Skip to main content
PMC home page
NASA Author Manuscripts logoLink to NASA Author Manuscripts
. Author manuscript; available in PMC: 2021 Apr 16.
Published in final edited form as: Geophys Res Lett. 2018 Dec 28;46(2):1049–1060. doi: 10.1029/2018GL081095

High resolution mapping of nitrogen dioxide with TROPOMI: First results and validation over the Canadian oil sands

Debora Griffin 1, Chris A McLinden 1, Folkert Boersma 2,3, Adam Bourassa 4, Enrico Dammers 1, Doug Degenstein 4, Henk Eskes 2, Lukas Fehr 4, Vitali Fioletov 1, Katherine Hayden 1, Shailesh K Kharol 1, Shao-Meng Li 1, Paul Makar 1, Randall V Martin 5, Cristian Mihele 1, Richard L Mittermeier 1, Nickolay Krotkov 6, Maarten Sneep 2, Lok N Lamsal 6,7, Mark ter Linden 2,8, Jos van Geffen 2, Pepijn Veefkind 2,9, Mengistu Wolde 10, Xiaoyi Zhao 1
PMCID: PMC8051066  NIHMSID: NIHMS1535818  PMID: 33867596

Abstract

TROPOMI, on-board the Sentinel-5 Precursor satellite is a nadir-viewing spectrometer measuring reflected sunlight in the ultraviolet, visible, near-infrared, and shortwave infrared spectral range. From these spectra several important air quality and climate-related atmospheric constituents are retrieved at an unprecedented high spatial resolution, including nitrogen dioxide (NO2). We present the first retrievals of TROPOMI NO2 over the Canadian Oil Sands, contrasting them with observations from the OMI satellite instrument, and demonstrate its ability to resolve individual plumes and highlight its potential for deriving emissions from individual mining facilities. Further, the first TROPOMI NO2 validation is presented, consisting of aircraft and surface in-situ NO2 observations, as well as ground-based remote-sensing measurements between March and May 2018. Our comparisons show that the TROPOMI NO2 vertical column densities are highly correlated with the aircraft and surface in-situ NO2 observations, and the ground-based remote-sensing measurements with a low bias (15–30 %) over the Canadian Oil Sands.

Plain Language Summary

Nitrogen dioxide (NO2) is a pollutant that is linked to respiratory health issues and has negative environmental impacts such as soil and water acidification. Near the surface the most significant sources of NO2 are fossil fuel combustion and biomass burning. With a recently launched satellite instrument (TROPOspheric Monitoring Instrument; TROPOMI) NO2 can be measured with an unprecedented combination of accuracy, spatial coverage, and resolution. This work presents the first TROPOMI NO2 measurements near the Canadian Oil Sands and shows that these measurements have an outstanding ability to detect NO2 on a very high horizontal resolution that is unprecedented for satellite NO2 observations. Further, these satellite measurements are in excellent agreement with aircraft and ground-based measurements.

1. Introduction

The ability to observe the near-surface abundance of air pollutants from space has drastically increased over the last two decades. The first space-borne ultraviolet-visible spectrometer focusing on tropospheric trace composition was the Global Ozone Monitoring Experiment (GOME), 1995–2011, with a pixel size of 40 × 320km2 [Burrows et al., 1999], followed by the SCanning Imaging Absorption spectroMeter for Atmospheric CartograpHY (SCIAMACHY), 2002–2012 (30 × 60km2) [Bovensmann et al., 1999] and the GOME-2 instruments (2006–present, 2012–present; 40×80km2) [Callies et al., 2000]. Until now, the previous standard was the Ozone Monitoring Instrument (OMI; 2004–present; 13 × 24 km2) [Levelt et al., 2006; Krotkov et al., 2016], that was able to resolve pollutant distributions at the urban scale (20–30 km), from which emissions could be inferred for large point sources [e.g. Lu and Streets, 2012; Ghude et al., 2013; de Foy et al., 2015]. However, with the launch of the European Space Agency’s (ESA) Sentinel-5 Precurser (S-5P) on October 13, 2017, space-borne measurements have entered a new era. In addition to spatial resolution, data product quality of the recorded spectra has significantly increased for this type of space-borne UV-vis spectrometer over the past two decades.

Here, we focus on the first retrieval results of tropospheric NO2 from TROPOMI. NO2 is an important pollutant gas that is linked to respiratory health issues [Health Canada, 2018] and has negative environmental impacts such as soil and water acidification [Environment Canada, 2018]. NO2 is emitted mainly as nitrogen monoxide (NO) by combustion processes, such as fossil fuel combustion and biomass burning, as well as lightning, where molecular nitrogen (N2) combines with molecular oxygen (O2) at extremely high temperatures. Due to the rapid cycling between them, NO and NO2 are often considered together as NOx (NO+NO2). Within a plume, the lifetime of NOx is relatively short, on the order of a few hours, and is in daytime primarily removed by reaction with the hydroxyl radical (OH). Due to this short lifetime, background NO2 concentrations can be orders of magnitude smaller than levels over polluted areas.

