Satellite Aerosol Optical Depth Research Articles

Winter and Summer PM2.5 Land Use Regression Models for the City of Novi Sad, Serbia

In this study, we describe the development of seasonal winter and summer (heating and non-heating season) land use regression (LUR) models for PM2.5 mass concentration for the city of Novi Sad, Serbia. The PM2.5 data were obtained through an extensive seasonal measurement campaign conducted at 21 locations in urban, urban/industrial, industrial and background areas in the period from February 2020–July 2021. At each location, PM2.5 samples were collected on quartz fibre filters for 10 days per season using a reference gravimetric pump. The developed heating season model had two predictors, the first can be associated with domestic heating over a larger area and the second with local traffic. These predictors contributed to the adjusted R2 of 0.33 and 0.55, respectively. The developed non-heating season model had one predictor which can be associated with local traffic, which contributed to the adjusted R2 of 0.40. Leave-one-out cross-validation determined RMSE/mean absolute error for the heating and non-heating season model were 4.04/4.80 μg/m3 and 2.80/3.17 μg/m3, respectively. For purposes of completeness, developed LUR models were also compared to a simple linear model which utilizes satellite aerosol optical depth data for PM2.5 estimation, and showed superior performance. The developed LUR models can help with quantification of differences between seasonal levels of air pollution, and, consequently, air pollution exposure and association between seasonal long-term exposure and possible health risk implications.

Ozone, nitrogen dioxide, and PM2.5 estimation from observation-model machine learning fusion over S. Korea: Influence of observation density, chemical transport model resolution, and geostationary remotely sensed AOD

High-resolution multi-component estimates of ground-level air pollutants are necessary for assessing their impacts to human health, agriculture, and ecosystems. We demonstrate a high-resolution fusion and downscaling approach over South Korea for May 2016 and May 2021. Daily 1 km fine particulate matter (PM2.5), ozone (O3), and nitrogen dioxide (NO2) concentrations are calculated at ground level using a random forest machine learning (ML) algorithm, with predictors including reanalysis meteorology, satellite aerosol optical depth (AOD), and gridded surface fields from chemical transport models (CTM). The ML model is tested for May 2016, coinciding with the Korea-United States Air Quality Study (KORUS-AQ) intensive field campaign, and for May 2021, to allow incorporation of observations from the Geostationary Environment Monitoring Spectrometer (GEMS). In the tests for May 2016, the correlation coefficients (R) and root mean squared errors (RMSE) relative to withheld observations of daily-averaged pollutants in 10-fold cross-validation are promising: 0.93 (5.5 μg/m3), 0.90 (5.5 ppbv), and 0.95 (4.7 ppbv) for PM2.5, O3, and NO2, respectively. Relative performance is assessed for alternate choices of predictors: (a) 80-km global reanalysis Copernicus Atmosphere Monitoring Service (CAMS) vs. 4-km regional Weather Research and Forecasting model coupled with Chemistry (WRF-Chem); (b) AOD polar-orbiting Moderate Resolution Image Spectroradiometer (MODIS) Multi-Angle Implementation of Atmospheric Correction (MAIAC) vs. geostationary GEMS; and (c) variations in surface observation density. This study is among the very first to incorporate both CTM and GEMS AOD for building surface high resolution multiple air pollution predictions over South Korea.

On the added value of satellite AOD for the investigation of ground-level PM2.5 variability

To assess and mitigate PM2.5 pollution, it is imperative to accurately characterize its spatiotemporal variability and distribution. As station measurements are spatially selective, additional data sources such as satellite observations or chemical transport models (CTMs) are indispensable for providing essential complementary information. With the increasing accessibility of high-quality reanalysis datasets on air pollutants from CTMs, the use of satellite-based observations with their limited spatial and temporal availability is more and more put into question. Model simulations present a clear advantage concerning spatiotemporal continuity. However, how reliable are these model simulations and how does the information they provide differ from satellite-derived data? By applying a random forest (RF) approach we derive PM2.5 concentrations for Germany based on observations of aerosol optical depth (AOD) and a variety of other atmospheric parameters and systematically compare the satellite-derived PM2.5 dataset to results from the CAMS regional model system. We discern the differences in the information they provide to demonstrate the added value of the satellite AOD-derived data for the investigation of long-term PM2.5 variability. The development in PM2.5 variability is analyzed on a yearly and seasonal base over the time period from 2018 to 2022, including the COVID-19 lockdown in spring 2020. We can demonstrate a better accuracy of the RF-based AOD model compared to CAMS reanalysis data with R2 values of 0.71 and 0.57, respectively. Moreover, we find that the AOD-based data provides more detail in spatial gradients and can better depict local and short-term events of enhanced PM2.5 pollution.

