How to reference these images/data in a publication

 

If you are using these images and/or data in a publication, please conctact us to verify and get the best data for your use. And use this for your reference:

Lamsal, L. N., Krotkov, N. A., Vasilkov, A., Marchenko, S., Qin, W., Yang, E.-S., Fasnacht, Z., Joiner, J., Choi, S., Haffner, D., Swartz, W. H., Fisher, B., and Bucsela, E.: OMI/Aura Nitrogen Dioxide Standard Product with Improved Surface and Cloud Treatments, Atmos. Meas. Tech. Discuss., https://doi.org/10.5194/amt-2020-200, in review, 2020.

 

General Information on deriving NO₂ from satellites

 

Air pollution is responsible for ~7 million premature deaths every year, nearly half of which is due to outdoor air pollution. The air pollutants linked closely to mortality are fine particulate matter (PM2.5), near surface ozone, sulfur dioxide, and nitrogen dioxide (NO₂). Nitrogen oxides (NOx = NO + NO₂) are produced during combustion processes (e.g., from vehicles, power plants, and factories). NO₂ is the main precursor of tropospheric ozone and nitrate aerosols, a component of PM2.5. It is designated as a criteria pollutant by the US Environmental Protection Agency (EPA) because of its negative effects on human health and the environment.

Satellite instruments afford a global view of the planet's air pollution. These instruments measure backscattered radiation from the sun in a wide spectral range from ultraviolet (UV) to infrared wavelengths. Sophisticated retrieval algorithms are used to convert the measured radiation to column pollution concentrations, such as a tropospheric column density of NO₂. Satellite measurements of pollutants have proven valuable for health and air quality applications.

NASA's Earth Observing System Aura OMI science team at Goddard Space Flight Center develops, maintains, and validates retrieval algorithms to derive NO₂ column densities from multiple satellite instruments. These data are publicly available at the Goddard Earth Sciences Data and Information Services Center (GES DISC) and the Aura Validation Data Center (AVDC).

NO₂ column observations from the Dutch-Finnish Ozone Monitoring Instrument (OMI) and ESA/EU Copernicus Sentinel 5 Precursor TROPOspheric Monitoring Instrument (TROPOMI) are available from October, 2004 and April, 2018, respectively. OMI is a UV-Visible wavelength spectrometer on the polar-orbiting NASA Aura satellite. Aura, launched on 15 July 2004, follows a sun-synchronous orbit with an equator crossing time near 13:45, local time. TROPOMI is a similar instrument with enhanced spatial resolution and additional spectral coverage. It was launched on board the European Space Agency/European Union (ESA/EU) Copernicus Sentinel-5 Precursor satellite on October 13, 2017.

For further reading:

https://airquality.gsfc.nasa.gov

https://aura.gsfc.nasa.gov

http://www.tropomi.eu/data-products/nitrogen-dioxide

 

 

Frequently Asked Questions (FAQs)

 

FAQs on OMI NO₂ time series and maps

 

What does "vertical column" mean, and what are the units of the NO₂ data?

 

Satellite instruments that measure NO₂ are sensitive to the number of molecules between the instrument and the Earth's surface (i.e., a Vertical Column Density or VCD). For NO₂ typically the amount high in the atmosphere (in the stratosphere) is removed in order to provide an estimate of the column in the lower part of the atmosphere (the troposphere), a Tropospheric Vertical Column Density or TVCD. The units of VCD or TVCD are molecules per square centimeter or molec./cm². Because these numbers are large, they may be divided by 10¹⁵. Units may also be given as moles per unit area.

 

Is the NO₂ vertical column the same as NO₂ emissions?