This study presents the first TROPOMI tropospheric NO2 retrievals (offline v.101) [van Geffen et al., 2018; Boersma et al., 2018] and demonstrates its ability to resolve individual plumes from space at an unprecedented spatial resolution. To demonstrate the quality of the TROPOMI NO2 product and its suitability for high-resolution mapping, aircraft and ground-based data over the Athabasca Oil Sands Region (AOSR) in north-eastern Alberta, Canada are used for its evaluation. The AOSR, and in particular the surface mining area at 57°N, is a remote area that has significant NOx emissions from from a variety of types of industrial sources, such as upgraders (a facility that converts bitumen, a viscous form of oil found in the oil sands, into synthetic crude oil) and heavy transportation vehicles [McLinden et al., 2012; Percy et al., 2012]. As such, this region is an ideal location for evaluating the TROPOMI NO2 retrieval product and its ability to capture localized enhancements of NO2, including individual plumes, and low background values over a relatively small domain. We also contrast these results with a similar analysis of lower resolution OMI observations to help demonstrate the importance of the finer spatial resolution in capturing horizontal gradients and small scale features.

The measurements shown here were taken between March and May, 2018 and for much of this period the surface was snow-covered (note that data for March is pre-operational and may be slightly noisier). In principle, this highly reflective surface enables a more accurate retrieval with larger signals and less dependence on the knowledge of profile shape [e.g., O’Byrne et al., 2010; McLinden et al., 2014]. Current retrieval algorithms can struggle due to the requirement for identifying the presence of snow and also correctly accounting for its altered reflectance properties. The surface mines of the AOSR are also surrounded by boreal forest that is associated with a very low reflectance, providing the additional challenge of a changing albedo over a small area.

2. Datasets

2.1. Satellite NO2 data

TROPOMI is the single payload on the S-5P satellite that has a Sun-synchronous orbit with a local overpass time of around 1:30pm and has near full-surface coverage on a daily basis [Veefkind et al., 2012; Hu et al., 2018]. The instrument contains four spectrometers, three that cover the Ultra-Violet-Near Infra-Red (UV-NIR) with two spectral bands at 270–500nm and 675–775nm, and one for the SWIR. The TROPOMI NO2 retrieval algorithm was developed by the Royal Netherlands Meteorological Institute (KNMI) and utilizes the bands of the UV-NIR spectrometer. The retrieval algorithm is based on the NO2 DOMINO retrieval previously used for OMI spectra [Boersma et al., 2011] with improvements made for all retrieval sub-steps [van Geffen et al., 2015; Lorente et al., 2017; Zara et al., 2018] within the QA4ECV project [Boersma et al., 2018; van Geffen et al., 2018]. A tropospheric vertical column density (VCD, or simply ‘column’) is provided by the algorithm, which represents the vertically integrated number of NO2 molecules per unit area between the surface and the tropopause (in units of mol/m2, with typical columns ranging from 10–200 μmol/m2). To determine the crucial air mass factor (AMF), a measure of the measurement’s sensitivity [e.g. Palmer et al., 2001; McLinden et al., 2014], the profile shape of the TM5-MP model is used (at 1 × 1° resolution) Williams et al. [2017], the surface albedo information is derived from a monthly OMI climatology (on a 0.5 × 0.5° resolution) [Kleipool et al., 2008]. TROPOMI uses a snow flag from the Near-real-time Ice and Snow Extent (NISE), and the albedo is set to 0.6 (over the AOSR) if the surface beneath is covered in snow or ice. For this study we use v1.01 [van Geffen et al., 2018], the first released offline version of the TROPOMI tropospheric NO2 columns (https://s5phub.copernicus.eu; http://www.tropomi.eu). Spatial resolution varies with across-track position and in this study the average pixel size is 6 × 7km. Pixels that are fully or partially covered by clouds were filtered, here we used 0.3 as a cut-off for the radiative cloud fraction (“cloud_fraction_crb_nitrogendioxide_window”).

OMI NO2 observations (NASA Standard Product, SP, version 3.1; Krotkov et al. [2017]; Marchenko et al. [2015]) are used in order to evaluate the relative performance of TROPOMI and OMI. The OMI data are filtered in a similar manner as TROPOMI with an additional restriction that only smaller pixels are considered (track positions 11–50), which are automatically removed in the SPv3.1 product.

2.2. In-situ measurements

Aircraft-borne in situ measurements were taken in April 2018 over the AOSR as part of Environment and Climate Change Canada’s (ECCC’s) aircraft campaign. Here, we use measurements taken during three flights, on April 5, 9, and 13, 2018 between 16:30UTC and 21:00UTC. The TROPOMI overpasses in this region are typically between 19:00 UTC and 21:00 UTC; on some days two overpasses occur that are approximately 100 min apart. During those flights, air parcels were sampled in the planetary boundary layer (PBL) between approximately 200 to 400 m agl (above ground level) with occasional spiral flights reaching up to 2300 m agl.

The aircraft in-situ NO2 measurements were taken on-board a National Research Council of Canada’s Convair 580 research aircraft using two Thermo Scientific Model 42i-TL (NO-NO2-NOx) Analyzers, modified to measure at 1Hz time resolution. The first analyzer measured NO directly, while the second analyzer measured the sum of NO and large fraction of NO2 (65 %), selectively converted to NO using a photolytic converter (Air Quality Design Inc.). NO2 was then calculated by difference and averaged to 3s, with an estimated detection limit of 0.2 ppbv.