The atmospheric aerosol spatial distribution and tropical intra-seasonal oscillations over the South Asian region

Intra-seasonal oscillations (ISO) are well known to modulate the weather phenomena which in turn are known to influence the atmospheric aerosol loading. This study investigates how aerosol loading is modulated in ISO spatio-temporal scales over the Indian region using long-term satellite aerosol optical depth data from Moderate Resolution Imaging Spectroradiometer (MODIS) sensor, onboard Terra Satellite. It is shown that Madden-Julian Oscillation (MJO) and Equatorial Rossby waves (ER) have the highest effect (15–20% of the mean) followed by Mixed-Rossby-gravity and Tropical depressions (MT), and Kelvin wave (KE) (5–15%). Further, a dipolar pattern in aerosol loading was observed, with poles over the Arabian Sea and the Bay of Bengal. These variabilities were found to be mainly driven by anomalous winds associated with the ISOs. Similar to aerosol, dipolar signatures in the atmospheric aerosol radiative forcing (ARF) were also observed with clearer patterns. However, the forcing poles are not centered exactly where aerosol poles were observed, indicating the effect of differential properties of aerosols on the aerosol radiative forcing. Quantitatively, at the surface level, modulation in ARF is up to 3 Wm-2 (15%) for MJO and ER, and up to 2 Wm-2 (5%) for KE and MT; in the atmosphere and at the top of the atmosphere, modulation is up to 2 Wm-2 (15%) for MJO and ER, and up to 1 Wm-2 (5%) for KE and MT.

Uncertainties from biomass burning aerosols in air quality models obscure public health impacts in Southeast Asia

Abstract. Models suggest that biomass burning causes thousands of premature deaths annually in Southeast Asia due to excessive exposure to particulate matter (PM) in smoke. However, measurements of surface air quality are sparse across the region, and consequently estimates for the public health impacts of seasonal biomass burning, are not well constrained. We use the nested GEOS-Chem model of chemistry and transport (horizontal resolution of 0.25°×0.3125°) to simulate atmospheric composition over Southeast Asia during the peak burning months of March and September in the moderate burning year of 2014. Model simulations with GEOS-Chem indicate that regional surface levels of PM2.5 (fine particulate matter with a diameter ≤2.5 µm) greatly exceed World Health Organization guidelines during the burning seasons, resulting in up to 10 000 premature deaths in a single month. However, the model substantially underestimates the regional aerosol burden compared to satellite observations of aerosol optical depth (AOD) (20 %–52 %) and ground-based observations of PM (up to 54 %), especially during the early burning season in March. We investigate potential uncertainties limiting the model representation of biomass burning aerosols and develop sensitivity simulations that improve model–measurement agreement in March (to within 31 %) and increase the estimated number of PM2.5-related premature deaths that month by almost half. Our modifications have a much smaller impact on the same metrics for September, but we find that this is due to canceling errors in the model. Compared to PM2.5 simulated directly with GEOS-Chem, PM2.5 derived from satellite AOD is less sensitive to model uncertainties and may provide a more accurate foundation for public health calculations in the short term, but continued investigation of uncertainties is still needed so that model analysis can be applied to support mitigation efforts. Further reduction of uncertainties can be achieved with the deployment of more aerosol measurements across Southeast Asia.

