since the profile shape directly influences the conversion from slant column to total NO2, through the air mass factor. There are only very few measurements of NO2 profiles available. The results of a novel instrument, namely the NO2 lidar developed at RIVM that is able to measure lower tropospheric profiles will be presented. The lidar system uses the DIAL technique to measure atmospheric trace gases. It is housed in a fully self-supporting mobile laboratory, so that measurements can be taken anywhere. The measured profiles that range from near ground level to about 2500 m will be compared with in-situ values obtained with in-situ NO2 monitors located on surface-level and at 200 m. In addition, we will use boundary layer measurements and MAXDOAS data obtained at surface level and at 200 m to interpret our findings. The presence of NO2 within 200 m is highlighted comparing the MAXDOAS results for downward and upward viewing directions from the top of the tower, and a differential column within the tower height is retrieved comparing the columns retrieved at both levels. In addition to the NO2 profiles, the measured lidar data contain unique information on the spatial homogeneity and the vertical and temporal variability of NO2. We did detailed investigations on the time-resolved three dimensional NO2 distributions. We will investigate the significance of the variability for the retrieval of NO2 columns by OMI.
	Ground-based measurements of tropospheric NO2-profiles Folkard Wittrock, Hester Volten, Ellen Brinksma, Hilke Oetjen, Anja Schönhardt, Daan Swart, Andreas Richter, John P. Burrows The major topic of this study is the description of tropospheric NO2 profiles derived from MAX-DOAS observations during the Dutch Aerosol and Nitrogen Dioxide Experiments for vaLIdation of OMI and SCIAMACHY (DANDELIONS). For (golden) days with good weather conditions these profiles have been compared to measurements from the novel RIVM NO2 lidar instrument (see Volten et al.) and to in situ observations close to the ground and on top of a 200 m mast. All measurements were performed during two periods at the Cabauw Experimental Site for Atmospheric Research (CESAR, 51.97°N, 4.93°E 0.70 m below mean sea level) which ran from May 8 - July 14, 2005, and September 1 - September 30, 2006, respectively. An important step in passive UV/vis remote sensing was the development from ground-based zenith sky observations to multi-axis (MAX)-DOAS measurements, which has enabled us to better validate findings from satellite observations and study the behaviour of important trace gases in the troposphere on a local scale. Recently, an automated optimal estimation based profile retrieval algorithm (BREAM) was developed for the MAX-DOAS measurements. The method first determines appropriate aerosol settings using measurements of the O4 columns and then inverts the profile of the absorber of interest from the trace gas slant columns. Now DANDELIONS has provided an unique opportunity to validate the retrieval by comparison with independent measurements of the nitrogen dioxide.

Session Modelling - Chair: Thomas Wagner

	INVITED TALK: Tropospheric NO_2 columns as a Top-down Constraint on NO_x Emission Inventories Randall Martin Tropospheric NO_2 columns are closely related to surface NO_x emissions. Factors affecting this relationship will be discussed in the context of providing top-down information on NO_x emissions. Examples will include emissions from fossil fuel combustion, biomass burning, soil microbial activity, and lightning. Sources of error will be examined. Validation needs and areas for further algorithmic development will be identified. Accuracy objectives will be proposed for the retrieval of tropospheric NO_2 with respect to the goal of emissions constraints.
	Distribution and trends of NOx sources inferred by inversion of one decade NO2 satellite columns Jenny Stavrakou, J.-F. Muller, F. Boersma, I. De Smedt, R. van der A The tropospheric NO2 columns retrieved by the GOME and SCIAMACHY satellite instruments between January 1997 and December 2006 are used together with the IMAGES CTM and its adjoint to construct a top-down inventory for anthropogenic, pyrogenic and natural NOx emissions. The influence of the emission updates on the chemical NOx lifetime is taken into account, and found to have a significant impact on the results. Anthropogenic emission trends are inferred over industrialized regions of the Northern Hemisphere. The largest emission increases are found over eastern China, and in particular in the Beijing area, whereas important emission decreases are calculated over the U.S., and to a lesser extent over Europe. The emission changes result in significant trends in surface ozone, amounting to 15%/decade over large parts of China in summertime.