Ground-based NO2 in-situ concentrations are continuously sampled at 17 locations in north-eastern Alberta as part of the Wood Buffalo Environment Association (WBEA; http://wbea.org/historical-monitoring-data/) [Percy et al., 2012]. NO2 concentrations (measured in a similar way as described above for the aircraft measurements, see supplementary material) are available every 5 min.

2.3. Pandora instrument

A type of ground-based spectrometer, known as Pandora, has been developed to help evaluate satellite NO2 retrievals [Herman et al., 2009]. The Pandora instrument measures direct sunlight in the UV-vis spectral range (280–525 nm). High-quality NO2 total columns (tropospheric + stratospheric) are obtained by using the differential optical absorption spectroscopy (DOAS) technique [Noxon, 1975; Platt and Stutz, 2008] in 400 to 440 nm range. The retrieved NO2 total columns data has a precision of ±2.7×1014 molec/cm2 [Herman et al., 2009]. A Pandora instrument has been deployed in Fort McKay in the AOSR (57.184°N, 111.64°W) since 2013 [Fioletov et al., 2016]. NO2 columns are available approximately every 1.5 min under unobscured Sun conditions.

2.4. Alternative AMF

We also re-calculate the AMFs and examine their impact on the TROPOMI tropospheric NO2 columns over the AOSR. The procedure is similar to that from McLinden et al. [2014] with the following differences: (1) GEM-MACH version 2, at 10 km resolution, updated emissions and using output at the time of the satellite overpass was used to provide the NO2 profile shape; (2) Interactive Multisensor Snow and Ice Mapping System (IMS) [Helfrich et al., 2007] at 4 km resolution was used to flag pixels with snow-cover; and the use of the SASKTRAN radiative transfer model [Bourassa et al., 2008; Zawada et al., 2015; Dueck et al., 2017] for the generation of the AMF look-up tables. SASKTRAN is employed here as it is more readily able to implement future planned improvements such as non-Lambertian surfaces. The cloud input needed for the estimation is taken from the TROPOMI product (“cloud_fraction_crb_nitrogendioxide_window”, “cloud_fraction_crb_nitrogendioxide_window”, “cloud_pressure_crb”).

An uncertainty is estimated for the alternative tropospheric columns (similar to McLinden et al. [2014]) by adding the uncertainty of the slant column density (provided in the S5P NO2 data product), the uncertainty of the stratospheric column density (provided in the S5P NO2 data product), and the uncertainty of the alternative AMF (assumed to be 20 %) in quadrature. The uncertainty of the AMF contains uncertainties from cloud fraction, cloud pressure, albedo, surface pressure and profile shape [McLinden et al., 2014].

3. TROPOMI over the Canadian oil sands

Before the launch of TROPOMI, OMI has been used extensively to study the distribution and evolution of NO2 over the Canadian oil sands McLinden et al. [2012, 2016]. However, due to the small spatial extent of the surface mining, which is commensurate with OMI’s pixel size, it was not possible to observe individual plumes. It was only possible to reliably map the average distribution by considering several years together, and emissions estimates were for the mining region as a whole [Fioletov et al., 2016].

Figure 1 a and b shows TROPOMI and OMI observations for a single day, April 9, 2018, over the AOSR surface mines. TROPOMI shows well defined plumes of significantly elevated NO2 originating from two distinct mining areas from which it should be possible to estimate NOx emissions. From the spatial location of the plumes in Fig. 1 a, it appears that the well-defined NO2 plume in the south originates from the bitumen upgrading stacks which are elevated sources, while the more diffuse enhancement in the north is consistent with the heavy hauler trucks, a surface source. OMI, by contrast shows much more modest enhancements for about a half-dozen pixels and captures none of the fine details making its spatial distribution impossible to interpret. Averaging TROPOMI over the three month period of interest using the oversampling technique of Fioletov et al. [2011] provides a distribution very similar to that of an OMI multi-year average, but with much better resolved spatial patterns and higher peak values (Fig. 1 c and d). The distribution shows a separation of the NO2 between the northern and southern mines.

Figure 1.

Figure 1.

Open in a new tab

NO2 measured by TROPOMI (a) and OMI (b) on April 9, 2018 over the surface mines of the AOSR (with NW winds). Averages (March-May, 2018) with an averaging radius of 5 km and 16 km was used for TROPOMI (c) and OMI (d), respectively. The black line traces the borders of the individual mining operations.

4. Evaluation of the NO2 product

4.1. Spiral flights

As noted in Sect. 2.2, spirals were flown (within ±2h of the TROPOMI overpass time) as high as 2300 m agl that penetrated the plume top. These types of aircraft measurements can be used to estimate the NO2 VCDs in the PBL, as well as determine AMFs based on the measured NO2 profiles. Using this true profile to calculate AMFs helps to identify errors in the retrieval, and further allows a more quantitative evaluation of AMFs derived using model inputs. The aircraft profile spirals began as low as approximately 200 m agl, measuring volume mixing ratios (VMRs) throughout the PBL. To compare these to TROPOMI, we estimated a tropospheric NO2 VCD using the aircraft measurements from those spirals. For this estimation, the aircraft measurements are binned into 100 m levels; between the surface and lowest aircraft altitude, we assume a well mixed layer and use the concentration of the lowest measurements. Above the maximum flight altitude we use NO2 VMRs from the GEOS-Chem model accounting for ∼ 1.8×1014 molec/cm2. Temperatures and pressures needed for the calculation were taken from the operational GEM-MACH model. During the three aircraft flights, seven spirals were flown. Figure 2a shows the in-situ measurements of some spiral flights along with the GEM-MACH NO2 VMRs (for the nearest grid-box at the closest hour). GEM-MACH reproduces the actual NO2 measurements very well in the PBL and Fig. 2 shows that the profile is captured well by the model. However, it does not always capture the height of the boundary layer and, as such, there is disagreement above approximately 1.5km, where NO2 VMRs drop to near zero in the model.

