An effective combination of the limb and nadir measurements of atmospheric composition can provide a new information about atmospheric composition. The ESA -funded project SUNLIT - Synergy of Using Nadir and Limb Instruments for Tropospheric ozone monitoring - is aimed at creating the tropospheric ozone column dataset using combination of total ozone column from OMI and TROPOMI with stratospheric ozone column dataset from all available limb-viewing instruments.
In the project, the methods for estimating stratospheric ozone contents by exploitation measurements from several limb-profile instruments are developed. The stratospheric ozone columns from different satellite instruments are compared with each other and with measurements by nadir instruments in cloudy conditions. A multi-instrument stratospheric ozone dataset, compatible with nadir total ozone column measurements, is created as an intermediate step of our project.
For developing an optimal method for interpolation of stratospheric ozone column from limb instruments on the fine-resolution grid of nadir instruments, we studied the variability of ozone field using both satellite data and simulations with the SILAM chemistry transport model.
The tropospheric ozone column is obtained using the residual method, i.e., by subtracting the stratospheric ozone column from total ozone column measurements. In our presentation, we show the results of the project analyses.

Area C - Board 255

The Copernicus missions Sentinel-4 (S4), Sentinel-5 (S5), and the Sentinel-5 Precursor (S5P) are dedicated to atmospheric composition and have been designed to serve the specific needs of the Copernicus Atmosphere Monitoring Service (CAMS). This presentation focuses on the S4 and S5 missions.

The atmospheric Sentinels have been conceived as elements of a constellation with complementary observational capabilities. The geostationary S4 mission will provide tropospheric composition data over Europe with hourly revisit time mainly for air quality applications. The target species of the Sentinel-4 mission include the key air quality parameters with pronounced temporal variability such as NO2, O3, SO2, HCHO, CHOCHO, and aerosols. The low-Earth orbit missions, S5 and S5P, will provide composition data with global coverage on a daily basis serving climate, air quality, and ozone/surface UV applications. The S5 and S5P target species include the S4 target species plus longer lived components such as CO, CH4, and stratospheric O3. The space component of the atmospheric Sentinels builds on heritage from instruments such as GOME, SCIAMACHY, GOME-2, and OMI and comprises imaging spectrometer instruments with significantly enhanced spatial resolution. While the S4 instrument covers the Ultra-violet, Visible, and Near infrared spectral range (S4/UVN) the S5 instrument covers additionally the Short wave infrared (S5/UVNS). S5 will be preceded by the S5P (S5 Precursor) mission that will bridge the data gap between OMI and S5. S4 establishes the European component of a constellation of geostationary instruments with similar observational capabilities together with the NASA mission TEMPO and the Korean mission GEMS.

ESA is responsible for the development of the protoflight models and the recurrent units of the S4/UVN and the S5/UVNS instruments and their payload data processors. For S4 and S5, the instruments and the Level-1b prototype processors are being built by a consortium led by Airbus Defence and Space. Consortia led by DLR and S&T are developing respectively the S4 Level-2 operational processor and the S5 Level-2 prototype processor. The S4/UVN instruments will be embarked on the Meteosat Third Generation-Sounder (MTG-S) platforms. Two S4/UVN instruments will be flown in sequence spanning an expected mission lifetime of 15 years. The S5/UVNS instruments will be embarked on the MetOp-Second Generation (MetOp-SG) satellite A platforms. Three S5/UVNS instruments will be flown in sequence spanning an expected mission lifetime of 21 years. The Flight Acceptance Reviews of the first Metop-SG and MTG-S satellites are expected to take place in 2021 and 2022, respectively. EUMETSAT will operate the S4/UVN and S5/UVNS instruments and will process the mission data up to Level-2.

An overview of the atmospheric composition capability of the two complementing missions will be presented.

Area C - Board 256

Nowadays a vast amount of operational satellite-based Earth Observations (EO) products are available, many of which have the potential to be useful for air quality applications. In order to better exploit these data, a three-year ESA funded project Satellite based Monitoring Initiative for Regional Air quality (SAMIRA) was established in 2016. The overall goal of SAMIRA is to improve regional and local air quality monitoring through synergetic use of satellite data, output from chemical transport models and data from in situ air quality monitoring networks. This is a collaborative effort of a team located in four countries (Poland, Romania, The Czech Republic, and Norway), where the capitals, the Gorj county in Romania, and the Silesia region, the border area between Poland and The Czech Republic, periodically suffer from air pollution.

We first concentrated on work with historical datasets, which cover the period June to September 2014. For this period WRF-Chem output was generated on a 1 km x 1 km local grid for the particular polluted areas mentioned above and on a 5 km x 5 km European grid for improved PM forecasts using in situ data assimilation. For spring 2019 a Near-Real-Time (NRT) demonstration is planned.

A first activity is the further development of an algorithm for the retrieval of aerosol optical depth (AOD) from the Spinning Enhanced Visible and InfraRed Imager (SEVIRI). Visualisation of SEVIRI AOD 15 min maps and their respective uncertainties are shown for the four countries. In a next step particulate matter (PM₂.₅) is derived from SEVIRI AOD using micro-physical properties from GADS/OPAC and mass mixing ratios of aerosol species from WRF-Chem model. Look-up tables (LUT) with extinction efficiencies for a range of possible mixtures of aerosols and different humidity conditions are generated. We show hourly PM₂.₅ maps obtained with this method and present comparisons with in situ air quality station data. Data fusion techniques are employed to utilize satellite products of atmospheric composition for European- and National scale air quality mapping. The additional benefit of satellite-based monitoring over existing monitoring techniques (in situ, models) is tested by combining hourly, daily and annual data using geostatistical methods and demonstrated for nitrogen dioxide (NO₂), sulphur dioxide (SO₂), and AOD/PM for rural and urban areas. Air quality applications, in particular within a city, require very high spatial resolutions, which are not yet available from satellites. In order to address this issue we have developed a spatial downscaling technique for satellite-based air quality products. The method applies a combination of area-to-point kriging and regression and essentially combines a high-resolution but often biased proxy dataset with the coarse-resolution but assumed to be unbiased satellite observations (from Aura/OMI and Sentinel-5/TROPOMI). We summarize the validation efforts for evaluating the quality of the generated products and activities linking to interested users, and the results from the planned NRT demonstration.

Area C - Board 257

Air pollution is a severe problem costing billions of euros each year in lost productivity and health expenses. Nowadays, more people die every year from air pollution than from malaria and HIV combined. Around 1.4 million people in China, 650,000 in India and 180,000 in Europe. Traditional methods to collect air quality data (e.g. ground monitoring data networks and airborne campaigns) are costly and limited in both timespan and geographical coverage. New technologies based on satellite data complement and fill the gaps of these traditional technologies providing innovative services at lower costs. The Tropomi instrument in particular, provides relevant insights due to its observations with higher spatial resolution and lower noise. Within the scope of the EU’s Research and Innovation programme Horizon 2020, AirQast is a self-sustainable air quality system that targets the demand for high quality and up-to-date data in many regions and cities. AirQast provides quality assured data of pollutants that have a high economic and social impact such as PM2.5, PM10, SO2, NO2 and CO, in the shape of updated emissions inventories, street level real-time monitoring, advanced forecasting systems and decision making tools to manage air quality events.

AirQast applications are based on different sources of data: Earth Observation, meteorological, monitoring ground and socioeconomic data for impact assessment (e.g. population distribution, open street maps). The Copernicus data include standard satellite products from European missions (Sentinel-3, Metop, Sentinel-5P) as well as data from Copernicus Atmosphere Monitoring Service (e.g. global air quality forecast and data reanalysis). These datasets are provided to the AirQast regional models CHIMERE of KNMI, SILAM of FMI, which assimilate these observations and create the variety of the regional AirQast products.

The study presents AirQast results at regional, urban and street level scales. Advanced algorithms have enabled the development of up-to-date emissions inventories in China by assimilating the observations from satellites and in-situ and merging them with the existing inventories finally ending up with the top-down model-based emission assessments. Satellite observations have been used to derive emissions for regions with 0.25 degree resolution. These top-down emissions have been combined with existing sector- split and statistical information from the ground. Based on these emission, an ensemble of regional air quality forecasts has been put in place. This forecast can be used to anticipate episodes of strong air pollution and warn the public in a timely manner.

At urban scale, EU-wide road traffic emission scenarios can be conducted for any city in the EU-28, thanks to the EU Traffic Data Mapper (ETDM) which provides estimates for the number of vehicles on all European roads based on COPERT 2014 mileage data, the OpenTransportMaps network and road typology, and proxy data based on earth observation data (e.g. population data). This data in conjunction with the CAMS European air quality reanalysis data facilities implementation across the EU.

The study also shows the performance of AirQast applied to street level in Amsterdam, where data from official ground stations is combined with low cost sensors installed by citizens and socio-economic data, with and an air quality distribution model for the smart interpolation of monitoring results as well as an improved 3-day forecast for NO2.

Preliminary results of the implementation of an EO-based air quality system are reviewed in terms of social impact and performance with respect to traditional methods at 3 different scales.

Area C - Board 258

Operational satellite-based Earth Observation products for air quality applications tend to have a relatively coarse spatial resolution of several kilometers to tens of kilometers. While the Sentinel-5P/TROPOMI instrument with its 7 km by 3.5 km footprint at nadir provides significantly improved spatial detail compared to previous satellite instruments, it is not yet sufficient for air quality applications within a city, where spatial gradients are often sharp and citizen exposure to air pollution varies significantly from street to street.

In order to address this issue we have developed a spatial downscaling technique for satellite-based air quality products. This was carried out as part of the ESA-funded project SAMIRA (“SAtellite based Monitoring Initiative for Regional Air quality”). The method applies geostatistics and a temporally static or dynamic high-resolution proxy dataset to downscale the satellite information from the original Level-2 pixel geometry to any arbitrary (usually regular) high-resolution pixel geometry. More specifically it uses area-to-point kriging and regression and can be used directly on Level-2 swath data without the need for prior gridding. It is further capable of providing pixel-level uncertainty estimates taking into account the original product uncertainty and the uncertainty introduced by the downscaling.

We demonstrate the methodology with data from Sentinel-5P/TROPOMI and Aura/OMI. We focus here primarily on nitrogen dioxide (NO2), but the algorithm is generic and can be readily used for other species. As spatial proxy datasets we use high-resolution output from the EPISODE and WRF-Chem models as temporally dynamic proxy data, as well as long-term average high-resolution NO2 maps for Europe as time-invariant proxy information. We demonstrate the algorithm for selected regions in Europe with significant air quality issues (Po Valley, Ruhr area, Silesia, Warsaw, and others).

Initial results indicate that the method provides qualitatively realistic high-resolution maps (at around 100-500 m spatial resolution), for which the absolute values are provided by the satellite observations and the sub-grid spatial patterns are inherited from the fine-scale spatial proxy dataset. Re-aggregating the downscaled results to the original satellite pixel geometry reproduces the original satellite-measured values. Quantitatively, a comparison against stations observations of surface NO2 shows that the correlation increases significantly for the downscaled maps compared to the correlation between the original satellite dataset and the station data. In addition, using Sentinel-5P/TROPOMI as a high-resolution reference to validate downscaled OMI/Aura data results in a substantial decrease in root mean square error for the downscaled product.