A deep learning-based imputation method for missing gaps in satellite aerosol products by fusing numerical model data

Satellite-based aerosol optical depth (AOD) products are commonly used in various aerosol-related studies, such as aerosol pollution mapping and aerosol-climate interactions. However, these satellite AOD products often suffer from significant missing gaps due to cloud cover and limitations in the retrieval algorithm. To address this issue, some studies take advantage of real-time seamless simulation of numerical models and successfully fill in these gaps by establishing a regression relationship between satellite AOD and numerical model AOD. However, these previous studies usually use satellite AOD retrievals as the regression target, which limits the accuracy of the imputation results by the original accuracy of satellite AOD retrievals and also consumes a considerable amount of time. To overcome these limitations, this study proposes a spatiotemporal imputation model called Bi-ConvRNN, which combines convolutional neural networks (CNN) and bidirectional recurrent neural networks (Bi-RNN). The model takes both satellite AOD retrievals and numerical model AOD data as input and utilizes the weighted mean squared error (MSE) loss function of multiple AOD datasets, e.g., ground-based data, satellite retrievals, and numerical simulation, as the optimization target to improve the imputation accuracy. The proposed model is evaluated using hourly COMS GOCI AOD products. In the independent test set, the AOD results generated by the Bi-ConvRNN model in the region containing GOCI AOD retrievals can break the accuracy of original GOCI AOD products with the accuracy improved from R2 = 0.70 [RMSE = 0.15] to R2 = 0.84 [RMSE = 0.11], and the filling accuracy, e.g. R2 = 0.79, [RMSE = 0.14], in the region without GOCI AOD retrievals are still better than those of the original GOCI AOD retrievals. Additionally, the Bi-ConvRNN model demonstrates satisfactory filling efficiency, requiring only 0.12 s to fill in the missing gaps of hourly GOCI AOD products per day. These results highlight the efficiency and reliability of the proposed model in filling the gaps in satellite AOD products, and the filled AOD results have great potential for further aerosol-related research.

Evaluation of WRF-Chem-simulated meteorology and aerosols over northern India during the severe pollution episode of 2016

Abstract. We use a state-of-the-art regional chemistry transport model (WRF-Chem v4.2.1) to simulate particulate air pollution over northern India during September–November 2016. This period includes a severe air pollution episode marked by exceedingly high levels of hourly PM2.5 (particulate matter having an aerodynamic diameter ≤ 2.5 µm) during 30 October to 7 November, particularly over the wider Indo-Gangetic Plain (IGP). We provide a comprehensive evaluation of simulated seasonal meteorology (nudged by ERA5 reanalysis products) and aerosol chemistry (PM2.5 and its black carbon (BC) component) using a range of ground-based, satellite and reanalysis products, with a focus on the November 2016 haze episode. We find the daily and diurnal features in simulated surface temperature show the best agreement followed by relative humidity, with the largest discrepancies being an overestimate of night-time wind speeds (up to 1.5 m s−1) confirmed by both ground and radiosonde observations. Upper-air meteorology comparisons with radiosonde observations show excellent model skill in reproducing the vertical temperature gradient (r>0.9). We evaluate modelled PM2.5 at 20 observation sites across the IGP including eight in Delhi and compare simulated aerosol optical depth (AOD) with data from four AERONET sites. We also compare our model aerosol results with MERRA-2 reanalysis aerosol fields and MODIS satellite AOD. We find that the model captures many features of the observed aerosol distributions but tends to overestimate PM2.5 during September (by a factor of 2) due to too much dust, and underestimate peak PM2.5 during the severe episode. Delhi experiences some of the highest daily mean PM2.5 concentrations within the study region, with dominant components nitrate (∼25 %), dust (∼25 %), secondary organic aerosols (∼20 %) and ammonium (∼10 %). Modelled PM2.5 and BC spatially correlate well with MERRA-2 products across the whole domain. High AOD at 550nm across the IGP is also well predicted by the model relative to MODIS satellite (r≥0.8) and ground-based AERONET observations (r≥0.7), except during September. Overall, the model realistically captures the seasonal and spatial variations of meteorology and ambient pollution over northern India. However, the observed underestimations in pollutant concentrations likely come from a combination of underestimated emissions, too much night-time dispersion, and some missing or poorly represented aerosol chemistry processes. Nevertheless, we find the model is sufficiently accurate to be a useful tool for exploring the sources and processes that control PM2.5 levels during severe pollution episodes.