	The use of satellite measurements for estimation of multi-annual changes in NOx emissions Igor Konovalov, M. Beekmann, J.P. Burrows, A. Richter It has been suggested in several recent studies that the available NO2 time series from satellite measurements can be used to study long-term changes in anthropogenic emissions of nitrogen oxides (NOx). It seems reasonable to expect that the most accurate quantitative estimates of the emission changes can be obtained when the measurement data are combined with model calculations which can account for the atmospheric transport and changes in meteorology. We present the first inverse modelling study of interannual changes of NOx emissions in Europe, the Mediterranean and Middle East. We combine the data for tropospheric NO2 columns derived from the long-term measurements performed by GOME and SCIAMACHY satellite instruments between 1996 and 2005 with calculations of the CHIMERE chemistry transport model performed on a 10 by 10 degree grid. From this data set, we estimate separately a linear trend of emissions in each grid cell and the year-to-year variability superimposed over the linear trend. In contrast to a more common inverse modelling approach, where the improved emission estimates are searched for as deviations from “expert” estimates of emissions, our original method yields measurement-based estimates in each grid cell that are subject to only some “global” constraints. Therefore, our estimates can be regarded as a measurement-based alternative to trends derived from emission cadastres. Validation of the results is performed by comparing CHIMERE calculations with independent data for near-surface NOx concentrations from the UK national monitoring network and ozone concentrations from the EMEP network. It is found that the use of our emission estimates in CHIMERE improves the agreement between the model calculations and measurement data in both cases. Our results confirm the generally negative trend of NOx emissions in Europe during the past decade. However, some considerable differences between the trends in the measurement-based and EMEP emission data are found for several countries, especially outside of Western Europe. It is found also that the NOx emission trends in several cities and regions (such as Moscow, Madrid, Eastern Ukraine) are strongly different from those averaged over a corresponding country. Among interesting features of interannual variations of NOx emissions is their strong decrease in Iraq in 2003.
	Evaluation of satellite NO2 columns over U. S. power plants using a regional atmospheric chemistry model Si-Wan Kim, A. Heckel, G. Frost, A. Richter, M.K. Trainer, J. Burrows, S. McKeen, E.-Y. Hsie The western U.S. has many isolated point sources of NOx emissions, which have been observed by various satellites such as GOME, SCIAMACHY, and OMI. Among these, Four Corners and San Juan Power Plants in New Mexico provide a good site to evaluate the satellite-retrieved NO2 columns. These are two of the highest NOx-emitting power plants in the U. S. They produce strong satellite NO2 column signals and are isolated from big cities. Because NOx emissions are independently monitored directly at each power plant, there is much less uncertainty in their known emissions than in the emissions from roads and urban areas. Therefore, differences between satellite-retrieved and model-calculated columns for these plants will depend on other factors besides errors in emission inventories. We carried out simulations with the Weather Research and Forecasting-Chemistry model (WRF-Chem) for the western US domain during the summer of 2005. Initial comparisons of the model and SCIAMACHY NO2 columns give good correlations between the two methods over the model domain, indicating that the model reproduces the relative spatial and temporal patterns of emissions across the entire domain. In the Four Corners and San Juan region (2° longitude x 1° latitude), however, summertime-average model NO2 columns are 30% larger than satellite NO2 columns. These discrepancies are bigger than the model NO2 variability resulting from the use of various chemical mechanisms. Some possible reasons for the satellite-model discrepancies will be discussed.

Session Applications - Chair: Ernest Hilsenrath

	INVITED TALK: Urban and Agricultural NOx Emissions Ronald Cohen, T.H. Bertram Analyses have yet to tap the full potential of the high space and time resolution of SCIAMACHY, OMI or GOME-II data. Using example from OMI and SCIAMACHY we describe observations of agricultural NOx emissions with time resolution of individual rain driven pulses. We use the daily coverage of OMI to describe day-of-week patterns in urban plumes that show both the well known decrease in weekend emissions but also a memory effect such that Monday, a weekday followed by a weekend is different than Wednesday, a weekday followed by a weekday. We will also discuss a NOx feedback on the NOx lifetime via control over OH in urban and agricultural plumes.
