Formaldehyde (HCHO) total column densities over the Mexico City metropolitan area (MCMA) were retrieved using two independent measurement techniques: multi-axis differential optical absorption spectroscopy (MAX-DOAS) and Fourier transform infrared (FTIR) spectroscopy. For the MAX-DOAS measurements, the software QDOAS was used to calculate differential slant column densities (dSCDs) from the measured spectra and subsequently the Mexican MAX-DOAS fit (MMF) retrieval code to convert from dSCDs to vertical column densities (VCDs). The direct solar-absorption spectra measured with FTIR were analyzed using the PROFFIT (PROFile FIT) retrieval code. Typically the MAX-DOAS instrument reports higher VCDs than those measured with FTIR, in part due to differences found in the ground-level sensitivities as revealed from the retrieval diagnostics from both instruments, as the FTIR and the MAX-DOAS information do not refer exactly to the same altitudes of the atmosphere. Three MAX-DOAS datasets using measurements conducted towards the east, west or both sides of the measurement plane were evaluated with respect to the FTIR results. The retrieved MAX-DOAS HCHO VCDs where 6
Megacities are in constant evolution, exhibiting continuous changes in territorial extension, population size and spatial redistribution, as well as in the types of socio-economic activities performed every day. In many cases the spatial growth is uneven, resulting in areas of the city being more prone to emissions or accumulation of pollutants due to chemical transformations or transport patterns influenced by meteorological conditions. For the specific case of the Mexico City metropolitan area (MCMA), the urban sprawl observed over the years has been topographically influenced, causing redensification processes due to the space-limited valley location of the MCMA
Formaldehyde (HCHO), a hazardous pollutant present mostly in the lower troposphere, is the most abundant carbonyl compound found in urban areas such as Mexico City. Due to its short lifetime of only a few hours, the quantitative determination of this gas gives an idea of the distribution of its sources
There are two main reactions that HCHO undergoes in the atmosphere, photolysis and reaction with OH
Deriving the global burden and emissions of many NMVOCs is a real challenge from the limited observations available; however, satellite HCHO observations can constrain their emissions in global chemistry transport models and thus provide a better understanding of their spatial distributions and temporal variability.
Remote sensing techniques are a useful complement in the quantification of gases by measuring the total atmospheric column amount along a line of sight. Spectrometers installed on board satellites, aircraft, balloons, vehicles or ground-based stations have the capacity to determine the atmospheric composition of gases and particles by observing their characteristic interaction with the radiation field. A common technique deployed from the ground is solar-absorption Fourier transform infrared (FTIR) spectroscopy, capable of quantifying vertical column densities and profiles of a wide range of gases
Some comparisons between FTIR and MAX-DOAS instruments have been done in the past. Surface HCHO was measured with two spectroscopic techniques by
In this study, we use a time series of more than 7 years to perform an unprecedented comparison (in terms of length and location) of the HCHO total vertical column amount measured with two independent techniques. Retrieval diagnostics from both the MAX-DOAS and FTIR results are used to characterize the difference in both measurement techniques and to improve the agreement and correlation between coincident data pairs (Sect.