Overall the method is a first step towards exploiting the Sentinel-5P/TROPOMI data for applications at the urban scale and further to bridge the scale gap between satellite monitoring of air quality and other upcoming methods such as dense low-cost sensor networks.

Area C - Board 259

Air quality mapping plays an important role in informing the public about air pollution levels in various areas. The most often used data sources for air quality mapping are in-situ measurements and chemical transport model (CTM) outputs. Next to these data sources, satellite resp. earth observation data show a large potential for the maps improvements. None of these datasets alone is ideal for mapping purposes due to either substantial data gaps, significant uncertainties and biases, or insufficient spatial resolution. Within the scope of the ESA funded SAMIRA project, we have developed methods to combine these data sources using data fusion, namely residual kriging.

The principle of the residual kriging method is multiple linear regression followed by interpolation of its residuals using the geostatistical method kriging. Due to the differences in the structure of the concentration field in urban and rural areas, the input data are combined separately for urban and rural measurement stations and then merged based on population density. After initial testing on historical data, we have aplied these methods in near real time mapping over the region of the Czech Republic to provide hourly maps of NO₂ concentrations in the resolution of 1 x 1 km. For this, we are combining in-situ measurements from the air quality database of the Czech Hydrometeorological Institute, outputs of the pre-operational runs of the chemical transport model CAMx and observations from Sentinel-5P/TROPOMI. The maps are evaluated in terms of root mean square error and bias automatically using the 'leave one out' cross-validation method.

Area C - Board 260

In the near future, the northern hemisphere will be closely monitored by three spectrometers, TEMPO, GEMS and Sentinel-4/UVN, that will make high spatial and temporal resolution, reflected sunlight radiance measurements, that will be used to infer air quality over the most populated regions of the planet. During the past decades many techniques have been developed to infer column concentrations of gases like O3 and NO2 from similar measurements made from low earth orbit and these techniques can be modified slightly to produce the same column integrated results using measurements made from a geostationary platform. However, the solar and viewing geometries associated with an instrument observing from geostationary orbit may be sufficiently different such that the data quality is compromised when measurements are made at high northern latitudes. This work will present preliminary results associated with data quality expected from TEMPO measurements of the northern Canadian oil sands. However, these results are equally applicable to Sentinel-4 measurements of Scandinavian cities at similar latitudes.
Our group at the University of Saskatchewan has developed a full simulator package to test ideas related to: spherical geometry radiative transfer; improved surface reflectance distributions within the radiative transfer model; improved spectral fitting; and improved air mass factors derived with more complicated radiative transfer modelling. This work has been funded by grants given by Environment and Climate Change Canada (ECCC) with a goal of enhancing the utility of the TEMPO measurements within the ECCC air quality forecast scheme. Within this poster, we will present preliminary results related to the sensitivity of both the precision and accuracy of the retrieved results given the computational requirements associated with these new ideas as compared to the computational time required by the current standard processing techniques.

Area C - Board 261

Air is indispensable for life. Air pollution threatens the public health but it’s impact on human health is poorly considered most of the time. There are many types of air pollution, and each have a different effect on human health. The most common type of air pollution in the developing countries is Nitrogen Di-oxide pollution. It irritates the respiratory system, making it difficult to breathe. It accelerates the formation of photo-chemical smog in urban area, which can have significant impacts on human health. In the developing countries, majority people do not want to spend more money for monitoring their surrounding air quality as most of the air quality measuring sensors are expensive. Moreover, to cover large area for air quality monitoring, the number of air quality sensors should increase. Therefore, few number of satellites have already launched to monitor air quality from space and more number of satellites will be planned in near future for air quality monitoring.
Sentinel 5P is new earth observation satellite for global atmospheric monitoring and satellite data are freely available. The spatial resolution of Sentinel 5P products is 7 km times 3.5 km. It measures total column of Ozone, Nitrogen Di-oxide (NO2), Sulfur Di-oxide, Formaldehyde, Aerosol, Carbon Mon-oxide, Methane and Clouds tropospheric concentration. Among this Nitrogen Dioxide (NO2) is considered in this study because the breathing in raised levels of nitrogen dioxide increases likelihood of respiratory problems. Nitrogen dioxide inflames the lungs cover cells, rise risk for Asthma and it can reduce immunity to lung infections. So, the idea is to justify the impact of tropospheric Nitrogen Dioxide (NO2) concentration on different lungs disease occurrences of any region.
Bangladesh has been selected for study area. According to Clean air and sustainable environment (CASE) project report 2018, Most of the city air of Bangladesh is ranged unhealthy to very unhealthy (Department of Environment, Government of the People’s Republic of Bangladesh). Because of unhealthy air and largest population density, the impact of air pollution on human health will be higher in Bangladesh.
To get spatial public health disease related data is not easily available for Bangladesh. The Local Health Bulletin data from Ministry of Health and Family Welfare in Bangladesh, is used for this study. The Local Health Bulletin composed of local government hospital facility information, annual hospital attended patient disease information including the number of patient of each disease. From top ten local diseases based on the local patient annual number, only the lungs diseases have been sorted from Local Health Bulletin for this analysis. The name of the lungs disease considered in this study are Asthma, Acute Severe Asthma, Chronic Obstructive Pulmonary Disease and Chronic Obstructive Pulmonary Disease with acute exacerbation. From the local health bulletin yearly data, the percentages of lungs disease occurrence have been calculated for fifty-four district in Bangladesh. It is important to mention that, district is the second largest administrative boundary in Bangladesh.
The Sentinel 5P total tropospheric column NO2 concentration data was downloaded from EO browser. The unit of the total tropospheric column concentration is molecule per square meter. Due to cloud coverage and availability of satellite data, only single date total tropospheric column NO2 concentration satellite data is used for this study. The basic assumption of this research is “Spatial variability of the total tropospheric column pollutant concentration is almost constant for a year”. As this research generalizes total tropospheric column pollutant concentration values by assumption, it is not correct to use those values for defining pollution threshold. The maximum cloud free coverage total tropospheric column NO2 concentration data has been chosen for this study. The total column NO2 concentration were calculated from Sentinel 5P product for sixty-four district in Bangladesh. The ordinary kriging interpolation using spherical model was applied in ArcGIS software for calculation here. The calculated total column NO2 concentration data for sixty-four districts were exported from ArcGIS into MS Excel.

In MS Excel, the annual lungs disease occurrence percentages data (dependent variable) were plotted on total air column NO2 concentration data of Bangladesh (independent variable). The Correlation Coefficient from Multiple R analysis and Coefficient of Determination from R Square were calculated for this two data variable namely annual lungs disease occurrence percentages data and total air column NO2 concentration data. The calculated Correlation Coefficient value is 0.34, indicating positive linear relationship between two data variables. Although, this research finds very low coefficient value but it establishes the truth that total air column NO2 concentration has proportional relationship with lung disease occurrences in Bangladesh. The obtained R Square value is 0.12. In other words, 12% of the dependent variables (annual lungs disease occurrence percentages data) are explained by the independent variables (total air column NO2 concentration data). That means total air column NO2 concentration has very weak linear relationship with lung disease occurrences. Here, the R Square value is underestimated because, this study uses only one hospital data (governmental district general hospital) for every district in Bangladesh.

This research is innovative because of its data use case and methodology. The resulted statistics is mostly underestimated from the reality because of public health sector data availability in Bangladesh. It is possible to obtain strong linear relationship between atmospheric NO2 concentration and lung disease occurrences if both public and private hospital patient data will be available for the proposed methodology. The important limitation of this study is to use only single date total air column NO2 concentration satellite data against annual lung disease occurrences. Due to time constraint, only few lungs disease occurrences data were considered for this study. For the future research, it is proposed to use multiple date total air column NO2 concentration satellite data for the same methodology. For, more accurate result the primary data collection of seasonal lungs disease occurrences is recommended for future researcher. However, this research identified the use case of Sentinel 5P satellite data for public health services in developing countries. It is expected that; this research outcome can increase the public awareness against air pollution in the developing countries. The proposed methodology can be used for other countries for research purposes.

Area C - Board 262

The Boundary-layer Air Quality Using Network of INstruments (BAQUNIN) project primary mandate is the validation of Atmospheric Composition Satellite products in Urban, Semi-Rural and Rural environment using a network of active (LIDAR, SODAR) and passive (e.g. Pandora, Brewer, sun-photometers) remote sensing instruments located in the centre of Rome (Sapienza University), in Tor Vergata (CNR-ISAC) and Montelibretti (CNR-IIA). Of particular importance is the Sapienza site, located in the centre of Rome and, thus, most suitable for urban pollution investigations. In this work, we propose a simple but effective method for the characterisation of trace gases horizontal gradients in Urban areas, using Pandora #117 multi-axes Sky observations acquired in the context of the BAQUNIN project. The method consists in retrieving “sequences” of tropospheric columns and surface concentrations of NO2 and HCHO for Sky measurements taken at six different azimuth angles. A direct-Sun measurement is performed after each Sky scan, and the sequence cycle is completed in less than 20 minutes, during which no significant change in the atmospheric state it is assumed. For each sequence, a set of pseudo-coordinates are defined as:
X(i) = D(i) sin(PAA(i))
Y(i) = D(i) cos(PAA(i))
Where i=1-6 is the azimuth position index, PAA is the Pandora pointing azimuth angle (0°, 60°, 120°, 180°, 240°, 300° degrees North), and D is the tropospheric column (or surface concentration) value retrieved for each Sky scan. The six (X,Y) pairs are then used to calculate the parameters describing a generic ellipse allowing the estimate of the related eccentricity and tilting angle, while the position of the barycentre is the pseudo-coordinate average. These three “Pandora inhomogeneity parameters”, hereafter called PIPs, condensate the information about the deviations from horizontal homogeneity, characterised by eccentricity=0, barycentre in the origin of pseudo-axes, and tilting angle not defined. The main advantage of such approach is that the PIPs values do not depend on species concentration values but only on their spatial distribution. The possible links between PIPs and wind speed/direction, atmospheric divergence and vorticity, estimated through Weather Research and Forecasting (WRF) Model simulations and averaged over a 10 km circle centred at Pandora #117 location (Rome downtown), have been investigated from two perspectives: a) to provide a geophysical explanation to the observations and b) to possibly rely on WRF model runs to forecast trace gases concentration inhomogeneity through an empirical parameterisation derived from Pandora data.
It should be stressed here that, the assessment of the impact of tropospheric species horizontal inhomogeneity on satellite retrievals plays a fundamental role in the context of TROPOMI Cal/Val activities.
Results of this investigation, including a detailed description of the methodology, will be reported and discussed in this work.

Area C - Board 263

The Smithsonian Astrophysical Observatory-led TEMPO instrument is the first NASA Earth Venture Instrument. It is to be launched between 2020 and 2021 to measure atmospheric pollution from Mexico City and Cuba to the Canadian oil sands, and from the Atlantic to the Pacific. It does this hourly at high spatial resolution, ~10 km2, to measure the key elements of air pollution chemistry. Geostationary daytime measurements capture the variability in the diurnal cycle of emissions and chemistry at sub-urban scale to improve emission inventories, monitor population exposure, and enable emission-control strategies. TEMPO is now completed and is awaiting selection of its commercial host satellite

TEMPO will measure UV/visible Earth reflectance spectra to retrieve ozone (O3, including tropospheric and planetary boundary layer O3). NO2, SO2, H2CO, C2H2O2, H2O, BrO, OClO, IO, aerosols, cloud parameters, and UVB radiation. It will track aerosol loading, transport and, transformation. It will provide near-real-time air quality products. TEMPO is the North American component of the upcoming global geostationary constellation for pollution monitoring, together with the European Sentinel-4 (TROPOMI) and the Korean Geostationary Environmental Monitoring Spectrometer (GEMS).