 

No. NO₂ vertical columns are observed from space and are the results of emissions of nitrogen oxides (NO_x = NO + NO₂) that are formed by fossil fuel combustion as well as fires, lightning, and in soil. With a lifetime of the order of hours near the surface, NO₂ is transported away from its sources and this transport depends on winds. Its lifetime varies with a number of factors including the amount of sunlight and other chemical species in the atmosphere. Therefore, NO₂ column amounts in the atmosphere can be highly variable and dependent on local emissions, chemistry and transport. NO₂ columns can serve as a proxy of NO_x emissions if averaged appropriately in time and space. However, if appropriate averaging is not undertaken and/or effective quality control (for example, to remove pixels substantially affected by clouds), a flaw interpretation may result (see the Copernicus article "Flawed estimates of the effects of lockdown measures on air quality derived from satellite observations").

 

What exactly is shown in the time-series plots?

The time-series show recent trends in NO2 for 1^o latitude x 1^o longitude grid boxes drawn around select global cities. The blue line in the top panel of the time series shows the median value of all good observations within +/-7 days of that date for the past several months, while the red line in the second panel shows the mean value for the same observations. The gray lines are similar to the blue and red lines but for a 5 year "baseline" (2015-2019) for comparison. To compute the "baseline" values shown, similar statistics as for 2020 were computed for each day within the range 2015-2019. The baseline values were then computed as a weighted mean of values for a given day over all years, where values were weighted by the number of points used to compute each median. The shading in the top panel represents the range between the 25^th and 75^th percentiles for the median values, while the shading in the second panel represents one standard deviation in each direction. The third panel shows a so-called "anomaly" defined as the difference between the 15 day running median and mean for the past several months and the baseline from 2015-2019. A negative value means the recent values are below the baseline. Confidence levels for anomaly were computed using the standard deviations computed over one year and assume a normal distribution. They may therefore be underestimated in cases on non-normal distributions. The bottom panel shows the number of samples going into a given 15 day subset of samples for the recent data. Gaps appear for dates in which a threshold of 3 days with good data for the 15 day period was not met. There may be no good data on a given day due to either no coverage by the satellite or because the data were filtered out due to heavy cloud cover, when the instrument is not able to measure NO₂ near the ground. 

Why are there differences between the time-series and the maps?

While the trends are generally similar between the time-series and maps, there are some differences due to the way that each are produced. The time-series plots are made by calculating the mean and median for all samples within the 15 day window. By doing this, we reduce some sampling noise that can come from day to day sampling difference and produce a smoother NO₂ trend. The maps were produced using high resolution daily gridded which is then averaged over the 15 day window to show the daily maps. Additionally, since the maps use high resolution data, there can be gaps for portions of the 1x1 degree grid box difference, whereas the time-series are sampled over the 1x1 degree box and therefore would not have the same gaps.

 

Why do you show both the median and the mean in the time series plots?

 

The calculation of the median can be more robust in the presence of outliers, which we have found mainly in large cities with high levels of NO₂ that shows high variability due to meteorology. The running median may therefore produce a more smooth time series, less affected by outliers, as compared with a running mean.

 

What exactly is shown in the animated time series of maps?

 

Maps show 15 day OMI tropospheric NO₂ means gridded at 0.1^o latitude by 0.1^o longitude resolution with a contour algorithm applied for a region around a given city. Animations thus show 15 day running means. The left panels show maps for 2020, middle shows maps for data averaged over the years 2015-2019 (the baseline) and right plots show the absolute differences. A percent difference is computed over a 1^o latitude by 1^o longitude box around the city center as indicated in the right panel. Data in gray indicate regions for which the gridding process yielded fewer than 3 days of good data over the 15 day averaging period.

 

Why use OMI when TROPOMI has higher resolution and why do OMI and TROPOMI sometimes show different results?

 

The OMI NO₂ baseline data (averaged over several years prior to 2020) provide important context for NO₂ data in 2020. Currently the data record of TROPOMI NO₂ prior to 2020 is limited. The spatial resolution for TROPOMI NO₂ has changed once over its record. In addition, the TROPOMI NO₂ retrieval algorithm may have changed over time and may not have been applied consistently for the whole record. For this reason, one must be cautious when using TROPOMI NO₂ data to detect changes over time. The OMI standard NO₂ product uses the same algorithm for the whole record and the OMI instrument has been very stable over time. The OMI NO₂ algorithm has undergone a substantial amount of validation (see information from publications tab) and is considered state-of-the-art. The TROPOMI NO₂ algorithm is based on the experience derived from OMI and other similar instruments. However, the data are processed by a different team and the algorithm differs in some respects from the current OMI standard NO₂ algorithm. Therefore, we should not expect perfect agreement between the NO₂ data from the two instruments.

