Latin American (LatAm) cities are grappling with elevated levels of gaseous and particulate pollutants, which are having detrimental effects on both the local ecosystem and human health. Of particular concern are aerosols with smaller diameters (lower or equal to 2.5 μm, PM2.5), known for their ability to penetrate deep into the respiratory system. While measurements in the region are increasing, they remain limited. This study addresses this gap by presenting the results of a comprehensive, year-long PM2.5 monitoring campaign conducted in six LatAm cities: Buenos Aires, São Paulo, Medellín, San José, Quito and Ciudad de México. Despite all six monitoring sites being urban, they exhibited significant variations in PM2.5 levels, as well as in the content and seasonal behavior of elemental carbon (EC) and organic carbon (OC). Estimations of secondary organic carbon (SOC) using the EC-tracer method revealed a notable SOC relevance across all cities: secondary organic aerosols (SOA) accounted in average for between 19 % to 48 % of the total carbonaceous matter. Source attribution, conducted through the Positive Matrix Factorization (PMF) model, highlights substantial contributions from gasoline and diesel traffic emissions (29 % to 49 % of total carbon, TC), regional biomass burning (21 % to 27 %), and SOA (20 % to 38 %) in all cities, with similar chemical signatures. Additionally, industrial emissions were significant in two cities (Medellín and San José), while two others experienced minor impacts from construction machinery at nearby sites (Buenos Aires and Quito). This comparative analysis underscores the importance of considering not only the thermal optical EC/OC fractions as tracers of sources but also the OC/EC ratio of the PMF factors. This dual approach not only adds depth to the research but also suggests varied methodologies for addressing this crucial environmental concern. This study lays the groundwork for future investigations into the factors influencing the content and seasonality of SOA in the region.

Having a prediction model for air quality at a low computational cost can be useful for research, forecasting, regulatory, and monitoring applications. This is of particular importance for Latin America, where rapid urbanization has imposed increasing stress on the air quality of almost all cities. In recent years, machine learning techniques have been increasingly accepted as a useful tool for air quality forecasting. Out of these, random forest has proven to be an approach that is both well-performing and computationally efficient while still providing key components reflecting the nonlinear relationships among emissions, chemical reactions, and meteorological effects. In this work, we employed the random forest methodology to build and test a forecasting model for the city of Buenos Aires. We used this model to study the deep decline in most pollutants during the lockdown imposed by the COVID-19 (COronaVIrus Disease 2019) pandemic by analyzing the effects of the change in emissions, while taking into account the changes in the meteorology, using two different approaches. First, we built random forest models trained with the data from before the beginning of the lockdown periods. We used the data to make predictions of the business-as-usual scenario during the lockdown periods and estimated the changes in concentrations by comparing the model results with the observations. This allowed us to assess the combined effects of the particular weather conditions and the reduction in emissions during the period when restrictions were in place. Second, we used random forest with meteorological normalization to compare the observational data from the lockdown periods with the data from the same dates in 2019, thus decoupling the effects of the meteorology from short-term emission changes. This allowed us to analyze the general effect that restrictions similar to those imposed during the pandemic could have on pollutant concentrations, and this information could be useful to design mitigation strategies. The results during testing showed that the model captured the observed hourly variations and the diurnal cycles of these pollutants with a normalized mean bias of less than 6 % and Pearson correlation coefficients of the diurnal variations between 0.64 and 0.91 for all the pollutants considered. Based on the random forest results, we estimated that the lockdown implied relative changes in concentration of up to −45 % for CO, −75 % for NO, −46 % for NO2, −12 % for SO2, and −33 % for PM10 during the strictest mobility restrictions. O3 had a positive relative change in concentration (up to an 80 %) that is consistent with the response in a volatile-organic-compound-limited chemical regime to the decline in NOx emissions. The relative changes estimated using the meteorological normalization technique show mostly smaller changes than those obtained by the random forest predictive model. The relative changes were up to −26 % for CO, up to −47 % for NO, −36 % for NO2, −20 % for PM10, and up to 27 % for O3. SO2 is the only species that had a larger relative change when the meteorology was normalized (up to 20 %). This points out the need for accounting not only for differences in emissions but also in meteorological variables in order to evaluate the lockdown effects on air quality. The findings of this study may be valuable for formulating emission control strategies that do not disregard their implication on secondary pollutants. We believe that the model itself can also be a valuable contribution to a forecasting system in the city and that the general methodology could also be easily applied to other Latin American cities as well. We also provide the first O3 and SO2 observational dataset in more that a decade for a residential area in Buenos Aires, and it is openly available at https://doi.org/10.17632/h9y4hb8sf8.1 (Diaz Resquin et al., 2021).