	Monitoring long range transport of tropospheric NO2 with OMI Bas Mijling, Ruud Dirksen, Ronald van der A Under favorable meteorological conditions, tropospheric NO2 can be transported fast enough (regarding its life time) to cross the Northern Atlantic Ocean, and affect overseas air quality. Due to predominant west winds in this region, long range transport usually occurs from the industrialized East coast of North America towards the West coast of Europe. To monitor the continental outflow of tropospheric NO2, we use data of the OMI instrument, taking advantage of its daily global coverage and its relatively high resolution (13 by 24 km at nadir). Retrievals of the slant column are done with the DOAS technique. From this, the vertical tropospheric column is derived using a combined modeling / retrieval / assimilation approach. The tropospheric NO2 data product of OMI, retrieved within the DOMINO project, includes a modeled ghost column to compensate for the shielding effect of clouds. For our purposes, we remove this ghost column to avoid that satellite observations done above cloudy scenes are “contaminated” with model results. Using this method we obtain an archive of daily images in which long range transport events of NO2 can clearly be distinguished. In order to quantify these events and to detect the more subtle structures in the column densities of the transported NO2, the quality of the DOMINO-data needs to be improved. With the newly reprocessed OMI dataset, collection 3, several artifact in the data are removed, such as the cross track gradient (causing a difference between eastern pixels and western pixels up to 0.8 1015 molecules/cm2) and clusters of unrealistically high NO2 values (caused by instrument saturation due to optically dense clouds). Further improvement can be made by applying a new destriping algorithm, which smoothes out the different offsets of the pixels in the detector array.
	The generation of a temporally consistent NO2 data record for ocean color work Wayne Robinson, Ziauddin Ahmad, Charles R. McClain Accurate ocean color retrievals depend on the accounting for the gaseous absorption affecting the visible radiances from 412 nm. up to the Near-IR. Recently, the Ocean Biology Processing Group at NASA Goddard Space Flight Center has been working on making an NO2 absorption correction for the visible bands with GOME, SCIAMACHY, and OMI NO2 as the data source. Differences in the NO2 from GOME and SCIAMACHY to OMI have been seen to cause discontinuities in the retrived water-leaving radiance and chlorophyll-a products. A correction has been developed to make a more consistent set of NO2 measurements so that these discontinuities are reduced. The development of this correction and the results of the correction will be discussed.	Paper
	Sources and fate of tropospheric NOx: What can we learn from satellite observations? Steffen Beirle Satellite observations provide several years of tropospheric NO2 data on global scale. Characteristic spatial and temporal patterns, as well as correlations with other quantities like fire or flash counts, have been successfully used to identify and quantify different sources of nitrogen oxides. Furthermore, satellite measurements also provide information on the evolution of the emitted NOx. By studying transport, information on lifetime can be gathered. This talk will give an overview on different approaches to the identification and quantification of different NOx sources, discussing their difficulties, limitations and results. In addition, the potential of studies on the NO2 transport for lifetime estimates will be demonstrated.
	OMI Tropospheric NO2 from Lightning in Observed Convective Events Kenneth Pickering, E. Bucsela, J. Gleason, P. Levelt Lightning is responsible for an estimated 10-20% of NOx emissions in the troposphere. In this study, we present evidence of lightning-generated NO2 (LNO2) using data from the Ozone Monitoring Instrument (OMI), which has observed tropospheric NO2 since its launch in 2004. Although LNO2 has been also reported in previous satellite studies from the Global Ozone Monitoring Experiment (GOME) and SCIAMACHY, OMI is better suited for such measurements by virtue of its higher resolution and daily global coverage. The LNO2 signal is clearly seen in OMI data on two days over and downwind of convective systems in the US Midwest in 2006. We also present an analysis of OMI data over northern Australia during the SCOUT-O3/ACTIVE field campaigns in November and December 2005. Both single- and multi-day averages are presented to examine possible LNO2 signals from individual diurnally recurrent convective events. In these events we compare the OMI signals with aircraft observations from the storm anvils.