TEMPO science studies may include: Solar-induced fluorescence from chlorophyll over land and in the ocean to study tropical dynamics, primary productivity and carbon uptake, to detect red tides, and to study phytoplankton; measurements of stratospheric O3 intrusions that cause air quality exceedances; measurements at peaks in vehicle travel to capture the variability in emissions from mobile sources; measurements of thunderstorm activity, including outflow regions to better quantify lightning NOx and O3 production; cropland measurements to follow the temporal evolution of emissions after fertilizer application and from rain-induced emissions from semi-arid soils; investigating the chemical processing of primary fire emissions and the secondary formation of volatile organic compounds and O3; prognostics of fire danger using VOCs, CO, and H2O; examining ocean halogen emissions and their impact on the oxidizing capacity of coastal environments; measuring spectra of nighttime lights, resolved by lighting type, as markers for human activity, energy conservation, and compliance with outdoor lighting standards intended to reduce light pollution.

Area C - Board 264

Vertical profiles of atmospheric variables are often obtained with the inversion of remote sensing observations performed by instruments operating on space-borne and airborne platforms, as well as from ground-based stations. When the same portion (or nearby portions) of atmosphere is observed more times by the same instrument or by different instruments the measurements can be combined in order to obtain a single vertical profile of improved quality with respect to that of the profiles retrieved from the single observation. The simultaneous retrieval from several observations is considered the most comprehensive way to combine different measurements of the same quantity, however, recently a new method, referred to as Complete Data Fusion, was proposed that, with simple implementation requirements, provides products of quality equivalent to that of the simultaneous retrieval products.
The adoption of the Complete Data Fusion has highlighted a problem that we believe to be common to simultaneous retrieval and data fusion. The measurements that we wish to fuse often present some inconsistencies, due to representations on different vertical grids, not perfect time and space coincidence and different forward model errors, which may spoil the quality of the fused profile. In order to apply the Complete Data Fusion method to inconsistent measurements without a degradation of the product, it is necessary to add to the measurement covariance matrices of each fusing profile a covariance matrix that takes into account these inconsistencies. Therefore, the main problem in the fusion of inconsistent measurements is the realistic estimate of these inconsistency covariance matrices.
We propose to use the value of the cost function that is minimized in the Complete Data Fusion to constrain the values of the inconsistency covariance matrices. To this purpose, we have analytically calculated the expected value of the cost function and its variance: the result is presented and is used in the proposed procedure.
Modelling the inconsistency covariance matrices with one parameter, we tune the value of this parameter using the constraint provided by the requirement of the value of the cost function being equal to the expected value and using the variance to assign an error to the obtained value of the parameter.
The use of Complete Data Fusion will be particularly useful for the analysis of the future atmospheric Sentinel missions of the Copernicus program. Indeed, the amount of data that will be available from these missions will pose technical challenges to many applications and the Complete Data Fusion can be used to reduce the number of products while maintaining the information content of the original datasets. For this reason, we test the proposed method on simulated measurements of ozone profiles obtained in the thermal infrared in the framework of the Sentinel 4 mission. In this context, the method is used to estimate the coincidence covariance matrices that take into account the variability of ozone when the fusing profiles refer to different times and space locations.
The merits of this new procedure in the case of this specific data set are presented and discussed.

Area C - Board 265

Regional models (e.g. COSMO) can simulate the atmospheric physical and chemical state at a high spatial resolution of less than 10 km. With the satellite observations till the end of 2017 (e.g. GOME, SCIAMACHY, OMI and GOME2), it was only possible to validate model performance and input emission inventories at coarser (> 25 km) resolution. The much improved spatial resolution (3.5 km x 7 km) of TROPOMI and even improved signal to noise ratio, now enables the model validation at sub-10 km spatial resolution, with minimal need of spatial and temporal averaging.
A 1-way nested MECO(3) coupled model system was set-up for Germany, in which the finer resolved domain receives the initial and boundary conditions online from the coarser regional or global model. Simulations at domains over Germany at ~7 km × 7 km spatial resolution and over Rhineland-Pfalz at ~2.2 km × 2.2 km spatial resolution were performed simultaneously for March - April 2018. Deep convection is resolved in the RP domain whereas shallow convection is parametrized based on the so-called Tiedtke scheme. We have used the MIME mechanism for chemical calculations within the model system. NO2 and HCHO tropospheric vertical columns were derived from the model simulated NO2 profile by taking into account averaging kernels and tropopauses level from the TROPOMI data products. For the application of averaging kernels, high spatial resolution a priori profiles from the regional model are used. Evaluation of three emission inventories (CMIP5, EDGAR-CIRCE and TNO-MACC III) was performed via intercomparison studies. We found out that the TNO-MACC III emission inventory yields the best agreement with satellite observations, if sector-dependent monthly scaling factors are applied to the emissions in order to account for the seasonal variability.
For the comparison with ground-based MAX-DOAS measurements, height-resolved 2-D box airmass factors were calculated using the McArtim (Monte Carlo Atmospheric radiative transfer model) and applied to the modelled trace gas profiles along elevation angles of the measurements. The derived modelled slant column densities (SCD) of NO2 and HCHO were directly evaluated against the measured SCD. This comparison procedure accounts for the complex dependency of the MAX-DOAS SCDs on the 3D (vertical and horizontal) trace gas distribution in the measurement direction.

Area C - Board 266

The Geostationary Environment Monitoring Spectrometer (GEMS) is a scanning UV-Vis spectrometer planned to measure concentration of ozone and aerosol with their precursors including NO2, SO2, and HCHO in high temporal (every 1hr, 8 times per day) and spatial resolutions (3.5 km N/S by 7.2km E/W at Seoul, Korea). The GEMS employs hyper-spectral imaging with 0.6nm resolution to observe solar backscatter radiation in the UV-Visible range (300-500nm). It will be onboard the GEO-KOMPSAT-2B (Korean Geostationary Earth Orbit Korea Multi-purpose SATellite-2B) in geostationary orbit in term from October 2019 to March 2020. Over the 10-year mission-life, the GEMS will serve to air quality products over East Asia and enhance our understanding of long-term climate change. Recently, PSR (Pre-Ship Review) of GEMS was completed successfully in January, 2018 and preflight test of GEO-KOMPSAT-2B is perfoming until launch. In addition, Level1 and Level2 algorithms are under the development and validation. Level2 algorithms for trace gases and aerosols show reasonable performances, but further improvement is required in hourly retrieval of air mass factor, in particular. Construction of ground station system will be completed by December, 2018, which plays a role in receiving, pre-processing, post-processing, and so on. After lunching GEMS, the synergy with Sentinel 5P, TEMPO, and Sentinel-4 will provide more reliable products and calibration/validation results. Futhermore, it is expected to obtain science and socio-economic benefit and to improve spatial and temporal coverage to monitor globalized pollutants via this formation-flying. Harmonized efforts for the concstellation are underway in terms of common basic requirements, standards, data product quality and cross participation of meeting under the framework of Committee on Earth Observation Satellite (CEOS) Atmospheric Composition Virtual Constellation (AC-VC). Regarding application plan, the GEMS data are expected to improve the accuracy of air quality forecasting and reduce current discrepancy between the predict model and observation. To be specific, GEMS data will be used to estimate reliable air pollutant emissions by using top-down methodology and to improve initial condition through data assimilation. In addition, Environmental Satellite Center will monitor air pollutants and causes of climate change to identify properties of spatio-temporal distribution and transboundary transport.

Area C - Board 267

In this work we present the first and longest zenith-sky mobile DOAS (Differential Optical Absorption Spectroscopy) measurements performed in Europe. The mobile DOAS observations were performed on board of a motor vehicle using an UV-Vis spectrometer during several days on 2016 covering Romania, Hungary, Austria, Germany, Belgium, Netherlands, Italy, France, Spain and Portugal. The target trace gases are: nitrogen dioxide (NO2) and sulfur dioxide (SO2). The mobile DOAS measurements performed on ground represents more than 5000km travelled on roads. The highest tropospheric NO2 (~1x1e17 molec./cm2) content was determined around Paris, France. The study also presents comparisons between tropospheric NO2 Vertical Column Density (VCD) deduced from mobile DOAS measurements and Ozone Monitoring Instrument observations.

Area C - Board 268

We present the L1 > L2 prototype algorithms for formaldehyde (HCHO) and sulphur dioxide (SO2) that have been developed in the framework of the ESA S4-L2 project in preparation of the Sentinel-4 mission. Sentinel-4 is an operational satellite foreseen for launch in the early 2020’s; it is
defined as an Ultraviolet Visible Near-infrared (UVN) spectrometer on each of the Meteosat Third Generation Sounder (MTG-S) platforms. It will be the first mission to monitor trace gas column densities and aerosol over Europe from a geostationary perspective, allowing for an hourly revisit time of the observed domain.

For HCHO the mission is of high interest as the high time resolution will allow to better study the diurnal variation of HCHO columns which is for the moment poorly known. Its lifetime being of the order of a few hours, HCHO concentrations in the boundary layer can be directly related to the release of short-lived hydrocarbons, which mostly cannot be observed directly from space. Furthermore, HCHO observations provide information on the chemical oxidation
processes in the atmosphere, including CO chemical production from CH4 and NMVOCs.

For SO2, it is expected that Sentinel-4 measurements will be able to detect low levels of anthropogenic SO2 and improve on the existing satellite data sets due to the combination of high spatial resolution (8x8 km²) and high temporal sampling (1 hour). These aspects are also of value in the monitoring of volcanic plumes. The S4 European domain hosts a number of active volcanoes (e.g. in Iceland and Italy), which can emit large quantities of SO2. S4 measurements of volcanic SO2 at hourly resolution are expected to help to netter determine the volcanic SO2 flux and will be a great asset in case of volcanic crises (SO2 is often used as a proxy for the presence of volcanic ash).

In this paper, we will address the specific challenges posed to the algorithms for both trace gases when observing in a geostationary geometry and discuss the new developments and performance with respect to existing algorithms for polar orbiters.

Area C - Board 269

The TROPOspheric Monitoring Instrument (TROPOMI) is a spectrometer measuring in the UV, visible, near-infrared and short-wave infrared, which allows the retrieval of trace gas species like O3, NO2, HCHO, SO2, CO, CH4 and aerosol aspects like the aerosol index. The Copernicus Sentinel-5P satellite, with TROPOMI as payload, was succesfully launched on 13 October 2017. The first operational data products, including NO2, have become publicly available in July 2018. Results from a first update of the NO2 retrieval are available since 24 October 2018. TROPOMI measurements of O3, CO, NO2, and also SO2 and HCHO, are currently investigated to be assimilated in the Copernicus Atmosphere Monitoring Service (CAMS).

TROPOMI has a full global coverage each day, but with a much improved resolution (3.5 x 7 km2) compared to the Ozone Monitoring Instrument (OMI). Because of the fine resolution, and because of the high signal to noise ratio compared to e.g. OMI, the TROPOMI observations are of great importance for estimating pollutant concentrations and emissions at the scale of smaller towns, individual power plants, wildfires and major infrastructures.

