Showing papers in "Atmospheric Measurement Techniques in 2020"

S5P TROPOMI NO2 slant column retrieval : Method, stability, uncertainties and comparisons with OMI

Abstract: . The Tropospheric Monitoring Instrument (TROPOMI), aboard the Sentinel-5 Precursor (S5P) satellite, launched on 13 October 2017, provides measurements of atmospheric trace gases and of cloud and aerosol properties at an unprecedented spatial resolution of approximately 7×3.5 km 2 (approx. 5.5×3.5 km 2 as of 6 August 2019), achieving near-global coverage in 1 d. The retrieval of nitrogen dioxide ( NO2 ) concentrations is a three-step procedure: slant column density (SCD) retrieval, separation of the SCD in its stratospheric and tropospheric components, and conversion of these into vertical column densities. This study focusses on the TROPOMI NO2 SCD retrieval: the retrieval method used, the stability of the SCDs and the SCD uncertainties, and a comparison with the Ozone Monitoring Instrument (OMI) NO2 SCDs. The statistical uncertainty, based on the spatial variability of the SCDs over a remote Pacific Ocean sector, is 8.63 µ mol m −2 for all pixels (9.45 µ mol m −2 for clear-sky pixels), which is very stable over time and some 30 % less than the long-term average over OMI–QA4ECV data (since the pixel size reduction TROPOMI uncertainties are ∼8 % larger). The SCD uncertainty reported by the differential optical absorption spectroscopy (DOAS) fit is about 10 % larger than the statistical uncertainty, while for OMI–QA4ECV the DOAS uncertainty is some 20 % larger than its statistical uncertainty. Comparison of the SCDs themselves over the Pacific Ocean, averaged over 1 month, shows that TROPOMI is about 5 % higher than OMI–QA4ECV, which seems to be due mainly to the use of the so-called intensity offset correction in OMI–QA4ECV but not in TROPOMI: turning that correction off means about 5 % higher SCDs. The row-to-row variation in the SCDs of TROPOMI, the “stripe amplitude”, is 2.15 µ mol m −2 , while for OMI–QA4ECV it is a factor of ∼2 ( ∼5 ) larger in 2005 (2018); still, a so-called stripe correction of this non-physical across-track variation is useful for TROPOMI data. In short, TROPOMI shows a superior performance compared with OMI–QA4ECV and operates as anticipated from instrument specifications. The TROPOMI data used in this study cover 30 April 2018 up to 31 January 2020.

The AERONET Version 3 aerosol retrieval algorithm, associated uncertainties and comparisons to Version 2

Abstract: . The Aerosol Robotic Network (AERONET) Version 3 (V3) aerosol retrieval algorithm is described, which is based on the Version 2 (V2) algorithm with numerous updates. Comparisons of V3 aerosol retrievals to those of V2 are presented, along with a new approach to estimate uncertainties in many of the retrieved aerosol parameters. Changes in the V3 aerosol retrieval algorithm include (1) a new polarized radiative transfer code (RTC), which replaced the scalar RTC of V2, (2) detailed characterization of gas absorption by adding NO2 and H2O to specify total gas absorption in the atmospheric column, specification of vertical profiles of all the atmospheric species, (3) new bidirectional reflectance distribution function (BRDF) parameters for land sites adopted from the MODIS BRDF/Albedo product, (4) a new version of the extraterrestrial solar flux spectrum, and (5) a new temperature correction procedure of both direct Sun and sky radiance measurements. The potential effect of each change in V3 on single scattering albedo (SSA) retrievals was analyzed. The operational almucantar retrievals of V2 versus V3 were compared for four AERONET sites: GSFC, Mezaira, Mongu, and Kanpur. Analysis showed very good agreement in retrieved parameters of the size distributions. Comparisons of SSA retrievals for dust aerosols (Mezaira) showed a good agreement in 440 nm SSA, while for longer wavelengths V3 SSAs are systematically higher than those of V2, with the largest mean difference at 675 nm due to cumulative effects of both extraterrestrial solar flux and BRDF changes. For non-dust aerosols, the largest SSA deviation is at 675 nm due to differences in extraterrestrial solar flux spectrums used in each version. Further, the SSA 675 nm mean differences are very different for weakly (GSFC) and strongly (Mongu) absorbing aerosols, which is explained by the lower sensitivity to a bias in aerosol scattering optical depth by less absorbing aerosols. A new hybrid (HYB) sky radiance measurement scan is introduced and discussed. The HYB combines features of scans in two different planes to maximize the range of scattering angles and achieve scan symmetry, thereby allowing for cloud screening and spatial averaging, which is an advantage over the principal plane scan that lacks robust symmetry. We show that due to an extended range of scattering angles, HYB SSA retrievals for dust aerosols exhibit smaller variability with solar zenith angles (SZAs) than those of almucantar (ALM), which allows extension of HYB SSA retrievals to SZAs less than 50 ∘ to as small as 25 ∘ . The comparison of SSA retrievals from closely time-matched HYB and ALM scans in the 50 to 75 ∘ SZA range showed good agreement with the differences below ∼0.005 . We also present an approach to estimate retrieval uncertainties which utilizes the variability in retrieved parameters generated by perturbing both measurements and auxiliary input parameters as a proxy for retrieval uncertainty. The perturbations in measurements and auxiliary inputs are assumed as estimated biases in aerosol optical depth (AOD), radiometric calibration of sky radiances combined with solar spectral irradiance, and surface reflectance. For each set of Level 2 Sun/sky radiometer observations, 27 inputs corresponding to 27 combinations of biases were produced and separately inverted to generate the following statistics of the inversion results: average, standard deviation, minimum and maximum values. From these statistics, standard deviation (labeled U27) is used as a proxy for estimated uncertainty, and a lookup table (LUT) approach was implemented to reduce the computational time. The U27 climatological LUT was generated from the entire AERONET almucantar (1993–2018) and hybrid (2014–2018) scan databases by binning U27s in AOD (440 nm), Angstrom exponent (AE, 440–870 nm), and SSA (440, 675, 870, 1020 nm). Using this LUT approach, the uncertainty estimates U27 for each individual V3 Level 2 retrieval can be obtained by interpolation using the corresponding measured and inverted combination of AOD, AE, and SSA.

Comparison of TROPOMI/Sentinel-5 Precursor NO 2 observations with ground-based measurements in Helsinki