	Monitoring air quality in an urban area using remote sensing techniques and in situ measurements. Louisa Kramer, Roland. J. Leigh, John. J. Remedios, Paul. S. Monks Monitoring urban air quality is an important issue and remotely sensed data from ground and space based instruments are being extensively used to study air pollution problems over urban and regional areas. The Leicester based UV/VIS Concurrent Multi-Axis Differential Optical Absorption Spectroscopy (CMAX-DOAS) system uses a remote sensing technique based on the concept of observing several viewing geometries simultaneously. It utilises an instrument that images from multiple viewing angles using a single CCD, giving an instrument that offers temporal resolution on every viewing angle of a minute or less. Global NO2 data have been retrieved from measurements performed by the Ozone Monitoring Instrument (OMI), launched onboard the NASA satellite Aura in July 2004. The relatively high spatial resolution of OMI makes it suitable for measurements of air quality on an urban scale. However, the comparison of near-surface NO2 measurements from chemiluminescence detectors situated in and around Leicester city centre with tropospheric NO2 columns from OMI and CMAX-DOAS demonstrates that NO2 emissions from a polluted urban area cannot be simplistically linked to the column measurement by a satellite instrument. Here we demonstrate a FOV-weighted correction to obtain a relationship between column measurements of NO2 from satellites to those at street level.

Poster Session - Chair: Thomas Wagner

	Strategies for tropospheric trace gas retrievals from satellite and comparison to model results Thomas Wagner, Steffen Beirle, Michael Grzegorski, Ulrich Platt The retrieval of tropospheric trace gas products from UV/vis satellite observations is a great challenge, because the sensitivity of such observations strongly dependens on several parameters, like height profile, surface albedo, and cloud and aerosol properties. Since usually not all of these quantities can be directly derived from the satellite observation itself, a-priori or model information is typically used in the retrieval of tropospheric trace gas products (e.g. tropospheric vertical column density or average concentration). This mixing of information from measurement and other sources, however, often complicates the interpretation of the derived trace gas products. Here we discuss different strategies to extract information on tropospheric trace gases from satellite information using different degrees of additional (e.g. model) information. We also discuss different ways for the comparison of measured trace gas data with model results.
	Poster:
	Impact of clouds on tropospheric trace gas retrievals Steffen Beirle, Tim Deutschmann, Michael Grzegorski, Ulrich Platt, Thomas Wagner Spectroscopic measurements from nadir-viewing satellite platforms allow the retrieval of column densities of several atmospheric trace gases. The retrieval of tropospheric columns is thereby strongly affected by clouds: Clouds shield boundary layer and lower tropospheric trace gases, leading to an underestimation of the actual column. On the other hand, the high albedo of clouds, as well as multiple scattering within the cloud, increase the visibility of trace gases at and above the cloud top. Cloud parameters like cloud fraction, cloud top height or cloud heterogeneity can also be directly deduced from satellite measurements, using intensity measurements and spectral absorption features of O2, O4 or the so-called “Ring-effect”. Here we analyze the dependency of tropospheric NO2 columns on several cloud parameters. This empirical study is complemented by theoretical radiative transfer modelling studies using the 3D-Monte-Carlo Model TRACY-2, that is in particular capable of modelling radiative transfer in clouds. With these investigations we check and improve our understanding on the different cloud effects on radiative transfer (shielding, path-length enhancement and albedo increase). Improved knowledge on the impact of clouds on trace gas columns allows to interpret clouded pixels, that are currently discarded in most analyses. Temporal or spatial variations of the observed dependencies of NO2 columns on cloud parameters hold additional information on e.g. the NO2 profile.