The Nitrogen Dioxide (NO2) tropospheric columns are retrieved using a newly implemented version of the OMI DOAS algorithm developed at KNMI combined with an integrated modelling-retrieval-assimilation approach to derive the air-mass factors and to estimate the stratospheric column (the DOMINO approach). This latter component is based on the TM5-MP chemistry-transport model operated at a resolution of 1x1 degree. Retrieval improvements from the EU QA4ECV project (www.qa4ecv.eu) have been included in the retrieval software. This will ensure consistency with the reprocessed NO2 datasets from this project, including OMI for the period 2004-2018. The near-real time NO2 data (NRTI) is available from the S5P operational processor at the German Aerospace Center (DLR). The off-line data product (OFFL) is produced at KNMI.

In our contribution we will summarize the components of the TROPOMI NO2 retrieval. The developments of the TROPOMI NO2 retrieval during the E1 and E2 phase, and an assessment of the quality of the NO2 product will be presented. Comparisons have been made with the OMI NO2 retrievals (QA4ECV NO2 product), with CAMS global and regional air quality models and MAX-DOAS/PANDORA observations. Candidate future improvements to the operational algorithm will be indicated. The benefits of the high-resolution TROPOMI NO2 observations for the monitoring of sources and air quality will be demonstrated based on a set of examples for various emission sources.

Figure 1: TROPOMI tropospheric NO2 column world map for August 2018.

Area C - Board 270

Ammonia plays a central role in many environmental threats, through either deposition to the Earth's surface or as a precursor of particulate matter. despite its importance our knowledge about the ammonia budget is rather poor. This knowledge may improve with the increasing amount of remote sensing products for ammonia. Remote sensing products for NO2 on the other hand have been developed for a much longer timeperiod, and proven useful for the derivation of emissions and trends. Here the spatial resolution of the products has made it hard to derive information on relevant urban scales.
In this study we have combined the information from the IASI ammonia product with the LOTOS-EUROS chemistry transport model to verify the current understanding of the ammonia emission distribution across Germany. The fit parameters between modelled and observed ammonia column are compared and related to the ones found for the Netherlands leading to the identification of regions in which systematic over- and underestimation of emissions is suspected. In the areas with high animal number densities the emissions are estimated to be too large whereas over the agricultural areas in the north-east the emissions seem to be too low. Available in-situ observations for ammonia and precipitation chemistry support these findings. Neglection of manure transport in the ammonia emission estimation and variable climate conditions in the country are identified as potential causes of the discrepancies. To improve the modelling of the ammonia budget an agricultural emission model should be developed with detailed spatial and temporal variability and constrained by remote sensing data.
A similar study as performed for NH3 will be performed in the coming months for NO2 using the tropospheric column observations from the TROPOMI instrument. The unprecedented high resolution of these new observations is expected to bring relevant information for exploration of the quantity and spatial distribution of NOx emissions. First results of this work will be presented at the symposium.

Area C - Board 271

Connecting the dots: a case study of detecting small isolated NOx sources with S5P (TROPOMI), pollution plumes, and daily emission inversions.
The launch of the Sentinel 5 Precursor mission TROPOMI in October 2017 marked the start of a new era in space observations of air quality. Lessons learned from previous satellite missions GOME, GOME-2, SCIAMACHY, and OMI, were used to design this first ‘next generation’ of UV-VIS satellite instruments, specifically focused on monitoring air quality.
In this talk, we present a case study of TROPOMI measurements of Nitrogen Dioxide (NO2) during spring 2018 plumes detected along the northern Siberian natural gas pipeline that connects the Urengoj natural gas field to Europe. These plumes were detected after a new snow-ice-filter was applied to TROPOMI data that allows for discriminating clouds from snow and ice, which we will briefly introduce. We show that the use of bright snow/ice pixels results in a much improved signal-to-noise, thereby allowing for detection of much smaller NO2 sources than previously possible.
These NO2 sources can be uniquely attributed to gas-compressor stations along the natural gas pipeline by visualizing TROPOMI NO2 images in Google-Earth. Gas-compressor stations are needed every 100 km or so to control the gas pressure in the pipeline. Gas compressor stations typically use gas-powered turbines, whose high temperature combustion of gas produces NOx.
We apply the algorithm for detection of Daily Emission estimates Constrained by Satellite Observations (DECSO) to derive daily emissions of these isolated sources, and discuss the potential of using the plume structure for deriving NO2 properties like lifetime and mixing.
We identify more of such plumes along other pipelines, and discuss what detection of these isolated sources and small plumes mean for detection of NO2 pollution plumes elsewhere in the world.
These results provide an exciting insight in future possibilities for monitoring air pollution, as they highlight what gains may be expected in case of further improving the signal-to-noise ratio of future missions (either satellites or HAPS), even though the S5P mission has just started.

Area C - Board 272

Clouds have a major impact on the retrieval of NO₂ tropospheric columns from space. The source of the cloud information is investigated here. For the TROPOMI instrument there are two primary options to retrieve cloud height information: the first is to use the oxygen A-band in the near infra-red at 760 nm (FRESCO or the standard S5P Cloud product), the second is to use the collision induced absorption feature near 477 nm in the visible (O₂-O₂, as used on OMI).

The O₂ A-band option has a stronger signal, but the relatively bright surface albedo in that window over land surface types (especially vegetation) is less than ideal for cloud retrieval. At the same time the wavelength difference between the cloud retrieval and the NO₂ retrieval window (405-465 nm) possibly introduces systematic biases.

The O₂-O₂ option avoids the wavelength difference, and performs the cloud retrieval with basically the same surface albedo as the NO₂ retrieval itself. The O₂-O₂ cloud retrieval is based on DOAS, much like the NO₂ retrieval. A complication in this option is the inclusion of the correct temperature profile, not because the O₂-O₂ cross sections are temperature dependent, but because the density depends on the temperature. A possible downside of the O₂-O₂ window is that it contains absorption features from other trace gases, such as O₃ and NO₂, while the O₂ A-band does not have overlapping trace gases.

An additional instrumental feature for TROPOMI is that the NIR and visible spectra come from different spectrometers that have a spatial mismatch with each other. While this is taken into account when using FRESCO for NO₂ airmass factor calculations - the cloud pressure is interpolated, and the cloud fraction is derived from the NO₂ window, using spectra for cloud information that originate from the same spectrometer as the NO₂ retrieval itself avoids the uncertainty in the spatial coregistration.

In this contribution we investigate the impact of the FRESCO and O₂-O₂ cloud products on NO₂. FRESCO is used operationally for the NO₂ retrieval, whereas the O₂-O₂ cloud product is processed specifically for this investigation.

Area C - Board 273

Nitrogen oxides, NOx (NOx = NO + NO2) play a key role in tropospheric chemistry. In addition to their directly harmful effects on the respiratory system of living organisms, they influence the levels of tropospheric ozone and contribute to acid rain and eutrophication of ecosystems. As they are produced in all combustion processes, they can serve as an indicator for anthropogenic air pollution. Urban areas exhibit strong emissions of NOx due to a high density of point source emitters, such as industrial facilities, as well as diffuse emissions caused by traffic. The dense population in cities leads to a strong exposure of the inhabitants to air pollution. Reduction of NOx emissions in densely populated areas is thus of public concern.

AirMAP, an airborne imaging DOAS instrument developed at the Institute of Environmental Physics of the University of Bremen, is capable of retrieving gapless trace gas maps of NO2 column densities at a spatial resolution better than 100 m and is operated on a campaign basis. Within the context of the ESA funded AROMAT / AROMAPEX campaigns held in 2014, 2015 and 2016, several mappings of Berlin’s horizontal NO2 distribution could be acquired. This dataset was validated against independent ground-based (MAX-DOAS) and other airborne imaging DOAS measurements (APEX, Spectrolite) and forms the basis for further studies.

This study focusses on the derivation of top-down NOx emissions of Berlin as a whole, but also for smaller areas on a sub-city scale. The results derived for the campaign days will be compared to current bottom-up emission inventories. Conclusions will be drawn on the quality of current emission inventories in relation to the measurement-based emission estimates and their uncertainties.

Area C - Board 274

Reactive nitrogen is a key nutrient and its availability in the environment, together with other nutrients’ availability, is controlling ecosystem productivity. The latter is linked to carbon dioxide removal from the atmosphere by the marine and terrestrial ecosystems and to food production. Atmospheric deposition of reactive nitrogen compounds has positive and negative impacts to ecosystems leading to fertilization or acidification and accumulation of excess nutrients, respectively. Several areas of the globe are estimated to be subject to high fluxes of deposited reactive nitrogen. However, model estimates are associated with high uncertainties.
The present study aims to reduce these uncertainties and increase the robustness of nitrogen atmospheric deposition flux estimates. Knowledge on the global present-day atmospheric deposition fluxes of reactive nitrogen is here summarized, accounting both for the oxygenated and reduced species as well as for the organic fraction of reactive nitrogen that has been, until recently, neglected.
The global 3-dimensional atmospheric chemistry transport model TM4-ECPL that accounts for multiphase chemistry, all major aerosol components and a very comprehensive atmospheric nitrogen cycle including organic nitrogen (Kanakidou et al., 2016) is here used in conjunction with satellite observations. The meteorology in this off-line model is taken from European Centre for Medium-Range Weather Forecasts ERA-interim reanalysis products. The simulated nitrogen deposition fluxes have been successfully evaluated against surface observations.
For the present study, satellite observations of NO2 and NH3 from the GOME-2, TROPOMI and IASI instruments are further used to evaluate model results. Atmospheric deposition fluxes derived from the satellite observations are compared with the simulated fluxes. Regions where significant improvements are achieved by reconciling model results and satellite observations are highlighted. The implications of the findings for the marine ecosystem functioning (Kanakidou et al., 2018) are discussed and put in perspective based on recent ocean biogeochemical modeling of the response of the oceans with regard to carbon sequestration and nitrous oxide emissions to the atmospheric inputs of nitrogen (Jickells et al., 2017).

Kanakidou M., et al, Past, Present and Future Atmospheric Nitrogen Deposition, Journal of the Atmospheric Sciences (JAS-D-15-0278) 73, 2039-2047, 2016.
Kanakidou M., et al., Aerosols in atmospheric chemistry and biogeochemical cycles of nutrients, Environ. Res. Lett. 13, 063004, doi.org/10.1088/1748-9326/aabcdb , 2018.
Jickells, T. D., et al., A reevaluation of the magnitude and impacts of anthropogenic atmospheric nitrogen inputs on the ocean, Global Biogeochem. Cycles, 31, doi:10.1002/2016GB005586, 2017.