Abstract: . We present a comparison between satellite-based TROPOMI (TROPOspheric Monitoring Instrument) NO2 products and ground-based observations in Helsinki (Finland). TROPOMI NO2 total (summed) columns are compared with the measurements performed by the Pandora spectrometer between April and September 2018. The mean relative and absolute bias between the TROPOMI and Pandora NO2 total columns is about 10 % and 0.12×1015 molec. cm −2 respectively. The dispersion of these differences (estimated as their standard deviation) is 2.2×1015 molec. cm −2 . We find high correlation ( r = 0.68 ) between satellite- and ground-based data, but also that TROPOMI total columns underestimate ground-based observations for relatively large Pandora NO2 total columns, corresponding to episodes of relatively elevated pollution. This is expected because of the relatively large size of the TROPOMI ground pixel ( 3.5×7 km) and the a priori used in the retrieval compared to the relatively small field-of-view of the Pandora instrument. On the other hand, TROPOMI slightly overestimates (within the retrieval uncertainties) relatively small NO2 total columns. Replacing the coarse a priori NO2 profiles with high-resolution profiles from the CAMS chemical transport model improves the agreement between TROPOMI and Pandora total columns for episodes of NO2 enhancement. When only the low values of NO2 total columns or the whole dataset are taken into account, the mean bias slightly increases. The change in bias remains mostly within the uncertainties. We also analyse the consistency between satellite-based data and in situ NO2 surface concentrations measured at the Helsinki–Kumpula air quality station (located a few metres from the Pandora spectrometer). We find similar day-to-day variability between TROPOMI, Pandora and in situ measurements, with NO2 enhancements observed during the same days. Both satellite- and ground-based data show a similar weekly cycle, with lower NO2 levels during the weekend compared to the weekdays as a result of reduced emissions from traffic and industrial activities (as expected in urban sites). The TROPOMI NO2 maps reveal also spatial features, such as the main traffic ways and the airport area, as well as the effect of the prevailing south-west wind patterns. This is one of the first works in which TROPOMI NO2 retrievals are validated against ground-based observations and the results provide an early evaluation of their applicability for monitoring pollution levels in urban sites. Overall, TROPOMI retrievals are valuable to complement the ground-based air quality data (available with high temporal resolution) for describing the spatio-temporal variability of NO2 , even in a relatively small city like Helsinki.

Introducing the 4.4 km spatial resolution Multi-Angle Imaging SpectroRadiometer (MISR) aerosol product

Abstract: . The Multi-angle Imaging SpectroRadiometer (MISR) instrument has been operational on the National Aeronautics and Space Administration (NASA) Earth Observing System (EOS) Terra satellite since early 2000, creating an extensive data set of global Earth observations. Here we introduce the latest version of the MISR aerosol products. The level 2 (swath) product, which is reported on a 4.4 km spatial grid, is designated as version 23 (V23) and contains retrieved aerosol optical depth (AOD) and aerosol particle property information derived from MISR's multi-angle observations over both land and water. The changes from the previous version of the algorithm (V22) have significant impacts on the data product and its interpretation. The V23 data set is created from two separate retrieval algorithms that are applied over dark water and land surfaces, respectively. Besides increasing the horizontal resolution to 4.4 km compared with the coarser 17.6 m resolution in V22 and streamlining the format and content, the V23 product has added geolocation information, pixel-level uncertainty estimates, and improved cloud screening. MISR data can be obtained from the NASA Langley Research Center Atmospheric Science Data Center at https://eosweb.larc.nasa.gov/project/misr/misr_table (last access: 11 October 2019). The version number for the V23 level 2 aerosol product is F13_0023. The level 3 (gridded) aerosol product is still reported at 0.5 ∘ × 0.5 ∘ spatial resolution with results aggregated from the higher-resolution level 2 data. The format and content at level 3 have also been updated to reflect the changes made at level 2. The level 3 product associated with the V23 level 2 product version is designated as F15_0032. Both the level 2 and level 3 products are now provided in NetCDF format.

Evaluating Sentinel-5P TROPOMI tropospheric NO2 column densities with airborne and Pandora spectrometers near New York City and Long Island Sound

Abstract: Airborne and ground-based Pandora spectrometer NOspan classCombining double low line"inline-formula"2/span column measurements were collected during the 2018 Long Island Sound Tropospheric Ozone Study (LISTOS) in the New York City/Long Island Sound region, which coincided with early observations from the Sentinel-5P TROPOspheric Monitoring Instrument (TROPOMI) instrument. Both airborne- and ground-based measurements are used to evaluate the TROPOMI NOspan classCombining double low line"inline-formula"2/span Tropospheric Vertical Column (TrVC) product v1.2 in this region, which has high spatial and temporal heterogeneity in NOspan classCombining double low line"inline-formula"2/span. First, airborne and Pandora TrVCs are compared to evaluate the uncertainty of the airborne TrVC and establish the spatial representativeness of the Pandora observations. The 171 coincidences between Pandora and airborne TrVCs are found to be highly correlated (span classCombining double low line"inline-formula"ir/i2Combining double low line/spanthinsp;0.92 and slope of 1.03), with the largest individual differences being associated with high temporal and/or spatial variability. These reference measurements (Pandora and airborne) are complementary with respect to temporal coverage and spatial representativity. Pandora spectrometers can provide continuous long-term measurements but may lack areal representativity when operated in direct-sun mode. Airborne spectrometers are typically only deployed for short periods of time, but their observations are more spatially representative of the satellite measurements with the added capability of retrieving at subpixel resolutions of 250thinsp;mthinsp;span classCombining double low line"inline-formula"×/spanthinsp;250thinsp;m over the entire TROPOMI pixels they overfly. Thus, airborne data are more correlated with TROPOMI measurements (span classCombining double low line"inline-formula"ir/i2Combining double low line0.96/span) than Pandora measurements are with TROPOMI (span classCombining double low line"inline-formula"ir/i2Combining double low line0.84/span). The largest outliers between TROPOMI and the reference measurements appear to stem from too spatially coarse a priori surface reflectivity (0.5span classCombining double low line"inline-formula"g /span) over bright urban scenes. In this work, this results during cloud-free scenes that, at times, are affected by errors in the TROPOMI cloud pressure retrieval impacting the calculation of tropospheric air mass factors. This factor causes a high bias in TROPOMI TrVCs of 4thinsp;%-11thinsp;%. Excluding these cloud-impacted points, TROPOMI has an overall low bias of 19thinsp;%-33thinsp;% during the LISTOS timeframe of June-September 2018. Part of this low bias is caused by coarse a priori profile input from the TM5-MP model; replacing thesespan idCombining double low line"page6114"/ profiles with those from a 12thinsp;km North American Model-Community Multiscale Air Quality (NAMCMAQ) analysis results in a 12thinsp;%-14thinsp;% increase in the TrVCs. Even with this improvement, the TROPOMI-NAMCMAQ TrVCs have a 7thinsp;%-19thinsp;% low bias, indicating needed improvement in a priori assumptions in the air mass factor calculation. Future work should explore additional impacts of a priori inputs to further assess the remaining low biases in TROPOMI using these datasets./.

Real-time pollen monitoring using digital holography

Abstract: . We present the first validation of the Swisens Poleno, currently the only operational automatic pollen monitoring system based on digital holography. The device provides in-flight images of all coarse aerosols, and here we develop a two-step classification algorithm that uses these images to identify a range of pollen taxa. Deterministic criteria based on the shape of the particle are applied to initially distinguish between intact pollen grains and other coarse particulate matter. This first level of discrimination identifies pollen with an accuracy of 96 % . Thereafter, individual pollen taxa are recognized using supervised learning techniques. The algorithm is trained using data obtained by inserting known pollen types into the device, and out of eight pollen taxa six can be identified with an accuracy of above 90 % . In addition to the ability to correctly identify aerosols, an automatic pollen monitoring system needs to be able to correctly determine particle concentrations. To further verify the device, controlled chamber experiments using polystyrene latex beads were performed. This provided reference aerosols with traceable particle size and number concentrations in order to ensure particle size and sampling volume were correctly characterized.