 

Where can I get answers to more FAQs about OMI?

 

This link gives more information about OMI including its swath and products.

 

What is the spatial resolution of OMI and TROPOMI measurements?

 

The spatial footprint of OMI is approximately 13 km by 24 km for the nadir pixels. The spatial resolution decreases (pixel size increases) towards the swath edge. A TROPOMI pixel is currently about 3.5 km by 5.5 km at nadir and the spatial resolution also decreases towards the swath edge but not as much as for OMI.

 

What is the temporal resolution of OMI and TROPOMI measurements?

 

TROPOMI generally has daily global coverage (makes a measurement once per day) while OMI covers a point on the ground about once every two days. At high latitudes, due to swath overlaps, it is possible for either satellite to make more than one measurement per day.

 

At what time of day do OMI and TROPOMI make measurements?

 

Both OMI and TROPOMI have overpasses near 13:30 local time which is near the nominal equator crossing time for both host satellites. However, this varies a bit day by day depending on which part of the satellite swath is covering a point on the ground and also varies by latitude as the satellite does not go directly over the poles but rather is in a slightly inclined orbit.

 

Can OMI and TROPOMI observe other species?

 

Yes, since these instruments measure spectra in a wide wavelength range, we can derive the amount of many other species that absorb radiation within the spectral range of these instruments. For instance, OMI can provide observations of ozone, NO₂, sulfur dioxide, water vapor, formaldehyde, cloud, and aerosol amounts. Because of its larger spectral coverage, TROPOMI can provide observation of other additional species such as carbon monoxide and methane.

 

Why are there gaps in the data?

 

When there is heavy cloud cover, it is not possible to derive NO₂ data near the surface below the clouds which affects any similar instrument. During certain times of the year in certain locations, there may be persistent cloud cover that limits satellite measurements. Even in the absence of clouds, OMI does not provide daily global coverage, but rather provides data over a given point about once every two days. TROPOMI provides global daily coverage over a given point (provided no heavy cloud cover). NO₂ data are provided for cases of thin or broken clouds as examination of these data has shown them to be of generally good scientific quality.

 

Why are there sudden jumps in the data?

 

The OMI time series provided on this site are smoothed over a number of days for a given location and may be averaged in space as well. This averaging process smooths out natural day to day variations that occur due to transport and dispersion of pollution plumes as well as random errors in the satellite measurements. This means that for each day, the value shown consists of an average of available satellite data on that day as well several days before and after that date. Even with this averaging process, not all of the variations are completely smoothed out. Please note that we display both mean values (averaged) as well as medians. The use of medians is more robust when outliers are present. We find that most outliers occur due to natural variations.

 

Why do the 2020 data appear noisier than the OMI multi-year (baseline) average?

 

The OMI baseline (average of 5 previous years of data prior to 2020) accounts for natural seasonal variations such as changes in the amount of time NO₂ resides in the atmosphere (which depends on the amount of sunlight and other factors). The baseline also accounts for other annual events such as the Lunar New Year celebrated in many countries in which factories may be shut down and normal traffic patterns altered. Averaging data over several years reduces the effects of variability due to weather patterns and random noise in satellite data.

 

Why do the 2020 data sometimes fall outside the baseline before the COVID-19 outbreak?

 

In the time series shown, estimates of the uncertainties are provided. These estimates include contributions from natural variability (e.g., weather variations) and random errors in satellite data. The values shown are based on several assumption such as normal distributions (which may or may not be the case) and are standard errors (based on one standard deviation) divided by square root of the number of points being analyzed. Therefore, the greater number of points, the better the estimate (assuming that errors are not correlated; note that the uncertainties do NOT account for any systematic errors). Statistically, we should expect that a small fraction of the time, values will naturally fall outside of this range. However, when we see values consistently falling outside the expected range for an extended period of time, we would consider this to be unusual and perhaps caused by an unexpected event such as traffic reductions that result from measures taken to reduce exposure to COVID-19.