Area C - Board 275

China is facing an important environmental issue with the strong degradation of air quality in the last decades. All the major pollutant concentrations exceed the thresholds recommended by the World Health Organization. In response to this, Chinese authorities have started to apply emission mitigation plans starting with the reduction of SO2 emissions since 2006 and with NOx emissions since 2010. Recent studies show a decrease of NO2 concentrations or tropospheric columns since 2013 and attribute it to the NOx emissions reduction.
In the framework of the French PolEASIA project, we propose to better quantify pollution precursor emissions (mainly NOx) and to evaluate the impact on pollutant variability and trends. To achieve this, we use the PYVAR-CHIMERE inverse modeling tool, based on the CHIMERE chemical transport model and a 4D variational assimilation scheme, to derive optimized NOX emission for China. The NO2 tropospheric columns observed from OMI are used as constraints. Gridded monthly NOx emissions are calculated at a fine spatial resolution of 0.5x0.5 degrees. The system have been evaluated first on the year 2015 where our optimized inventory shows a decrease of 12% of the NOx emissions over the Northern East China compared to 2010 HTAP emission inventory. The comparison with in-situ measurements shows that the use of our optimized inventory increases the reliability of the CHIMERE model in simulating NO2 surface concentration over the Northern East China at both the annual and monthly scales. The PYVAR-CHIMERE system will be applied to optimize NOx emissions from 2008 to 2016. Trend analyses of the NOx emissions over the Northern East China will be presented.

Area C - Board 276

Sulphur dioxide (SO2) is an important atmospheric pollutant, of natural (volcanic) and anthropogenic origin, and a precursor of secondary sulphate aerosols, which have an impact on air quality, the climate system, the environment and ecosystems. Satellite-based observation of SO2 is a tool for the monitoring of its three-dimensional distributions, the understanding of its atmospheric evolution and processes, the identification of its sources and the estimation of its emission fluxes. At present, the satellite-based observation of SO2 is based on the exploitation of two independent spectral regions:
• the region of its absorption bands in the ultraviolet (UV), at wavelength longer than 360 nm;
• the region of its absorption bands in the thermal infrared (TIR), within the roto-vibrational bands around 1200 and 1350 cm-1.
The UV and TIR absorption contain a very limited but partially complementary information on the vertical distribution of the SO2, especially at the lowest altitudes, i.e. the most important for air quality and for the interaction with the biosphere.

Here we present a suite of sensitivity analyses of nadir UV/VIS and TIR satellite observations to the vertical distribution of SO2, based on pseudo-observations in the mentioned sensitive UV and TIR spectral regions. Test instruments will be GOME-2 (UV/VIS) and IASI (TIR). This work is built upon the Observing Systems Simulation Experiments (OSSE) framework. Pseudo-reality scenarios are generated for both volcanic (a number of moderate Mount Etna eruptions) and anthropogenic sources (Asian pollution scenarios), with the CHIMERE chemistry/transport model. Then, the observations of GOME-2 and IASI are simulated with realistic simulators, mimicking their radiometric noise, spectral resolution, observation geometry and other instrumental parameters. The sensitivity and the information content on SO2 of these observations are analysed and are here critically discussed. Finally, first tests of multi-spectral integration UV+TIR are here presented for the first time and put in the general context of multi-spectral/instrumental synergy.

Area C - Board 277

We have developed a theoretical basis for the observation of volcanic secondary sulphate aerosols (SSA) from thermal infrared (TIR) nadir-viewing satellite instruments [1]. The sensitivity of TIR observations to the chemical composition (sulphuric acid mixing ratio) and the size distribution (effective number concentration and radius) of idealised SSA layers has been analysed. Both high-spectral resolution, IASI (Infrared Atmospheric Sounding Interferometer)-like and broadband radiometers, MODIS (Moderate Resolution Imaging Spectroradiometer) and SEVIRI (Spinning Enhanced Visible and InfraRed Imager)-like pseudo-observations are generated using forward radiative transfer calculations performed with the 4A (Automatized Atmospheric Absorption Atlas) radiative transfer model. Then, the impact of the extinction of these idealised aerosol layers on the brightness temperature spectra observed by the different simulated satellite instruments has been analysed. The dependence of the aerosol spectral signature to the chemical and microphysical properties of the layers, as well as the role of interfering parameters, in particular co-existing sulphur dioxide [2], are here discussed. The information content (degrees of freedom and retrieval uncertainties) of synthetic satellite observations is estimated for the different instrumental configurations. High spectral resolution and broadband spectral features approaches are proposed and discussed. First tests on the detection and tracking of SSA plumes from both explosive events with stratospheric injection of the plumes (Nabro volcano eruption, summer 2011, analysed using the high-temporal-resolution geostationary SEVIRI observations [3]) and smaller tropospheric events (different recent medium-sized eruptive events of Mount Etna, observed by IASI) have been performed. Finally, future perspectives on the combined estimation of SO2 and SSA amounts in volcanic plumes, as well as multi-spectral/instrumental coupling, for volcanic plumes observations are discussed.

References:
[1] P. Sellitto and B. Legras, Sensitivity of thermal infrared nadir instruments to the chemical and microphysical properties of UTLS secondary sulfate aerosols, Atmospheric Measurement Techniques, 9, 115-132, 2016.
[2] H. Guermazi, P. Sellitto, M. M. Serbaji, B. Legras, F. Rekhiss, Assessment of the combined sensitivity of nadir TIR satellite observations to volcanic SO2 and sulphate aerosols after a moderate stratospheric eruption, MDPI Geosciences, 7, 84; doi:10.3390/geosciences7030084, 2017.
[3] P. Sellitto, G. Sèze, B. Legras, Secondary sulphate aerosols and cirrus clouds detection with SEVIRI during Nabro volcano eruption, International Journal of Remote Sensing, 38(20), 5657-5672, 2017.

Area C - Board 278

The Eyjafjallajökull eruption in 2010 has shown the necessity to forecast volcanic ash plumes for several sectors, such as aviation, tourism, etc. Volcanic ash prediction is based on dispersion models initialized with ad hoc estimated emissions from the volcano under consideration. So far systems such as that run by the Copernicus Atmosphere Monitoring Service (CAMS) which provides atmospheric composition forecasts at the global and European scale up to day five, initialized via analyses, have only been used in research studies to understand the role that data assimilation can have to improve the prediction of volcanic plumes. The challenges are many. For example, most systems including the CAMS system, assimilate column-integrated Aerosol Optical Depth (AOD). In that case, even with a well performing model, the altitude information is missing. However, the injection height is one of the most important variables to predict the direction of the plume. Moreover even under the best conditions, it is difficult to estimate the altitude and the composition of the volcanic ashes. The assimilation of profiling data such as those provided by lidar instruments can provide the missing information and make the prediction more accurate.
The lidar measurements are a well known source of aerosol profile information. In recent years the ECMWF’s Integrated Forecast System (IFS) has been developed with the aim to assimilate different aerosol lidar information (backscatter, attenuated backscatter and extinction profiles) at different wavelengths and for different instruments (ground-based or space-borne). These developments have been carried out also in the context of the EUNADICS-AV project which aims at developing a system to provide consistent and coherent information to aviation authorities, airlines and pilots in the event of a natural disaster affecting the airspace. As part of this effort, ECMWF has run some pilot assimilations studies using CALIOP (Cloud-Aerosol LIdar with Orthogonal Polarization) data, ground-based data from EARLINET (European Aerosol Research LIdar NETwork) and a combination of both to investigate the potential of this type of data data in the assimilation and prediction of volcanic ash. In particular, two icelandic eruptions, Eyjafjallajokull eruption of 2010 and Grímsvötn eruption of 2011, were investigated. Results will be shown relative to these assimilation studies.

Area C - Board 279

The CO₂ slicing technique has been widely used to obtain the heights of meteorological clouds from satellite data. Research has shown that it has the capability to obtain the heights of volcanic ash clouds in the troposphere; an important variable used to minimise the risk posed by volcanic ash clouds to aviation and used as an input into models of ash cloud propagation. The technique has only been applied to ash for measurements by the Moderate Resolution Imaging Spectroradiometer. This study adapted the CO₂ slicing technique for the Infrared Atmospheric Sounding Interferometer (IASI), a hyperspectral sensor with hundreds of channels within the CO₂ absorption band. Simulated ash spectra were first used to select a subset of the most appropriate channels. It was then demonstrated that the subset of channels was capable of successfully retrieving the ash cloud-top pressure using simulated spectra. The results indicated a strong match between the simulated and retrieved pressures except for ash clouds close to the surface or with low optical depths, The technique was applied to IASI spectra obtained during the Eyjafjallajökull and Grímsvötn eruptions in Iceland in 2010 and 2011 respectively. Both eruptions emitted ash plumes into the troposphere, and were widely studied with other instruments, making them appropriate choices for this application. The CO₂ slicing results were compared against those from an optimal estimation (OE) scheme developed for IASI which also retrieves the ash optical depth and effective radius. In general, the CO₂ slicing results were slightly higher than those from the OE method. The Cloud-Aerosol Lidar with Orthogonal Polarization satellite instrument was used for validation. In some cases, the CO₂ slicing and OE techniques both produce results which closely resemble the backscatter plots from the LiDAR, but there are clear cases where one outperforms the other. Overall the CO₂ slicing technique is shown to perform better. Despite this technique being widely used for meteorological cloud, this is the first application to volcanic ash with IASI. It is a fast tool which could be used to get a first approximation of the ash height, potentially for use for hazard mitigation, and which could also be used as an a priori for other retrieval techniques.

Area C - Board 280

With one year of operational data from the Sentinel 5-Precursor Tropospheric Monitoring Instrument (TROPOMI) UV Aerosol Index (UVAI) data, the full seasonal cycle of variability in sources of UV-absorbing aerosol can now be examined. Owing to its high spatial resolution, the TROPOMI UVAI has proven to be well-suited for capturing absorbing aerosol plume dynamics in great detail from all expected sources including biomass burning, desert dust, and volcanic eruptions. Several examples will be presented to illustrate how the UVAI can be used for detailed plume tracking and analysis.

In addition to this known behavior, there are also some unexpected features in the UVAI that have arisen due to the high spatial resolution. Effects related to shallow waters, bright surfaces, and clouds including cloud shadows are clearly observable in the dataset and warrant special consideration. Several approaches will be presented and compared to that of other missions. The aerosol index is very sensitive to absolute changes in radiance and irradiance and as such is a good indicator of the quality of the instrument and its calibration. Trends in global, regional and seasonal means will be presented to illustrate how UVAI tracks with changes in the TROPOMI instrument. In advance of the release of the updated TROPOMI Level 1b data in June 2019, some comparisons of the UVAI as derived using the current operational processing and the update L1b input will be shown.

Lastly, the TROPOMI UVAI data has been verified and validated via satellite-to-satellite intercomparisons with other instruments including the Ozone Mapping and Profiler Suite (OMPS) and Ozone Monitoring Instrument (OMI). Several examples will be given to illustrate how the aerosol index varies per instrument for a variety of regional scenes. An outlook will be given for future developments of the TROPOMI UVAI and how it can be used in conjunction with other TROPOMI data products like Aerosol Layer Height (ALH).

Area C - Board 281

During the dry seasons in some parts of the world, strong wildfires with pronounced diurnal cycle (e.g., grass fires mainly occurring during the daylight hours) exist in great numbers concentrated in relatively small areas. For example, in Africa in the same comparatively small area (e.g. 150 x 150 km2), hundreds of grass fires can occur on a daily basis. Each fire is comparatively small but, due to their large numbers, they produce significant amounts of gaseous and particulate matter emissions.

With the utilization of MSG/SEVIRI geostationary satellite data and support of orbital satellite data (MODIS, VIIRS), we present a weather-based method to forecast these emissions, and also showing that they are fairly predictable if an areal total emission is considered (the individual events are still largely unpredictable). The used emission prediction formalism is built within a phenomenological framework and based on the direct proportionality between the emissions and temporal integral of fire radiative power (FRP) (i.e., fire radiative energy).