Laboratory evaluation of particle-size selectivity of optical low-cost particulate matter sensors

Abstract: . Low-cost particulate matter (PM) sensors have been under investigation as it has been hypothesized that the use of low-cost and easy-to-use sensors could allow cost-efficient extension of the currently sparse measurement coverage. While the majority of the existing literature highlights that low-cost sensors can indeed be a valuable addition to the list of commonly used measurement tools, it often reiterates that the risk of sensor misuse is still high and that the data obtained from the sensors are only representative of the specific site and its ambient conditions. This implies that there are underlying reasons for inaccuracies in sensor measurements that have yet to be characterized. The objective of this study is to investigate the particle-size selectivity of low-cost sensors. Evaluated sensors were Plantower PMS5003, Nova SDS011, Sensirion SPS30, Sharp GP2Y1010AU0F, Shinyei PPD42NS, and Omron B5W-LD0101. The investigation of size selectivity was carried out in the laboratory using a novel reference aerosol generation system capable of steadily producing monodisperse particles of different sizes (from ∼0.55 to 8.4 µ m) on-line. The results of the study show that none of the low-cost sensors adhered to the detection ranges declared by the manufacturers; moreover, cursory comparison to a mid-cost aerosol size spectrometer (Grimm 1.108, 2020) indicates that the sensors can only achieve independent responses for one or two size bins, whereas the spectrometer can sufficiently characterize particles with 15 different size bins. These observations provide insight into and evidence of the notion that particle-size selectivity has an essential role in the analysis of the sources of errors in sensors.

First validation of Aeolus wind observations by airborne Doppler wind lidar measurements

Abstract: . Soon after the launch of Aeolus on 22 August 2018, the first ever wind lidar in space developed by the European Space Agency (ESA) has been providing profiles of the component of the wind vector along the instrument's line of sight (LOS) on a global scale. In order to validate the quality of Aeolus wind observations, the German Aerospace Center (Deutsches Zentrum fur Luft- und Raumfahrt e.V., DLR) recently performed two airborne campaigns over central Europe deploying two different Doppler wind lidars (DWLs) on board the DLR Falcon aircraft. The first campaign – WindVal III – was conducted from 5 November 2018 until 5 December 2018 and thus still within the commissioning phase of the Aeolus mission. The second campaign – AVATARE (Aeolus Validation Through Airborne Lidars in Europe) – was performed from 6 May 2019 until 6 June 2019. Both campaigns were flown out of the DLR site in Oberpfaffenhofen, Germany, during the evening hours for probing the ascending orbits. All together, 10 satellite underflights with 19 flight legs covering more than 7500 km of Aeolus swaths were performed and used to validate the early-stage wind data product of Aeolus by means of collocated airborne wind lidar observations for the first time. For both campaign data sets, the statistical comparison of Aeolus horizontal line-of-sight (HLOS) observations and the corresponding wind observations of the reference lidar (2 µm DWL) on board the Falcon aircraft shows enhanced systematic and random errors compared with the bias and precision requirements defined for Aeolus. In particular, the systematic errors are determined to be 2.1 m s −1 (Rayleigh) and 2.3 m s −1 (Mie) for WindVal III and −4.6 m s −1 (Rayleigh) and −0.2 m s −1 (Mie) for AVATARE. The corresponding random errors are determined to be 3.9 m s −1 (Rayleigh) and 2.0 m s −1 (Mie) for WindVal III and 4.3 m s −1 (Rayleigh) and 2.0 m s −1 (Mie) for AVATARE. The Aeolus observations used here were acquired in an altitude range up to 10 km and have mainly a vertical resolution of 1 km (Rayleigh) and 0.5 to 1.0 km (Mie) and a horizontal resolution of 90 km (Rayleigh) and down to 10 km (Mie). Potential reasons for those errors are analyzed and discussed.

TROPOMI-Sentinel-5 Precursor formaldehyde validation using an extensive network of ground-based Fourier-transform infrared stations

Abstract: . TROPOMI (the TROPOspheric Monitoring Instrument), on board the Sentinel-5 Precursor (S5P) satellite, has been monitoring the Earth's atmosphere since October 2017 with an unprecedented horizontal resolution (initially 7 km 2×3.5 km 2 , upgraded to 5.5 km 2×3.5 km 2 in August 2019). Monitoring air quality is one of the main objectives of TROPOMI; it obtains measurements of important pollutants such as nitrogen dioxide, carbon monoxide, and formaldehyde (HCHO). In this paper we assess the quality of the latest HCHO TROPOMI products versions 1.1.(5-7), using ground-based solar-absorption FTIR (Fourier-transform infrared) measurements of HCHO from 25 stations around the world, including high-, mid-, and low-latitude sites. Most of these stations are part of the Network for the Detection of Atmospheric Composition Change (NDACC), and they provide a wide range of observation conditions, from very clean remote sites to those with high HCHO levels from anthropogenic or biogenic emissions. The ground-based HCHO retrieval settings have been optimized and harmonized at all the stations, ensuring a consistent validation among the sites. In this validation work, we first assess the accuracy of TROPOMI HCHO tropospheric columns using the median of the relative differences between TROPOMI and FTIR ground-based data (BIAS). The pre-launch accuracy requirements of TROPOMI HCHO are 40 %–80 %. We observe that these requirements are well reached, with the BIAS found below 80 % at all the sites and below 40 % at 20 of the 25 sites. The provided TROPOMI systematic uncertainties are well in agreement with the observed biases at most of the stations except for the highest-HCHO-level site, where it is found to be underestimated. We find that while the BIAS has no latitudinal dependence, it is dependent on the HCHO concentration levels: an overestimation ( + 26 ± 5 %) of TROPOMI is observed for very low HCHO levels ( 2.5 × 10 15 molec. cm −2 ), while an underestimation ( - 30.8 % ± 1.4 %) is found for high HCHO levels ( > 8.0 × 10 15 molec. cm −2 ). This demonstrates the great value of such a harmonized network covering a wide range of concentration levels, the sites with high HCHO concentrations being crucial for the determination of the satellite bias in the regions of emissions and the clean sites allowing a small TROPOMI offset to be determined. The wide range of sampled HCHO levels within the network allows the robust determination of the significant constant and proportional TROPOMI HCHO biases (TROPOMI = + 1.10 ± 0.05 × 10 15 + 0.64 ± 0.03 × FTIR; in molecules per square centimetre). Second, the precision of TROPOMI HCHO data is estimated by the median absolute deviation (MAD) of the relative differences between TROPOMI and FTIR ground-based data. The clean sites are especially useful for minimizing a possible additional collocation error. The precision requirement of 1.2×1016 molec. cm −2 for a single pixel is reached at most of the clean sites, where it is found that the TROPOMI precision can even be 2 times better (0.5– 0.8×1015 molec. cm −2 for a single pixel). However, we find that the provided TROPOMI random uncertainties may be underestimated by a factor of 1.6 (for clean sites) to 2.3 (for high HCHO levels). The correlation is very good between TROPOMI and FTIR data ( R=0.88 for 3 h mean coincidences; R=0.91 for monthly means coincidences). Using about 17 months of data (from May 2018 to September 2019), we show that the TROPOMI seasonal variability is in very good agreement at all of the FTIR sites. The FTIR network demonstrates the very good quality of the TROPOMI HCHO products, which is well within the pre-launch requirements for both accuracy and precision. This paper makes suggestions for the refinement of the TROPOMI random uncertainty budget and TROPOMI quality assurance values for a better filtering of the remaining outliers.

Assessing the accuracy of low-cost optical particle sensors using a physics-based approach.