Figure 2.

Figure 2.

Open in a new tab

(a) In-situ aircraft measurements and GEM-MACH model output of the three of the seven spiral flights that are within retrieved TROPOMI pixels. The right figure shows the mean difference (and standard deviation) between the GEM-MACH model and the aircraft measurements from all seven spirals. Aircraft columns versus (b) the original and (c) the alternative AMF TROPOMI NO2 columns. The solid red and the black dashed lines represent the line of best fit and the 1:1 line, respectively. (d) AMFs, including the original TROPOMI (AMF TROPOMI; green), the re-calculated (AMF EC; blue), and the calculated “true” AMF from aircraft measurements using the TROPOMI aldebo (True 1; yellow) the MODIS albedo (True 2; red).

The comparison with the aircraft spirals indicates a low bias for TROPOMI (approximately −18 % compared to the aircraft measurements). This bias nearly disappears (<5 %) if the alternative AMFs are applied (panel c). Figure 2d shows the “true” AMF estimated using the albedo as used in the TROPOMI input file (orange), and using the MODIS albedo (red). The comparison (Fig. 2d) shows that the AMFs (using the MODIS albedo) from the aircraft measurements are very close to the estimated AMFs (blue) using the GEM-MACH model, and that TROPOMI tropospheric AMFs (green) are likely overestimated in this region (above snow). It also shows that a wrong profile shape over the AOSR and an albedo that is too high can cancel each other’s effects, leading to a TROPOMI AMF value close to the “true” AMF (red). This highlights the benefits of using of a high-resolution regional model which reproduces AMFs close to the true/measured AMFs.

4.2. Spatial correlation with the aircraft measurements

While tropospheric NO2 VCDs can be estimated from the spirals, only seven are available for comparison with TROPOMI NO2 VCDs. In this section, we evaluate the ability of TROPOMI to detect localized enhancements of tropospheric NO2 on the scale of its pixel size, using all pixels and aircraft measurements to evaluate the ability of TROPOMI to capture spatial gradients. We average the all aircraft in-situ measurements of NO2 for the three flights that are sampled below 1 km agl (to ensure the data collected is in the PBL), taken within ±2h within the overpass time, and that are within the area of the individual TROPOMI pixel. For all three flights the surface below was covered completely in snow and ice (Fig. S2). On average, over 150 aircraft in-situ NO2 measurements with an integration time of 3s were taken within one TROPOMI pixel, and some pixels contain as many as 700. The aircraft in-situ and the TROPOMI measurements (Fig. 3) correlate well, with a correlation coefficient R ≈ 0.7, showing that TROPOMI captured the gradients and localized enhancements of NO2 on it’s small pixel size of approximately 6 × 7km. To be able to quantitatively compare the satellite VCDs to the in-situ VMRs, we convert the average in-situ measurements (at the average aircraft altitude, h, below 1000 m agl) into tropospheric VCDs using the daily output from the GEM-MACH model:

VCDaircraft=Caircraft(h)CGEM−MACH(h)×VCDGEM−MACH,PBL+VCDGEOS−Chem,freetrop.

Figure 3.

Figure 3.

Open in a new tab

(a) Map around the AOSR showing the NO2 VMRs as sampled by the aircraft (for clarity of the plot the averages of 30 data points are shown) and TROPOMI tropospheric NO2 VCD pixels for April 5, 2018. TROPOMI tropospheric NO2 columns versus aircraft in-situ measurements (b), the aircraft “ VCDs” (Eq. 1) (c), and the aircraft “VCDs” versus the alternative TROPOMI columns (d). For the regression analysis, we use a geometric mean analysis with y = sx + i (s and i values are indicated in the plots, as well as the correlation coefficient, R and the number of points, N).

The results (Fig.3 c) indicate a low bias of approximately 15 % for the TROPOMI tropospheric NO2 VCDs compared to the aircraft measurements, with a mean difference of −5.3e14 molec/cm2 (−21 %; TROPOMI-aircraft) and a random error of approximately 0.97e15 molec/cm2. After applying the correction to the TROPOMI VCDs (as described in Sect.2.4) the bias is reduced to 4 %, leading to an accuracy of −3.3e14 molec/cm2 (−13 %; TROPOMI-aircraft), and a random error of 1.27e15 molec/cm2. The improved slope in Fig. 3 d is expected given that the same profile shape is now used for both the scaling of aircraft concentrations and in the calculations of AMFs.The correlation between the aircraft measurements and the TROPOMI retrievals does not change using the alternative AMFs.