In this section we describe the two independent measurement techniques, based on FTIR spectroscopy and MAX-DOAS, used to retrieve HCHO vertical column densities over two measurement sites. One of the sites is in the south of the MCMA, at the Universidad Nacional Autónoma de México (UNAM) campus – National Autonomous University of Mexico – on the roof of the Centro de Ciencias de la Atmósfera (CCA-UNAM, lat 19.32, long
The UNAM station is equipped with a Fourier transform infrared spectrometer (FTIR) from Bruker Optics (model Vertex 80) that measures solar-absorption spectra at different spectral regions with mercury–cadmium–telluride (MCT) and indium–gallium–arsenide (InGaAs) detectors and five band-pass filters placed on a rotating wheel. The interferometer has a maximal optical-path difference of 12 cm and continually records spectra at 0.075 cm
At the Altzomoni remote site, a high-resolution FTIR (Bruker Optics, IFS120/5 HR) is operated remotely with a microwave antenna that allows us to have communication with the station. This instrument allows a maximal optical-path difference of 257 cm, recording spectra typically at 0.005 cm
Vertical column densities (VCDs) are retrieved from solar-absorption FTIR spectra in four spectral microwindows in the region between 2763 and 2782 cm
A MAX-DOAS instrument, designed and built by the Spectroscopy and Remote Sensing Group at CCA-UNAM, was used to conduct sky measurements in the UV–Vis (ultraviolet–visible) region of the electromagnetic spectrum. The MAX-DOAS instrument, which has been collecting data since 2013, is installed on the roof of CCA-UNAM (same location as the FTIR-Vertex instrument) and forms part of a small network
The MAX-DOAS has a theoretical field of view of 0.31
Before conducting retrievals, spectra are filtered with the objective to remove all spectra either with light conditions that are too low (10
DOAS analysis settings for HCHO slant column density retrieval.
HCHO VCDs were retrieved using the Mexican MAX-DOAS fit (MMF) code
To run MMF retrievals, the absorption cross section was taken at a wavelength in between the range of the wavelength interval used for the QDOAS retrieval: for O
Three different versions of HCHO VCDs were retrieved using the MMF code: V1 retrieved VCDs from MAX-DOAS measurements conducted towards the east (telescope's azimuth angle of 85
For V1, V2 and V3 the same a priori is used both for the trace gas and for the aerosol. For V3, the “scan” is simply treated as consisting of two different azimuth directions. The V1, V2 and V3 retrievals are performed independent of each other and differ in the definition of a “scan”, where V3 contains all pointing directions from V1 and V2 together. A single vertical profile is retrieved in both directions for V3, so assuming horizontal homogeneity. This assumption clearly is not fulfilled; however, it is also not fulfilled in a single viewing direction, since the effective light path is around 5–20 km. As pointed out in the paper, the advantage of using both directions is a higher information content, while the disadvantage is a more rigorous breakdown of the homogeneity assumption.
Hourly means of HCHO vertical column densities (VCDs) from the MAX-DOAS instrument at UNAM (V1, magenta; V2, cyan; and V3, blue) and the solar-absorption FTIR instrument at UNAM (red) and at Altzomoni (green). The measurements in Mexico City at the UNAM campus (red, magenta, cyan and blue) are 1 order of magnitude higher than the background measurements at the mountain site of Altzomoni (green). The same applies to their variance. The difference between the MAX-DOAS and FTIR measurements at UNAM is mainly explained by their averaging kernel and is discussed in detail in the text. The altitude ranges covered by each instrument are as follows: for the FTIR instrument at UNAM, VCDs of 2–16 km; for the FTIR instrument at Altzomoni, VCDs of 4–16 km; and for the MAX-DOAS instrument at UNAM, VCDs of 2–5 km.
A large dataset of measurements taken at the UNAM site between January 2013 and May 2020 allowed us to study the diurnal and seasonal variability of HCHO. Figure
Figure
Diurnal cycle from hourly averaged HCHO VCDs from FTIR (red), MAX-DOAS V3 (blue), MAX-DOAS V1 (magenta) and MAX-DOAS V2 (cyan) measurements at UNAM. Vertical lines represent the standard deviation of all conducted measurements during that hour.
In Fig.
Seasonal FTIR (red), MAX-DOAS V3 (blue), MAX-DOAS V1 (magenta) and MAX-DOAS V2 (cyan) cycles at UNAM. Vertical lines represent the standard deviation of all conducted measurements during each month.
In order to assess the horizontal inhomogeneity of HCHO in the MCMA, an average distribution map of HCHO was constructed from data between 2005 and 2018 from the Ozone Monitoring Instrument (OMI) satellite instrument and is presented in Fig.
