High resolution aerosol optical depth retrieval over urban areas from Landsat-8 OLI images

doi:10.1016/j.atmosenv.2021.118591

Atmospheric Environment

Volume 261, 15 September 2021, 118591

https://doi.org/10.1016/j.atmosenv.2021.118591 Get rights and content

Highlights

●
Urban AOD at 30 m spatial resolution are retrieved from Landsat-8 OLI images.
●
Surface anisotropy and aerosol type variation are represented in retrieval.
●
Detailed AOD gradient in urban area could be represented.
●
Aerosol emission sources are identified with high resolution AOD dataset.

Abstract

The satellite-retrieved aerosol optical depth (AOD) provides unique estimation of aerosol loading across a continuous space. However, current AOD products with a relatively coarse resolution (≥1 km) can hardly capture the details in urban areas with large spatial AOD gradients. To address this issue, here we developed a novel retrieval algorithm for retrieving AOD with extra fine spatial resolution (30 m) from Landsat-8 satellite OLI images. In the algorithm, the three land surface reflectance (LSR) estimation schemes and four aerosol types are adopted to improve the retrieval accuracy. The algorithm is applied on Beijing and Wuhan city in China during 2014–2019. Results suggest that the retrieved AOD with the new algorithm exhibits good agreement with the ground-based measured AOD (R² = 0.920), and 81.63% of the AODs fall within the expected error line with a root-mean-square error of 0.112. Moreover, the high-resolution AOD products also successfully identified polluted sources and discrepancy of aerosol loading over different land cover types in two megacities in China, implying its great potential on detecting fine aerosol emission sources over the complex urban area. Such improvement of the new algorithm for retrieving 30 m AOD developed in this study demonstrates its substantial potentials in supporting further studies of air pollution management and human exposure at extra-fine spatial scale.

Graphical abstract

Introduction

Accurate estimation of aerosol loading in the atmosphere is crucial for assessing its impacts on the climate, ecosystem, and human health (Huang et al., 2015; Kaufman et al., 2002; Partanen et al., 2018; Shiraiwa et al., 2017; Stocker et al., 2013; Wang et al., 2014). Aerosol optical depth (AOD) is defined as the cumulative extinction coefficient of the aerosols over the vertical column of a specific cross section. It can be retrieved from ground-based or satellite-based observations. Compared to ground-based remote sensing (Dubovik and King, 2000; Eck et al., 2019; Holben et al., 1998; Welton et al., 2002), the satellite-based remote sensing AOD produce with much wider spatial coverage enables us to observe near-real time aerosol loading thus to monitor the dynamic changes of air pollutants in urban areas.

Satellite image was initiated to retrieve AOD over ocean using NOAA Advanced Very High Resolution Radiometer (AVHRR) visible bands data in the mid-1970s (Fraser, 1976). Later, a series of satellite sensors were used for AOD retrieval and analysis over both land and ocean, including the Visible/Infrared Imager Radiometer Suite (VIIRS) (Jackson et al., 2013), the Moderate resolution Imaging Spectrometer (MODIS) (Hsu et al., 2013; Kaufman et al. 1997a, 1997b; Levy et al., 2010), the Advanced Himawari Imagers (AHI) (Shi et al., 2018), the Geostationary the Operational Environmental Satellite (GOES) (He et al., 2019). Meanwhile, corresponding AOD retrieval algorithms were developed for addressing the retrievals at certain land surface types, including the Dark Target (DT) algorithm (Kaufman et al. 1997a, 1997b; Levy et al., 2010) designed for retrieving AOD at densely vegetation covered area, and Deep Blue (DB) algorithm (Hsu et al. 2004, 2013) designed for bright area with high surface reflectance. Advanced retrieval algorithms was developed to improve the resolution and efficiency, such as improving the resolutions Multi-Angle Implementation of Atmospheric Correction (MAIAC) algorithm (Lyapustin et al., 2011; Martins et al., 2017), and Simplified Aerosol Retrieval Algorithm (SARA) (Bilal et al., 2013). Traditional satellite-based remote sensing AOD product was mostly used to support the aerosol-climate studies over a large spatial scale, thus have spatial resolutions ranging from several kilometers to dozens of kilometers.

Along with the serious air pollution in developing countries like China and India, an urgent need arises for monitoring urban aerosols with large spatial gradients which need high resolution (<1 km) AOD products. The development of higher-resolution satellite sensors like Landsat 5–8 (Chen et al., 2020; Hagolle et al., 2015; Tian et al., 2018; Wei et al., 2017a), Sentinel-2A/B (Li et al., 2019a; Obregón et al., 2019), GF-1 (Sun et al., 2017), HJ-1 (Fan and Qu, 2019; Li et al., 2012) enables us to retrieve AOD at higher spatial resolution to 1 km or even 10 m. For example, Li et al., (2019b) retrieved AOD with the spatial resolution at 30 m and 10 m for Landsat-8 and Sentinel-2A via the modified Land Surface Reflectance Code (LaSRC). Müller-Wilm et al. (Müller-Wilm, 2016) developed the Sentinel Radiative Transfer Atmospheric Correction (SEN2COR) Code to retrieve AOD at 10 m, 20 m, and 60 m with DT method and interpolate to the entire scene for Sentinel-2 image.

However, retrievals of high spatial resolution AOD at urban areas are still challenging due to lack of proper retrieval algorithm. First, the accurate AOD retrieval at the urban area with large aerosol loading requires adequate representation of aerosol optical and microphysical properties (Jeong et al., 2005; Kaufman et al., 1997a), because the aerosol type varies greatly across space and time, especially in urban areas with complex aerosol sources and composition. However, in the current operational AOD products (such as MOD04), the aerosol types are simply assumed based on global spatial clustering analysis (Remer et al., 2005). Such of aerosol types cannot represent the regional aerosol characterizations, resulted in large systemic errors for AOD retrieving. Therefore, frequent variation of aerosol type in urban area should be considered in retrieval by using long-term optical properties from in-situ observations (Wei et al. 2017b, 2019b). Second, accurate characterization of land surface reflectance (LSR) is also important in retrieving AOD at urban areas with complex surface types (Mi et al., 2007) since calculation of atmospheric path reflectance needs to separate the LSR from the top of atmosphere (TOA) reflectance. Previous studies suggested that the error in estimated LSR (Kaufman et al., 1997b) can be enlarged by 10 times in AOD retrievals. The influence of surface anisotropy on the LSR estimation was also suggested to be an important factor particularly for retrieving at urban area with high resolution. However, most previous retrieval algorithms for the high-resolution sensors assumed the surface as lambert and ignored the influence of surface anisotropy.

To address the issues above, here we developed a novel algorithm for retrieving high -resolution (30 m) AOD over urban areas from the Landsat-8 satellites with improved representation of aerosol type and considered surface anisotropy. The study areas and data used in this study are described in Section 2, followed by development of retrieval algorithm in Section 3. The evaluation of the retrieval algorithm and detailed analysis of retrieved AOD products are discussed in Section 4.

Section snippets

Study area

We selected both Beijing and Wuhan cities as our target regions (Fig. 1) by considering their relatively heavy air pollution and complex aerosol composition, which are good examples for examining the applicability of the algorithm at different regions and land surfaces. Beijing (Fig. 1b) as the capital of China, is located in the northern part of North China Plain, and mainly distributed in the southeastern plain with a large amount of farmland in the neighborhood, while the west, north and

Aerosol retrieval methodology

Reliable surface reflectance monitoring and aerosol type assumption are two key factors that determine the accuracy of AOD retrievals. Here we developed a new approach to calculate the LSR at three different surface types, and the aerosol types were estimated into four types based on monthly averaged local aerosol properties observations. A look-up table is built using 6SV radiative transfer model for AOD retrievals. The processes of the cloud mask, gaseous absorption correction, Rayleigh

Validation with ground-based measurement AOD

The ground-based AOD measurements are used to examine the reliability of the AOD retrievals from satellite images as lots of previous studies (Holben et al., 2001; Kaufman et al., 1997b; Wei et al., 2019a). Since the ground-based measurement processing algorithms does not provide AOD at the wavelength of 550 nm, we use the AODs at the wavelength of 440, 675 nm and Ångström exponent provided by the ground-based measurement sites to calculate AOD at 550 nm with the Ångström exponent algorithm.

Conclusions

In this study, we proposed an aerosol retrieval algorithm considering surface anisotropy for Landsat-8 OLI images. With improvements on several key parts in AOD retrievals including surface reflectance estimation, cloud screening, and the selection of aerosol type, the retrieval AODs exhibit higher accuracy compared with the previous AOD products, particularly in urban areas with complex surface types and successfully identified polluted sources and discrepancy of aerosol loading over different

CRediT authorship contribution statement

Hao Lin: Methodology, Formal analysis, Validation, Writing – original draft. Siwei Li: Conceptualization, Methodology, Resources, Supervision, Formal analysis, Funding acquisition, Writing – review & editing. Jia Xing: Methodology, Formal analysis, Writing – review & editing. Tao He: Methodology, Formal analysis, Writing – review & editing. Jie Yang: Data curation, Methodology, Validation, Writing – review & editing. Qingxin Wang: Data curation, Investigation, Methodology, Validation, Software.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgements

This research was funded by the National Natural Science Foundation of China (41975022) and Foundation for Innovative Research Groups of the Hubei Natural Science Foundation (2020CFA003). We are very grateful to NASA, USGS for their free remote sensing data and algorithm code. We would like to thank for Chinese academy of sciences and Tsinghua University provide the GLC_FCS30 and GAIA data.

References (66)

M. Bilal et al.
A Simplified high resolution MODIS Aerosol Retrieval Algorithm (SARA) for use over mixed surfaces
Rem. Sens. Environ.
(2013)
Z. Dong et al.
Regional transport in Beijing-Tianjin-Hebei region and its changes during 2014–2017: the impacts of meteorology and emission reduction
Sci. Total Environ.
(2020)
P. Gong et al.
Annual maps of global artificial impervious area (GAIA) between 1985 and 2018
Rem. Sens. Environ.
(2020)
B.N. Holben et al.
AERONET - a federated instrument network and data archive for aerosol characterization
Rem. Sens. Environ.
(1998)
J. Lee et al.
Characteristics of aerosol types from AERONET sunphotometer measurements
Atmos. Environ.
(2010)
L.L. Li et al.
An improved dark object method to retrieve 500m-resolution AOT (Aerosol Optical Thickness) image from MODIS data: a case study in the Pearl River Delta area, China
ISPRS J. Photogrammetry Remote Sens.
(2014)
Y.J. Li et al.
High-resolution aerosol remote sensing retrieval over urban areas by synergetic use of HJ-1 CCD and MODIS data
Atmos. Environ.
(2012)
Z. Li et al.
Evaluation of Landsat-8 and Sentinel-2A aerosol optical depth retrievals across Chinese cities and implications for medium spatial resolution urban aerosol monitoring
Rem. Sens.
(2019)
C. Lin et al.
Using satellite remote sensing data to estimate the high-resolution distribution of ground-level PM2.5
Rem. Sens. Environ.
(2015)
N. Luo et al.
Improved aerosol retrieval algorithm using Landsat images and its application for PM10 monitoring over urban areas
Atmos. Res.
(2015)

N.C. Hsu et al.

Aerosol properties over bright-reflecting source regions

IEEE Trans. Geosci. Rem. Sens.

(2004)

R.J. Huang et al.

High secondary aerosol contribution to particulate pollution during haze events in China

Nature

(2015)

Cited by (15)