This study was majorly inspired by Grange et al 2019 to use a random forest model (RF) to train, validate and predict the air quality concentration in a megacity of Argentina. Although the methodology is not new, this study has the potential to have a significant improvement to fill the data gap given the reason that some monitored trace gas concentrations were lacking but the pollutants are becoming a concern for local authorities. (line 145150) I would like to propose a few major revisions for the authors of this manuscript. After done with that improvement, I think it would be strong enough to publish on earth system science data:

The multidisciplinary project Prediction of Air Pollution in Latin America and the Caribbean (PAPILA) is dedicated to the development and implementation of an air quality analysis and forecasting system to assess pollution impacts on human health and economy. In this context, a comprehensive emission inventory for South America was developed on the basis of the existing data on the global dataset CAMS-GLOB-ANT v4.1 (developed by joining CEDS trends and EDGAR v4.3.2 historical data), enriching it with data derived from locally available emission inventories for Argentina, Chile, and Colombia. This work presents the results of the first joint effort of South American researchers and European colleagues to generate regional maps of emissions, together with a methodological approach to continue incorporating information into future versions of the dataset. This version of the PAPILA dataset includes CO, NOx, NMVOCs, NH3, and SO2 annual emissions from anthropogenic sources for the period 2014–2016, with a spatial resolution of 0.1∘ × 0.1∘ over a domain that covers 32–120∘ W and 34∘ N–58∘ S. The PAPILA dataset is presented as netCDF4 files and is available in an open-access data repository under a CC-BY 4 license: https://doi.org/10.17632/btf2mz4fhf.3 (Castesana et al., 2021). A comparative assessment of PAPILA–CAMS datasets was carried out for (i) the South American region, (ii) the countries with local data (Argentina, Colombia, and Chile), and (iii) downscaled emission maps for urban domains with different environmental and anthropogenic factors. Relevant differences were found at both country and urban levels for all the compounds analyzed. Among them, we found that when comparing PAPILA total emissions versus CAMS datasets at the national level, higher levels of NOx and considerably lower levels of the other species were obtained for Argentina, higher levels of SO2 and lower levels of CO and NOx for Colombia, and considerably higher levels of CO, NMVOCs, and SO2 for Chile. These discrepancies are mainly related to the representativeness of local practices in the local emission estimates, to the improvements made in the spatial distribution of the locally estimated emissions, or to both. Both datasets were evaluated against surface concentrations of CO and NOx by using them as input data to the WRF-Chem model for one of the analyzed domains, the metropolitan area of Buenos Aires, for summer and winter of 2015. PAPILA-based modeling results had a smaller bias for CO and NOx concentrations in winter while CAMS-based results for the same period tended to deliver an underestimation of these concentrations. Both inventories exhibited similar performances for CO in summer, while the PAPILA simulation outperformed CAMS for NOx concentrations. These results highlight the importance of refining global inventories with local data to obtain accurate results with high-resolution air quality models.