Average HCHO total column distribution map over the MCMA between 2005 and 2018. The columnar HCHO distribution is reconstructed from OMI measurements on board the Aura satellite with a daily Mexico City overpass at around 14:00 LT (spatial distribution is only representative for this time). Color bar units are in molec. cm
Comparison between FTIR and MAX-DOAS measurements conducted at the UNAM measurement site. The panels in the first and second rows correspond to the VCDs retrieved from MAX-DOAS measurements conducted towards the eastern (V1) and western (V2) measurement sides, respectively. For the third row panel, corresponding to the V3 data product, the VCDs are retrieved including both measurement sides. The panels in the fourth row correspond to the comparison between FTIR and MAX-DOAS V1 measurements during the morning and FTIR and MAX-DOAS V2 measurements during the afternoon. The linear regression when forced to zero (red) and not constrained (green) is presented. Black lines represent the
A detailed comparison between VCDs retrieved using the MAX-DOAS and FTIR measurement techniques was conducted and is explained in this section. The correlation between the coincident hourly mean vertical columns from FTIR and MAX-DOAS measured at UNAM are shown in Fig.
For the correlation plots presented in the middle column of Fig.
To further investigate the large HCHO inhomogeneity already shown in Fig.
The hourly differences between VCDs computed using the eastern or western sides of the scanning plane is investigated further; therefore simulated VCDs were calculated in order to compare them with measured VCDs. Simulated VCDs east–west differences are the result of the different amount of information in the retrievals in V1 and V2. The true profile has much higher HCHO concentrations in the polluted mixing layer than what the a priori information reflects. The retrieval using both sides of the measurement plane contains more information originating from the measurements and allows the result on an optimal-estimation-based retrieval to be less close to the a priori information. The V1 and V2 retrievals do not always have the same amount of information, as the filtering criteria for the spectra do not act similarly throughout the day, and spectral measurements are selected in an unbalanced way. The factors affecting the uneven amount of information used in the retrievals include permanent or temporal obstacles and the time-dependent probability of saturation of the spectra when viewing towards or close to the sun. This means that even if the atmosphere around the measurement site would be perfectly homogeneous in the horizontal plane, the columns retrieved using V1 and V2 datasets might be slightly different. We try to explain this by the different sensitivities and their averaging kernels (AK, AK
Correlation plots between HCHO VCDs retrieved using dSCDs measured towards the
Histograms showing degrees of freedom and frequencies for V1, V2 and V3 retrievals.
Equation
Here we choose
If the errors
If we assume that the retrieved profile
This expected or simulated difference, which evidently depends on the time of the day, is calculated according to Eq. (
Average differences between the retrieved columns using the east and west measurement sides, as function of the hour of the day, are presented in the blue line. The red line simulates how the information available from the retrievals would produce a difference in the columns, assuming that the retrieval of the V3 dataset using both measurement sides best describes the atmospheric state (see text for details). The black points are the difference between the found differences and the “forecasted” differences and should show the part of the difference which might be related to a real inhomogeneity and a gradient between the east and the west HCHO concentrations in the mixing layer. The error bars represent the standard error and show that the measurement amount is large enough to calculate a statistically significant mean difference for each day by hour, and the grouping for different hours is necessary.
The averaging kernels from the V1 and V2 retrievals allow us to estimate and forecast a difference because of their different sensitivities. This effect, dominant after 15:00 LT, most likely depends on the number of dSCDs available for the MMF retrievals that could be significantly reduced, as the sun is closer to the viewing angles and does not pass the filtering criteria due to saturation. Alternatively, the forward model in QDOAS could be having more difficulties in explaining the measured spectra so that the errors in the retrieved dSCDs (typically 15
Hourly FTIR versus MAX-DOAS comparisons at UNAM between 09:00 and 16:00 LT.
As can be seen in Fig.