	Poster: Presentation:
	OMI and SCIAMACHY NO2 validation by MultiAxis and DirectSun DOAS observations during the DANDELIONS campaigns Gaia Pinardi, Michel Van Roozendael, Francois Hendrick, Caroline Fayt, Christian Hermans, Alexis Merlaud, Martine De Mazière The DANDELIONS (Dutch Aerosol and Nitrogen Dioxide Experiments for vaLIdation of OMI and SCIAMACHY) project has been set up in the Netherlands with the aim to contribute to the validation of OMI (Ozone Monitoring Instrument), SCIAMACHY (Scanning Imaging Absorption SpectroMeter for Atmospheric CartographY) and AATSR (Advanced Along-Track Scanning Radiometer) measurements of aerosols and nitrogen dioxide (NO2). Two measurement campaigns took place in Cabauw (52° N, 5° E), first from May to July 2005, and second in September 2006. These gathered several types of complementary ground-based measurement techniques, such as in-situ samplers, LIDAR, MAXDOAS and Direct sun instruments. In this work, total and tropospheric NO2 columns derived from a combination of ground-based Multi-Axis DOAS and Direct Sun DOAS measurements, are compared to OMI and SCIAMACHY NO2 products obtained during both campaigns. The Multi-Axis DOAS (MAXDOAS) technique rely on UV-visible scattered sunlight observations, whereby NO2 absorption can be quantified using the Differential Optical Absorption Spectroscopy (DOAS) technique. By scanning viewing angles successively from zenith to the horizon, atmospheric light paths of increasing length into the lower troposphere are sampled, so that the measured NO2 columns can be vertically resolved, providing independent information on the tropospheric and stratospheric contents. Complementary to MAXDOAS observations, Direct Sun measurements are characterized by a well defined geometrical path and therefore provide accurate total NO2 columns. The aim of the study is to assess the overall agreement between ground-based and satellite data sets, and to investigate the relative performance of various OMI and SCIAMACHY NO2 retrieval schemes. Analysis are carried out with respect to several issues, among them the sensitivity of satellite measurements to clouds being present in the field of view (including the role of the ghost column), the importance of co-location mismatch effects and, for OMI, the swath angle dependency.
	Poster: Presentation:
	Multi-model ensemble simulations of tropospheric NO2 compared with GOME retrievals for the year 2000 Twan van Noije, H.J. Eskes, F.J. Dentener, D.S. Stevenson, K. Ellingsen, M.G. Schultz, O. Wild, et al. We present a systematic comparison of tropospheric NO2 from 17 global atmospheric chemistry models with three state-of-the-art retrievals from the Global Ozone Monitoring Experiment (GOME) for the year 2000. The models used constant anthropogenic emissions from IIASA/EDGAR3.2 and monthly emissions from biomass burning based on the 1997–2002 average carbon emissions from the Global Fire Emissions Database (GFED). Model output is analyzed at 10:30 local time, close to the overpass time of the ERS-2 satellite, and collocated with the measurements to account for sampling biases due to incomplete spatiotemporal coverage of the instrument. We assessed the importance of different contributions to the sampling bias: correlations on seasonal time scale give rise to a positive bias of 30–50% in the retrieved annual means over regions dominated by emissions from biomass burning. Over the industrial regions of the eastern United States, Europe and eastern China the retrieved annual means have a negative bias with significant contributions (between –25% and +10% of the NO2 column) resulting from correlations on time scales from a day to a month. We present global maps of modeled and retrieved annual mean NO2 column densities, together with the corresponding ensemble means and standard deviations for models and retrievals. The spatial correlation between the individual models and retrievals are high, typically in the range 0.81–0.93 after smoothing the data to a common resolution. On average the models underestimate the retrievals in industrial regions, especially over eastern China and over the Highveld region of South Africa, and overestimate the retrievals in regions dominated by biomass burning during the dry season. The discrepancy over South America south of the Amazon disappears when we use the GFED emissions specific to the year 2000. The seasonal cycle is analyzed in detail for eight different continental regions. Over regions dominated by biomass burning, the timing of the seasonal cycle is generally well reproduced by the models. However, over Central Africa south of the Equator the models peak one to two months earlier than the retrievals. We further evaluate a recent proposal to reduce the NOx emission factors for savanna fires by 40% and find that this leads to an improvement of the amplitude of the seasonal cycle over the biomass burning regions of Northern and Central Africa. In these regions the models tend to underestimate the retrievals during the wet season, suggesting that the soil emissions are higher than assumed in the models. In general, the discrepancies between models and retrievals cannot be explained by a priori profile assumptions made in the retrievals, neither by diurnal variations in anthropogenic emissions, which lead to a marginal reduction of the NO2 abundance at 10:30 local time (by 2.5–4.1% over Europe). Overall, there are significant differences among the various models and, in particular, among the three retrievals. The discrepancies among the retrievals (10–50% in the annual mean over polluted regions) indicate that the previously estimated retrieval uncertainties have a large systematic component. Our findings imply that top-down estimations of NOx emissions from satellite retrievals of tropospheric NO2 are strongly dependent on the choice of model and retrieval.
