Sounding the Troposphere from Space

The Kyoto Protocol, recently ratified by the EU, calls for a reduction in the emission of greenhouse gases in the period 2008 to 2012. However, global greenhouse gas emissions and absorptions, sources and sinks, are not well known. There is a large discrepancy between bottom-up emission estimates, derived from national government energy, transport, agricultural, etc. figures, and top-down estimates derived from atmosphere concentration distributions. Many parts of the world are poorly monitored (southern hemisphere, oceans) and some countries do not take part in the treaty. Measurements from ENVISAT, the ESA earth observation satellite presently in operation, are expected to improve this situation. The SCIAMACHY instrument on ENVISAT is able to probe the troposphere where the emission and absorption of greenhouse gases takes place. space measurements, modelling and ground-based data will yield the required level of accuracy. The gases to be investigated are CO₂, CH₄, N₂O, O₃, H₂O, CO and NO₂. The focus will be on methane and carbon monoxide because of their importance for the Kyoto Protocol and their relatively good feasibility prospect. This TROPOSAT work forms part of a wider European collaboration, the EVERGREEN project of the European Commission 5th Framework Programme on Environment and Sustainable Development, to be carried out in the period 2003 to 2006.

Aerosol retrieval from GOME data indicates that new aerosol classes can be found, provided that their optical properties are well defined. Validated results may also be used in climatology studies of the aerosol. The current procedure used for GOME can be also used for SCIAMACHY data, implementing the algorithms and the aerosol classes.

A new approach for retrieving ozone profiles from ERS2-GOME spectral data has been developed, which relies on feed-forward neural networks to perform the data inversion. By using GOME spectral data from selected wavelength regions, instrument (e.g. scan angle), geolocation and UKMO temperature profile data as input, neural networks have been trained to determine the ozone profile from 1–60 km geopotential height in a one-step inversion scheme. In order to train neural networks, an extensive database of collocated GOME and ozone profile measurements is necessary. Ozone profiles from sondes collected by the World Ozone and Ultraviolet Data Centre (WOUDC), as well as HALOE, SAGE II and POAM III limb measurements have been utilised for this purpose, constituting about 70000 training collocations. While training takes a certain amount of time, it is only needed once, the actual retrieval process is by a factor of 103 to 105 faster than classical methods.The NNORSY retrieval results indicate that the neural network successfully extracts information on the tropospheric ozone distribution from the GOME spectra. Overall, there is good agreement with collocated ozone-sondes and a reasonable agreement of large scale ozone field structures in comparison with a tropospheric chemistry and transport model. However, due to the structure of the training data set, some inaccuracies remain at extreme SZAs and over the oceans, where there is little sonde data available.The method is thus not yet optimised for tropospheric ozone retrieval. In the future, we plan to improve error checking and cloud treatment, dataset distribution and possibly training procedures, i.e. by using specialised networks. Adaptation to upcoming sensors is also envisioned, and should be possible with relatively little effort.

Progress on the development of retrieval algorithms for tropospheric nitrogen dioxide and ozone from the GOME, SCIAMACHY, OMI and GOME-2 instruments is outlined. New techniques are reported that significantly improve on the existing products. A better insight in the accuracy has been achieved. GOME data has been made available through web access.

In the last decade, algorithms for the retrieval of aerosol properties from various space-borne sensors have been developed. Usually, these properties are available on continental or global scales. The objective of this study is to retrieve information on aerosol parameters on regional or local scales, which are better adapted to the Swiss landscape. Our study focuses on the Multi-Angle Imaging Spectro-Radiometer (MISR) onboard the NASA satellite TERRA, observing the earth with 9 cameras at different view angles and at 4 spectral bands. Georectified radiances at 275 m and 1100m ground resolution are available. An aerosol operational product is provided, which includes climatological aerosol mixtures and gridded aerosol optical depths at 17.6 km ground spatial resolution. This is not sufficiently accurate for the complex landscape of Switzerland.Data of one block of the aerosol operational product has been processed (September 11th, 2000). The operational aerosol optical depths (AODs) of the green band (557 nm) were compared with AODs measured by the sunphotometer at Bern. This data is within the uncertainties of the MISR product.For the retrieval of aerosol properties on a smaller scale, we simulated the MISR radiances for 2 water targets and 2 vegetated surfaces by using the atmospheric radiation transfer code MOOTRAN 4. Standard atmospheric models and aerosol compositions implemented in MOOTRAN 4 as well as continental aerosol mixtures provided by the MISR operational product were included. A new algorithm is proposed, which allows the simultaneous retrieval ofthe surface reflectance, ρ_surf, and the aerosol extinction coefficient, β_e (550 nm), at surface level. Aerosol mixtures provided by the MISR operational product were used for this study. The green band AOD retrieved by our algorithm for a site in Lake Neuchatel equals 0.145 ± 0.009, which is in excellent agreement with the sun photometer data (0.143 ± 0.008). The operational AOD for the same site was found to be 0.134 ± 0.262. At the 3 other locations the AODs of two algorithms agree within their uncertainties as well.

Ozone is one of the most important trace gases and it plays a variety of important roles in the earth's atmosphere. The direct and indirect effects of ozone on atmospheric chemistry, on the earth's climate system, and on the UV surface fluxes depend not only on the vertically integrated ozone amount but also on its vertical distribution. To study the related scientific issues on a global scale, in 1995 the European Space Agency (ESA) launched the Global Ozone Monitoring Experiment (GOME) on board of the second European Remote Sensing Satellite (ERS-2). GOME is a nadir viewing grating spectrometer measuring at around 0.2 nm spectral resolution in the spectral range 240 to 790 nm. More information about the GOME instrument is given in the overview paper by Burrows et al. 1999.

Algorithms for the retrieval of aerosol properties over land and over sea have been developed by the TNO Physics and Electronics Laboratory (TNO-FEL) for several instruments, such as AVHRR (Veefkind et al., 1998a), GOME, ATSR-2 (Veefkind et al. 1998a, b; 1999) and OMI (Torres et al. 2002). OMI will be launched in 2004. An algorithm for SCIAMACHY is developed for application over water and over land (Kusmierzcyk-Michulec and De Leeuw 2002a).In these algorithms, different techniques are applied, depending on the instrument properties . The current emphasis is on the Along Track Scanning Radiometer (ATSR-2) on board the ESA Earth Remote Sensing satellite ERS-2. ATSR-2 offers two views, nadir and forward. The nadir view is used over water, where simple assumptions can be used to separate the surface reflectance from the aerosol effects. Over land both views are used. Initially, the single and dual view algorithms have been applied over the US east coast and Europe, with a relative simple description of the aerosol composition in terms of sea spray and “continental” aerosol. However, over other areas the situation can be more complicated due to the occurrence of other aerosol types with different optical properties, such as black carbon, dust and biomass aerosol. In particular when none of these aerosol types dominates it is difficult to determine the influence of each of them on the aerosol optical depth (AOD).This challenging problem has been studied using data from the INDOEX (India and Indian Ocean, see Figure 3.7.1) experiments in February and March 2000 and the SAFARI experiments over South Africa in September 2000. In the INDOEX area the aerosol consists of a complicated mixture of five or more aerosol types. An algorithm has been developed with look up tables (LUTs) that describe the influence of these aerosols on the radiation received by the ATSR-2. A result from this algorithm is presented in Fig. 3.7.1, which is a composite of retrieved AOD for the duration of the INDOEX experiment. Over land the dual view algorithm has been used, over water the single view algorithm. The algorithm also provides the aerosol type, which over land is anthropogenic which changes gradually to maritime as the air mass is advected over sea. The Ångström coefficients have been used to determine the spatial distribution of Be in the area using the method described in (2002b). Results from these retrievals compare favourably with ground-based data, as described in Robles Gonzalez et al. (2003a). The algorithm is now further developed for biomass aerosol, using data from the SAFARI experiments. Initial results are encouraging (Schmid et al., 2003).Apart from the INDOEX and SAFARI experiments, ATSR-2 retrieved AOD maps over Europe are available for October 1995, August 1997, and April, May and October 1999. The ATSR-2 algorithm is being converted to a user-friendly algorithm that can be used to handle large data sets to retrieve aerosols over extended areas and long periods. It will be applied for AATSR on ENVISAT. Results will be madeavailable on the TEMISwebsite (www.temis.nl). Fig. 3.7.1Composite map of the mean AOD at 0.659 μm retrieved using ATSR-2 data for February and March, 1999 over the INDOEX area. Over white areas no data is present

The Measurements of Pollution in the Troposphere (MOPITT) instrument was launched on December 18th, 1999, onboard the NASA EOS-TERRA satellite. It is a nadir viewing infra-red radiometer which targets measurements of carbon monoxide (CO) and methane (CH₄) using gas correlation spectroscopy to improve its spectral sensitivity. The CO profile measurements can exploit both up-welling thermal radiance in the 4.6 μm fundamental band and reflected solar radiance in the 2.3 μm band; CH₄ column measurements are made at 2.2 μm. This project focuses on CO measurements from MOPITT and the development of an off-line retrieval algorithm which can be employed to determine the intrinsic accuracy of the system in particular regimes, such as profiles of pollution export regions relevant to Europe. This project was a new but essential activity in the TROPOSAT project supporting studies in a number of TROPOSAT activities including “Inverse modelling of satellite data using the IMAGE model”, “Tracing atmospheric pollution with satellite measurements”, “Validation of methane and carbon monoxide data from SCIAMACHY”, and “Satellite validation using ground-based spectroscopic techniques”.The basis of the study was the examination of the MOPITT CO profile data and their sensitivity to CO values at various altitudes through the troposphere. This sensitivity, as expressed through the averaging kernel, was examined via an off-line data inversion system (retrieval algorithm) and via the operational MOPITT retrieval system. The off-line scheme has been developed to allow the factors which influence the averaging kernel to be investigated for case situations. Results have shown that this scheme, currently implemented for the four MOPITT difference channels only, produces averaging kernels similar to those derived in the operational processor. Both systems demonstrate that averaging kernels are a necessary part of the interpretation of MOPITT data and are essential to comparisons of profiles from models/correlative observations.

In the retrieval of tropospheric NO₂ columns from space borne UV/visible measurements, the separation between stratospheric and tropospheric contributions to the signal is an essential step. Up to now, the stratospheric NO₂ column has been estimated using measurements over unpolluted regions and assuming that stratospheric NO₂ columns do not depend on longitude. In this study, several novel approaches to extract the tropospheric contribution are presented, and first applications to GOME and SCIAMACHY data are shown. The techniques studied are separation of stratosphere and troposphere using the temperature dependence of the absorption cross-section of NO₂, use of the wavelength dependence of the sensitivity to tropospheric absorptions, and use of a stratospheric chemical transport model to simulate the stratospheric NO₂.

The CRyogenic Infrared Spectrometers and Telescopes for the Atmosphere instrument (CRISTA) aboard the Shuttle Palette Satellite (SPAS) was developed to provide global measurements of atmospheric infrared emissions with increased horizontal resolution. The moderate spectral resolution (λ/Δλ = 500) of the instrument is sufficient to obtain valuable information on the distribution of trace species at low altitudes, e.g. in the upper troposphere. This report gives a brief description of the retrieval of H₂O in the tropopause region and summarises some results obtained for the second CRISTA mission in August 1997. The retrieved H₂O values are in good agreement with simultaneous observations made by the FISH experiment onboard the Falcon aircraft and with coincident MOZAIC observations. The data product is available via www.crista.uni-wuppertal.de.

This contribution has been concerned with three different activities. First, global N0₂ data records from HALOE, POAM-II and POAM-III satellite instruments complemented by ground-based data sets from the NDSC, have been investigated for mutual consistency and exploited to build a global time-latitude stratospheric NO₂ profile climatology that may be used for tropospheric NO2 retrieval from ERS2-G0ME. Second, a BrO total column product derived from GOME has been created over the total present lifetime of the ERS-2 satellite (1996 to 2002) and made available to the scientific community through a dedicated web site. In addition, a prototype improved GOME BrO retrieval algorithm has been designed with the aim to determine better global tropospheric BrO columns from the combined use of GOME data and three-dimensional model data, and to provide the stratospheric reference needed in a tropospheric residual approach. Third, ground-based multi-axis DOAS measurements have been conducted at the Observatoire de Haute Provence, (OHP, 44 °N) between January 2001 and June 2002 in support of satellite retrieval activities. Results obtained demonstrate the high sensitivity of the measurements to tropospheric amounts of NO₂ and HCHO, as well as the potential of the technique to unravel small mid-latitude free-tropospheric BrO contents. As to the latter point, comparisons between combined GOME-SLIMCAT and MAX-DOAS tropospheric BrO evaluations suggest that approximately half of the mid-latitude BrO total column resides in the troposphere.

ENVISAT with MIPAS (Michelson Interferometer for Passive Atmospheric Sounding) (Fischer and Oelhaf, 1996) on board was successfully launched on 1st March, 2002. Since 24th March 2002, MIPAS has been measuring high-quality mid-infrared limb emission spectra with 0.035 cm-1 spectral resolution. The non-operational scientific level-2 data analysis of MIPAS/ENVISAT data to be performed at IMK has been developed further. The work of this project was focussed on various retrieval aspects in the upper troposphere region. On the basis of synthetic spectra, the information content of the spectra was analysed with respect to about 45 trace species, temperature and microphysical properties of cirrus clouds. The retrieval strategy for more than 30 trace species was set up, verified, and fine-tuned within test retrievals. Up to 20 species in the upper troposphere region can be detected with total accuracy (including all relevant systematic error sources) better than ±100 %, many of them better than ±10 %. Among these species are constituents which are of utmost interest for the understanding of the global and regional ozone budget in the upper troposphere, like acetone, NMHCs, PAN, or HO₂. For the first time the impact of scattering of mid-infrared tropospheric radiation into the line-of-sight by cirrus particles was modelled with high spectral resolution. Spectrally high resolved signatures from tropospheric radiation scattered into the line-of-sight allow an improvement in the retrievability of microphysical parameters of cirrus clouds. The observation and retrieval scenarios to be applied for the derivation of tropical water vapour volume mixing ratio (VMR) close to the tropical tropopause was optimised with respect to accuracy versus vertical resolution. The achievable accuracy and vertical resolution is high enough to contribute to questions referring to processes in the tropical tropopause layer. First available MIPAS spectra were inspected in order to verify the predictions achieved on the basis of simulated measurements.

The hydroxyl radical plays an important role in the tropospheric chemistry. Due to its low concentrations and high reactivity it is hard to measure in-situ. Using indirect methods, like satellite remote sensing data, enables us to make an estimate of the global oxidising capacity. For this study we used GOME satellite data. The primary production rates of OH were calculated for the first three days of August 1999, for a wide latitude band between 70 °N and 55 °S. We also indicate the uncertainty that is connected with the production rate values due to the uncertainties in the input data. The calculated primary OH production rate was found to be highest around 2 km (0.7 ppt s-1) and its uncertainty ranges between 20 and 50 %, mostly due to the high standard deviation in the ozone climatology used for the calculation.

The development of a DOAS based ground validation station for satellite based atmospheric sensor instruments GOME and SCIMACHY at the Institute of Atomic Physics and Spectroscopy, University of Latvia, is now underway. The activity has to look to the future by: — the development of co-operation with the European research community in atmospheric science contributing to the valuable infrastructure for common needs; and— a building up of the capacity of the team (through research training of students) to become a strong partner in the future.

An algorithm to generate vertically resolved actinic flux maps has been developed and demonstrated in a stand-alone mode. A semi-empirical method has been devised and tested to derive routinely the aerosol optical thickness (AOT) from satellite-measured “Top of Atmosphere” (TOA) radiance in the visible region, over sites previously documented by ground measurements. It has been successfully demonstrated over three very different sites and using data from both the VEGETATION and SeaWiFS satellite instruments.

This project focuses on the determination of cloud properties (like geometric cloud fraction and average cloud to height) from satellite observations and on the quantification of the corresponding cloud influence on tropospheric trace gas products derived from satellites (see e.g.Burrows et al. 2000; Richter et al. 2002; Wagner et al. 2002a). The investigations first concentrate on GOME and will later also be applied to SCIAMACHY on ENVISAT.The most dominant effects of clouds are (a) that they shield the atmosphere below the cloud cover, and (b) that their albedo is typically significantly larger compared to the earth’s surface. Because of these effects the determination of quantitative tropospheric trace gas products depends strongly on the knowledge of cloud properties.Cloud algorithms which already exist are based on spatially resolved intensity measurements (Polarisation measurement devices, PMD, see ESA (1995)) as well as on the determination of the optical depth of the OrA band (see e.g. (1994), Kjoelemeijer and Stammes 1999). However, both quantities show important shortcomings, especially over snow and ice surfaces, which strongly limit their applicability.In this study we investigate a large variety of cloud sensitive parameters measured by GaME. Besides the “traditional” cloud sensitive parameters mentioned above we also investigate the polarisation of the measured light and in particular of various absorption bands of the oxygen dimer O₄ (at 360, 380, 477, 577 and 630 nm (Greenblatt et al. 1990)) as well as the filling-in of the solar Fraunhofer lines by inelastic Raman scattering. We demonstrate that several of these quantities are well suited for the characterisation of clouds, especially over snow and ice surfaces.

With the new series of nadir viewing UV/visible satellite instruments SCIAMACHY/ENVISAT, GOME-2 and GOME-3 aboard METOP-1 and 2, succeeding the current and extremely successful GOME/ERS-2 mission, continuation of global ozone profile measurements in the next two decades will be possible. In the continuous UV/visible spectral range of these instruments (240–600 nm) the absorption of ozone and the multiple scattering cross-section varies several orders of magnitude, which makes it possible to derive, in addition to column abundance, height resolved ozone distribution from the respective nadir spectra using appropriate optimal estimation inversion schemes (Hoogen et al. 1999a, 1999b). However, the ozone information content in the UV/visible spectrum is dominated by lower stratospheric ozone making the tropospheric retrieval a quite challenging task. This project aims at improving the determination of tropospheric ozone by adding additional geophysical information contained in the spectra into the profile retrieval. In order to distinguish the profile retrieval from the total column retrieval, the advanced optimal estimation scheme is referred to as the full retrieval method (FURM).

The emphasis in 2002 has been on the further development of aerosol modelling within the 3-D Eulerian chemistry-transport model LOTOS. The aim is to extend the model with all aerosol compounds contributing to the Aerosol Optical Depth (AOD), so that comparison with and (later) data assimilation of retrieved AOD fields from satellites is possible. Within the year 2002 no data assimilation of AOD fields from satellites has been performed, mainly due to a lack of funding. However, the model was extended substantially and extensive validation has taken place.

Recently, the fully coupled chemistry-climate model ECHAM4.L39(DLR)/CHEM has been developed (Hein et al. 2001). To check the abilities and deficiencies of the model system a detailed comparison is strongly required with a variety of distinct observations. In this project, an intensive comparison of NO₂ tropospheric columns derived from ERS-2 GOME and results of the interactively coupled model system has been carried out. Nearly five years of GOME measurements (January 1996 to August 2000) have been used. The length of this data series is sufficient to enable a comparison based on climatological averages and with global coverage, focussing on the geographical distribution of the tropospheric NO₂, for the first time. A new approach of analysing regional differences (i.e. on continental scales) calculating individual averages for different environments provides more detailed information about specific NO_X sources and of their seasonal variations. The results obtained enable the validity of the model NO₂ source distribution and the assumptions used to separate tropospheric and stratospheric parts of the NO₂ column amount from the satellite measurements to be investigated. Observations and model results show a qualitative agreement, but also some differences in detail. The method employed to assess the tropospheric column of NO₂ from GOME yields some disadvantages, which must be considered.

The focus of our TROPOSAT-related work has been mainly on comparisons of tropospheric NO₂ with output from our global chemistry-transport model MATCH-MPIC. Global monthly NO₂ distributions from MATCH-MPIC were compared with GOME NO₂ columns in von Kuhlmann (2001). A closer focus on NO₂ over Asia has been taken by a joint PhD student with the University of Bremen (T. Kunhikrishnan); the first part of this study has been aimed at improving our understanding of NO_X in the troposphere over Asia (especially India), and at developing an improved estimate of the NO_X emissions rate from this region, through the use of sensitivity runs with MATCH-MPIC along with GOME tropospheric NO₂ columns. We first verified that the retrieval assumptions from GOME (stratospheric zonal symmetry and negligible troposphere NO₂ in the Pacific reference sector) hold well for MATCH-MPIC. We then showed that the model simulates the mean NO₂ amounts in the region fairly well, and that the NO_X lifetime of about 15 hours computed by the model, agrees well with the lifetime derived from the decay of NO₂ off the west coast of India. Taken together, these imply that the current total magnitude of NO_X emissions in MATCH-MPIC is reasonable. While 60–70 % of the NO_X in the lower troposphere over India comes from Indian emissions, in the upper troposphere the NO_X comes mainly from outside the region during the winter months and from convective transport and lightning during the summer monsoon.

This contribution aims at investigation and analysis multidimensional data sets, primarily atmospheric data sequences in time by using techniques like the extended structure tensor Haußecker can be used. Usually the structure tensor is used to detect motion in image sequences, but can be extended to detect other model-based parameters of dynamic processes like exponential decay rates or diffusion constants. The work of the second year includes preparation of the data on which the described algorithms will be applied.

Ozone in the troposphere is controlled by stratospheric-tropospheric exchange (STE) and in-situ production which depends on the concentration of precursors such as the nitrogen oxides NO_X. Due to the short tropospheric lifetime of NO_X, its global distribution strongly corresponds to the distribution of emissions. We want to quantify the relative contributions and the geographic distributions of individual NO_X emission sources using a variety of satellite data. Nitrogen dioxide NO₂ measured by the GOME satellite has been compared to night-time light emissions observed from space. These light emissions can serve as a proxy for emissions of NO_X from fossil fuel combustion. It turns out that the light density at the earth’s surface shows a better correlation with tropospheric NO₂ measured by GOME than the estimated anthropogenic emissions in the EDGAR database which are widely used in global chemistry models. Recently satellite datasets of global lightning flash frequencies (LIS/OTD) and fire counts (ATSR) became available. With the satellite datasets of light density, lightning intensity, fire counts, and NO₂ column density we hope to improve the current knowledge of nitrogen oxide emissions.

Satellite measurements of tropospheric composition are becoming of increasing importance. Important precursors of ozone such as carbon monoxide and NO₂ can now be measured. Careful consideration of the methods used to compare satellite data to models has been required because of such issues as the method to calculate the tropospheric column of species which have substantial columns in the troposphere, averaging kernels and the overpass time of the satellite for those species which have a strong diurnal variation.

Tropospheric NO₂ columns retrieved from GOME were compared to column densities derived from aircraft NO₂ profile measurements. For a case study with clear sky conditions in central Europe, a very good agreement was found between the two methods. The in situ measurements yield a tropospheric NO₂ column of (4.2 ± 1.7)·x 1015 molec./cm2, whereas the GOME data result in columns of (3.5 ± 0.9) ·x 1015 molec./cm2 for near-real-time and (4.1 ± 1.0)·x 1015 molec./cm2 and dedicated analyses. The most important uncertainty in this study for the NO₂ columns from the aircraft measurements is caused by the lack of data in the lower boundary layer. The uncertainties for the GOME columns in this case are dominated by the assumptions made for the air-mass factor calculation. This work is the first independent validation of tropospheric NO₂ columns from satellite instrumentation. Further validation at other seasons and regions including a more comprehensive sampling of the boundary layer by the aircraft is needed.

Tropospheric nitrogen oxides (NO_X) play a key role in tropospheric photochemistry, being a limiting factor of tropospheric ozone production. NO_X have various sources with highly uncertain magnitudes. Therefore, the main task of our work is to combine GOME satellite image sequences with Lagrangian transport models to determine tropospheric NO_X sources. We focussed on case studies on different NO_X sources. So far we investigated biomass burning, NO_X industrial, NO_X and lightning NO_X. To simulate the atmospheric transport of these emissions we used the Lagrangian particle dispersion model FLEXPART (see http://www.fw.tum.de/EXT/LST/METEO/stohl).Firstly, we investigated transport of NO_X emissions from Canadian forest fires during August 1998 (Spichtinger et al. 2001), using a dispersion model calculations, Total Ozone Mapping Spectrometer (TOMS) aerosol index data, and tropospheric NO₂ columns derived from the GOME satellite data. We tracked an NO_X plume from forest fire hot spots, via the Atlantic Ocean, to the west coast of Europe. An NO₂ plume, comparable in magnitude to values over major anthropogenic emission regions, was found in the GOME data over the largest fire. This plume could be traced to Greenland on subsequent days. A weaker signal, very likely also due to the forest fires, was detected over the Atlantic Ocean and even close to Europe.Secondly, we studied intercontinental transport of nitrogen oxides from South African power plants to Australia in May 1998. This episode was also simulated with FLEXPART, which used NO_X emissions from the Global Emissions Inventory Activity (GEIA). Emission densities of the highly industrialised Highveld are among the highest in the world. Additionally, lightning emissions were added by utilising LIS data. Lightning NO_X amounted to around 10% of the simulated concentrations. Recently, the forest fire analysis was extended to a climatology of the fire seasons 1997 and 1998 with a focus on Siberian forest fire emissions. The fire season (May to October) of 1998 was, in contrast to 1997, very intense and dominated by very high fire activity in both Canada and Siberia. Transport of forest fire emissions was simulated with FLEXPART over the whole burning season 1998. NO_X was emitted from Siberian and Canadian forest fires according to the fire information from ATSR. The FLEXPART simulation shows that the strong boreal forest fires in 1998 enhanced both the TOMS aerosol index as well as GOME tropospheric column densities.Due to considerable uncertainties in the satellite data, in the chemical partitioning of NO_y, and wet and dry removal thereof, and unknown variations in fire intensity the results are rather qualitative. Climatalogical investigations and the quantification of NO_X emissions are our main future tasks.

Tropospheric NO₂ columns from GOME have been compared with a ground-based system covering the profile with stations situated at different heights. The mode of the observed ground-based concentrations yields a tropospheric column estimate, which is consistent with GOME by factor 2. The correlation is not significant, probably because of the highly variable representativity of the ground stations and coarse resolution of GOME.The added value of satellite data for air pollution analysis has been identified and demonstrated with case studies. For example, high pollution observed from GOME residing over Switzerland and adjacent France has been investigated. Pollution tracing employing trajectories allowed us to identify the source of this pollution to be in the region of Belgium, Netherlands and adjacent Germany.

The ultimate objective of the CARIBIC project is to create a database of in-situ measurements of a large number of trace species (~ 60) and aerosol properties in the upper troposphere and lower stratosphere (UTLS) which may be used to validate satellite measurements. Altogether 45 CARIBIC return flights were executed between November 1997 and May 2002 when the Boeing 767 of LTU International Airways was decommissioned. The CO and O₃ data for all flights were checked and incorporated into the CARIBIC data bank established on the basis of CERA2 (Climate and Environmental data Retrieval and Archive system) database designed by PIK, DKRZ, and AWL A new CARIBIC container with an extended instrumentation is being built for an Airbus A340-600 of Lufthansa AG. It is hoped to resume the measurements in September 2003 and to continue them for a period of 10 years.

The Global Ozone Monitoring Experiment (GOME) observes back scattered upwelling radiance from the atmosphere and the extra terrestrial solar irradiance between 240 and 790 nm. Inversion of the data yields the total atmospheric columns of atmospheric trace constituents (gases, aerosol and cloud). The Lightning Imaging Sensor (LIS) is an instrument aboard the Tropical Rainfall Measuring Mission (TRMM) satellite measuring lightning flashes. The objectives of this study have been to investigate the relative contribution of lightning and convective uplifting of pollutants from the boundary layer to the amount of NO₂ above clouds, using GOME and LIS data. The first results of a case study, a thunderstorm nearby the coast of Madagascar, shows that the NO_X production by lightning for this thunderstorm is in the range of 1–4 x 1025 molecules per flash.

This report is made on behalf of the Geophysica community, which comprises about 30 research groups across Europe, and forms part of the APERCU initiative for the exploitation of ENVISAT data.During the APE-THESEO mission in 1999, members of the Geophysica team retrieved data from Meteosat-5, and used this in combination with data from the Geophysica, to unravel the sometimes subtle links between convection and high-level tropical clouds. Satellite and in-situ data were used to distinguish non-convective tropopause cirrus from underlying convective anvils. It was shown that tropopause cirrus could be generated by convective systems that do not in themselves reach to the tropopause.Scientific campaigns, primarily for validation, but also testing the use of ENVISAT data synergistically with Geophysica data, took place in Italy, in July and October 2002. Geophysica flights probed the upper troposphere and lower stratosphere over southern Europe and the Mediterranean, a region that has been little studied. Small-scale structure was seen in long-lived tracers, indicative of quasi-isentropic stirring in both the upper troposphere and lower stratosphere, and rapid vertical mixing in the troposphere. This small-scale structure is also seen in three-dimensional Lagrangian transport model results. Resolving such small-scale structure will be a severe test of the ENVISAT instruments, but the intention is to follow the structures that ENVISAT instruments can resolve, and that are seen by Geophysica instruments, in time and space, to improve our understanding of transport and chemistry in the region.

Vertical inversion of ground-based high-resolution Fourier transform infrared (FTIR) solar absorption spectra is a promising tool to derive height-resolved information about a large number of atmospheric species. In particular it allows the determination of tropospheric abundances of gases that are important actors in tropospheric chemistry and in the earth's climate problem. Today, only few data sets of tropospheric abundances are available; satellite data are being released, but validation is still needed.To exploit the vertical inversion technique of FTIR data fully, its capabilities must be characterised properly . A full characterisation was performed for the retrieval of ozone profiles from FTIR spectra recorded at the International Scientific Station of the Jungfraujoch (ISSJ, 46.5 °N, 8° E, 3580 m a.s.l.), using the SFIT2 code based on the empirical implementation of the Optimal Estimation Method (OEM) (Rodgers 1976, Rodgers 2000).It was shown that an optimal retrieval sensitivity over the altitude range 3.6 km to 40 km was obtained when performing the retrieval in the micro-window 1000–1005 cm-1. The vertical resolution of the retrieval is of the order of 8 km. The retrieval has been characterised in terms of the error budget on the total column abundances and the partial column abundances in four independent layers, that are the 3.6–12 km, 12–18 km, 18–24 km and 24–40 km altitude ranges. The error budgets have been estimated based on the Rodgers theoretical framework, and the results have been compared by statistical comparisons of the FTIR retrieved data with independent correlative data from Dobson measurements, ozone soundings, lidar and microwave data, covering more than 4 years of data (June 1996 to November 2000). The largest offset of the FTIR height-resolved ozone data in comparison with the correlative data is of order 4 %, in the 18–24 km layer. No systematic biases have been found in the troposphere and the upper troposphere and lower stratosphere, UTLS. The dispersion of the relative differences, if any, is never larger than half of the natural ozone variability.Having established a reliable method for the retrieval characterisation for ozone, for which many correlative datasets exist, a similar procedure has been applied to the retrievals and associated characterisation of carbon monoxide (CO). CO profiles and total columns have been retrieved from FTIR spectra at the Jungfraujoch between January 1997 and May 2001. The results have been compared with in-situ surface data and, for the period between March 2000 and May 2001, with (version 2) CO profile and column data from the MOPITT instrument, Measurement of Pollution in the Troposphere, that was launched on December 18th, 1999 onboard the TERRA satellite and for which validation is still in progress. It has been found that the FTIR retrievals have a good sensitivity in the boundary layer where they compare well with in-situ surface measurements on a monthly basis. Taking into account the different sensitivities and height resolutions of the MOPITT and ground-based FTIR retrievals, we conclude that they agree well in daytime; night-time MOPITT measurements are higher by about 4 %. One mustn’t forget that the present study is based on version 2 MOPITT data and that the situation may be slightly improved for the version 3 data that are beginning to be released.

Retrieval of tropospheric trace constituent concentrations from satellite instrumentation can be augmented by the use of multiple datasets. This process can involve the use of a multitude of data sets of various atmospheric parameters, covering a wide range of spatial and temporal scales, which are currently available from measurements on space-, aircraft- and ground-based platforms. Algorithms that combine data from all or some of these platforms have the potential to give enhanced products. In this work, progress towards developing multi-platform methods for the derivation of tropospheric composition is reported. The work has two themes. Firstly, algorithms for synergistic multi-satellite derivations of tropospheric concentrations have been investigated. Secondly, ground-based instrumentation has been developed and advanced in order to validate the initial multi-satellite derivations and provide new data for potential multi-platform tropospheric retrievals.