Estimation of pan-European, daily total, fine-mode and coarse-mode Aerosol Optical Depth at 0.1° resolution to facilitate air quality assessments

Aerosol Optical Depth (AOD) data derived from satellites is crucial for estimating spatially-resolved PM concentrations, but existing AOD data over land remain affected by several limitations (e.g., data gaps, coarser resolution, higher uncertainty or lack of size fraction data), which weakens the AOD-PM relationship. We developed a 0.1° resolution daily AOD data set over Europe over the period 2003–2020, based on two-stage Quantile Machine Learning (QML) frameworks. Our approach first fills gaps in satellite AOD data and then constructs three components' models to obtain reliable full-coverage AOD along with Fine-mode AOD (fAOD) and Coarse-mode AOD (cAOD). These models are based on AERONET (AErosol RObotic NETwork) observations, Gap-filled satellite AOD, climate and atmospheric composition reanalyses. Our QML AOD products exhibit better quality with an out-of-sample R2 equal to 0.68 for AOD, 0.66 for fAOD and 0.65 for cAOD, which is 23–92 %, 11–13 % and 115–132 % higher than the corresponding satellite or reanalysis products, respectively. Over 91.6 %, 81.6 %, and 88.9 % of QML AOD, fAOD and cAOD predictions fall within ±20 % Expected Error (EE) envelopes, respectively. Previous studies reported that a weak satellite AOD-PM correlation across Europe (Pearson correlation coefficient (PCC) around 0.1). Our QML products exhibit higher correlations with ground-level PMs, particularly when broadly matched by size: AOD with PM10, fAOD with PM2.5, cAOD with PM coarse (R = 0.41, 0.45 and 0.26, respectively). Different AOD fractions more effectively distinct PM size fractions, than total AOD. Our QML aerosol dataset and models pioneer full-coverage, daily high-resolution monitoring of fine-mode and coarse-mode aerosols, effectively addressing existing AOD challenges for further PMs exposures' estimations. This dataset opens avenues for more in-depth exploration of the impacts of aerosols on human health, climate, visibility, and biogeochemical processes, offering valuable insights for air quality management and environmental health risk assessment.

A Deep-Learning and Transfer-Learning Hybrid Aerosol Retrieval Algorithm for FY4-AGRI: Development and Verification over Asia

The Advanced Geosynchronous Radiation Imager (AGRI) is a mission-critical instrument for the Fengyun series of satellites. AGRI acquires full-disk images every 15 min and views East Asia every 5 min through 14 spectral bands, enabling the detection of highly variable aerosol optical depth (AOD). Quantitative retrieval of AOD has hitherto been challenging, especially over land. In this study, an AOD retrieval algorithm is proposed that combines deep learning and transfer learning. The algorithm uses core concepts from both the Dark Target (DT) and Deep Blue (DB) algorithms to select features for the machine-learning (ML) algorithm, allowing for AOD retrieval at 550 nm over both dark and bright surfaces. The algorithm consists of two steps: ①A baseline deep neural network (DNN) with skip connections is developed using 10 min Advanced Himawari Imager AODs as the target variable, and ②Sunphotometer AODs from 89 ground-based stations are used to fine-tune the DNN parameters. Out-of-station validation shows that the retrieved AOD attains high accuracy, characterized by a coefficient of determination (R2) of 0.70, a mean bias error (MBE) of 0.03, and a percentage of data within the expected error (EE) of 70.7%. A sensitivity study reveals that the top-of-atmosphere reflectance at 650 and 470 nm, as well as the surface reflectance at 650 nm, are the two largest sources of uncertainty impacting the retrieval. In a case study of monitoring an extreme aerosol event, the AGRI AOD is found to be able to capture the detailed temporal evolution of the event. This work demonstrates the superiority of the transfer-learning technique in satellite AOD retrievals and the applicability of the retrieved AGRI AOD in monitoring extreme pollution events.