	Poster: Presentation:
	Using MERIS data to calculate NO2 airmass factors Joana Leitao Alexandre, A. Richter, A. Heckel, J.P. Burrows, W. von Hoyningen-Huene, A. Kokhanovsky, T. Dinter Nowadays there are a handful of satellite instruments (GOME, GOME-2, SCHIAMACHY, OMI) providing measurements that can be inverted to retrieve trace gas distributions in the atmosphere. These measurements are of great importance to investigate the global distribution of pollutants as is the case of ozone (O3), nitrogen dioxide (NO2) and several others. With these global fields one can identify emission sources and analyse long-term trends of pollutant concentrations. As in all remote sensing techniques, the retrieval of tropospheric columns of NO2 from the satellite measurements is based on several assumptions that in one way or another contribute to the uncertainty in the final retrieval. The improvement of the a priori assumptions used as well as the characterisation of the uncertainties is a main concern to obtain the correct values of NO2 present in the troposphere. To obtain the vertical column of NO2 one must divide the slant column obtained from satellite measurement by an airmass factor (AMF). This AMF is dependent on many aspects such as: geometry and wavelength of measurement, vertical distribution of the species, surface albedo, aerosol loading and clouds. While some of these factors are well known others are highly uncertain and variable. In this poster, we investigate the possibility to use data from the MERIS instrument to provide measured values for aerosol optical thickness, spectral surface direct reflectance and possibly also cloud cover to be used in the computation of airmass factors for SCIAMACHY retrievals. Like SCIAMACHY, MERIS is also operating on ENVISAT and provides data which are collocated with SCIAMACHY measurements in space and time. This approach has the potential to replace current values which are either from climatologies or low resolution atmospheric models by measured values and therefore could provide more realistic retrievals, in particular over regions where rapid changes of aerosols are frequently verified. As this study is still work in progress, the presentation will focus on sensitivity studies and selected examples.
	Poster: Presentation:
	the effect of updating reaction rate data on the tropospheric NO2 column simulated by TM4 as compared t GOME retreivals for the year 2000 Jason Williams, Twan van Noije, Henk Eskes, Ronald van der A Here we present a systematic comparison of tropospheric NO2 columns calculated using the global chemistry transport model TM4 with three state-of-the-art retrievals from the Global Ozone Monitoring Experiment (GOME) for the year 2000. Since the original study of van Noije et al (2006) the reaction rate parameters in TM4 have been updated using the latest recommendations (Atkinson et al, 2004, 2006; Sander et al, 2006). Here we investigate the differences introduced into the annual mean tropospheric NO2 columns by adopting these updated chemical rates. In general, there are decreases in the NO2 column density in regions influenced by biomass burning activity and increases in the NO2 column density for regions influenced by industrial NOx emissions. These changes are of the order of ~10% of the total tropospheric NO2 column. The reason for these differences is the increase in the reservoir species ORGNIT and a substantial modification in [HO2], respectively. Seasonal decompositions reveal that for regions influenced by a biomass burning season, this improves the agreement between NO2 columns retrieved by GOME and those calculated by TM4. However, for regions dominated by strong NOx emissions the improvements do not reconcile the large differences which occur between the model and the GOME retrievals.
	Poster: Presentation:
	Comparison between tropospheric NO2 vertical columns by GOME and surface NO2 mixing ratio by the air-monitoring network over Japan Katsuyuki Noguchi, H. Itoh, T. Shibasaki, S. Hayashida, I. Uno, A. Richter, J. P. Burrows We compared the tropospheric NO2 vertical columns observed by GOME [Richter et al., 2005] and the surface NO2 mixing ratio by the air-pollution monitoring network composed of more than 1000 stations over Japan during 1996-2003. The comparison of GOME and surface measurements showed almost no long-term trends of NO2 over Japan. The comparison also showed the similar seasonal variations of NO2, which were asymmetric with a rapid increase in fall and a slow decrease in spring. These consistencies suggest that the GOME successfully observed the NO2 behavior in the lower troposphere over Japan. We also examined how we should manipulate the data of GOME and surface measurements. Due to a large coverage of one pixel of GOME swath data, we should carefully select the swath pixels which correspond to the surface measurements when we average the GOME-NO2 from those pixels. We will discuss the effect of the GOME swath pixel selection on the averages of GOME-NO2.
	Poster: Presentation: