Nitrogen dioxide (NO2) is an important trace gas in the Earth's atmosphere. In the stratosphere, NO2 is strongly related to halogen compound reactions and ozone destruction . In the troposphere, nitrogen oxides (NOx=NO2+NO) serve as a precursor of zone in the presence of volatile organic compounds (VOCs) and of secondary aerosol through gas-to-particle conversion . As a prominent air pollutant affecting human health and ecosystems, large amounts of NO2 are produced in the boundary layer by industrial processes, power generation, transportation, and biomass burning over polluted hot spots. For instance, a strong growth of NO2 during the past 2 decades has caused severe air pollution problems for China, with the largest NO2 columns in 2011; since then, cleaner techniques and stricter controlling have been applied to reduce the NO2 pollution . An increase in NO2 concentrations due to economic growth is also found over India, with a peak in 2012 . Despite the decrease in NOx emissions in Europe, around half of European Union member states still exceed the air quality standards, mainly caused by diesel car emissions .

NO2 column measurements have been provided by satellite instruments, e.g. Global Ozone Monitoring Experiment (GOME) , SCanning Imaging Absorption SpectroMeter for Atmospheric CHartographY (SCIAMACHY) , Ozone Monitoring Instrument (OMI) , and Global Ozone Monitoring Experiment-2 (GOME-2) . NO2 observations will be continued by the new generation instruments with high spatial resolution such as TROPOspheric Monitoring Instrument (TROPOMI) launched in October 2017; and by geostationary missions such as Sentinel-4 . The GOME-2 instrument, which is the main focus of this study, is included on a series of MetOp satellites as part of the EUMETSAT Polar System (EPS). The first GOME-2 was launched in October 2006 aboard the MetOp-A satellite, and a second GOME-2 was launched in September 2012 aboard MetOp-B. The consistent long-term dataset will be further extended by the third GOME-2 on the upcoming MetOp-C platform (to be launched in September 2018). NO2 measurements from GOME-2 have been widely used to characterise the distribution, evolution, or transport of NO2 e.g., to estimate the NOx emission e.g., and to interpret VOC levels, ozone variation, or anthropogenic aerosol loading e.g..

The GOME-2 total and tropospheric NO2 products are generated using the GOME Data Processor (GDP) algorithm at the German Aerospace Center (DLR). The retrieval algorithm has been first described by as implemented in the GDP version 4.4 and was later updated to the current operational version 4.8 . The NO2 retrieval for GOME-2 follows a classical three-step scheme.

First, the total NO2 slant columns (namely the concentration integrated along the effective light path from the sun through the atmosphere to the instrument) are derived using the differential optical absorption spectroscopy (DOAS) method . The DOAS technique is a least-squares method fitting the molecular absorption cross sections to the measured GOME-2 sun-normalised radiances provided by the EUMETSAT's processing facility. The fit is applied on the data within a fitting window optimised for NO2. As analysed by and in the Quality Assurance for Essential Climate Variables (QA4ECV; http://www.qa4ecv.eu) project, extension of the fitting window for GOME-2 increases the signal-to-noise ratio and hence improves the NO2 slant column error. The total NO2 slant columns depend on the viewing geometry and also on parameters such as surface albedo and the presence of clouds and aerosol loads. They are therefore converted to initial total NO2 vertical columns through division by a stratospheric air mass factor.

Second, the stratospheric contribution is estimated and separated from the NO2 slant columns (referred to as “stratosphere–troposphere separation”). The GDP 4.8 algorithm applies a modified reference sector method, which uses measurements over clean regions to estimate the stratospheric NO2 columns based on the assumptions of longitudinally invariable stratospheric NO2 layers and of negligible tropospheric NO2 abundance over the clean areas. The modified reference sector method defines a global pollution mask to remove potentially polluted regions and applies an interpolation over the unmasked areas to derive the stratospheric NO2 columns. As a result of using a fixed pollution mask, the modified reference sector method in GDP 4.8 has larger uncertainties over polluted areas, because limited amount of information over continents is used. To overcome the shortcomings, the STRatospheric Estimation Algorithm from Mainz (STREAM) method has been developed for the TROPOMI instrument and was also successfully applied on GOME, SCIAMACHY, OMI, and GOME-2 measurements. Also belonging to the modified reference sector method, STREAM defines not a fixed pollution mask but rather weighting factors for each observation to determine its contribution to the stratospheric estimation.

Third, the tropospheric NO2 vertical columns are calculated from the tropospheric slant columns by an air mass factor (AMF) calculation, which contributes the largest uncertainty to the NO2 retrieval, particularly over polluted regions . The AMFs are determined with a radiative transfer model (RTM) and stored in a look-up table (LUT) requiring ancillary information such as surface albedo, vertical shape of the a priori NO2 profile, clouds, and aerosols. Improvements in the RTM and LUT interpolation scheme, the ancillary parameters, and the cloud and aerosol correction approach have been reported for the OMI instrument e.g., which in principle are beneficial for similar satellite instruments like GOME-2.

In this paper, a new algorithm to retrieve the total and tropospheric NO2 for the GOME-2 instruments is described, which includes improvements in each of the three algorithm steps introduced above. The improved algorithm will be implemented in the next version of GDP (referred to as GDP  4.9 hereafter). We briefly introduce the GOME-2 instrument (Sect. ) and the current operational (GDP 4.8) total and tropospheric NO2 retrieval algorithm (Sect. ). We present the improvements to the DOAS slant column retrieval (Sect. ), the stratosphere–troposphere separation (Sect. ), and the AMF calculation (Sect. ). Finally, we show an end-to-end validation of the tropospheric NO2 dataset using ground-based multiple-axis DOAS (MAXDOAS) datasets with different pollution conditions (Sect. ).

GOME-2 is a nadir-scanning UV–VIS spectrometer aboard the MetOp-A and MetOp-B satellites (referred to as GOME-2A and GOME-2B throughout this study) with a satellite repeating cycle of 29 days and an equator crossing time of 19:30 local time (LT) (descending node). The GOME-2 instrument measures the Earth's backscattered radiance and extraterrestrial solar irradiance in the spectral range between 240 and 790 nm. The morning measurements from GOME-2 provide a better understanding of the diurnal variations of the NO2 columns in combination with afternoon observations from for example the OMI, and TROPOMI instruments (13:30 LT). The default swath width of GOME-2 is 1920 km, enabling a global coverage in ∼1.5 days. The default ground pixel size is 80 km ×40 km in the forward scan, which remains almost constant over the full swath width. In a tandem operation of MetOp-A and MetOp-B from July 2013 onwards, a decreased swath of 960 km and an increased spatial resolution of 40 km ×40 km are employed by GOME-2A. See for more details on instrument design and performance.

The operational GOME-2 NO2 product is provided by DLR in the framework of EUMETSAT's Satellite Application Facility on Atmospheric Composition Monitoring (AC-SAF). The product processing chain starts with the level 0 to 1b processing within the core ground segment at EUMETSAT in Darmstadt (Germany), where the raw instrument (level 0) data are converted into geolocated and calibrated (level 1b) (ir)radiances by the GOME-2 Product Processing Facility (PPF). The level 1b (ir)radiances are disseminated through the EUMETCast system to the AC-SAF processing facility at DLR in Oberpfaffenhofen (Germany) and further processed using the Universal Processor for UV/VIS Atmospheric Spectrometers (UPAS) system. Broadcasted via EUMETCast, WMO GTS, and the Internet, the resulting level 2 near-real-time total column products including NO2 columns can be received by user communities 2 h after sensing. Offline and reprocessed GOME-2 level 2 and consolidated products are also provided within 1 day by DLR, which can be ordered via FTP server and the EUMETSAT Data Centre (https://acsaf.org/, last access: 1 February 2019).

The first main step of the retrieval algorithm is the DOAS technique, which is applied to determine the total NO2 slant columns from the (ir)radiance spectra measured by the instrument. Based on the Beer–Lambert law, the DOAS fit is a least-squares inversion to isolate the trace gas absorption from the background processes, e.g. extinction resulting from scattering on molecules and aerosols, with a background polynomial P(λ) at wavelength λ: ln⁡I(λ)+offset(λ)I0(λ)=-∑gSgσg(λ)-αRR(λ)-P(λ). The measurement-based term is defined as the natural logarithm of the measured earthshine radiance spectrum I(λ) divided by the daily solar irradiance spectrum I0(λ). The intensity offset correction offset(λ), which describes the additional contributions such as stray light in the spectrometer to the measured intensity, is modelled using a zero-order polynomial with the polynomial coefficient as the fitting parameter. The spectral effect from the absorption of species g is determined by the fitted slant column density Sg and associated absorption cross section σg(λ). An additional term with the Ring scaling factor αR and the Ring reference spectrum R(λ) describes the filling-in effect of Fraunhofer lines by rotational Raman scattering (the so-called Ring effect). The GDP 4.8 algorithm adopts a wavelength range of 425–450 nm to ensure prominent NO2 absorption structures and controllable interferences from other absorbing species, e.g. water vapour (H2Ovap), ozone (O3), and oxygen dimer (O4). Table  gives an overview of the DOAS settings for the current operational GDP 4.8 algorithm, the improved version 4.9 algorithm (see Sect. ), and the algorithm used in the QA4ECV product (see Sect. ).

The second component in the retrieval is the calculation of initial total vertical column densities Vinit using a stratospheric AMF (Mstrat) conversion: Vinit=SMstrat. Given the small optical thickness of NO2, Mstrat can be determined as Mstrat=∑lml(b)xlcl∑lxl, with ml the box-air-mass factors (box-AMFs) in layer l, xl the altitude-dependent sub-columns from a stratospheric a priori NO2 profiles climatology , and cl a correction coefficient to account for the temperature dependency of NO2 cross section . The calculation of Vinit assumes negligible tropospheric NO2 and hence uses only the stratospheric a priori NO2 profiles to derive AMF. The box-AMFs ml are derived using the multilayered multiple scattering LInearized Discrete Ordinate Radiative Transfer (LIDORT) RTM and stored in a LUT as a function of various model inputs b, including GOME-2 viewing geometry, surface pressure, and surface albedo. The surface albedo is described by the Lambertian equivalent reflectivity (LER). The surface LER climatology used in the GDP 4.8 algorithm is derived from combined TOMS–GOME measurements for the years 1979–1993 with a spatial resolution of 1.25∘ long. × 1.0∘ lat.

In the presence of clouds, the calculation of Mstrat adopts the independent pixel approximation based on GOME-2 cloud parameters: Mstrat=ωMstratcloud+(1-ω)Mstratclear, with ω being the cloud radiance fraction, Mstratcloud the cloudy-sky stratospheric AMF, and Mstratclear the clear-sky stratospheric AMF. Mstratcloud and Mstratclear are derived with Eq. () with Mstratcloud mainly relying on the cloud pressure and the cloud albedo. The ω value is derived from the cloud fraction cf: ω=cfIcloud(1-cf)Iclear+cfIcloud, where Icloud is the radiance for a cloudy scene and Iclear for a clear scene. Icloud and Iclear are calculated using LIDORT, depending mostly on the GOME-2 viewing geometry, surface albedo, and cloud albedo. From GOME-2, cf is determined with the Optical Cloud Recognition Algorithm (OCRA) by separating a spectral scene into cloudy contribution and cloud-free background, and the cloud pressure and the cloud albedo are derived using the Retrieval Of Cloud Information using Neural Networks (ROCINN) algorithm by comparing simulated and measured radiance in and near the O2 A band . Applied in the NO2 retrieval in GDP 4.8, the latest version 3.0 of the OCRA applies a degradation correction on the GOME-2 level 1 measurements as well as corrections for viewing angle and latitudinal dependencies. A new cloud-free background is constructed from 6 years of GOME-2A measurements from the years 2008–2013. The updated OCRA also includes an improved detection and removal of sun glint that affects most of the GOME-2 orbits. Version 3.0 of ROCINN applies a forward RTM calculation using updated surface albedo climatology and spectroscopic data as well as a new inversion scheme based on Tikhonov regularisation . The computation time of ROCINN is optimised with a smart sampling method .

The next retrieval step is the separation of stratospheric and tropospheric components from the initial vertical total columns, namely the stratosphere–troposphere separation. Since no direct stratospheric measurements are available for GOME-2, a spatial filtering algorithm is applied to estimate the stratospheric NO2 columns in GDP 4.8. The spatial filtering algorithm belongs to the modified reference sector method, which uses total NO2 columns over clean regions to approximate the stratospheric NO2 columns based on the assumption of longitudinally invariable stratospheric NO2 layers and of negligible tropospheric NO2 abundance over the clean areas. The spatial filtering algorithm uses a pollution mask to filter the potentially polluted areas (tropospheric NO2 columns larger than 1×1015 molec cm2), followed by a low-pass filtering (with a zonal 30∘ boxcar filter) on the initial total columns of the unmasked areas, and afterwards a removal of a tropospheric background NO2 (1×1014 molec cm2) from the derived stratospheric columns.

Finally, the tropospheric NO2 columns Vtrop can be computed as Vtrop=MstratMtrop×T, where Mstrat is the stratospheric AMF in Eq. (), Mtrop is the tropospheric AMF, and T is the tropospheric residues (T=Vinit-Vstrat). Mtrop is determined using Eqs. () and () with tropospheric a priori NO2 profiles. The calculation of Mtrop relies on the same model parameters as of Mstrat, but the dependency on the parameters like surface albedo and cloud properties as well as on the a priori NO2 profiles is much stronger. The GDP 4.8 adopts the tropospheric a priori NO2 profiles from a run of the global chemistry transport model MOZART version 2 with anthropogenic emissions from the EDGAR2.0 inventory for the early 1990s. The monthly average vertical profiles are calculated from MOZART-2 data from the year 1997 for the overpass time of GOME-2 (09:30 LT) with a resolution of 1.875∘ long × 1.875∘ lat.

A larger 425–497 nm wavelength fitting window for the DOAS method is implemented in the GDP 4.9 to retrieve the NO2 slant columns, which improves the signal-to-noise ratio by including more NO2 absorption structures. Compared to the extended 405–465 nm range, as employed by the QA4ECV GOME-2 NO2 product and used in the NO2 retrieval for OMI instrument , the 425–497 nm fitting window has stronger sensitivity to NO2 columns in the boundary layer because the importance of scattering decreases with wavelength . In this study, the slant columns are derived using QDOAS software developed at the Belgian Institute for Space Aeronomy (BIRA-IASB)

Note that the derived slant columns are scaled by geometric AMFs to correct for the angular dependencies of GOME-2 measurements in this section.

The calculation of tropospheric NO2 requires an estimation and removal of the stratospheric contribution to the initial total NO2 columns. In our GDP 4.9 retrieval, the stratosphere–troposphere separation algorithm STREAM has been adapted to GOME-2 measurements. Belonging to the modified reference sector method, STREAM uses initial total NO2 columns with negligible tropospheric contribution, i.e. unpolluted measurements at remote areas and cloudy measurements at medium altitudes, to derive the stratospheric NO2 columns. Based on a tropospheric NO2 climatology and the GOME-2 cloud product, STREAM calculates weighting factors for each satellite pixel to define the contribution of initial total columns to the stratospheric estimation; potentially polluted pixels are weighted low instead of being totally masked out in the GDP 4.8 spatial filtering method, cloudy observations at medium altitudes are given higher weights because they directly provide the stratospheric information, and the weights are further adjusted in a second iteration if pixels suffer from large biases in the tropospheric residues. Depending on these weighting factors, stratospheric NO2 fields are derived by weighted convolution on the daily initial total columns using convolution kernels. The convolution kernels are wider at lower latitudes due to the longitudinal homogeneity assumption of stratospheric NO2 and narrower at higher latitudes to reflect the stronger natural variations. To remove the biases in the weighted convolution resulting from the large latitudinal gradients, a latitudinal correction is applied on the initial total columns: the latitudinal dependencies of initial total NO2 are calculated over the clean Pacific, removed from the initial total NO2 before weighted convolution, and added back to the estimated stratospheric columns afterwards. However, we found that longitudinal variations of NO2 concentration resulted in biases in the latitudinal correction and hence in the stratospheric estimation. For the adaptation of STREAM to GOME-2 measurements, the performance of STREAM is analysed using synthetic GOME-2 NO2 observations (see Sect. ), and an improved latitudinal correction is applied (see Sect. ).

The validation of NO2 data derived from the GOME-2 GDP algorithm is part of the validation activities done at BIRA-IASB in the AC-SAF context . An end-to-end validation approach is usually performed for each main release and summarised in validation reports that can be found on AC-SAF validation website (http://cdop.aeronomie.be/validation/valid-reports, last access: 1 February 2019). This includes several steps, such as

the DOAS analysis results, cloud property retrievals, and AMF evaluations by confrontation of GOME-2 retrievals to other established satellite retrievals and AMF evaluations;

NO2 columns retrieved from measurements of the GOME-2 aboard the MetOp-A and MetOp-B platforms have been successfully applied in many studies. The abundance of NO2 is retrieved from the narrow band absorption structures of NO2 in the backscattered and reflected radiation in the visible spectral region. The current operational retrieval algorithm (GDP 4.8) for total and tropospheric NO2 from GOME-2 was first introduced by , and an improved algorithm (GDP 4.9) is described in this paper.

To calculate the NO2 slant columns, a larger 425–497 nm wavelength fitting window is used in the DOAS fit to increase the signal-to-noise ratio. Absorption cross sections are updated and a linear intensity offset correction is applied. The long-term and in-orbit variations of GOME-2 slit function are corrected by deriving effective slit functions with a stretched preflight GOME-2 slit function and by including a resolution correction function as a pseudo absorber cross section in the DOAS fit, respectively. Compared to the GDP 4.8 algorithm, the NO2 columns from GDP 4.9 are higher by ∼1–3×1014 molec cm2 (up to 27 %) and the NO2 slant column noise is lower by ∼24 %. In addition, the effect of using a new version (6.1) of the GOME-2 level 1b data has been analysed in our NO2 algorithm. The application of new GOME-2 level 1b data largely reduces the offset between GOME-2A and GOME-2B NO2 columns by removing calibration artefacts in the GOME-2B irradiances (due to Xe-line contaminations in the calibration key data). Compared to the GOME-2 NO2 product from the QA4ECV project, the NO2 columns from GDP 4.9 show good consistency and the NO2 slant column noise is ∼14 %–28 % smaller, indicating a good overall quality of the improved DOAS retrieval.

The stratosphere–troposphere separation algorithm STREAM, which was designed for TROPOMI, was optimised for GOME-2 instrument. Compared to the spatial filtering method used in the GDP 4.8, STREAM provides an improved treatment of polluted and cloudy pixels by defining weighting factors for each measurement depending on polluted situation and cloudy information. For the adaption to GOME-2 measurements, the performance of STREAM is analysed by applying it to synthetic GOME-2 data and by comparing the difference between estimated and original stratospheric fields. Applied to synthetic GOME-2 data calculated by a RTM using C-IFS model data, the estimated stratospheric NO2 columns from STREAM show good consistency with the a priori truth. A slight overestimation by ∼1–2×1014 molec cm2 is found over lower latitudes, and larger differences of up to ∼5×1014 molec cm2 are found at higher latitudes. To reduce the biases over the subtropical regions in winter, an improved latitudinal correction is used in STREAM. Applied to GOME-2 measurements, the updated STREAM successfully separates the stratospheric and tropospheric contribution over polluted regions, especially in the Northern Hemisphere. Compared to the current method in the GDP 4.8, the use of STREAM slightly decreases the stratospheric NO2 columns by ∼1×1014 molec cm2 in general and largely reduces the overestimation over polluted areas.

To improve the calculation of NO2 AMF, a new box-AMF LUT was generated using the latest version of the VLIDORT RTM with an increased number of reference points and vertical layers to reduce interpolation errors. The new GOME-2 surface LER climatology used in this study is derived with a high resolution of 1∘ long. ×1∘ lat. (0.25∘ long. ×0.25∘ lat. at coastlines) and an improved LER algorithm based on observations for 2007–2013. Daily a priori NO2 profiles, obtained from the chemistry transport model TM5-MP, capture the short-term variability in the NO2 fields with a resolution of 1∘ long. × 1∘ lat. A large impact on the retrieved tropospheric NO2 columns (more than 10 %) is found over polluted areas.

The uncertainty in our GDP 4.9 NO2 slant columns is 4.4×1014 molec cm2, calculated from the average slant column error using a statistical method described in Sect. . The uncertainty in the GOME-2 stratospheric columns is ∼4–5×1014 molec cm2 for polluted conditions based on the daily synthetic GOME-2 data and ∼1–2×1014 molec cm2 for monthly averages. The uncertainty in the tropospheric AMFs is estimated to be in the 10 %–45 % range, considering the use of updated box-AMF LUT and improved surface albedo climatology and a priori NO2 profiles, resulting in a total uncertainty in the tropospheric NO2 columns likely in the range of 30 %–70 % for polluted conditions.

An end-to-end validation of the improved GOME-2 GDP 4.9 dataset was performed by comparing the GOME-2 tropospheric NO2 columns with BIRA-IASB ground-based MAXDOAS measurements. The validation was illustrated for different MAXDOAS stations (Beijing, Bujumbura, OHP, Réunion, Uccle, and Xianghe) covering urban, suburban, and background situations. Taking Xianghe station as an example, the GDP 4.9 dataset shows a similar seasonal variation in the tropospheric NO2 columns as the MAXDOAS measurements with a relative difference of -5.8 % (i.e. -2.7×1015 molec cm2 in absolute) and a correlation coefficient of 0.91 for GOME-2A, indicating good agreement. The Xianghe site, by its suburban nature, is the best site for validation. At the other sites, mean biases range from (-47 %; -16×1015 molec cm2) for Beijing, (-76 %, -74 %; -3.6×1015 molec cm2, -2.8×1015 molec cm2) for Bujumbura, (-25 %, -17 %; -0.9×1015 molec cm2, -0.5×1015 molec cm2) for OHP, (-64 %, -47 %; -1.5×1015 molec cm2, -0.8×1015 molec cm2) for Réunion, and (-43 %, -40 %; -5×1015 molec cm2, -4.2×1015 molec cm2) for Uccle. Réunion and Bujumbura are difficult sites for validation, due to their valley and mountain nature, while urban sites Beijing and Uccle show similar relative results. A smaller absolute bias is found at the rural OHP station. Compared to the current operational GDP 4.8 product, the GDP 4.9 dataset is a significant improvement. Although GOME-2 measurements are still underestimating the tropospheric NO2 columns with respect to the ground data, the absolute and relative differences with the different MAXDOAS stations are smaller, both for the original comparisons and for the comparisons with the smoothed MAXDOAS columns.

In the future, the AMF calculation will be further improved, since uncertainty in AMF is one dominating source of errors in the tropospheric NO2 retrieval, especially over polluted areas. The surface bidirectional reflectance distribution function (BRDF) effect will be included using a direction-dependent LER climatology from GOME-2 (L. Gijsbert Tilstra, personal communication, 2018) to describe the angular distribution of the surface reflectance. Aerosol properties will be considered explicitly in the RTM calculation using ground-based aerosol observations from, e.g. MAXDOAS instruments, Mie scattering lidars, or sun photometers operated by the AErosol RObotic NETwork (AERONET). A priori NO2 profiles from different global and regional models will help to analyse the effect of spatial resolution, temporal resolution, and emissions on the tropospheric NO2 retrieval for GOME-2. Furthermore, the NO2 algorithm will be adapted to measurements from the TROPOMI instrument with a spatial resolution as high as 7 km ×3.5 km.