DIRECT AND INDIRECT CAUSAL EFFECTS OF AIR POLLUTANTS ON ALZHEIMER’S DISEASE RISK AND DISPARITIES

Abstract Ambient air contaminants have been associated with higher risk of Alzheimer’s disease (AD), related dementia, and cognitive impairment; however, mechanisms that underlay these effects remain unclear. Pre-existing arterial hypertension, diabetes mellitus, cerebrovascular disease, and renal disease have a strong impact on AD risk and disparities. These diseases, in turn, are associated with exposure to air contaminants, including particulate matter (PM2.5). We applied causal mediation analysis in the formulation of VanderWeele which provides causal estimates of the direct and indirect (through the effects on pre-existing diseases) effects of PM2.5 exposure on the risk of AD. We used Administrative Health Insurance Claims from 5%-Medicare (N=6,042,239; 1991-2020) to identify the date of onset for AD and risk related diseases such as arterial hypertension, diabetes, cerebrovascular, and kidney diseases that were considered mediators in these analyses. Information on ambient PM2.5 at the zip code level were derived from the Socioeconomic Data and Applications Center (SEDAC) that incorporates air monitoring data, satellite aerosol optical depth, meteorological conditions, chemical transport model simulations, and land-use variables. The estimates supported the hypotheses on the linear dependence of the dose-response function with higher effects for Males vs. Females, White vs. Black subpopulations, and stroke-belt states vs. other U.S. regions. We found that the contributions of indirect effects were stronger for younger older adults and reached up to 13% for cerebrovascular disease, 10% for arterial hypertension, 5% for diabetes, and 4% for chronic kidney disease. Analysis in subpopulations demonstrated health disparities in the effects of PM2.5 on AD risk.

Addressing Biases in Ambient PM2.5 Exposure and Associated Health Burden Estimates by Filling Satellite AOD Retrieval Gaps over India.

Ambient PM2.5 exposure statistics in countries with limited ground monitors are derived from satellite aerosol optical depth (AOD) products that have spatial gaps. Here, we quantified the biases in PM2.5 exposure and associated health burden in India due to the sampling gaps in AOD retrieved by a Moderate Resolution Imaging Spectroradiometer. We filled the sampling gaps and derived PM2.5 in recent years (2017-2022) over India, which showed fivefold cross-validation R2 of 0.92 and root mean square error (RMSE) of 11.8 μg m-3 on an annual scale against ground-based measurements. If the missing AOD values are not accounted for, the exposure would be overestimated by 19.1%, translating to an overestimation in the mortality burden by 93,986 (95% confidence interval: 78,638-110,597) during these years. With the gap-filled data, we found that the rising ambient PM2.5 trend in India has started showing a sign of stabilization in recent years. However, a reduction in population-weighted exposure balanced out the effect of the increasing population and maintained the mortality burden attributable to ambient PM2.5 for 2022 (991,058:798,220-1,183,896) comparable to the 2017 level (1,014,766:812,186-1,217,346). Therefore, a decline in exposure alone is not sufficient to significantly reduce the health burden attributable to ambient PM2.5 in India.

Diurnal time representation of MODIS, VIIRS, MISR, and AHI over Asia and Oceania

Retrievals of satellite aerosol optical depth (AOD) products are used for studies of air pollution, energy balance, and climate change. However, few studies have evaluated the diurnal time representation (DTR) of satellite AOD products. In this study, we proposed a statistically valid threshold method to investigate the DTR of Moderate Resolution Imaging Spectroradiometer (MODIS), Visible Infrared Imaging Radiometer Suite (VIIRS), Multi-angle Imaging Spectroradiometer (MISR), and Advanced Himawari Imager (AHI) AOD products by comparing their retrievals to the daily AERONET AOD values and explored the ability of AHI and AERONET to capture daily aerosol variability at 54 sites in Asia and Oceania from July 2015 to December 2019. For polar-orbiting satellites, based on collocations with AERONET measurements, MISR-derived AOD showed high consistency with daily average AERONET AOD, with a correlation coefficient (R) of 0.94 and 80.85% matchups falling into the envelopes of expected error (EE, ± (0.05 + 0.15 × AODAERONET)). Compared with single MODIS products, a combination of satellite data from the Terra and Aqua MODIS dark target (DT) algorithm might obtain a dataset with better DTR (R = 0.94 and %Within EE = 72.10). The satellite-retrieved AOD did a good job to represent the daily mean AERONET AOD, because AERONET AOD averaged within a ± 30 min time window of different satellite overpasses closely mirrored the daily mean AERONET AOD. Furthermore, the seasonal average diurnal variations (SADV) of AERONET AOD were changing insignificant, except for the urban coverage area in the Southern Hemisphere and equatorial climate. For geostationary satellite, the DTR of AHI was relatively poorer in comparison to the other sensors (R = 0.84, %Within EE = 58.00), and AHI AOD was seriously underestimated in South Asia and overestimated in Oceania. Besides, the SADVs of AHI AOD presented several patterns summarized as wave type, unimodal type, stable fluctuation type, and other types, which meant that AHI AOD retrievals did not accurately capture the diurnal variability of aerosols. This study provides a reference for the aerosol community to select satellite AOD products with appropriate DTR.