The model uses meteorological variables only and showed high potential in predicting emissions from grass fires in the western parts of Central Africa, especially in Angola and the Congo which have served as our test areas. In the model, the globe is formed by a longitude-latitude grid divided into rectangular-shaped areas at spatial resolution of 1.44 degrees squared and at temporal resolutions of 15 minutes for fires (SEVIRI) and 3 hours for weather (ERA-Interim dataset of ECMWF was used). Based on the weather forecast for several past days and that for the coming days, the expected SEVIRI FRP data pattern for each day is mimicked using the logistic growth and decay. The parameters of the equation are identified from the recent retrospective fire record and showed good stability for several ddays forwards. The predictions of the model are based on the knowledge of areal characteristics (i.e. grid cell parametrization) and correlation between the weather and FRP maxima of cloudless days.

Our ultimate goal is to make reasonable predictions about fire-induced emissions from wildfires all over the world. Thus, using the same formalism, we are hoping to expand the model to cover also the longer-terms of forecasts and all wildfires globally.

Area C - Board 282

Currently the possibilities offered by the COPERNICUS Sentinel missions are just partially explored to provide consistent and statistically reliable information about volcanic cloud quantification and dispersion in the atmosphere. Such information are crucial for aviation safety and civil protection purposes therefore new tools to exploit satellite observations are required. Satellite Remote sensing data represent a unique opportunity to monitor the volcanic emission on large areas with adequate spatial, temporal and spectral resolution. In the late 1980s it was shown that volcanic ash clouds can be discriminated from meteorological clouds on remote sensing images by considering two specific thermal infrared channels of multispectral satellite-borne sensors. Volcanic ash particles of absorb more infrared radiation around 11 µm, while water droplets or ice particles absorb more around 12 μm. Thus, the brightness temperature difference (BTD) of a multispectral image between channels centred at 11 μm (Tb,11) and 12 μm (Tb,12) will be negative for ash clouds, and positive for meteorological clouds (Prata, 1989).
On this finding, several works, aimed to characterise eruption phenomena by a number of satellite missions have been based. Such studies have proven the key role of remote sensing data as a primary information source to detect volcanic ash clouds and retrieve ash particles properties like effective radius-Re, optical depth-AOD and the total column mass-Ma in the ash plume.
In this context, one of the most consolidated approach (named in the following Look Up Table-LUT approach) for volcanic ash retrievals using multispectral satellite instruments is implemented by the simulation of the Top Of Atmosphere (TOA) radiances computed using radiative transfer models (Wen and Rose, 1994, Prata and Grant, 2001, Yu et al., 2002, Corradini et al., 2009; 2010; Prata and Prata, 2012). The ash AOD and Re are retrieved by bilinear interpolation by considering the couple BTD-Tb,11 of each ash pixel detected from the image and the inverted arches curves BTD vs Tb,11 computed using radiative transfer models. The ash mass is then estimated by using the simplified formula introduced by Wen and Rose in 1994. The same approach can be also used for the ice/wvp retrievals by only substituting the ash optical properties with the ice/wvp optical properties.
The use of radiative transfer models has two main drawbacks: the need of several inputs (often identified with the term ancillary information, that are the surface temperature and emissivity, the atmospheric profiles of PTH, the cloud geometry and the particles optical properties) and the processing time to generate a set of Look Up Tables with a given parametrization.
Moreover, every RTM input parameter is affected by an uncertainty that cause errors in the ash retrievals implemented by the LUT approach. A 40% of total ash mass retrieval error has been estimated, following the sensitivity study carried out by Corradini et al. (2008) which considered the uncertainties of many parameters such as atmospheric profiles, plume geometry, surface temperature and emissivity and ash type.
To reduce the computation time, necessary for the real time monitoring of the volcanic clouds, different retrievals procedures have been developed. One example is the Volcanic Plume Retrieval (VPR) approach (Pugnaghi et al., 2013; Guerrieri et al., 2015; Pugnaghi et al., 2016). VPR approximates the relationship between the satellite Top Of Atmosphere (TOA) radiances with the transmittances using a set of parameters adapted for a specific area and volcanic cloud particle type (ash or ice). From the transmittances computed at 11 and 12 µm, the ash AOD, Re and Ma can be derived. An important characteristic of this procedure is that the only input required for the generation of all the ash products is the Volcanic Cloud Top Height (VCTH).
The application of NNs to detect volcanic ash has significant interest to reduce the need of human interpretation of the ash detection maps as those generated by the BTD. Moreover, the overlapping of ash plume and meteorological clouds causes a mixed behaviour making difficult to identify the “ashy pixels” and even more difficult the retrieval of ash parameters. These problems have been carefully addressed by the NNs proposed methodology that identify the ashy pixels also considering the surfaces underlying the ash cloud (e.g. ash on sea, ash on meteorological clouds etc.). Moreover, the principal surfaces like sea, ice, and land are identified on the satellite image.
The NNs approach for parameter retrieval is based on the coefficients computation of the relationship existing between the Top Of Atmosphere (TOA) spectral radiance and the volcanic ash. The NN coefficients estimation is realized in the learning phase by examples composed of training patterns, i.e. RTM spectral radiances simulation samples, and the corresponding ash values.
The application of NNs is here proposed to address some still opened issues not completely solved by standard techniques, as the need of several input parameters and the processing time. Moreover, being a supervised machine learning technique, NNs can exploit the data information in a very effective way taking advantage of their non-linear interpolation capabilities and may outperform approaches based on linear interpolation.
In this study the NNs framework to characterize a volcanic eruption is provided considering the NASA MODIS and Copernicus Sentinel-3 SLSTR data acquired on eruptions from different volcanoes. The application of NNs to detect the volcanic ashpresence in the atmosphere and to retrieve ash quantitative parameters has already proven the effectiveness of the approach in our previous studies (Picchiani et al., 2011 and 2015). In the presence research is presented the recent improvements in the application of the NNs solutions. The latter are achieved considered the possibility to combine RTM and NNs to generate more general ash parameters retrieval algorithms and the evaluation of the possibility to discriminate between volcanic and meteorological ice/water vapour clouds. The development of such discrimination is particularly important being the volcanic ice/water vapour clouds potentially very dangerous for the aircrafts. These particles, in fact, could cover a possible layer of ash present in the cloud. The results of the study are compared in terms of accuracy and reliability with the outcomes of standard BTD and LUT approaches in order to show the effectiveness of the NNs algorithms.

Corradini, S. et al., 2008. Mt. Etna tropospheric ash retrieval and sensitivity analysis using moderate resolution imaging spectroradiometer measurements. J. Appl. Remote Sens. 2, 023550–20.
Corradini, S. At al., 2009. Retrieval of SO2 from thermal infrared satellite measurements: correction procedures for the effects of volcanic ash. Atmospheric Measurement Techniques 2, 177–191.
Corradini S., et al., 2010. Volcanic ash and SO2 in the 2008 Kasatochi eruption: Retrievals comparison from different IR satellite sensors. JGR, 115, D00L21. doi:10.1029/2009JD013634.
Picchiani, M. Et al., 2011. Volcanic ash detection and retrievals using MODIS data by means of neural networks. Atmospheric Measurement Techniques 4, 2619–2631.

Picchiani, M., Chini, M., Corradini, S., Merucci, L., Piscini, A. and Del Frate, F., 2015. Neural network multispectral satellite images classification of volcanic ash plumes in a cloudy scenario. Annals of Geophysics, 57.
Prata, A. J. 1989. Radiative transfer calculations for volcanic ash clouds. Geophys. Res. Lett. 16(11), 1293-1296. doi:10.1029/GL016i011p01293.
Prata, A.J. and Grant, I.F., 2001. Retrieval of microphysical and morphological properties of volcanic ash plumes from satellite data: Application to Mt Ruapehu, New Zealand. Quarterly Journal of the Royal Meteorological Society, 127(576), pp.2153-2179.
Prata, A.J., Prata, A.T., 2012. Eyjafjallajökull volcanic ash concentrations determined using Spin Enhanced Visible and Infrared Imager measurements. Journal of Geophysical Research 117, 24 PP.
Yu, T., Rose, W.I. and Prata, A.J., 2002. Atmospheric correction for satellite‐based volcanic ash mapping and retrievals using “split window” IR data from GOES and AVHRR. Journal of Geophysical Research: Atmospheres, 107(D16), pp.AAC-10.
Wen, S., Rose, W.I., 1994. Retrieval of sizes and total masses of particles in volcanic clouds using AVHRR bands 4 and 5. Journal of Geophysical Research -Atmosphere 99, 5421–5431.

Area C - Board 283

Biomass burning emissions over the Amazon vary due to climate variability (e.g. El Nino) and changes in the policy that aims to abate deforestation. In an earlier study, we showed that carbon uptake by vegetation was reduced in 2010 compared to 2011, but that the magnitude of the decrease strongly depended on the estimated 2010 and 2011 biomass burning emissions. Specifically, we derived that the El Nino induced 2010 Amazon drought resulted in a total reduction of carbon uptake of 0.24 to 0.50 PgC/yr and turned the balance from carbon sink to source.
In this study, we extend the period to more recent years (2010-2016). Starting from fire emission estimates from the Global Fire Emissions Database version 4 (GFED4) and the Global Fire Assimilation System (GFAS), we employ satellite-observed CO columns from the Infrared Atmospheric Sounding Interferometer (IASI) to optimise fire emissions in our TM5-4DVAR system. This system was recently successfully applied to the 2015 Indonesian fires (Nechita-Banda et al., 2018). Results reveal that IASI and MOPITT provide a strong constraint on CO biomass burning emissions, and significantly modify the timing and magnitude of the GFED and GFAS prior estimates (Figure 1).
A a next step, we will propagate the derived CO emissions over the Amazon to the prior CO2 fire emissions of the Carbon-Tracker South America system. Using this prior, Carbon-Tracker assimilates CO2 surface observations to derive the net carbon exchange with the Amazonian biosphere and the atmosphere. This net exchange and its variability will reveal the sensitivity of the Amazon to climate variability. A comparison to independent data (biosphere models, Solar Induced Fluorescence (SIF) from satellites) will also be presented.

References:
1. van der Laan Luijkx, I. T. et al. Response of the Amazon carbon balance to the 2010 drought derived with CarbonTracker South America. Global Biogeochemical Cycles 29, 1092–1108 (2015).
2. Nechita-Banda, N. et al. Monitoring emissions from the 2015 Indonesian fires using CO satellite data. Philosophical Transactions of the Royal Society of London B: Biological Sciences 373, 20170307 (2018).

Area C - Board 284

In this talk, we discuss the joint assimilation of satellite atmospheric carbon monoxide (CO) concentration and aerosol optical depth (AOD) within a chemical transport model with the aim of quantifying aerosol emissions from wildfires and assessing the downwind fire plume impact on air quality.

Wildfires are a growing hazard in the western U.S., where burn areas have increased by a factor of 6 since 1970, and big western wildfires are increasing in frequency. This comes with severe economic and societal impacts. Wildfire emissions contain fine particulate matter (e.g. PM2.5), which causes cardiovascular and respiratory problems and impacts air quality (AQ) on the continental scale. Links have also been demonstrated between smoke exposure and the demand for medical services, making this a significant concern for public health. Large fires impact not only the local environment, but also affect AQ far downwind through long-range transport of primary pollutants and production of secondary pollutants such as tropospheric ozone. The relative importance of fire in emissions of aerosol and ozone precursors is growing, and improving estimates of wildfire pollutant emissions, understanding subsequent transport processes, and predicting the AQ impact are therefore priorities.