In the previous section, it was shown that analyzing the MAX-DOAS viewing directions independently can in part explain the large horizontal inhomogeneity around the UNAM urban site. We now investigate the behavior in the correlation between MAX-DOAS and FTIR data for different hours of the day and how it can be affected by the sensitivities of both instruments and the changing vertical distribution of the HCHO profiles.
Figure
The relation between the MAX-DOAS and the FTIR VCDs is described by the scatter (the Pearson correlation coefficient), the slope and constant bias. As we already have seen in the comparison between the MAX-DOAS V3 data product (both sides) with respect to the single sides, having a slope of 1.0 does not ensure that both retrievals are correct and similar to the true atmospheric state, but it rather means that both sensitivities are similar.
Based on Eq. (
Neither the retrieved FTIR profile (1.1) nor the MAX-DOAS profile retrieval (
Here we try to evaluate the consistency of the two retrievals differently, starting with Eq. (
The average of the product of the columns of both instruments is theoretically given by the following equation; for that purpose we introduce the errors
To simplify the interpretation, we assume that the averaging kernels of both instruments are more or less constant and independent in
and therefore the Pearson correlation coefficient and the slope are formally calculated:
If the correlation plot is limited to just 1 h as shown in Fig.
The variability of the concentration profile with a fixed shape (
The individual plots in Fig.
In order to investigate background HCHO levels, HCHO VCDs retrieved from measurements conducted with the high-resolution FTIR spectrometer (Bruker Optics, IFS120/5 HR) in Altzomoni are presented in Fig.
As in the case of the UNAM measurement site, the diurnal and seasonal HCHO cycles were calculated. Hourly HCHO VCDs at Altzomoni show a steady increase during the day, with a smaller growth rate from 14:00 to 17:00.
The seasonal cycle shows a maximum during September, while the lowest HCHO VCDs values occur during December. The background HCHO VCDs at Altzomoni (Fig.
Hourly
In this contribution we present a comparison between HCHO total column densities retrieved from two independent measurement techniques: MAX-DOAS and solar-absorption FTIR. Both measurement techniques, although based on spectroscopy, exhibit a very different measurement strategy and geometry. Despite these differences, a good agreement was obtained between both instruments. Due to the versatility of the retrieval code used to process the MAX-DOAS data, VCDs were retrieved using measurements conducted towards different viewing directions. Retrieval products were obtained employing measurements conducted exclusively towards the east or the west or using measurements conducted towards both sides of the measurement station. Considering the FTIR results as the reference, MAX-DOAS VCDs from these datasets where 6
Reasons for the overestimation of the MAX-DOAS over the FTIR results are attributed to an enhanced ground level (lowest few kilometers of the atmosphere) sensitivity of the former with respect to the latter. However, the intrinsic differences between the two measurement techniques could also account for the discrepancies found in this study. In the first place, both measurement techniques have different sampling geometries and strategies. The MAX-DOAS instrument measures spectra at different elevation angles, leading to an altitude-averaged measurement in the lower atmosphere. From these measurements, HCHO VCDs are computed using a numerical code – MMF
Moreover, this research provided the opportunity to study in more detail horizontal HCHO inhomogeneities in the MCMA, identifying diurnal and seasonal variabilities of the abundance of HCHO total columns. In the future this could be used to further study primary versus secondary HCHO in the MCMA and develop specific analysis strategies focused on the identification and disaggregation of freshly emitted and secondary produced HCHO in the boundary layer of the MCMA. Satellite-based data have been used to corroborate the spatial inhomogeneity of the HCHO total column over the MCMA as shown in Fig.
Identifying and characterizing horizontal inhomogeneities with respect to the abundance of molecules present in air can also be of service when making decisions regarding location and azimuth measurement angles for future MAX-DOAS stations in the MCMA. Future work includes studying horizontal inhomogeneities at other stations of the MAX-DOAS network as well as horizontal inhomogeneities of other chemical species, such as nitrogen dioxide, which is routinely retrieved as well from the spectra measured by the MAX-DOAS instruments located in the MCMA.