A new dataset of erodibility in dust source for WRF-Chem model based on remote sensing and soil texture - Application and Validation

The erodibility factor is an essential parameter in the WRF-Chem model that determines the amount of dust emission. However, current models that use erodibility data are assumed static and lack sufficient characterization of dust source regions' highly heterogeneous and dynamic nature. This results in significant errors and uncertainties in the dust simulation. To address this issue, we propose a new approach that involves developing a physically based erodibility dataset using soil moisture, vegetation coverage, soil texture, and land use data. This new dataset covers the entire globe with a 1 km resolution, which can represent dust sources in finer detail. The results of the dust simulations indicated that the new dataset considerably improved the model's overall performance. The evaluation results showed that the root-mean-square error (RMSE) of PM10 simulated with the new erodibility data was reduced by 32.4%, and the correlation coefficient (R) was increased by 82.4% compared to the default data. Additionally, the simulated spatial distribution of aerosol optical depth (AOD) is closer to that of the satellite AOD product. Our new data provide a more accurate description of the erodibility of dust source regions, refine the parameterization of dust emissions in the model, and ultimately improve dust prediction.

Evaluation and Comparison of Multi-Satellite Aerosol Optical Depth Products over East Asia Ocean.

The atmosphere over the ocean is an important research field that involves multiple aspects such as climate change, atmospheric pollution, weather forecasting, and marine ecosystems. It is of great significance for global sustainable development. Satellites provide a wide range of measurements of marine aerosol optical properties and are very important to the study of aerosol characteristics over the ocean. In this study, aerosol optical depth (AOD) data from seventeen AERONET (Aerosol Robotic Network) stations were used as benchmark data to comprehensively evaluate the data accuracy of six aerosol optical thickness products from 2013 to 2020, including MODIS (Moderate-resolution Imaging Spectrometer), VIIRS (Visible Infrared Imaging Radiometer Suite), MISR (Multi-Angle Imaging Spectrometer), OMAERO (OMI/Aura Multi-wavelength algorithm), OMAERUV (OMI/Aura Near UV algorithm), and CALIPSO (Cloud-Aerosol Lidar and Infrared Pathfinder Satellite Observation) in the East Asian Ocean. In the East Asia Sea, VIIRS AOD products generally have a higher correlation coefficient (R), expected error within ratio (EE within), lower root mean square error (RMSE), and median bias (MB) than MODIS AOD products. The retrieval accuracy of AOD data from VIIRS is the highest in spring. MISR showed a higher EE than other products in the East Asian Ocean but also exhibited systematic underestimation. In most cases, the OMAERUV AOD product data are of better quality than OMAERO, and OMAERO overestimates AOD throughout the year. The CALIPSO AOD product showed an apparent underestimation of the AOD in different seasons (EE Below = 58.98%), but when the AOD range is small (0 < AOD < 0.1), the CALIPSO data accuracy is higher compared with other satellite products under small AOD range. In the South China Sea, VIIRS has higher data accuracy than MISR, while in the Bohai-Yellow Sea, East China Sea, Sea of Japan, and the western Pacific Ocean, MISR has the best data accuracy. MODIS and VIIRS show similar trends in R, EE within, MB, and RMSE under the influence of AOD, Angstrom exponent (AE), and precipitable water. The study on the temporal and spatial distribution of AOD in the East Asian Ocean shows that the annual variation of AOD is different in different sea areas, and the ocean in the coastal area is greatly affected by land-based pollution. In contrast, the AOD values in the offshore areas are lower, and the aerosol type is mainly clean marine type aerosol. These findings can help researchers in the East Asian Ocean choose the most accurate and reliable satellite AOD data product to better study atmospheric aerosols' impact and trends.