Abstract: . Low-cost sensors for measuring particulate matter (PM) offer the ability to understand human exposure to air pollution at spatiotemporal scales that have previously been impractical. However, such low-cost PM sensors tend to be poorly characterized, and their measurements of mass concentration can be subject to considerable error. Recent studies have investigated how individual factors can contribute to this error, but these studies are largely based on empirical comparisons and generally do not examine the role of multiple factors simultaneously. Here, we present a new physics-based framework and open-source software package (opcsim) for evaluating the ability of low-cost optical particle sensors (optical particle counters and nephelometers) to accurately characterize the size distribution and/or mass loading of aerosol particles. This framework, which uses Mie theory to calculate the response of a given sensor to a given particle population, is used to estimate the fractional error in mass loading for different sensor types given variations in relative humidity, aerosol optical properties, and the underlying particle size distribution. Results indicate that such error, which can be substantial, is dependent on the sensor technology (nephelometer vs. optical particle counter), the specific parameters of the individual sensor, and differences between the aerosol used to calibrate the sensor and the aerosol being measured. We conclude with a summary of likely sources of error for different sensor types, environmental conditions, and particle classes and offer general recommendations for the choice of calibrant under different measurement scenarios.

A global analysis of climate-relevant aerosol properties retrieved from the network of Global Atmosphere Watch (GAW) near-surface observatories

Abstract: . Aerosol particles are essential constituents of the Earth's atmosphere, impacting the earth radiation balance directly by scattering and absorbing solar radiation, and indirectly by acting as cloud condensation nuclei. In contrast to most greenhouse gases, aerosol particles have short atmospheric residence times, resulting in a highly heterogeneous distribution in space and time. There is a clear need to document this variability at regional scale through observations involving, in particular, the in situ near-surface segment of the atmospheric observation system. This paper will provide the widest effort so far to document variability of climate-relevant in situ aerosol properties (namely wavelength dependent particle light scattering and absorption coefficients, particle number concentration and particle number size distribution) from all sites connected to the Global Atmosphere Watch network. High-quality data from almost 90 stations worldwide have been collected and controlled for quality and are reported for a reference year in 2017, providing a very extended and robust view of the variability of these variables worldwide. The range of variability observed worldwide for light scattering and absorption coefficients, single-scattering albedo, and particle number concentration are presented together with preliminary information on their long-term trends and comparison with model simulation for the different stations. The scope of the present paper is also to provide the necessary suite of information, including data provision procedures, quality control and analysis, data policy, and usage of the ground-based aerosol measurement network. It delivers to users of the World Data Centre on Aerosol, the required confidence in data products in the form of a fully characterized value chain, including uncertainty estimation and requirements for contributing to the global climate monitoring system.

Assessment of the quality of TROPOMI high-spatial-resolution NO 2 data products in the Greater Toronto Area

Abstract: . The TROPOspheric Monitoring Instrument (TROPOMI) aboard the Sentinel-5 Precursor satellite (launched on 13 October 2017) is a nadir-viewing spectrometer measuring reflected sunlight in the ultraviolet, visible, near-infrared, and shortwave infrared spectral ranges. The measured spectra are used to retrieve total columns of trace gases, including nitrogen dioxide ( NO2 ). For ground validation of these satellite measurements, Pandora spectrometers, which retrieve high-quality NO2 total columns via direct-sun measurements, are widely used. In this study, Pandora NO2 measurements made at three sites located in or north of the Greater Toronto Area (GTA) are used to evaluate the TROPOMI NO2 data products, including a standard Royal Netherlands Meteorological Institute (KNMI) tropospheric and stratospheric NO2 data product and a TROPOMI research data product developed by Environment and Climate Change Canada (ECCC) using a high-resolution regional air quality forecast model (in the air mass factor calculation). It is found that these current TROPOMI tropospheric NO2 data products (standard and ECCC) met the TROPOMI design bias requirement (< 10 %). Using the statistical uncertainty estimation method, the estimated TROPOMI upper-limit precision falls below the design requirement at a rural site but above in the other two urban and suburban sites. The Pandora instruments are found to have sufficient precision (< 0.02 DU) to perform TROPOMI validation work. In addition to the traditional satellite validation method (i.e., pairing ground-based measurements with satellite measurements closest in time and space), we analyzed TROPOMI pixels located upwind and downwind from the Pandora site. This makes it possible to improve the statistics and better interpret the high-spatial-resolution measurements made by TROPOMI. By using this wind-based validation technique, the number of coincident measurements can be increased by about a factor of 5. With this larger number of coincident measurements, this work shows that both TROPOMI and Pandora instruments can reveal detailed spatial patterns (i.e., horizontal distributions) of local and transported NO2 emissions, which can be used to evaluate regional air quality changes. The TROPOMI ECCC NO2 research data product shows improved agreement with Pandora measurements compared to the TROPOMI standard tropospheric NO2 data product (e.g., lower multiplicative bias at the suburban and urban sites by about 10 %), demonstrating benefits from the high-resolution regional air quality forecast model.

Intercomparison of NO2, O4, O3 and HCHO slant column measurements by MAX-DOAS and zenith-sky UV-visible spectrometers during CINDI-2

Abstract: . In September 2016, 36 spectrometers from 24 institutes measured a number of key atmospheric pollutants for a period of 17 d during the Second Cabauw Intercomparison campaign for Nitrogen Dioxide measuring Instruments (CINDI-2) that took place at Cabauw, the Netherlands (51.97 ∘ N, 4.93 ∘ E). We report on the outcome of the formal semi-blind intercomparison exercise, which was held under the umbrella of the Network for the Detection of Atmospheric Composition Change (NDACC) and the European Space Agency (ESA). The three major goals of CINDI-2 were (1) to characterise and better understand the differences between a large number of multi-axis differential optical absorption spectroscopy (MAX-DOAS) and zenith-sky DOAS instruments and analysis methods, (2) to define a robust methodology for performance assessment of all participating instruments, and (3) to contribute to a harmonisation of the measurement settings and retrieval methods. This, in turn, creates the capability to produce consistent high-quality ground-based data sets, which are an essential requirement to generate reliable long-term measurement time series suitable for trend analysis and satellite data validation. The data products investigated during the semi-blind intercomparison are slant columns of nitrogen dioxide ( NO2 ), the oxygen collision complex ( O4 ) and ozone ( O3 ) measured in the UV and visible wavelength region, formaldehyde (HCHO) in the UV spectral region, and NO2 in an additional (smaller) wavelength range in the visible region. The campaign design and implementation processes are discussed in detail including the measurement protocol, calibration procedures and slant column retrieval settings. Strong emphasis was put on the careful alignment and synchronisation of the measurement systems, resulting in a unique set of measurements made under highly comparable air mass conditions. The CINDI-2 data sets were investigated using a regression analysis of the slant columns measured by each instrument and for each of the target data products. The slope and intercept of the regression analysis respectively quantify the mean systematic bias and offset of the individual data sets against the selected reference (which is obtained from the median of either all data sets or a subset), and the rms error provides an estimate of the measurement noise or dispersion. These three criteria are examined and for each of the parameters and each of the data products, performance thresholds are set and applied to all the measurements. The approach presented here has been developed based on heritage from previous intercomparison exercises. It introduces a quantitative assessment of the consistency between all the participating instruments for the MAX-DOAS and zenith-sky DOAS techniques.