4.3. Comparison with ground-based measurements

To evaluate TROPOMI’s ability to capture NO2 distributions over a longer time and spatial scale than that covered by the aircraft flights, comparisons are made with measurements from a ground-based in-situ network (WBEA). This allows for a longer time period and larger area as for the aircraft data, discussed in the previous section. The results are displayed in Fig. 4a for measurements between March and May 2018, comparing the TROPOMI measurements from individual pixels to the averaged WBEA NO2 measurements (±30 min of the TROPOMI overpass). Only the TROPOMI pixels that cover the stations are used (due to the small size of the TROPOMI pixel and the distance between the WBEA stations, one pixel only covers one station). Overall, there is a good correlation between the TROPOMI tropospheric VCDs and the surface measurements (R= 0.65). To examine the theoretical correlation with surface measurements, we utilize output from an offline run of GEM-MACH at 2.5 km resolution. The correlation between model VCD and model surface VMR without any noise is roughly 0.6 (Fig. S5) and only decreases slightly with increased smoothing. Considering the observed correlation, with measurement noise, is comparable to this or slightly higher than that from these noise-free simulations, it appears that the reduction in correlation from unity is mainly due to the mismatch in the qualities being compared (column vs. surface) and little from the noise in either type of observation.

Figure 4.

Figure 4.

Open in a new tab

TROPOMI tropospheric NO2 columns versus the ground-based in-situ (a) and remote-sensing Pandora measurements (b-c) for: (b) the original TROPOMI columns, and (c) the alternative columns.

For the comparison with the TROPOMI NO2 tropospheric columns, we average the Pandora data (±30 min of the TROPOMI overpass) and subtract the corresponding TROPOMI NO2 stratospheric columns to obtain a tropospheric Pandora columns. A good agreement was found between the TROPOMI and the Pandora columns (Fig. 4b). The theoretical correlation is expected around 0.95 with a slope of 0.85–0.9 at TROPOMI resolution (Fig. S5). The regression analysis shows a low bias for the TROPOMI tropospheric NO2 VCDs with a slope of approximately 0.7. The slope increases to 0.75 when the alternative AMF is applied, which is only 0.1 lower than the expected slope (Fig. 4c); the correlation, however, slightly decreases.

While most of the ground in March and April was covered by snow, there are snow-free days at the end of April and in May over the AOSR. However, even for the snow-free days a small low bias of the original TROPOMI NO2 VCDs remains over the AOSR. Also for snow-free days, we found that the MODIS albedo is approximately half of the climatological OMI albedo used as input for the original AMFs (Fig. S4). This low albedo appears to offset the effect of the TM5 model at 1° resolution which has the effect of diluting emissions and leading to profiles that are not sufficiently weighted to near-surface NO2. The alternative and original TROPOMI NO2 VCDs agree within the estimated uncertainties, however, the uncertainties associated with the original NO2 tropospheric columns, which can be close to 100 % over snow, can be much reduced by higher-resolution input (Fig. S6).

5. Discussion and Conclusions

This study presents the first results of TROPOMI NO2 VCDs over the AOSR. We found that TROPOMI has an outstanding ability to detect tropospheric NO2 on a very high horizontal resolution. The noise of the individual TROPOMI pixels is quite low, and without averaging there is a high correlation (∼ 0.8 − 0.7) between the TROPOMI observations and the aircraft in-situ measurements in the boundary layer that sampled air across the TROPOMI pixels. As promised, the TROPOMI tropospheric NO2 product can retrieve good quality information on a 6 × 7 km (and even smaller) pixel size that is unprecedented for satellite-remote sensing observations.

Further, TROPOMI displays a high correlation with ground-based in-situ and remote sensing measurements that is unprecedented for satellite-remote sensing measurements of NO2. Its predecessor, OMI (NO2 SPv3.1), only shows low to moderate correlation with ground-based in-situ (R = 0.35) and remote-sensing (R = 0.06) measurements if the same method is applied (supplementary material). For the ground-based in-situ and remote-sensing measurements, there is a good correlation (R ≈ 0.68) with TROPOMI NO2. Again, this is a significant improvement to previous satellite and ground-based in-situ comparisons where measurements are typically averaged over years to achieve a good correlation [McLinden et al., 2014] or the correlation is moderate (R ≈ 0.4) if no corrections are applied [Kramer et al., 2008].

Quantitatively we found a low bias of −15 to −30 % of the TROPOMI tropospheric NO2 columns over the AOSR that was consistent for all of the comparisons carried out in this study. Previously similar discrepancies have been identified for OMI, which were associated with issues of input data used to estimate the tropospheric AMF [McLinden et al., 2014]. Here, we re-calculated the AMFs using higher resolution input data. This includes, a high-resolution model, GEM-MACH (10 × 10 km resolution), to estimate the NO2 profile shape for the lower part of the troposphere, a better identification of snow-covered surfaces (IMS, on 4 × 4 km resolution) and the surface reflectivity from the MODIS satellite. This improved input data provided more precise information on surface cover and more appropriate albedo, if snow is present. Using the alternative tropospheric NO2 VCDs, the low bias is reduced to between 0 and −25 %. Thus, the low bias is not due to the retrieval itself, but is associated with the input data used for the tropospheric AMF calculation. A low bias can be seen for both, snow-covered and snow-free surfaces, indicating that the issue is due to both, an incorrect albedo over snow and an inaccurate NO2 profile shape over the AOSR. While using high-resolution input improves the tropospheric AMF and as such the tropospheric NO2 VCDs, this correction is not as significant as previously seen for OMI [McLinden et al., 2014]. For differences where the recalculated NO2 columns increased by a factor of two compared to the original, we only see a 5 − 15 % improvement of the bias when the alternative AMFs were used. We found that the TROPOMI tropospheric NO2 v1.01 is well within the requirements of 25–50 % bias. Based on our analysis, we recommend to improve the albedo (particularly over snow), as well as the snow flag information for the global product. Secondly, regional products could be developed, using high-resolution model information to improve the NO2 a priori. Furthermore, we conclude that the TROPOMI stratospheric NO2 column estimates work well.

Overall, we found that the data quality, noise level and resolution of these satellite-borne NO2 VCDs is unprecedented. These high-resolution TROPOMI NO2 observations can allow to pin-point even kilometer-scale emission sources and estimate localized emissions from industry, cities or fires on a monthly and likely daily basis.

Supplementary Material

SM

Key Points:

  • First evaluation of the TROPOMI NO2 retrieval product

  • The quality of the TROPOMI NO2 data is unprecedented and captures variation on a very high spatial resolution

  • TROPOMI tropospheric NO2 retrievals can be corrected with higher resolution input data

Acknowledgments

The authors would like to thank the Wood Buffalo Environmental Association (WBEA) for the provision of their in situ data. The Sentinel 5 Precursor TROPOMI Level 2 product is developed with funding from the Netherlands Space Office (NSO) and processed with funding from the European Space Agency (ESA). We acknowledge the NASA Earth Science Division for funding of OMI NO2 product development and analysis, and the Air Quality Research Division support teams and the National Research Council aircraft pilots and technical support team for the aircraft measurement campaign. These measurements were carried out as part of the Oil Sands 2018 aircraft measurement campaign project, funded by the Oil Sands Monitoring (OSM) program by the Governments of Alberta and Canada.

References

  1. Boersma KF, Eskes HJ, Dirksen RJ, van der A RJ, Veefkind JP, Stammes P, Huijnen V, Kleipool QL, Sneep M, Claas J, Leitão J, Richter A, Zhou Y, and Brunner D. (2011), An improved tropospheric NO2 column retrieval algorithm for the Ozone Monitoring Instrument, Atmos. Meas. Tech, 4, 1905. [Google Scholar]
  2. Boersma KF, Eskes HJ, Richter A, De Smedt I, Lorente A, Beirle S, van Geffen JHGM, Zara M, Peters E, Van Roozendael M, Wagner T, Maasakkers JD, van der A J, Nightingale J, De Rudder A, Irie H, Pinardi G, Lambert J-C, and Compernolle (2018), Improving algorithms and uncertainty estimates for satellite NO2 retrievals: Results from the Quality Assurance for Essential Climate Variables (QA4ECV) project, Atmospheric Measurement Techniques Discussions, 2018, 1–70, doi: 10.5194/amt-2018-200. [DOI] [Google Scholar]
  3. Bourassa A, Degenstein D, and Llewellyn E. (2008), SASKTRAN: A spherical geometry radiative transfer code for efficient estimation of limb scattered sunlight, Journal of Quantitative Spectroscopy and Radiative Transfer, 109(1), 52 – 73, doi: 10.1016/j.jqsrt.2007.07.007. [DOI] [Google Scholar]
  4. Bovensmann H, Burrows JP, Buchwitz M, Frerick J, Noël S, Rozanov VV, Chance KV, and Goede APH (1999), SCIAMACHY: Mission Objectives and Measurement Modes, Journal of the Atmospheric Sciences, 56(2), 127–150, doi:. [DOI] [Google Scholar]
  5. Burrows JP, Weber M, Buchwitz M, Rozanov V, Ladstätter-Weißenmayer A, Richter A, DeBeek R, Hoogen R, Bramstedt K, Eichmann K-U, Eisinger M, and Perner D. (1999), The Global Ozone Monitoring Experiment (GOME): Mission Concept and First Scientific Results, Journal of the Atmospheric Sciences, 56(2), 151–175, doi:. [DOI] [Google Scholar]
  6. Callies J, Corpaccioli E, Eisinger M, Hahne A, and Lefebvre A. (2000), GOME-2-Metop’s second-generation sensor for operational ozone monitoring, 102, 28–36. [Google Scholar]
  7. de Foy B, Lu Z, Streets DG, Lamsal LN, and Duncan BN (2015), Estimates of power plant NOx emissions and lifetimes from OMI NO2 satellite retrievals, Atmospheric Environment, 116, 1 – 11, doi: 10.1016/j.atmosenv.2015.05.056. [DOI] [Google Scholar]
  8. Dueck SR, Bourassa AE, and Degenstein DA (2017), An efficient algorithm for polarization in the SASKTRAN radiative transfer framework, Journal of Quantitative Spectroscopy and Radiative Transfer, 199, 1 – 11, doi: 10.1016/j.jqsrt.2017.05.016. [DOI] [Google Scholar]
  9. Environment Canada (2018), Air pollution: drivers and impacts, http://www.ec.gc.ca/indicateurs-indicators/default.asp?lang=En&n=D189C09D-1;last accessed: 24 September 2018.
  10. Fioletov VE, McLinden CA, Krotkov N, Moran MD, and Yang K. (2011), Estimation of SO2 emissions using OMI retrievals, Geophysical Research Letters, 38(21), doi: 10.1029/2011GL049402. [DOI] [Google Scholar]