Biases in estimated ambient PM2.5 exposure and associated health burden in India due to satellite aerosol optical depth sampling gaps

Biases in estimated ambient PM2.5 exposure and associated health burden in India due to satellite aerosol optical depth sampling gaps

Empirical Correlation Weighting (ECW) Spatial Interpolation Method for Satellite Aerosol Optical Depth Products by MODIS AOD over Northern China in 2016

Satellite aerosol products are pivotal in studies of regional air quality and global climate change. Compared with accurate in situ observations, satellite measurements provide valuable large-scale atmospheric information. However, limitations such as clouds and retrieval assumptions result in a significant number of missing values in satellite aerosol optical depth (AOD) products, which severely hampers the representativeness. To address this issue, spatial interpolation of the AOD data is necessary to improve data coverage. In this study, one year of AOD observation data from the MODIS C6.1 version was applied to analyze the spatiotemporal correlated characteristics. The statistical parameters were used as dynamic interpolation weights to develop a novel interpolation method called empirical correlation weighting (ECW) based on MODIS AOD over Northern China in 2016. The ECW interpolation results were obtained at a 0.05° resolution (~5 km). The results showed that the spatial coverage of the Deep Blue (DB) and Dark Target (DT) products increased from 43.88% to 70.65% and from 15.04% to 32.62%, respectively. The reconstruction of the ECW method illustrated good agreement with original values in three cases and in two experimental areas. The mean absolute error (MAE) and root mean square error (RMSE) in the two experiments were 0.1171 and 0.0809, and 0.1212 and 0.0838, respectively, indicating that the ECW exhibited the better accuracy than ordinary Kriging (OK) and Thin Plate Spline (TPS). The AERONET validation results indicated that the values of RMSE and MAE were slightly higher after interpolation compared with those before interpolation, maintaining relatively low values, 0.241 and 0.257, 0.140 and 0.150, respectively.

Enhancing the reliability of hindcast modeling for air pollution using history-informed machine learning and satellite remote sensing in China

Despite the availability of numerous satellite-based machine-learning methods for supplementing air quality monitoring data, models estimating ground-level PM2.5 concentrations have proven unsatisfactory at a daily scale during periods without PM2.5 observations, with validated R2 values for daily estimates in the literature typically below 0.55. In this study, we introduce a novel ensemble learning-based hindcast modeling method that incorporates information on predictors in periods without PM2.5 observations into the modeling process to enhance the reliability of daily estimates for those unmonitored historical years. Our proposed method constructs the annually varying relationship between PM2.5 and the predictors during periods without PM2.5 observations, differing from previous hindcast modeling procedures where the PM2.5-predictors relationship established was assumed to be the same between periods with and without PM2.5 observations. Using PM2.5 heavily polluted northern China from 2013 to 2020 as an example, our history-informed machine-learning hindcast model outperformed the state-of-the-art method by a large margin, improving the leave-one-year-out cross-validation (CV) R2 [ root-mean-square error (RMSE)] from 0.56 [34.66 μg/m3] to 0.63 [31.79 μg/m3] a daily scale, while achieving a comparable sample-based 10-fold CV R2 [(RMSE] of 0.90 [16.85 μg/m3]. In addition, incorporating satellite aerosol optical depth (AOD) into the hindcast modeling further improved the historical PM2.5-predictors relationship and provided more reliable estimates during periods without PM2.5 observations, increasing the leave-one-year-out CV R2 [RMSE] from 0.61 [30.31 μg/m3] for the model without AOD to 0.65 [28.56 μg/m3]. Consequently, our proposed hindcast modeling framework enables the production of more reliable PM2.5 estimates during periods without PM2.5 observations and holds promise for application to other regions and air pollutants with longer predictor data compared to the response variable data.