Intercomparison of wind observations from the European Space Agency's Aeolus satellite mission and the ALADIN Airborne Demonstrator

Abstract: . Shortly after the successful launch of the European Space Agency's wind mission Aeolus, co-located airborne wind lidar observations were performed in central Europe; these observations employed a prototype of the satellite instrument – the ALADIN (Atmospheric LAser Doppler INstrument) Airborne Demonstrator (A2D). Like the direct-detection Doppler wind lidar on-board Aeolus, the A2D is composed of a frequency-stabilized ultra-violet (UV) laser, a Cassegrain telescope and a dual-channel receiver to measure line-of-sight (LOS) wind speeds by analysing both Mie and Rayleigh backscatter signals. In the framework of the first airborne validation campaign after the launch and still during the commissioning phase of the mission, four coordinated flights along the satellite swath were conducted in late autumn of 2018, yielding wind data in the troposphere with high coverage of the Rayleigh channel. Owing to the different measurement grids and LOS viewing directions of the satellite and the airborne instrument, intercomparison with the Aeolus wind product requires adequate averaging as well as conversion of the measured A2D LOS wind speeds to the satellite LOS (LOS*). The statistical comparison of the two instruments shows a positive bias (of 2.6 m s −1 ) of the Aeolus Rayleigh winds (measured along its LOS*) with respect to the A2D Rayleigh winds as well as a standard deviation of 3.6 m s −1 . Considering the accuracy and precision of the A2D wind data, which were determined from comparison with a highly accurate coherent wind lidar as well as with the European Centre for Medium-Range Weather Forecasts (ECMWF) model winds, the systematic and random errors of the Aeolus LOS* Rayleigh winds are 1.7 and 2.5 m s −1 respectively. The paper also discusses the influence of different threshold parameters implemented in the comparison algorithm as well as an optimization of the A2D vertical sampling to be used in forthcoming validation campaigns.

MAX-DOAS measurements of tropospheric NO 2 and HCHO in Munich and the comparison to OMI and TROPOMI satellite observations

Abstract: . We present two-dimensional scanning Multi-AXis Differential Optical Absorption Spectroscopy (MAX-DOAS) observations of nitrogen dioxide ( NO2 ) and formaldehyde (HCHO) in Munich. Vertical columns and vertical distribution profiles of aerosol extinction coefficient, NO2 and HCHO are retrieved from the 2D MAX-DOAS observations. The measured surface aerosol extinction coefficients and NO2 mixing ratios derived from the retrieved profiles are compared to in situ monitoring data, and the surface NO2 mixing ratios show a good agreement with in situ monitoring data with a Pearson correlation coefficient ( R ) of 0.91. The aerosol optical depths (AODs) show good agreement as well ( R = 0.80 ) when compared to sun photometer measurements. Tropospheric vertical column densities (VCDs) of NO2 and HCHO derived from the MAX-DOAS measurements are also used to validate Ozone Monitoring Instrument (OMI) and TROPOspheric Monitoring Instrument (TROPOMI) satellite observations. Monthly averaged data show a good correlation; however, satellite observations are on average 30 % lower than the MAX-DOAS measurements. Furthermore, the MAX-DOAS observations are used to investigate the spatiotemporal characteristic of NO2 and HCHO in Munich. Analysis of the relations between aerosol, NO2 and HCHO shows higher aerosol-to-HCHO ratios in winter, which reflects a longer atmospheric lifetime of secondary aerosol and HCHO during winter. The analysis also suggests that secondary aerosol formation is the major source of these aerosols in Munich.

A review and framework for the evaluation of pixel-level uncertainty estimates in satellite aerosol remote sensing

Abstract: . Recent years have seen the increasing inclusion of per-retrieval prognostic (predictive) uncertainty estimates within satellite aerosol optical depth (AOD) data sets, providing users with quantitative tools to assist in the optimal use of these data. Prognostic estimates contrast with diagnostic (i.e. relative to some external truth) ones, which are typically obtained using sensitivity and/or validation analyses. Up to now, however, the quality of these uncertainty estimates has not been routinely assessed. This study presents a review of existing prognostic and diagnostic approaches for quantifying uncertainty in satellite AOD retrievals, and it presents a general framework to evaluate them based on the expected statistical properties of ensembles of estimated uncertainties and actual retrieval errors. It is hoped that this framework will be adopted as a complement to existing AOD validation exercises; it is not restricted to AOD and can in principle be applied to other quantities for which a reference validation data set is available. This framework is then applied to assess the uncertainties provided by several satellite data sets (seven over land, five over water), which draw on methods from the empirical to sensitivity analyses to formal error propagation, at 12 Aerosol Robotic Network (AERONET) sites. The AERONET sites are divided into those for which it is expected that the techniques will perform well and those for which some complexity about the site may provide a more severe test. Overall, all techniques show some skill in that larger estimated uncertainties are generally associated with larger observed errors, although they are sometimes poorly calibrated (i.e. too small or too large in magnitude). No technique uniformly performs best. For powerful formal uncertainty propagation approaches such as optimal estimation, the results illustrate some of the difficulties in appropriate population of the covariance matrices required by the technique. When the data sets are confronted by a situation strongly counter to the retrieval forward model (e.g. potentially mixed land–water surfaces or aerosol optical properties outside the family of assumptions), some algorithms fail to provide a retrieval, while others do but with a quantitatively unreliable uncertainty estimate. The discussion suggests paths forward for the refinement of these techniques.

An overview of and issues with sky radiometer technology and SKYNET

Abstract: . This paper is an overview of the progress in sky radiometer technology and the development of the network called SKYNET. It is found that the technology has produced useful on-site calibration methods, retrieval algorithms, and data analyses from sky radiometer observations of aerosol, cloud, water vapor, and ozone. A formula was proposed for estimating the accuracy of the sky radiometer calibration constant F0 using the improved Langley (IL) method, which was found to be a good approximation to observed monthly mean uncertainty in F0 , around 0.5 % to 2.4 % at the Tokyo and Rome sites and smaller values of around 0.3 % to 0.5 % at the mountain sites at Mt. Saraswati and Davos. A new cross IL (XIL) method was also developed to correct an underestimation by the IL method in cases with large aerosol retrieval errors. The root-mean-square difference (RMSD) in aerosol optical thickness (AOT) comparisons with other networks took values of less than 0.02 for λ≥500 nm and a larger value of about 0.03 for shorter wavelengths in city areas and smaller values of less than 0.01 in mountain comparisons. Accuracies of single-scattering albedo (SSA) and size distribution retrievals are affected by the propagation of errors in measurement, calibrations for direct solar and diffuse sky radiation, ground albedo, cloud screening, and the version of the analysis software called the Skyrad pack. SSA values from SKYNET were up to 0.07 larger than those from AERONET, and the major error sources were identified as an underestimation of solid viewing angle (SVA) and cloud contamination. Correction of these known error factors reduced the SSA difference to less than 0.03. Retrievals of other atmospheric constituents by the sky radiometer were also reviewed. Retrieval accuracies were found to be about 0.2 cm for precipitable water vapor amount and 13 DU (Dobson Unit) for column ozone amount. Retrieved cloud optical properties still showed large deviations from validation data, suggesting a need to study the causes of the differences. It is important that these recent studies on improvements presented in the present paper are introduced into the existing operational systems and future systems of the International SKYNET Data Center.