High loadings of carbonaceous aerosols from wood smoke in the atmosphere of Beijing from 2015 to 2017: Implications for energy transition policy
2024, Environmental Pollution
Recently, biomass has been regarded as a promising option for solid energy in China, which is promoted in the residential sector and firing power plants. We collected 200 PM_2.5 samples (particulate matter with a aerodynamic diameter smaller than 2.5 μm) at multi-sites across Beijing from three individual sampling cases from 2015 to 2017. The levels of OC, OC fractions, EC, EC fractions, as well as K⁺ were measured. Then, we adopted the Positive Matrix Factorization 5.0 to apportion the sources of carbonaceous aerosols. The source apportionment results were compared with the estimates of source contribution using the bottom-up technical method with the latest emission inventories after the Action Plan was put into effect in 2013. Our results demonstrate that high pollution of carbonaceous aerosols originated from wood smoking based on the receptor modeling and bottom-up technical method in Beijing from 2015 to 2017. Future energy transition policy should focus on the technologies and regulations for reducing emissions from renewable biomass fuel combustion. This study highlights the importance of regulations that address emissions controls on fuels replacing coal combustion to meet the needs to mitigate air pollution from primary energy use.
Modeling the impacts of 2D/3D urban structure on PM<inf>2.5</inf> at high resolution by combining UAV multispectral/LiDAR measurements and multi-source remote sensing images
2024, Journal of Cleaner Production
Understanding the impacts of 2D/3D urban structure on PM_2.5 is critical for protecting resident health and sustainable development. However, accurate understanding is limited by the lack of high-resolution PM_2.5 simulation in cities with sparse monitoring stations and the simultaneously consideration of different landscape types both in 2D and 3D directions. Using Yinchuan City as a case study, an integrated simulation model combining satellite AOD, NDVI and meteorological data was developed using random forest (RF) algorithm to simulate city-scale PM_2.5 variation at 30 m resolution. Then, the most critical 2D/3D urban metrics in altering PM_2.5 and their impacts were explored through RF analysis. Finally, optimal urban structure was identified using Bayesian Network and multi-scenario analysis. The results indicated that 1) the established RF model could effectively simulate within-city variation of PM_2.5 at 30 m with an R² of 0.75. Thus, it is feasible to use the abundant temporal information and high-resolution AOD to map city-scale PM_2.5 and overcome the limitation of spare monitoring stations. 2) The emission-related metric of distance to polluting enterprise and city center had the largest impacts on PM_2.5, followed by NDVI, building height, impervious surface area and shape, and UGS green volume, area, shape and connectivity. Optimized 3D urban structure for reducing PM_2.5 was proposed from the three aspects of emission source location, UGS pattern and impervious surface. This study deepens the understanding of how 2D/3D urban structure impact PM_2.5 and provides scientific references for optimization possibility of urban air quality in both horizontal and vertical directions.
Direct aerosol optical depth retrievals using MODIS reflectance data and machine learning over East Asia
2023, Atmospheric Environment
Anthropogenic aerosols have rapidly increased since the industrial revolution and are harmful to human health. Moderate Resolution Imaging Spectroradiometer (MODIS) data are critical for retrieving aerosol properties worldwide. However, current MODIS aerosol optical depth (AOD) products require extensive computations and a precalculated lookup table. This study proposes assumption-free high-resolution AOD retrieval models based on the light gradient boosting machine method using MODIS data and ground-based observations over East Asia. The models were developed with three spatial resolutions: 250 m, 500 m, and 1 km. The results showed that 77.8% of the 250 m AOD values were within the MODIS expected error (EE) range, while 76.5%, 76.3%, and 70.08% of the 500 m, 1 km, and Multi-Angle Implementation of Atmospheric Correction (MAIAC) AOD values were within the EE range, respectively. Furthermore, an analysis of the time series and detailed spatial distribution of the proposed model-derived AOD based on data from the Korea–United States Air Quality campaign demonstrated the excellent quality of the 250 m AOD via further validation using a spatially independent dataset. The Shapley Additive exPlanations analysis identified the sensor zenith angle and top-of-atmosphere reflectance of the blue band as the key contributors to the models. In addition, while MAIAC has limited spatial coverage, the spatial frequency of the proposed direct AOD retrieval was nearly 1.5-times higher than that of the MAIAC AOD. Our findings confirmed that machine learning-based high-resolution AOD estimates can be obtained using only satellite data.
72-hour real-time forecasting of ambient PM<inf>2.5</inf> by hybrid graph deep neural network with aggregated neighborhood spatiotemporal information
2023, Environment International
The observation-based air pollution forecasting method has high computational efficiency over traditional numerical models, but a poor ability in long-term (after 6 h) forecasting due to a lack of detailed representation of atmospheric processes associated with the pollution transport. To address such limitation, here we propose a novel real-time air pollution forecasting model that applies a hybrid graph deep neural network (GNN_LSTM) to dynamically capture the spatiotemporal correlations among neighborhood monitoring sites to better represent the physical mechanism of pollutant transport across the space with the graph structure which is established with features (angle, wind speed, and wind direction) of neighborhood sites to quantify their interactions. Such design substantially improves the model performance in 72-hour PM_2.5 forecasting over the whole Beijing-Tianjin-Hebei region (overall R² increases from 0.6 to 0.79), particularly for polluted episodes (PM_2.5 concentration > 55 µg/m³) with pronounced regional transport to be captured by GNN_LSTM model. The inclusion of the AOD feature further enhances the model performance in predicting PM_2.5 over the sites where the AOD can inform additional aloft PM_2.5 pollution features related to regional transport. The importance of neighborhood site (particularly for those in the upwind flow pathway of the target area) features for long-term PM_2.5 forecast is demonstrated by the increased performance in predicting PM_2.5 in the target city (Beijing) with the inclusion of additional 128 neighborhood sites. Moreover, the newly developed GNN_LSTM model also implies the “source”-receptor relationship, as impacts from distanced sites associated with regional transport grow along with the forecasting time (from 0% to 38% in 72 h) following the wind flow. Such results suggest the great potential of GNN_LSTM in long-term air quality forecasting and air pollution prevention.
Evaluation of LaSRC aerosol optical depth from Landsat-8 and Sentinel-2 in Guangdong-Hong Kong-Macao greater bay area, China
2022, Atmospheric Environment
Citation Excerpt :
The mainstream AOD retrieval algorithms are performed on per pixel using the LUT based on the radiative transfer model. It is unavoidable that some outliers occur in the AOD results due to the BRDF (Bidirectional Reflectance Distribution Function), mixed pixels, cloud contamination (e.g., cloud, cloud shadows and cloud adjacent effect) and other factors (Lin et al., 2021a; Su et al., 2021). Hence, it is critical to discard the outliers, which skew the average and calculate the average satellite AODs in a more reasonable way.
The aerosol optical depth (AOD) is a key factor to assess atmospheric environment, and accurate monitoring of it is conducive to the research and environmental protection. The Land Surface Reflectance Code (LaSRC) algorithm has been used to retrieve AOD from fine resolution satellite instruments, including the Landsat-8 and Sentinel-2. This study evaluated the AODs against the in situ data (i.e., Aerosol Robotic Network and Sun-Sky Radiometer Observation NETwork measurements). The temporal window and wind speed were jointly applied to determine the spatial window size (i.e., 6 km) centered on the in situ site. And a three-step strategy for calculating the average of satellite retrieved AODs within the spatial window was proposed in this study. As a result, both STEPs 2 and 3 had improvements compared to STEP 1 and the three-step strategy achieved the best performance in terms of the highest coefficient of determination (R²: 0.754) and within expected error (EE: 60.81%), the lowest root mean square error (RMSE: 0.150) and mean absolute error (MAE: 0.113). Then the contemporaneous AOD matchups were used for validation and evaluation of the LaSRC retrieved AODs. The Sentinel-2 AOD retrievals outperformed the Landsat-8 AOD retrievals with lower RMSE (0.138 and 0.173, respectively), lower MAE (0.102 and 0.136, respectively), higher R² (0.818 and 0.707, respectively) and higher within EE (64.36% and 53.19%, respectively). However, there are 31.91% of Landsat-8 matchups falling below EE and 32.67% of Sentinel-2 collocations falling above EE, which were mainly detected at HKP and HKS sites. And the site-scale validation results revealed that the LaSRC algorithm performed well over inland urban areas, but should be developed further over coastal and water areas in the next generation.
Spatiotemporal retrieval of the aerosol optical thickness using Landsat 8 OLI imagery for Indian urban area
2024, Air Quality, Atmosphere and Health