Markers of economic activity in satellite aerosol optical depth data

This study investigated the impact of COVID-19 lockdowns on satellite aerosol optical depth (AOD), to explore the hypothesis that if changes in economic activity are seen in emissions of NO2, an aerosol precursor, then AOD should change commensurably. We developed a technique to filter AOD data to isolate changes associated with anthropogenic emissions. Overall, in 37 of the 43 cities that were identified as top oxides of nitrogen (NO x ) emitters from their transportation sectors, AODs decreased by 21.2% ± 7.8%, 18.9% ± 11.7%, 27% ± 12.4%, 22.9% ± 7.6% in the United States, India, western Europe, and China, respectively—an average of 22.4% ± 7.4%. In contrast, AODs increased on average by 11.7% ± 8.4% in Taiwan, where economic stimulus was used as a strategy during the pandemic. This analysis implies NO x and volatile organic compounds emissions reductions from the transportation sector can be targeted, and by transitioning 6 million light duty vehicles from gasoline to electricity, the US can achieve 21% improvement in AOD.

Developing high-resolution PM2.5 exposure models by integrating low-cost sensors, automated machine learning, and big human mobility data

Introduction: Traditional methods to estimate exposure to PM2.5 (particulate matter with less than 2.5 µm in diameter) have typically relied on limited regulatory monitors and do not consider human mobility and travel. However, the limited spatial coverage of regulatory monitors and the lack of consideration of mobility limit the ability to capture actual air pollution exposure.Methods: This study aims to improve traditional exposure assessment methods for PM2.5 by incorporating the measurements from a low-cost sensor network (PurpleAir) and regulatory monitors, an automated machine learning modeling framework, and big human mobility data. We develop a monthly-aggregated hourly land use regression (LUR) model based on automated machine learning (AutoML) and assess the model performance across eight metropolitan areas within the US.Results: Our results show that integrating low-cost sensor with regulatory monitor measurements generally improves the AutoML-LUR model accuracy and produces higher spatial variation in PM2.5 concentration maps compared to using regulatory monitor measurements alone. Feature importance analysis shows factors highly correlated with PM2.5 concentrations, including satellite aerosol optical depth, meteorological variables, vegetation, and land use. In addition, we incorporate human mobility data on exposure estimates regarding where people visit to identify spatiotemporal hotspots of places with higher risks of exposure, emphasizing the need to consider both visitor numbers and PM2.5 concentrations when developing exposure reduction strategies.Discussion: This research provides important insights for further public health studies on air pollution by comprehensively assessing the performance of AutoML-LUR models and incorporating human mobility into considering human exposure to air pollution.

Research progress, challenges, and prospects of PM2.5 concentration estimation using satellite data

Satellite data are vital for understanding the large-scale spatial distribution of particulate matter (PM2.5) due to their low cost, wide coverage, and all-weather capability. Estimation of PM2.5 using satellite aerosol optical depth (AOD) products is a popular method. In this paper, we review the PM2.5 estimation process based on satellite AOD data in terms of data sources (i.e., inversion algorithms, data sets, and interpolation methods), estimation models (i.e., statistical regression, chemical transport models, machine learning, and combinatorial analysis), and modeling validation (i.e., four types of cross-validation (CV) methods). We found that the accuracy of time-based CV is lower than others. We found significant differences in modeling accuracy between different seasons ( p &lt; 0.01) and different spatial resolutions ( p &lt; 0.01). We explain these phenomena in this article. Finally, we summarize the research process, present challenges, and future directions in this field. We opine that low-cost mobile devices combined with transfer learning or hybrid modeling offer research opportunities in areas with limited PM2.5 monitoring stations and historical PM2.5 estimation. These methods can be a good choice for air pollution estimation in developing countries. The purpose of this study is to provide a basic framework for future researchers to conduct relevant research, enabling them to understand current research progress and future research directions.