Evaluation and calibration of a low-cost particle sensor in ambient conditions using machine-learning methods

Abstract: . Particle sensing technology has shown great potential for monitoring particulate matter (PM) with very few temporal and spatial restrictions because of its low cost, compact size, and easy operation. However, the performance of low-cost sensors for PM monitoring in ambient conditions has not been thoroughly evaluated. Monitoring results by low-cost sensors are often questionable. In this study, a low-cost fine particle monitor (Plantower PMS 5003) was colocated with a reference instrument, the Synchronized Hybrid Ambient Real-time Particulate (SHARP) monitor, at the Calgary Varsity air monitoring station from December 2018 to April 2019. The study evaluated the performance of this low-cost PM sensor in ambient conditions and calibrated its readings using simple linear regression (SLR), multiple linear regression (MLR), and two more powerful machine-learning algorithms using random search techniques for the best model architectures. The two machine-learning algorithms are XGBoost and a feedforward neural network (NN). Field evaluation showed that the Pearson correlation ( r ) between the low-cost sensor and the SHARP instrument was 0.78. The Fligner and Killeen (F–K) test indicated a statistically significant difference between the variances of the PM 2.5 values by the low-cost sensor and the SHARP instrument. Large overestimations by the low-cost sensor before calibration were observed in the field and were believed to be caused by the variation of ambient relative humidity. The root mean square error (RMSE) was 9.93 when comparing the low-cost sensor with the SHARP instrument. The calibration by the feedforward NN had the smallest RMSE of 3.91 in the test dataset compared to the calibrations by SLR (4.91), MLR (4.65), and XGBoost (4.19). After calibrations, the F–K test using the test dataset showed that the variances of the PM 2.5 values by the NN, XGBoost, and the reference method were not statistically significantly different. From this study, we conclude that a feedforward NN is a promising method to address the poor performance of low-cost sensors for PM 2.5 monitoring. In addition, the random search method for hyperparameters was demonstrated to be an efficient approach for selecting the best model structure.

Measurements of PM 2.5 with PurpleAir under atmospheric conditions

Abstract: . The PurpleAir PA-II unit is a low-cost sensor for monitoring changes in the concentrations of particulate matter (PM) of various sizes. There are currently more than 10 000 PA-II units in use worldwide; some of the units are located in areas where no other reference air monitoring system is present. Previous studies have examined the performance of these PA-II units (or the sensors within them) in comparison to a co-located reference air monitoring system. However, because PA-II units are installed by PurpleAir customers, most of the PA-II units are not co-located with a reference air monitoring system and, in many cases, are not near one. This study aims to examine how each PA-II unit performs under atmospheric conditions when exposed to a variety of pollutants and PM 2.5 concentrations (PM with an aerodynamic diameter smaller than 2.5 µ m), when at a distance from the reference sensor. We examine how PA-II units perform in comparison to other PA-II units and Environmental Protection Agency (EPA) Air Quality Monitoring Stations (AQMSs) that are not co-located with them. For this study, we selected four different regions, each containing multiple PA-II units (minimum of seven per region). In addition, each region needed to have at least one AQMS unit that was co-located with at least one PA-II unit, all units needed to be at a distance of up to 5 km from an AQMS unit and up to 10 km between each other. Correction of PM 2.5 values of the co-located PA-II units was implemented by multivariate linear regression (MLR), taking into account changes of temperature and relative humidity. The fit coefficients, received from the MLR, were then used to correct the PM 2.5 values in all the remaining PA-II units in the region. Hourly PM 2.5 measurements from each PA-II unit were compared to those from the AQMSs and other PA-II units in its region. The correction of the PM 2.5 values improved the R -squared value ( R2 ), root-mean-square error (RMSE), and mean absolute error (MAE) and slope values between all units. In most cases, the AQMSs and the PA-II units were found to be in good agreement (75 % of the comparisons had a R2>0.8 ); they measured similar values and followed similar trends; that is, when the PM 2.5 values measured by the AQMSs increased or decreased, so did those of the PA-II units. In some high-pollution events, the corrected PA-II had slightly higher PM 2.5 values compared to those measured by the AQMS. Distance between the units did not impact the comparison between units. Overall, the PA-II unit, after corrections of PM 2.5 values, seems to be a promising tool for identifying relative changes in PM 2.5 concentration with the potential to complement sparsely distributed monitoring stations and to aid in assessing and minimizing the public exposure to PM.

Effect of aerosol composition on the performance of low-cost optical particle counter correction factors

Abstract: . There is considerable interest in using low-cost optical particle counters (OPCs) to supplement existing routine air quality networks that monitor particle mass concentrations. In order to do this, low-cost OPC data need to be comparable with particle mass reference instrumentation; however, there is currently no widely agreed upon methodology to accomplish this. Aerosol hygroscopicity is known to be a key parameter to consider when correcting particle mass concentrations derived from low-cost OPCs, particularly at high ambient relative humidity (RH). Correction factors have been developed that apply κ -Kohler theory to correct for the influence of water uptake by hygroscopic aerosols. We have used datasets of co-located reference particle measurements and low-cost OPC (OPC-N2, Alphasense) measurements, collected in four cities on three continents, to explore the performance of this correction factor. We provide evidence that the elevated particle mass concentrations, reported by the low-cost OPC relative to reference instrumentation, are due to bulk aerosol hygroscopicity under different RH conditions, which is determined by aerosol composition and, in particular, the levels of hygroscopic aerosols (sulfate and nitrate). We exploit measurements made in volcanic plumes in Nicaragua, which are predominantly composed of sulfate aerosol, as a natural experiment to demonstrate this behaviour in the ambient atmosphere; the observed humidogram from these measurements closely resembles the calculated pure sulfuric acid humidogram. The results indicate that the particle mass concentrations derived from low-cost OPCs during periods of high RH ( >60 %) need to be corrected for aerosol hygroscopic growth. We employed a correction factor based on κ -Kohler theory and observed that the corrected OPC-N2 PM 2.5 mass concentrations were within 33 % of reference measurements at all sites. The results indicated that a κ value derived in situ (using suitable reference instrumentation) would lead to the most accurate correction relative to co-located reference instruments. Applying a κ values from the literature in the correction factor also resulted in improved OPC-N2 performance, with the measurements being within 50 % of the reference values. Therefore, for areas where suitable reference instrumentation for developing a local correction factor is lacking, using a literature κ value can result in a reasonable correction. For locations with low levels of hygroscopic aerosols and low RH values, a simple calibration against gravimetric measurements (using suitable reference instrumentation) would likely be sufficient. Whilst this study generated correction factors specific for the Alphasense OPC-N2 sensor, the calibration methodology developed is likely amenable to other low-cost PM sensors.

SegCloud: a novel cloud image segmentation model using a deep convolutional neural network for ground-based all-sky-view camera observation

Abstract: . Cloud detection and cloud properties have substantial applications in weather forecast, signal attenuation analysis, and other cloud-related fields. Cloud image segmentation is the fundamental and important step in deriving cloud cover. However, traditional segmentation methods rely on low-level visual features of clouds and often fail to achieve satisfactory performance. Deep convolutional neural networks (CNNs) can extract high-level feature information of objects and have achieved remarkable success in many image segmentation fields. On this basis, a novel deep CNN model named SegCloud is proposed and applied for accurate cloud segmentation based on ground-based observation. Architecturally, SegCloud possesses a symmetric encoder–decoder structure. The encoder network combines low-level cloud features to form high-level, low-resolution cloud feature maps, whereas the decoder network restores the obtained high-level cloud feature maps to the same resolution of input images. The Softmax classifier finally achieves pixel-wise classification and outputs segmentation results. SegCloud has powerful cloud discrimination capability and can automatically segment whole-sky images obtained by a ground-based all-sky-view camera. The performance of SegCloud is validated by extensive experiments, which show that SegCloud is effective and accurate for ground-based cloud segmentation and achieves better results than traditional methods do. The accuracy and practicability of SegCloud are further proven by applying it to cloud cover estimation.