It is worth mentioning that these types of strong spatial heterogeneity scenarios have been observed in different areas of the planet and specific studies of atmospheric constituents have been or are currently being performed in other urbanized and densely populated areas such as North America
The quantified diurnal and seasonal variability of HCHO as well as the characterized horizontal inhomogeneity in the MCMA, presented in this study, could be attributed to direct emissions or secondary formation of HCHO from released precursors from anthropogenic and/or biogenic sources that form part of the MCMA and constantly influence its atmospheric composition. Identification of either primary emissions or secondary formation of HCHO is outside the scope of this study; however, based on the results presented here and previous research conducted by
In terms of further characterizing HCHO horizontal inhomogeneity, the Tropospheric Emissions: Monitoring of Pollution (TEMPO) instrument
The MAX-DOAS and FTIR data can be accessed via
CR was responsible for the QDOAS retrieval setup and parameter choices, for the setup and running of the MMF code, for processing HCHO OMI data, and data analysis and interpretation. CR wrote parts of the Abstract and parts of Sects. 2, 3 and 4. CG was responsible for developing and optimizing various HCHO retrievals for Altzomoni and UNAM and contributed to the development of the NDACC retrieval strategy, which was finally applied. CG took care of the measurements and focused on the separation of freshly emitted HCHO and secondary produced HCHO in the boundary layer of Mexico City. WS wrote parts of Sects. 2 and 3. MMF was responsible for the retrieval MMF code development and testing, the retrieval chain setup from spectra to profiles, and the retrieval parameter choices for MMF and software support. MMF wrote parts of Sect. 2. DR was responsible for running parts of the MMF code. CAMR was responsible for processing HCHO OMI data and assisted in the creation of Fig. 4. AB provided technical support for instruments and data management. MG was involved in the data interpretation and wrote Sect. 1 and parts of Sects. 2, 3 and 4. WS, AB and MG were responsible for running the FTIR retrievals at UNAM and Altzomoni and for data analysis and interpretation. TB and FH provided the IFS120/5 HR spectrometer located in Altzomoni and developed the setup of the spectrometer and solar tracker. They provided assistance by bringing the spectrometer in Altzomoni into operation and trained the Mexican group in the operation and alignment of the IFS120/5 HR. FH developed the retrieval code PROFFIT9 and LINEFIT and provided continuous support during its use. In terms of specific figures contributions, Figs. 1, 2 and 3 were created by WS and CR; Fig. 4 was created by CR and CAMR; Figs. 5 and 6 were created by WS; Figs. 7, 8 and 9 were created by WS, DR and CR; and Figs. 10 and 11 were created by WS.
The authors declare that they have no conflict of interest.
Arne Krueger, Josué Arellano, Alfredo Rodríguez, Delibes Flores, Miguel Angel Robles and Omar López are thanked for their technical assistance. We thank Caroline Fayt (caroline.fayt@aeronomie.be), Michel Van Roozendael (michelv@aeronomie.be) and Thomas Danckaert for the free use of the QDOAS software, and we thank Robert Spurr for free use of the VLIDORT radiative transfer code package. We thank Agustin García Reynoso for important and fruitful discussions about HCHO in Mexico City. We thank the Mexican Solarimetric Service for their effort in establishing and maintaining the AERONET Mexico City site. We thank the University of Wyoming Department of Atmospheric Science for providing the sounding data on
This research has been supported by the DGAPA-UNAM (grant nos. TA100418, IN111418, IN107417 and IA101620), the CONACYT (grant no. 290589) and the INECC (grant no. INECC-A1-002-2019). Cristina A. Mendoza-Rodríguez received financial support from Consejo Nacional de Ciencia y Tecnología (CONACYT) through a graduate studies grant (no. CVU 956921). Wolfgang Stremme received financial support from DGAPA-PASPA.
This paper was edited by Michel Van Roozendael and reviewed by two anonymous referees.