Continental-scale coverage of fire intensity using fire radiative power (FRP) is available from Terra/Aqua/MODIS (MODerate-resolution Imaging Spectroradiometer) and Suomi NPP/VIIRS (Visible Infrared Imaging Radiometer Suite), and the evolution of aerosol resulting from fire emissions can be tracked by MODIS and other sensors. The most reliable satellite aerosol product is AOD, but this an integrated total column quantity. Other information is required to infer either surface level aerosol to estimate fire emissions and AQ impacts, or the aerosol burden higher in the free troposphere where long-range transport most readily occurs. This talk discusses leveraging strong correlations between primary emissions of wildfire carbonaceous aerosol and CO to inform on the aerosol vertical distribution. Terra/MOPITT (Measurement Of Pollution in The Troposphere) multispectral CO retrievals have demonstrated unique, reliable, independent profile information close to the surface and also higher in the free troposphere.

We present results from the Data Assimilation Research Testbed (DART) system that has been developed to assimilate MOPITT CO in the global-scale chemistry-climate model CAM-Chem. The assimilation technique allows inference on various fire model state variables such as CO emissions, and also aerosol species resulting from fires such as organic and black carbon. The benefit of joint assimilation of MODIS AOD and MOPITT CO profile for estimating emissions from the 2015 Washington and Indonesian fire cases will be discussed, along with the ability of the ensemble approach to infer information on the aerosol vertical distribution, both close to the fire sources and far downwind.

Area C - Board 286

The challenge in precisely quantifying the sources and sinks of atmospheric carbon dioxide (CO₂) is that the CO₂ taken up by plants during photosynthesis cannot be distinguished from the CO₂ released by plants and micro-organisms during respiration. It has been shown that carbonyl sulfide (OCS) can be used as a proxy for the photosynthetic uptake of CO₂. The relationship between the draw-down of OCS and CO₂ has been quantified for various species of plants and for different ecosystems, the results of which have been used to analyse this relationship on a continental scale. The aim of this project is to both quantify the location and magnitude of the sources and sinks of atmospheric OCS, and to use these data to infer photosynthetic uptake of CO₂ by vegetation on a global scale.

An initial step in this project is to retrieve OCS total columns from top-of-atmosphere radiances, as measured by the Infrared Atmospheric Sounding Interferometer (IASI) instruments, onboard each of the MetOp satellites. In order to assess the information content from an IASI OCS retrieval, retrievals from simulated IASI spectra have been performed, generated by the Reference Forward Model (RFM), using the University of Leicester IASI retrieval Scheme (ULIRS). Where the RFM is a radiative transfer model and the ULIRS is an optimal estimation retrieval scheme, utilising 30 equidistant pressure levels and a floating pressure grid. Progress has been made in improving the sensitivity of the satellite measurement to the lower troposphere, which is essential if the retrieved concentrations are to be used in surface flux inversions.

The 3D chemical transport model TOMCAT will be adapted to include emissions of OCS, carbon disulphide (CS₂) and dimethyl sulfide (C₂H₆S), as well as their sinks to provide a detailed representation of OCS chemistry. Available satellite measurements from the Atmospheric Chemistry Experiment – Fourier Transform Spectrometer instrument will be used to verify experimental model runs in the early stages of the project, as well as using surface flask (NOAA-ESRL) and column (NDACC) measurements. Later in the project a version of TOMCAT capable of calculating flux inversions will be used to create a new flux inventory of the global OCS budget.

By using IASI spectra from October 2006 to the present day a 12-year on-going global satellite dataset of atmospheric OCS will be generated. When analysed in combination with TOMCAT, this project aims to improve current uncertainties in the OCS budget, such as the magnitude and location of the ocean source of OCS. Furthermore, these data will provide essential insight into carbon cycle processes, including the potential to quantify gross primary productivity on a global scale.

Area C - Board 287

Carbon monoxide (CO) is an important atmospheric constituent affecting air quality and methane (CH₄) is the second most important greenhouse gas contributing to human-induced climate change. Detailed and continuous observations of these gases are necessary to better assess their impact on climate and atmospheric pollution. While surface and airborne measurements are able to accurately determine atmospheric abundances on local scales, global coverage can only be achieved using satellite instruments.

The TROPOspheric Monitoring Instrument (TROPOMI) on board the Sentinel-5P, which was successfully launched in October 2017, is a spaceborne nadir viewing imaging spectrometer measuring solar radiation reflected by and radiated from the earth in a push-broom configuration. It has a swath width of 2600 km on the Earth's surface and covers wavelength bands between the ultraviolet (UV) and the shortwave infrared (SWIR) combining a high spatial resolution of 7×7 km² with daily global coverage.

Abundances of atmospheric CO and CH₄ can be retrieved from TROPOMI's radiance measurements in the 2.3 μm spectral range of the shortwave infrared part of the solar spectrum with high sensitivity to all altitude levels including the lowest atmospheric layers where the variability is largest. Due to its wide swath in conjunction with high spatial and temporal resolution, TROPOMI enables the determination of column-averaged mole fractions of both gases with unprecedented level of detail on a global scale potentially introducing new areas of application. For instance, large emission sources can be detected in a single overpass. Here we present recent results for both trace gases obtained using the scientific retrieval algorithm Weighting Function Modified DOAS (WFM-DOAS).

Area C - Board 288

Satellite observations from instruments such as GOME, SCIAMACHY, GOME-2 and OMI have been used for more than two decades to monitor bromine monoxide (BrO) distributions on regional to global scales. To continue and improve daily global trace gas observations with an unprecedented spatial resolution, the TROPOspheric Monitoring Instrument (TROPOMI) onboard the Copernicus Sentinel-5 Precursor platform was launched in October 2017.
In this study, sensitivity tests were performed to find an optimal wavelength range for TROPOMI Differential Optical Absorption Spectroscopy (DOAS) BrO retrievals under various measurement scenarios by a systematic investigation of retrieval results in different wavelength intervals. Negative BrO SCDs, large deviations from the expected BrO SCDs and high fitting errors occur at shorter wavelengths when strong absorption structures of O₃ and SO₂ are included. In the case of DOAS retrievals over the cloudy background region and the volcanic plume, higher fitting errors were found at wavelengths longer than 362 nm due to the wavelength dependency of the Ring effect. From these sensitivity tests, an optimised wavelength range for TROPOMI BrO retrievals was determined, and applying this spectral window, TROPOMI shows excellent performances with much smaller fitting RMS and lower random scatter of BrO columns than OMI and GOME-2B. In addition to the good retrieval results, TROPOMI BrO columns show good agreement and consistency with data from those two existing satellite sensors, suggesting that TROPOMI data can be used for extending the long-term data set of space-based BrO observations.
Individual TROPOMI overpasses provide a more detailed view of BrO explosion events and variabilities in different source regions than previous sensors which provide only limited information due to lower spatial resolution. In this presentation, IUP Bremen TROPOMI BrO results for various source regions such as the polar sea ice region, salt marshes, and volcanoes are presented, and developments of small-scale BrO enhancements that can be investigated from the unprecedented small footprint of TROPOMI of 3.5x7 km² are discussed.

Area C - Board 289

Ultraviolet (UV) radiation is the key driver of the chemistry in the troposphere. UV-radiation reaching Earth’s surface dissociates molecules into reactive species that enter atmospheric chemical cycles and controls their photolysis rates. Examples of species that photodissociate include many important trace constituents of the troposphere, such as NO₂ and O₃. The photolysis rates of these species, denoted as J(NO₂) and J(O₃), are tightly connected to the formation of secondary air pollution. One of the main pathways for surface-level ozone formation is the photolysis of NO₂. On the other hand, the photolysis of O₃ is related to the formation of radicals, e.g. hydroxyl radical (OH), that plays an important role in secondary new particle formation since it is involved in the formation of sulphuric acid. Considering the new particle formation in highly polluted air many open questions still remain, but recently it has been shown that new particle formation events e.g. in Shanghai are generally photochemically induced, which highlights the importance of UV radiation as a key parameter in the process.

One of the applications of the UV-type satellite instruments, such as GOME-2, OMI, and most recently TROPOMI, has been to monitor UV radiation reaching the Earth’s surface. The operational surface UV- algorithms provide e.g. estimates of noontime erythemal and vitamin-D UV dose rates. The satellite-based surface UV data has been successfully used in various studies considering the health effects on humans, but overall much less attention has been given to the potential of using satellite data to study atmospheric chemistry and the formation of secondary pollution. Photolysis rates for different gases can be retrieved from UV satellite observations, and currently J(NO₂) and J(O₃) are available from GOME-2 instrument. These photolysis rates can give valuable information on global conditions favorable for secondary pollution formation, but they can also be combined with other satellite observations, e.g. NO₂, to estimate ozone pollution formation rate. In this work we will analyse satellite-based photolysis rates of NO₂ and O₃ from GOME-2 and demonstrate the potential of using TROPOMI observations for such applications.

Area C - Board 290

The Asian monsoon impacts the lives of more than a billion people, and exerts the great influence over global climate. It becomes a pressing concern that the monsoon convection system coupled to anthropogenic activities. It is important to assess the ozone variation over the Asian-Pacific region (AP) and impacts of East Asian Summer Monsoon (EASM). It is beneficial to understand the correlative relationship between the human activity and climate change in the region with the complex interaction.

Ozone plays a very important role because it protects any living organisms at the Earth's surface against the harmful solar UVB and UVC radiation. In the stratosphere, ozone plays a critical role in the energy budget because it absorbs both solar UV and terrestrial IR radiation. Furthermore, ozone in the tropopause acts as a strong greenhouse gas, and increasing ozone trends at these altitudes contribute to climate change. Tropospheric column ozone is a key trace gas owing to its important roles in atmospheric chemistry, air quality, and climate change. Despite its relatively small quantity, however, tropospheric ozone is the primary precursor for hydroxyl radicals, which control the lifetimes of many gas species in the troposphere.

The daily dataset of coherent total column ozone were created from FY3/TOU from 2008 to 2017. Using total column ozone from FY3/TOU and tropospheric and stratospheric column ozone from AURA OMI/MLS satellite data, the seasonal variations of the climatological ozone in the region (40ºE-160ºE, 0-60ºN) are analyzed for the total, tropospheric and stratospheric column, respectively. Differences of ozone and circulation pattern between strong and weak East Asian summer monsoon year are also investigated. It is revealed that the combination of the monsoon anomaly and the upper westerly jet anomaly affects obviously distribution of the tropospheric and stratospheric ozone during summer in the East Asia and the western Pacific regions.

Area C - Board 291

RAL's ozone profile retrieval scheme for the GOME-class of solar uv/vis backscatter spectrometer has unique sensitivity to tropospheric ozone, which led to its selection for nadir ozone profile retrieval from this class of sensor in ESA's Climate Change Initiative (CCI) and inclusion in the Tropospheric Ozone Assessment Report (TOAR). The JASMIN/CEMS computing facility at RAL has enabled the production of full-mission global data sets from GOME-1, SCIAMACHY, OMI and GOME-2A & 2B, resulting in over 20 years of height-resolved dataset for ozone from 1995-2016, spanning both stratosphere and troposphere.