Rain event detection in commercial microwave link attenuation data using convolutional neural networks

Abstract: . Quantitative precipitation estimation with commercial microwave links (CMLs) is a technique developed to supplement weather radar and rain gauge observations. It is exploiting the relation between the attenuation of CML signal levels and the integrated rain rate along a CML path. The opportunistic nature of this method requires a sophisticated data processing using robust methods. In this study we focus on the processing step of rain event detection in the signal level time series of the CMLs, which we treat as a binary classification problem. This processing step is particularly challenging, because even when there is no rain, the signal level can show large fluctuations similar to that during rainy periods. False classifications can have a high impact on falsely estimated rainfall amounts. We analyze the performance of a convolutional neural network (CNN), which is trained to detect rainfall-specific attenuation patterns in CML signal levels, using data from 3904 CMLs in Germany. The CNN consists of a feature extraction and a classification part with, in total, 20 layers of neurons and 1.4×105 trainable parameters. With a structure inspired by the visual cortex of mammals, CNNs use local connections of neurons to recognize patterns independent of their location in the time series. We test the CNN's ability to recognize attenuation patterns from CMLs and time periods outside the training data. Our CNN is trained on 4 months of data from 800 randomly selected CMLs and validated on 2 different months of data, once for all CMLs and once for the 3104 CMLs not included in the training. No CMLs are excluded from the analysis. As a reference data set, we use the gauge-adjusted radar product RADOLAN-RW provided by the German meteorological service (DWD). The model predictions and the reference data are compared on an hourly basis. Model performance is compared to a state-of-the-art reference method, which uses the rolling standard deviation of the CML signal level time series as a detection criteria. Our results show that within the analyzed period of April to September 2018, the CNN generalizes well to the validation CMLs and time periods. A receiver operating characteristic (ROC) analysis shows that the CNN is outperforming the reference method, detecting on average 76 % of all rainy and 97 % of all nonrainy periods. From all periods with a reference rain rate larger than 0.6 mm h −1 , more than 90 % was detected. We also show that the improved event detection leads to a significant reduction of falsely estimated rainfall by up to 51 %. At the same time, the quality of the correctly estimated rainfall is kept at the same level in regards to the Pearson correlation with the radar rainfall. In conclusion, we find that CNNs are a robust and promising tool to detect rainfall-induced attenuation patterns in CML signal levels from a large CML data set covering all of Germany.

Consistency and structural uncertainty of multi-mission GPS radio occultation records

Abstract: . Atmospheric climate monitoring requires observations of high quality that conform to the criteria of the Global Climate Observing System (GCOS). Radio occultation (RO) data based on Global Positioning System (GPS) signals are available since 2001 from several satellite missions with global coverage, high accuracy, and high vertical resolution in the troposphere and lower stratosphere. We assess the consistency and long-term stability of multi-satellite RO observations for use as climate data records. As a measure of long-term stability, we quantify the structural uncertainty of RO data products arising from different processing schemes. We analyze atmospheric variables from bending angle to temperature for four RO missions, CHAMP, Formosat-3/COSMIC, GRACE, and Metop, provided by five data centers. The comparisons are based on profile-to-profile differences aggregated to monthly medians. Structural uncertainty in trends is found to be lowest from 8 to 25 km of altitude globally for all inspected RO variables and missions. For temperature, it is < 0.05 K per decade in the global mean and < 0.1 K per decade at all latitudes. Above 25 km, the uncertainty increases for CHAMP, while data from the other missions – based on advanced receivers – are usable to higher altitudes for climate trend studies: dry temperature to 35 km, refractivity to 40 km, and bending angle to 50 km. Larger differences in RO data at high altitudes and latitudes are mainly due to different implementation choices in the retrievals. The intercomparison helped to further enhance the maturity of the RO record and confirms the climate quality of multi-satellite RO observations towards establishing a GCOS climate data record.

A new TROPOMI product for tropospheric NO 2 columns over East Asia with explicit aerosol corrections

Abstract: . We present a new product with explicit aerosol corrections, POMINO-TROPOMI, for tropospheric nitrogen dioxide (NO2) vertical column densities (VCDs) over East Asia, based on the newly launched TROPOspheric Monitoring Instrument with an unprecedented high horizontal resolution. Compared to the official TM5-MP-DOMINO (OFFLINE) product, POMINO-TROPOMI shows stronger concentration gradients near emission source locations and better agrees with MAX-DOAS measurements (R2 = 0.75, NMB = 0.8 % versus R2 = 0.68, NMB = 41.9 %). Sensitivity tests suggest that implicit aerosol corrections, as in TM5-MP-DOMINO, lead to underestimations of NO2 columns by about 25 % over the polluted Northern East China region. Reducing the horizontal resolution of a priori NO2 profiles would underestimate the retrieved NO2 columns over isolated city clusters in western China by 35 % but with overestimates by more than 50 % over many offshore coastal areas. The effect of a priori NO2 profiles is more important under calm conditions.

CLIMCAPS observing capability for temperature, moisture, and trace gases from AIRS/AMSU and CrIS/ATMS

Abstract: . The Community Long-term Infrared Microwave Combined Atmospheric Product System (CLIMCAPS) retrieves vertical profiles of temperature, water vapor, greenhouse and pollutant gases, and cloud properties from measurements made by infrared and microwave instruments on polar-orbiting satellites. These are AIRS/AMSU on Aqua and CrIS/ATMS on Suomi NPP and NOAA20; together they span nearly 2 decades of daily observations (2002 to present) that can help characterize diurnal and seasonal atmospheric processes from different time periods or regions across the globe. While the measurements are consistent, their information content varies due to uncertainty stemming from (i) the observing system (e.g., instrument type and noise, choice of inversion method, algorithmic implementation, and assumptions) and (ii) localized conditions (e.g., presence of clouds, rate of temperature change with pressure, amount of water vapor, and surface type). CLIMCAPS quantifies, propagates, and reports all known sources of uncertainty as thoroughly as possible so that its retrieval products have value in climate science and applications. In this paper we characterize the CLIMCAPS version 2.0 system and diagnose its observing capability (ability to retrieve information accurately and consistently over time and space) for seven atmospheric variables – temperature, H2O , CO, O3 , CO2 , HNO3 , and CH4 – from two satellite platforms, Aqua and NOAA20. We illustrate how CLIMCAPS observing capability varies spatially, from scene to scene, and latitudinally across the globe. We conclude with a discussion of how CLIMCAPS uncertainty metrics can be used in diagnosing its retrievals to promote understanding of the observing system and the atmosphere it measures.