  11. Fioletov VE, McLinden CA, Cede A, Davies J, Mihele C, Netcheva S, Li S-M, and O’Brien J. (2016), Sulfur dioxide (SO2) vertical column density measurements by Pandora spectrometer over the Canadian oil sands, Atmospheric Measurement Techniques, 9(7), 2961–2976, doi: 10.5194/amt-9-2961-2016. [DOI] [Google Scholar]
  12. Ghude SD, Kulkarni SH, Jena C, Pfister GG, Beig G, Fadnavis S, and van der A RJ (2013), Application of satellite observations for identifying regions of dominant sources of nitrogen oxides over the Indian Subcontinent, Journal of Geophysical Research: Atmospheres, 118(2), 1075–1089, doi: 10.1029/2012JD017811. [DOI] [Google Scholar]
  13. Health Canada (2018), Human Health Risk Assessment for Ambient Nitrogen Dioxide, https://www.canada.ca/en/health-canada/services/publications/healthy-living/human-health-risk-assessment-ambient-nitrogen-dioxide.html; last accessed: 24 September 2018.
  14. Helfrich SR, McNamara D, Ramsay BH, Baldwin T, and Kasheta T. (2007), Enhancements to, and forthcoming developments in the Interactive Multisensor Snow and Ice Mapping System (IMS), Hydrological Processes, 21(12), 1576–1586, doi: 10.1002/hyp.6720. [DOI] [Google Scholar]
  15. Herman J, Cede A, Spinei E, Mount G, Tzortziou M, and Abuhassan N. (2009), NO2 column amounts from ground-based Pandora and MFDOAS spectrometers using the direct-sun DOAS technique: Intercomparisons and application to OMI validation, Journal of Geophysical Research: Atmospheres, 114(D13), doi: 10.1029/2009JD011848. [DOI] [Google Scholar]
  16. Hu H, Landgraf J, Detmers R, Borsdorff T, de Brugh JA, Aben I, Butz A, and Hasekamp O. (2018), Toward Global Mapping of Methane With TROPOMI: First Results and Intersatellite Comparison to GOSAT, Geophysical Research Letters, 45(8), 3682–3689, doi: 10.1002/2018GL077259. [DOI] [Google Scholar]
  17. Kleipool QL, Dobber MR, de Haan JF, and Levelt PF (2008), Earth surface reflectance climatology from 3 years of omi data, Journal of Geophysical Research: Atmospheres, 113(D18), doi: 10.1029/2008JD010290. [DOI] [Google Scholar]
  18. Kramer LJ, Leigh RJ, Remedios JJ, and Monks PS (2008), Comparison of OMI and ground-based in situ and MAX-DOAS measurements of tropospheric nitrogen dioxide in an urban area, Journal of Geophysical Research: Atmospheres, 113(D16), doi: 10.1029/2007JD009168. [DOI] [Google Scholar]
  19. Krotkov NA, McLinden CA, Li C, Lamsal LN, Celarier EA, Marchenko SV, Swartz WH, Bucsela EJ, Joiner J, Duncan BN, Boersma KF, Veefkind JP, Levelt PF, Fioletov VE, Dickerson RR, He H, Lu Z, and Streets DG (2016), Aura OMI observations of regional SO2 and NO2 pollution changes from 2005 to 2015, Atmospheric Chemistry and Physics, 16(7), 4605–4629, doi: 10.5194/acp-16-4605-2016. [DOI] [Google Scholar]
  20. Krotkov NA, Lamsal LN, Celarier EA, Swartz WH, Marchenko SV, Bucsela EJ, Chan KL, Wenig M, and Zara M. (2017), The version 3 omi no2 standard product, Atmospheric Measurement Techniques, 10(9), 3133–3149, doi: 10.5194/amt-10-3133-2017. [DOI] [Google Scholar]
  21. Levelt PF, van den Oord GHJ, Dobber MR, Mälkki A, Visser H, de Vries J, Stammes P, Lundell JOV, and Saari H. (2006), The Ozone Monitoring Instrument, IEEE Trans. Geosci. Remote Sens, 44, 1093. [Google Scholar]
  22. Lorente A, Folkert Boersma K, Yu H, Dörner S, Hilboll A, Richter A, Liu M, Lamsal LN, Barkley M, De Smedt I, Van Roozendael M, Wang Y, Wagner T, Beirle S, Lin J-T, Krotkov N, Stammes P, Wang P, Eskes HJ, and Krol M. (2017), Structural uncertainty in air mass factor calculation for NO2 and HCHO satellite retrievals, Atmospheric Measurement Techniques, 10(3), 759–782, doi: 10.5194/amt-10-759-2017. [DOI] [Google Scholar]
  23. Lu Z, and Streets DG (2012), Increase in NOx emissions from Indian thermal power plants during 1996âĂŞ2010: Unit-based inventories and multisatellite observations, Environmental Science & Technology, 46(14), 7463–7470, doi: 10.1021/es300831w. [DOI] [PubMed] [Google Scholar]
  24. Marchenko S, Krotkov NA, Lamsal LN, Celarier EA, Swartz WH, and Bucsela EJ (2015), Revising the slant column density retrieval of nitrogen dioxide observed by the ozone monitoring instrument, Journal of Geophysical Research: Atmospheres, 120(11), 5670–5692, doi: 10.1002/2014JD022913. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. McLinden CA, Fioletov V, Boersma KF, Krotkov N, Sioris CE, Veefkind JP, and Yang K. (2012), Air quality over the Canadian oil sands: A first assessment using satellite observations, Geophysical Research Letters, 39(4), doi: 10.1029/2011GL050273. [DOI] [Google Scholar]