A reprocessing of data has been enabled under the Copernicus Climate Change (C3S) project and work is underway to reconcile these data time series. We present some of the retrieval scheme advancements and highlights of the latest development version of the dataset, including comparisons with coupled chemistry climate models, chemical transport models and MACC/CAMS analyses.

The scheme is also operating in near real time and showed elevated ozone over Europe during Spring/Summer 2018 compared to 2017. Analysis of these and other events will be presented.

Area C - Board 292

Tropospheric ozone is a pollutant that damages ecosystems and triggers human health problems. Ozone concentrations are highly variable over time and across the troposphere, which poses clear challenges to deepen our understanding of the processes involved in the production and transport of ozone. Further progress depends on the availability of instruments capable of measuring tropospheric ozone and its distribution at finer spatio-temporal scales. The TROPOMI instrument on the Sentinel-5p platform, launched into an early afternoon polar orbit in October 2017, combines a high spatial resolution, a large swath width and the spectral measurement characteristics required to deliver tropospheric ozone data records at unprecedented detail. The first of these products was released during Fall 2018. It consists in 0.5° (latitude) by 1° (longitude) resolved daily maps of 3-day moving mean values of the tropospheric ozone column between 20°S and 20°N, and it is computed using the convective-cloud method (CCD). A second data product consists in maps of tropical upper tropospheric ozone mixing ratio at a coarser spatial and temporal resolution. It is based on a cloud slicing algorithm (CSA), which is currently being fine-tuned for release in the very near future. A third data product, also under development at the time of this abstract, consists of the vertical profile of ozone concentration in the global troposphere and stratosphere, retrieved with the classical Optimal Estimation (OE) technique.

We present here an assessment of the quality of the first year of the different tropospheric ozone column data sets retrieved from Sentinel-5p TROPOMI measurements, carried out within the context of ESA’s Sentinel-5p Mission Performance Center (MPC) and the S5PVT AO project CHEOPS-5p (Validation of Copernicus Height-resolved Ozone data Products from Sentinel-5p TROPOMI using global sonde and lidar networks, #28587). The first stage of this analysis consists of an inspection of the tropospheric ozone fields for structures which are potentially introduced in the measurement process. Sampling effects, in particular, are a possible source of uncertainty as the CCD product is derived from binned TROPOMI total ozone column data. Another structure introduced by retrieval assumptions would be the dependence of the quality of Sentinel-5p retrieved total column data to cloud parameters. In a second stage the satellite data are confronted to quality-assured ozonesonde and –tentatively– ground-based lidar measurements from the NDACC, SHADOZ and TOLNET networks. These well-characterized observational data records serve as a reference to evaluate the bias and uncertainty of the Sentinel-5p data, and their dependence on influence quantities. The study concludes with an assessment of the compliance of Sentinel-5p tropospheric ozone data with respect to mission and user requirements for key data applications.

Area C - Board 293

The Sentinel 5 Precursor (S5P) satellite was launched into a polar orbit in October 2017, carrying the TROPOMI instrument. S5P has sun synchronous orbit with an equator crossing time of 13:30 LT. TROPOMI achieves an almost daily coverage, due to the wide swath width of 2600 km. The near-real-time (NRTI) ozone total column is based on the two step DOAS approach, consisting of a slant column retrieval and an iterative AMF-calculation. The stratospheric column from the Aura microwave Limb sounder (MLS) is subtracted from the NRTI S5P total column to gain the tropospheric columns.
The stratospheric column results from 4D-Var assimilation of Aura MLS ozone profiles. MLS profiles are assimilated by the Belgian Assimilation System for Chemical ObsErvations (BASCOE, Lefever et al., 2015, ACP) and are delivered with a time frequency of 3 hours. BASCOE stratospheric columns are interpolated linearly in time and space to TROPOMI total column locations before subtraction. In order to avoid potential errors only cloud free TROPOMI data were used in this study.
After a brief introduction of the method the data will be compared to other tropospheric ozone data such as OMPS or sondes. The study focuses on the first year highlighting some tropospheric ozone hotspots like megacities.

Area C - Board 294

Tropospheric ozone is a key species for tropospheric chemistry and air quality. Its monitoring is essential to quantify sources, transport, chemical transformation and sinks of atmospheric pollution. Accurate data are required for understanding and predicting chemical weather. Space-borne observations are very promising for these concerns, especially those using the infrared measurements. Current missions (IASI/MetOp, TES/Aura) will be followed by the new missions in the near future. However, their sensitivity near the surface remains poor and advanced retrieval methods are needed to access to the information from the lowest layers of the troposphere.
In the case of the retrieval of vertical atmospheric profiles from optical measurements, ill-conditioning produces oscillations in the retrieved profiles beyond the error margins defined by the mapping of the measurement noise onto the solution. As for any regularization scheme, a crucial step is the choice of the strength of the applied constraint. This choice depends on the measurement errors and on the sensitivity of the measurements to the target parameters at the different altitudes. For this reason a self-adapting and altitude-dependent regularization scheme is likely preferable over a fixed strength determined a priori, on the basis of sensitivity tests. Such a scheme was already introduced in 2009 and applied to atmospheric profiles retrieved from MIPAS/ENVISAT.
The implementation of this method on nadir retrievals required the appropriated definition of the target function used to optimize the constraint for lower tropospheric retrievals. The challenge for this new retrieval algorithm is to limit the use of a priori constraints to the minimal amount needed to perform the inversion. Since the sensitivity of the observations to the ozone amount in the lowest layers depends on the atmospheric and surface conditions, it is crucial for the inversion algorithm to tune accordingly the contribution of the a priori information.
A statistical analysis based on the simulated retrievals shows that the new algorithm leads to an enhanced sensitivity for measurements at the lower troposphere, which improves biases, spatial correlation, and error estimation for the retrieved lower tropospheric ozone concentrations. The new retrieval scheme was tested for instrumental specifications of the future IASI-NG/MetOp-SG-A instrument. Nevertheless, it may be applied for any other spectral measurements.

Area C - Board 295

Terrestrial ecosystems are a major carbon sink, and so act to mitigate anthropogenic climate change. However, vegetation in many areas are damaged by exposure to tropospheric O₃, which is formed from anthropogenic nitrogen oxides (NO, NO₂) and volatile organic compound (VOC) emissions. Damaged vegetation cannot absorb as much atmospheric CO₂, which would result in more anthropogenic CO₂ remaining in the atmosphere, and so a worsening of climate change. In addition, O₃ exposure also decreases crop yields, and therefore poses an additional threat to future global food security in a warming world.

Previous investigations into O₃ exposure on vegetation have relied on long-term in-situ studies, involving direct fumigation of plants at a constant concentration. Such investigations are costly and extremely geographically limited, and do not cover most of the tropics and emerging economies. Additionally, poorly constrained parameters such as phenotype-specific reactions also affect the uncertainty of derived estimates of O₃-related damage.

Satellite datasets from the ESA Climate Change Initiative (CCI) and Third Party missions provide long-term global monitoring of atmospheric composition and plant productivity, spanning over 20 years of near-continuous observations. These datasets could be combined with existing models of land-atmosphere processes to better estimate the rate of degradation of the terrestrial carbon sink, and to provide more useful metrics on crop losses stemming from O₃ exposure.

In this presentation we will discuss initial progress made towards estimating O₃-induced Gross Primary Productivity (GPP) and crop yield losses using ESA CCI datasets. First, CCI datasets of tropospheric O₃ profiles from various nadir-viewing satellite instruments (e.g. OMI) were compared with in-situ measurements and models of surface O₃ in agricultural regions affected by chronic poor air quality (e.g. Eastern USA), in order to determine which are the most representative of surface conditions.

Next, we created multiple regression models (MLRs) where satellite-derived GPP was regressed against surface O₃ and other dependent parameters (e.g. temperature, soil moisture). Similar models were also derived to investigate the impact of surface O₃ on crop yield. We will also discuss initial progress made towards combining satellite datasets for plant physiology (e.g. soil moisture, leaf area index) with the Joint UK Land Environment Simulator (JULES) land surface model in order to provide more robust estimates of GPP and crop yield losses to compare with the MLRs.

Area C - Board 296

The scientific and industrial communities are being confronted with a strong increase of Earth Observation (EO) satellite missions and related data. This is the case for the Atmospheric Sciences communities, with the already launched Copernicus Sentinel 3 and Sentinel-5 Precursor and the upcoming Sentinel-4, and 5, and ESA’s Earth Explorers scientific satellites ADM-Aeolus and EarthCARE. The challenge is not only to manage the large volume of data generated by each mission / sensor, but also to manage the data variety. Tools are needed to be able to rapidly and trustfully identify, from all available datasets of a specific region for a specific timeframe, all available products for a selected field (e.g. Ozone, trace gases) and prepare these data into a format that is ready to be extracted and used /analysed (Analysis-Ready Data, ARD). Creating synergies among the different datasets will be key to exploit the full potential of the available information. In summary, there is a need of an “intelligent” packaging of subsets of the available data tailored to the users’ needs.
The scope of the “Atmospheric Mission Data Packaging” (AMiDA) project is to design, implement and demonstrate the functionalities of an infrastructure for access and distribution of a wide variety of EO data in the field of Atmospheric Sciences: heritage, current, and future missions will be managed by the platform, to allow the users accessing, visualizing and downloading a meaningful subset of this growing data stream.
The main AMiDA platform will be based on the existing TAMP platform ([1], [2]) that already allows accessing and manipulating a large variety of satellite, model and ground measurements data. The platform will be empowered with an effective spatial and temporal homogenization module and packaging module, that will allow creating, from heterogeneous data sources (e.g. SO2 total column data from different satellites and numerical models) a single data structure (data cube) that will permit simultaneous exploitation of the various data sources. A product descriptor will be associated to the result keeping track of all original products and related metadata, processing parameters and results metadata. The resulting data cube can be exploited using a platform (web application, Jupyter notebook) as well as being downloaded by the user.
A comprehensive demonstration campaign will be performed through five main use cases (UC) to demonstrate the capability of AMiDA to improve the usability of various satellite, model and ground measurement data. Three use cases deal with aerosol products: while UC1 intercompares various space borne aerosol products to consolidate the GRASP algorithm, the scope of UC5 is to create a synergistic aerosol product from different input data sources to retrieve more aerosol parameters with higher spatio-temporal resolution. UC4 is designated for data assimilation of satellite-based aerosol products for model forecast purposes: a relevant Sahara dust episode will be chosen to test the AMiDA system with the inclusion and analysis of different observational and modelling data sets. Similarly, UC3 foresees to utilize AMiDA to visualize in-situ and satellite retrievals simultaneously: Pandora spectrometers measuring amounts of trace gas in the atmosphere for air quality monitoring and Earth Observation satellite validation will be visualized together with OMI and TROPOMI satellite retrievals. Lastly, UC2 deals with the synergistic use of multiple complementary data sets for UV radiation monitoring: the platform will be used to incorporate available satellite data (clouds, ozone, snow height, albedo) into the workflow of generating UV maps.
[1] TAMP platform landing page http://vtpip.zamg.ac.at/
[2] TAMP introductory video https://www.youtube.com/watch?v=xWiy8h1oXQY

Area C - Board 297