Overview: Estimating and reporting uncertainties in remotely sensed atmospheric composition and temperature

Abstract: . Remote sensing of atmospheric state variables typically relies on the inverse solution of the radiative transfer equation. An adequately characterized retrieval provides information on the uncertainties of the estimated state variables as well as on how any constraint or a priori assumption affects the estimate. Reported characterization data should be intercomparable between different instruments, empirically validatable, grid-independent, usable without detailed knowledge of the instrument or retrieval technique, traceable and still have reasonable data volume. The latter may force one to work with representative rather than individual characterization data. Many errors derive from approximations and simplifications used in real-world retrieval schemes, which are reviewed in this paper, along with related error estimation schemes. The main sources of uncertainty are measurement noise, calibration errors, simplifications and idealizations in the radiative transfer model and retrieval scheme, auxiliary data errors, and uncertainties in atmospheric or instrumental parameters. Some of these errors affect the result in a random way, while others chiefly cause a bias or are of mixed character. Beyond this, it is of utmost importance to know the influence of any constraint and prior information on the solution. While different instruments or retrieval schemes may require different error estimation schemes, we provide a list of recommendations which should help to unify retrieval error reporting.

Validation of TROPOMI tropospheric NO 2 columns using dual-scan multi-axis differential optical absorption spectroscopy (MAX-DOAS) measurements in Uccle, Brussels

Abstract: . Ground-based multi-axis differential optical absorption spectroscopy (MAX-DOAS) measurements of aerosols and tropospheric nitrogen dioxide ( NO2 ) were carried out in Uccle (50.8 ∘ N, 4.35 ∘ E), Brussels, during 1 year from March 2018 until March 2019. The instrument was operated in both the UV and visible wavelength ranges in a dual-scan configuration consisting of two sub-modes: (1) an elevation scan in a fixed viewing azimuthal direction (the so-called main azimuthal direction) pointing to the northeast and (2) an azimuthal scan in a fixed low elevation angle (2 ∘ ). By applying a vertical profile inversion algorithm in the main azimuthal direction and a parameterization technique in the other azimuthal directions, near-surface NO2 volume mixing ratios (VMRs) and vertical column densities (VCDs) were retrieved in 10 different azimuthal directions. The dual-scan MAX-DOAS dataset allows for partly resolving the horizontal distribution of NO2 around the measurement site and studying its seasonal variations. Furthermore, we show that measuring the tropospheric NO2 VCDs in different azimuthal directions improves the spatial colocation with measurements from the Sentinel-5 Precursor (S5P), leading to a reduction of the spread in validation results. By using NO2 vertical profile information derived from the MAX-DOAS measurements, we also resolve a systematic underestimation in S5P NO2 data due to the use of inadequate a priori NO2 profile shape data in the satellite retrieval.

Effects of the prewhitening method, the time granularity, and the time segmentation on the Mann–Kendall trend detection and the associated Sen's slope

Abstract: . The most widely used non-parametric method for trend analysis is the Mann-Kendall test associated with the Sen's slope. The Mann-Kendall test requires serially uncorrelated time series, whereas most of the atmospheric processes exhibit positive autocorrelation. Several prewhitening methods have been designed to overcome the presence of lag-1 autocorrelation. These include a prewhitening, a detrending and/or a correction for the detrended slope and the original variance of the time series. The choice of which prewhitening method and temporal segmentation to apply has consequences for the statistical significance, the value of the slope and of the confidence limits. Here, the effects of various prewhitening methods are analyzed for seven time series comprising in-situ aerosol measurements (scattering coefficient, absorption coefficient, number concentration and aerosol optical depth), Raman Lidar water vapor mixing ratio and the tropopause and zero degree levels measured by radio-sounding. These time series are characterized by a broad variety of distributions, ranges and lag-1 autocorrelation values and vary in length between 10 and 60 years. A common way to work around the autocorrelation problem is to decrease it by averaging the data over longer time intervals than in the original time series. Thus, the second focus of this study is evaluation of the effect of time granularity on long-term trend analysis. Finally, a new algorithm involving three prewhitening methods is proposed in order to maximize the power of the test, to minimize the amount of erroneous detected trends in the absence of a real trend and to ensure the best slope estimate for the considered length of the time series.

A machine-learning-based cloud detection and thermodynamic-phase classification algorithm using passive spectral observations

Abstract: . We trained two Random Forest (RF) machine learning models for cloud mask and cloud thermodynamic-phase detection using spectral observations from Visible Infrared Imaging Radiometer Suite (VIIRS) on board Suomi National Polar-orbiting Partnership (SNPP). Observations from Cloud-Aerosol Lidar with Orthogonal Polarization (CALIOP) were carefully selected to provide reference labels. The two RF models were trained for all-day and daytime-only conditions using a 4-year collocated VIIRS and CALIOP dataset from 2013 to 2016. Due to the orbit difference, the collocated CALIOP and SNPP VIIRS training samples cover a broad-viewing zenith angle range, which is a great benefit to overall model performance. The all-day model uses three VIIRS infrared (IR) bands (8.6, 11, and 12 µ m), and the daytime model uses five Near-IR (NIR) and Shortwave-IR (SWIR) bands (0.86, 1.24, 1.38, 1.64, and 2.25 µ m) together with the three IR bands to detect clear, liquid water, and ice cloud pixels. Up to seven surface types, i.e., ocean water, forest, cropland, grassland, snow and ice, barren desert, and shrubland, were considered separately to enhance performance for both models. Detection of cloudy pixels and thermodynamic phase with the two RF models was compared against collocated CALIOP products from 2017. It is shown that, when using a conservative screening process that excludes the most challenging cloudy pixels for passive remote sensing, the two RF models have high accuracy rates in comparison to the CALIOP reference for both cloud detection and thermodynamic phase. Other existing SNPP VIIRS and Aqua MODIS cloud mask and phase products are also evaluated, with results showing that the two RF models and the MODIS MYD06 optical property phase product are the top three algorithms with respect to lidar observations during the daytime. During the nighttime, the RF all-day model works best for both cloud detection and phase, particularly for pixels over snow and ice surfaces. The present RF models can be extended to other similar passive instruments if training samples can be collected from CALIOP or other lidars. However, the quality of reference labels and potential sampling issues that may impact model performance would need further attention.

An improved TROPOMI tropospheric HCHO retrieval over China

Abstract: . We present the improved retrieval of TROPOspheric Monitoring Instrument (TROPOMI) tropospheric formaldehyde (HCHO) over China. The new retrieval optimizes the slant column density (SCD) retrieval and air mass factor (AMF) calculation for TROPOMI observations of HCHO over China. HCHO SCDs are retrieved using the basic optical differential spectroscopy (BOAS) technique, while AMFs are calculated with a priori HCHO profile from a higher resolution regional chemistry transport model. Compared to the operational product, the new TROPOMI HCHO retrieval shows better agreement with the ground based Multi-AXis Differential Optical Absorption Spectroscopy (MAX-DOAS) measurements in China. The operational product in general overestimates HCHO VCDs by 14.01 %, while the improved HCHO only shows an underestimation of 3.67 %. The improvements are mainly related to the AMF calculation with higher resolution a priori profile (61.11 %), while the SCD retrieval only shows a minor effect of 0.15 %. The improved HCHO is also used to investigate the spatial-temporal characteristic of HCHO over China. The result shows that HCHO VCDs reach maximum in summer and minimum in winter. High HCHO VCDs mainly located over populated areas, i.e., Sichuan Basin, Central and Eastern China, indicating a significant contribution of anthropogenic emissions. The result indicates the improved TROPOMI HCHO is more suitable for the analysis of regional and city scale pollution in China.