  26. McLinden CA, Fioletov V, Boersma KF, Kharol SK, Krotkov N, Lamsal L, Makar PA, Martin RV, Veefkind JP, and Yang K. (2014), Improved satellite retrievals of NO2 and SO2 over the Canadian oil sands and comparisons with surface measurements, Atmospheric Chemistry and Physics, 14(7), 3637–3656, doi: 10.5194/acp-14-3637-2014. [DOI] [Google Scholar]
  27. McLinden CA, Fioletov V, Krotkov NA, Li C, Boersma KF, and Adams C. (2016), A Decade of Change in NO2 and SO2 over the Canadian Oil Sands As Seen from Space, Environmental Science & Technology, 50(1), 331–337, doi: 10.1021/acs.est.5b04985. [DOI] [PubMed] [Google Scholar]
  28. Noxon JF (1975), Nitrogen dioxide in the stratosphere and troposphere measured by ground-based absorption spectroscopy, Science, 189(4202), 547–549, doi: 10.1126/science.189.4202.547. [DOI] [PubMed] [Google Scholar]
  29. O’Byrne G, Martin RV, van Donkelaar A, Joiner J, and Celarier EA (2010), Surface reflectivity from the Ozone Monitoring Instrument using the Moderate Resolution Imaging Spectroradiometer to eliminate clouds: Effects of snow on ultraviolet and visible trace gas retrievals, Journal of Geophysical Research: Atmospheres, 115(D17), doi: 10.1029/2009JD013079. [DOI] [Google Scholar]
  30. Palmer PI, Jacob DJ, Chance K, Martin RV, Spurr RJD, Kurosu TP, Bey I, Yantosca R, Fiore A, and Li Q. (2001), Air mass factor formulation for spectroscopic measurements from satellites: Application to formaldehyde retrievals from the Global Ozone Monitoring Experiment, Journal of Geophysical Research: Atmospheres, 106(D13), 14,539–14,550, doi: 10.1029/2000JD900772. [DOI] [Google Scholar]
  31. Percy KE, Hansen MC, and Dann T. (2012), Air Quality in the Athabasca Oil Sands Region, in Volume 11: Alberta Oil Sands - Energy, Industry and the Environment, 1 ed., Elsevier. [Google Scholar]
  32. Platt U, and Stutz J. (2008), Differential Optical Absorption Spectroscopy: Principles and Applications, Springer, Berlin, Germany. [Google Scholar]
  33. van Geffen JHGM, Boersma KF, Van Roozendael M, Hendrick F, Mahieu E, De Smedt I, Sneep M, and Veefkind JP (2015), Improved spectral fitting of nitrogen dioxide from OMI in the 405–465 nm window, Atmospheric Measurement Techniques, 8(4), 1685–1699, doi: 10.5194/amt-8-1685-2015. [DOI] [Google Scholar]
  34. van Geffen JHGM, Eskes HJ, Boersma KF, Maasakkers JD, and Veefkind JP (2018), TROPOMI ATBD of the total and tropospheric NO2 data products, issue 1.2.0 ed., s5P-KNMI-L2–0005-RP. [Google Scholar]
  35. Veefkind J, Aben I, McMullan K, Forster H, de Vries J, Otter G, Claas J, Eskes H, de Haan J, Kleipool Q, van Weele M, Hasekamp O, Hoogeveen R, Landgraf J, Snel R, Tol P, Ingmann P, Voors R, Kruizinga B, Vink R, Visser H, and Levelt P. (2012), TROPOMI on the ESA Sentinel-5 Precursor: A GMES mission for global observations of the atmospheric composition for climate, air quality and ozone layer applications, Remote Sensing of Environment, 120, 70 – 83, doi: 10.1016/j.rse.2011.09.027, the Sentinel Missions - New Opportunities for Science. [DOI] [Google Scholar]
  36. Williams JE, Boersma KF, Le Sager P, and Verstraeten WW (2017), The high-resolution version of TM5-MP for optimized satellite retrievals: description and validation, Geoscientific Model Development, 10(2), 721–750, doi: 10.5194/gmd-10-721-2017. [DOI] [Google Scholar]
  37. Zara M, Boersma KF, De Smedt I, Richter A, Peters E, van Geffen JHGM, Beirle S, Wagner T, Van Roozendael M, Marchenko S, Lamsal LN, and Eskes HJ (2018), Improved slant column density retrieval of nitrogen dioxide and formaldehyde for omi and gome-2a from qa4ecv: intercomparison, uncertainty characterisation, and trends, Atmospheric Measurement Techniques, 11(7), 4033–4058, doi: 10.5194/amt-11-4033-2018. [DOI] [Google Scholar]
  38. Zawada DJ, Dueck SR, Rieger LA, Bourassa AE, Lloyd ND, and Degenstein DA (2015), High-resolution and monte carlo additions to the sasktran radiative transfer model, Atmospheric Measurement Techniques, 8(6), 2609–2623, doi: 10.5194/amt-8-2609-2015. [DOI] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

SM
NIHMS1535818-supplement-SM.pdf (1.4MB, pdf)

RESOURCES