Water vapour measurements from the MOZAIC (Measurement of Ozone and Water Vapour by Airbus In-Service Aircraft) aircraft platform in combination with satellite observations of sea surface temperature (SST) and cloud top temperature (CTT) have been used to investigate the control mechanisms of upper tropospheric humidity (UTH) content in regions of tropical cumulonimbus (Cb) convection over the equatorial Atlantic Ocean. We developed a concept, whereby MOZAIC air samples of convective origin are linked with the location of convection by back tracing with 3-D trajectories in combination with satellite observations of the cloud top temperature along the trajectories. At this location the SST can be related to the water vapour measured by MOZAIC. The concept has been successfully employed for the tropical Atlantic region using a five year MOZAIC record of UTH observations (from 1996 to 2000). The relative humidity in the outflow of Cb convection decreases in time much more slowly than subsidence under cloud-free conditions would predict. Under the assumption that there is no mixing along the trajectory and the specific humidity stays constant, the derived mean cooling rates in the outflow region subject to subsidence are 0.55 K per day, 1.24 K per day and 0.22 K per day over the periods 0–15 h, 15 h-24 h and 24–48 h, respectively, between convection and sampling at the aircraft. The results of the first and third period are more than a factor two lower compared with clear sky radiative transfer model results. The most plausible explanation is either a humidity or a energy source originated from the possible presence of local sub-visible cirrus clouds or by entrainment of moist air through sub-scale mixing processes.

Ground-based solar FTIR spectrometry at the NDSC Primary Station “Zugspitze” Germany (47.4 °N, 11.0 °E, 2964 m a.s.l.), hitherto mainly utilised for stratospheric monitoring, is investigated with respect to the tropospheric information that can be attained. The goals are to: i) validate and, ii) complement satellite data from ground by the FTIR technique together with complementary ground-based instrumentation.

The MOZAIC program was designed to collect ozone and water vapour data, using automatic equipment installed on board five long-range Airbus A340 aircraft flying regularly all over the world since August 1994 (Marenco et al. 1998). From ozone data recorded at cruise levels during a 2-year period (September 1994 to August 1996), the first accurate ozone climatology at 9–12 km altitude has been generated (Thouret et al. 1998a). From now on, we are providing different “elaborated” products such as the tropospheric ozone columns and the horizontal climatology with data referred to the tropopause altitude. We have chosen to use the tropopause altitude as the reference to get rid of its seasonal variations. Thus, we have access to the upper tropospheric ozone and to the lower stratospheric ozone distributions. In this first approach, we have chosen only to represent and analyse the measurements recorded at mid northern latitudes. In this study, we defined the tropopause as a mixing zone 30 mb thick centred on the surface PV = 2 PVU. Another set of climatologies is now available for the levels “tropopause ±15 mb” and “tropopause ±45 mb”. In the frame of TROPOSAT, this new set of climatologies demonstrates that we have started a development for future comparisons with the SCIAMACHY instrument, for example. The 8 first years of the MOZAIC program has allowed a first assessment of the inter-annual variability of ozone both in the free troposphere and in the UT/LS to be made. The results are surprisingly high (about 2 %/year). The year 1998 appears as a positive anomaly. Further studies have started to explain such a high increase of ozone in the troposphere and the lower stratosphere at northern mid-latitudes.

The first results from the retrieval of methane and carbon monoxide from the SCIAMACHY instrument are now available. The first results look promising although the validation work has, at the time of writing, just started. The actual quality of the measurements will be established during the coming 1 to 2 years by comparison with other satellite measurements, with chemical tracer model calculations and with measurements from ground-based stations. The technique of data assimilation will be used for extended validation and interpretation of the satellite data.

This effort mainly involved end-user support activities for the TROPOSAT community, for obtaining supportive data from the EMEP and NILU/NADIR databases. The EMEP database (Co-operative programme for monitoring and evaluation of the long-range transmission of air pollutants in Europe) contains long term NO₂, ozone, SO₂, VOC and aerosol measurements from about 100 European sites that can be used for selected test cases. Data available on the NILU/NADIR database includes ozone sondes for “case study” satellite and model validation of ozone in the free troposphere over Europe. In addition, European NDSC ground-based total column DOAS and FTIR measurements can assist in the determination of pollution episodes. Such data can be used for “case study” validation of satellite instruments such as GOME, SCIAMACHY and MIPAS. The data can also be used for model validation when used in combination with various modelling activities within the TROPOSAT project.

The four-dimensional variational (4D-var) data assimilation method has used for emission rate optimisation by ingesting real surface data in tandem with monthly averaged satellite retrievals of tropospheric NO₂ columns. As emission rates are not directly observable, concentration measurements of emitted species and ozone as product species, can be used for their estimation, given an inversion procedure of a full chemistry transport model (CTM).Hence, space-time assimilation algorithms can also be taken for the optimisation of other modelled parameters, like emission rates, deposition velocities, boundary values and others.In polluted areas, however, emission rates can be of higher importance on a time scale larger than half a day. Emission rate estimation with advanced modelling technique s is an active field of research since the work of (1974), who introduced the variational calculus for the solution of the problem. Misspecified emission parameters are an important error source for tropospheric CTMs. (2000) demonstrated for the first time, that, with suitably chosen regularisation rules, emission rate estimates of precursor species of ozone can be inferred with the variational calculus by assimilation of ozone observations. In this study the adjoint University of Cologne EURAD CTM is used for inversion. As observational input direct surface ozone measurements and monthly mean NO₂ columns are assimilated to investigate the potential benefits from direct retrievals of tropospheric columns. The results suggest that, with monthly averaged values of tropospheric columns, it is still the surface data of ozone measurements which ae more helpful in improving the ensuing forecasts.