 

General FAQs

 

The following were adapted from the review paper of Duncan et al. (2014). This paper contains all references listed below.

 

How does a satellite instrument measure gases and aerosols?

Most satellite instruments that collect data relevant for air quality (AQ) applications are "passive". ("Active" instruments, such as lidars or radars, send a signal and detect the portion of the signal that returns). Passive instruments detect electromagnetic radiation from the Sun that is absorbed and reemitted, reflected, and scattered by the Earth's surface and atmosphere. The incoming radiation passes through a spectrometer, a device that measures energy intensity as a function of wavelength, to create a spectrum of wavelengths that are then detected. When individual photons strike the instrument's detector, the energy is converted into electrons as a way of measuring the amount of incoming energy at various wavelengths. The infrared (IR), visible, and ultraviolet (UV) regions of the electromagnetic spectrum contain the most useful wavelengths for observing pollutants relevant for AQ applications as these gases and aerosols absorb IR wavelengths (e.g., water vapor) or scatter and absorb visible and UV wavelengths (e.g., dust, ozone, SO₂, NO₂).

 

Can satellites measure "nose-level" concentrations?

The short answer is "no" because the majority of satellite instruments that measure pollutants of interest to the AQ community are downward-looking, providing limited information on the vertical structure of the pollutant in the atmosphere. Satellite instruments that measure ozone, NO₂, formaldehyde, and SO₂ detect the number of molecules between the instrument and the Earth's surface (i.e., a Vertical Column Density or VCD). Nevertheless, these data are highly useful in many AQ applications, including for inferring "nose-level" concentrations with the help of chemistry models and ground-based in-situ instrument networks. For more in-depth discussions of issues associated with detecting surface concentrations from space, beyond what is presented here, please see Fishman et al., 2008, Martin, 2008, and Hoff and Christopher (2009).

 

Are there satellite data for surface NO₂?

 

While it is not feasible to measure surface NO₂ with current satellite instruments (e.g., Bovensmann et al., 1999), tropospheric NO₂ vertical column density (VCD) is readily detected. The NO₂ VCD serves as an effective proxy for NO_x (=NO+NO₂) and correlates well with surface levels of NO₂ in industrialized regions (e.g., Leue et al., 2001, Velders et al., 2001). Most of the NO₂ VCD is found near its emission sources because its chemical lifetime is short (i.e., hours to about a day depending on meteorological conditions) and its background level is low relative to the level in industrialized areas. Like ozone, there is a significant contribution to the NO₂ VCD from the stratosphere, but it can be subtracted as part of satellite retrieval processing (e.g., Bucsela et al. (2013) and references therein). Even if the stratospheric portion is not removed, the local gradients in the VCD are associated with gradients near the surface in polluted regions as the distributions of NO₂ in the stratosphere are rather uniform. This subtraction is already done in the level 2 (swath) and level 3 (gridded) OMI tropospheric NO₂ products. Airborne measurements over polluted areas suggest that the portion of the NO₂ VCD in the boundary layer could be over 75% of the tropospheric VCD over land and (Martin et al., 2004; Bucsela et al., 2008). Ordóñez et al. (2006) demonstrated a strong correlation between the tropospheric NO₂ VCD and surface NO₂ observations. Lamsal et al., 2008, Lamsal et al., 2010 developed a method for estimating surface NO₂ from OMI tropospheric NO₂ VCD data, finding that their OMI-derived surface NO₂ concentrations were well correlated with surface measurements, both temporally (r = 0.3-0.8) and spatially (r = 0.76). Knepp et al. (2013) found that VCD data from a ground-based suntracking spectrometer system, which is similar to OMI VCD, compared well to "nose-level" in-situ NO₂ data collected nearby when the daily cycle of boundary layer mixing was taken into account.