The COVID-19 (COronaVIrus Disease 2019) pandemic provided the unique opportunity to evaluate the role of a sudden and deep decline in air pollutant emissions in the ambient air of numerous cities worldwide. Argentina, in general, and the Metropolitan Area of Buenos Aires (MABA), in particular, were under strict control measures from March to May 2020. Private vehicle restrictions were intense, and primary pollutant concentrations decreased substantially. To quantify the changes in CO, NO, NO2, PM10, SO2 and O3 concentrations under the stay-at-home orders imposed against COVID-19, we compared the observations during the different lockdown phases with both observations during the same period in 2019 and concentrations that would have occurred under a business-as-usual (BAU) scenario under no restrictions. We employed a Random Forest (RF) algorithm to estimate the BAU concentration levels. This approach exhibited a high predictive performance based on only a handful of available indicators (meteorological variables, air quality concentrations and emission temporal variations) at a low computational cost. Results during testing showed that the model captured the observed daily variations and the diurnal cycles of these pollutants with a normalized mean bias (NMB) of less than 11 % and Pearson correlation coefficients of the diurnal variations of between 0.65 and 0.89 for all the pollutants considered. Based on the Random Forest results, we estimated that the lockdown implied concentration decreases of up to 47 % (CO), 60 % (NOx) and 36 % (PM10) during the strictest mobility restrictions. Higher O3 concentrations (up to 87 %) were also observed, which is consistent with the response in a VOC-limited chemical regime to the decline in NOx emissions. Relative changes with respect to the 2019 observations were consistent with those estimated with the Random Forest model, but indicated that larger decreases in primary pollutants and lower increases in O3 would have occurred. This points out to the need of accounting not only for the differences in emissions, but also in meteorological variables to evaluate the lockdown effects on air quality. The findings of this study may be valuable for formulating emission control strategies that do not disregard their implication on secondary pollutants. The data set used in this study and an introductory machine learning code are openly available at https://data.mendeley.com/datasets/h9y4hb8sf8/1 (Diaz Resquin et al., 2021).

Equivalent black carbon (EBC) mass concentrations in the fine inhalable fraction of airborne particles (PM2.5) were determined using a 7-wavelength Aethalometer for 17 months, between November 2014 and March 2016, for a suburban location of the Metropolitan Area of Buenos Aires (MABA), Argentina. In addition to describing seasonal and diurnal black carbon (BC) cycles for the first time in this region, the relative contributions of fossil fuel and remote and local biomass burning were determined by distinguishing different carbonaceous components based on their effect on light attenuation for different wavelengths. Trajectory analyses and satellite-based fire products were used to illustrate the impact of long-range transport of particles emitted by non-local sources. EBC data showed a marked diurnal cycle, largely modulated by traffic variations and the height of the boundary layer, and a seasonal cycle with monthly median EBC concentrations (in μg/m3) ranging from 1.5 (February) to 3.4 (June). Maximum values were found during winter due to the combination of prevailingly stable atmospheric conditions and the increase of fossil fuel emissions, derived primarily from traffic and biomass burning from the domestic use of wood for heating. The use of charcoal grills was also detected and concentrated during weekends. The average contribution of fossil fuel combustion sources to EBC concentrations was 96%, with the remaining 4% corresponding to local and regional biomass burning. During the entire study period, only two events were identified during which EBC concentrations attributed to regional biomass burning accounted for over 50% of total EBC; these events demonstrate the relevance of agricultural and forestry activities that take place far from the city yet whose emissions can affect the urban atmosphere of the MABA.

Air quality networks need revision and optimisation as instruments and network requirements, both scientific and societal, evolve over time. Assessing and optimising the information content of a monitoring network is a non-trivial problem. Here, we introduce a methodology formulated in a variational framework using an air quality model to simulate the dispersion of carbon monoxide (CO) as a passive tracer at the city scale. We address the specific case of adding or removing stations, and the more general situation of optimally distributing a given number of stations in a domain taking into account transport patterns and spatial factors such as population density and emission patterns. We consider three quality indicators: precision gain, information gain and degrees of freedom for a signal. These metrics are all functions of the singular values of the sensitivity matrix that links emissions and observations in the variational framework. We illustrate the application of the methodology in the case of Santiago (33.5°S, 70.5°W, 500 m a.s.l.), a city of ca. 7 million inhabitants with significant pollution levels. We deem information gain as the best of the above indicators for this case. We then quantify the actual evolution of Santiago's network and compare it with the optimal configuration suggested by our methodology and with results previously obtained using a statistical approach. The application is restricted to diurnal and summer conditions, for which the dispersion model shows a good agreement with observations. The current method offers advantages in that it allows extending a network to include new sites, and it explicitly considers the effects of dispersion patterns, and desired weighting functions such as emission fluxes and population density. We find that Santiago's air quality has improved two-fold since 1988, regarding CO under diurnal summer conditions. Still, according to our results, the current configuration could be improved by integrating more suburban stations in the southwest of the basin.