The major contribution to TROPOSAT from Chalmers is to participate in the development of validation strategies for tropospheric satellite data products. Since 1994 the group has been operating a ground-based high resolution Fourier transform spectrometer for solar spectroscopic measurements at the Harestua site (60 °N, 11 °E) in southern Norway. Since 1996 the activity has been part of the global network NDSC. The work involves development of optimal strategies for comparing total columns and profiles from satellites with corresponding entities derived from ground based FTIR spectra, hereby taking into account the different vertical resolution, averaging kernels and viewing geometries of the two techniques, and focusing on the troposphere. During the time frame of TROPOSAT ground based high resolution solar FTIR has successfully been applied to retrieve atmospheric columns of HCl, HF, O₃ and CH₄ and we are presently working on CO. We have also actively participated in measurements for satellite validation related to the satellites MOPITT, ODIN and ENVISAT. During 2001 and 2002 a total of about 100 measurement days with high quality data has been obtained. So far no official validation results are available, partly due to problems with the satellite retrieval algorithms, partly problems with the ground-based algorithms.In a second approach, a strategy for satellite and model validation is being developed based on a mobile solar UV-visible/IR system. Direct and scattered sunlight is being used to derive total columns of tropospheric molecules on a grid-size comparable to the one obtained from satellites and used in GCM models. Such a mobile solar UV-visible/IR system has been developed and successfully tested for measurements of industrial emissions, emissions from ship traffic and volcanic gas plumes. It was also used on a field campaign in Italy in an EU project aimed at satellite validation of formaldehyde. These data are presently in the process of being evaluated. An additional, very compact, system has been developed utilising scattered sunlight. This system has been applied in measurements of volcanic SO₂ emissions from a large number of volcanoes in Europe, the West Indies and central America.

Satellite remote sensing techniques together with column measurements are increasingly being used for the estimation of tropospheric aerosols and trace gases. As both are often strongly linked to ground sources, the main fraction within a vertical column is likely to exist in the planetary boundary layer. Scattering of light on aerosols in the same altitude range where the substance is expected, makes the estimation of trace gas mixing ratios in the lower troposphere more difficult. Coexistence of trace gases and light scattering aerosols in the planetary boundary layer is, therefore, a challenge for satellite derived trace gas retrieval algorithms, and whether it is possible to compare satellite with ground based remote sensing or airborne in situ measurements depends on the knowledge about the vertical distribution of gases and aerosols. Altitude resolved remote sensing is limited to a few chemical compounds and aerosol properties and not sufficient for a quantitative description of the vertical profiles. For more detailed studies with independent measurement techniques, in situ investigations with airborne platforms are required. A system suitable to be used on small aircraft has been developed for vertical profile measurements. It consists of sensors for actinic UV radiation, aerosol properties and trace gas components and can be flown on a variety of small aircraft reaching altitudes well above the planetary boundary layer (PBL).

Validation of SCIAMACHY is essential to ensure the quality of the measurements. The activities are co-ordinated by the SCIAMACHY Validation and Interpretation Group (SCIAVALIG).There are four distinct phases of validation of SCIAMACHY data products: the preparatory phase, the commissioning phase, the main validation phase and the long-term validation phase. We are currently in the main validation phase, which will last until the end of 2003.ENVISAT was successfully launched in March 2002. The commissioning phase lasted until six months after launch. After the functional testing, the first nominal measurements of the earth shine spectrum have been performed. The validation activities in this phase have focused on a thorough inspection of the level 1 data products and quick-look comparisons and consistency checks for level 2 products.

Test measurements with an airborne ozone DIAL, based on a Nd:YAG pumped KTP-OPO are reported from a campaign over southern Germany in late August 2002. The ability of the instrument to measure tropospheric ozone concentrations with high spatial resolution has been proven. In the near future, it is planned to extend the set-up for SO₂ measurement and add some channels to the existing system for aerosol back-scattering.

The GIRPAS (Groupe Infra-Rouge de Physique Atmosphérique et Solaire) of the University of Liège (ULg) has continued to produce total vertical column abundances of over 20 telluric constituents above the international scientific station on the Jungfraujoch, Switzerland, as part of its commitment to long-term investigations of the variability and secular evolution of the earth's atmosphere above central Europe. Those species of primary relevance for quantifying and understanding stratospheric chemistry processes controlling the ozone layer depletion, stabilisation, as well as its anticipated recovery (thanks to the successive Montreal Protocol updates) are being studied within the frame of the NDSC (Network for the Detection of Stratospheric Change) of which the Jungfraujoch is a primary site for atmospheric monitoring at northern mid-latitudes. The others are source gases of natural and/or anthropogenic origin, affecting both the greenhouse strength as well as the oxidising capacity of the troposphere and are, thus, of interest to the aims of the Kyoto Protocol. The Jungfraujoch studies of the latter and their specific provision for the validation of past, ongoing and coming space sensors of the atmosphere are definitely in line with TROPOSAT objectives, i.e. seeking “Synergistic use of different instrumentation and platforms for tropospheric measurements”, and supporting the “Development of validation strategies for tropospheric satellite data products”. An original set of volume mixing-ratio profiles retrieved automatically from space-based infrared solar occultation measurements by the ATMOS instrument is reported, as it demonstrates that such profiles can be extended reliably down throughout the free troposphere for a large number of important source gases.