View all citing articles on Scopus

View full text

High resolution aerosol optical depth retrieval over urban areas from Landsat-8 OLI images

Highlights

Abstract

Graphical abstract

Introduction

Section snippets

Study area

Aerosol retrieval methodology

Validation with ground-based measurement AOD

Conclusions

CRediT authorship contribution statement

Declaration of competing interest

Acknowledgements

Rem. Sens. Environ.

Sci. Total Environ.

Rem. Sens. Environ.

Rem. Sens. Environ.

Atmos. Environ.

ISPRS J. Photogrammetry Remote Sens.

Atmos. Environ.

Rem. Sens.

Rem. Sens. Environ.

Atmos. Res.

Atmos. Pollut. Res.

Atmos. Pollut. Res.

Remote Sens. Environ.

Atmos. Environ.

Atmos. Environ.

Ground-based aerosol climatology of China: aerosol optical depths from the China Aerosol Remote Sensing Network (CARSNET) 2002–2013

Atmos. Chem. Phys.

Large contribution of meteorological factors to inter-decadal changes in regional aerosol optical depth

Atmos. Chem. Phys.

Spatial distribution of aerosol microphysical and optical properties and direct radiative effect from the China Aerosol Remote Sensing Network

Atmos. Chem. Phys.

Retrieval of fine-resolution aerosol optical depth (AOD) in semiarid urban areas using landsat data: a case study in Urumqi, NW China

Rem. Sens.

Assessing PM2.5 exposures with high spatiotemporal resolution across the continental United States

Environ. Sci. Technol.

Atmospheric correction inter-comparison exercise

Rem. Sens.

A flexible inversion algorithm for retrieval of aerosol optical properties from Sun and sky radiance measurements

J. Geophys. Res. Atmos.

AERONET remotely sensed measurements and retrievals of biomass burning aerosol optical properties during the 2015 Indonesian burning season

J. Geophys. Res. Atmos.

Retrieval of high spatial resolution aerosol optical depth from HJ-1 A/B CCD data

Rem. Sens.

Satellite measurement of mass of Sahara dust in the atmosphere

Appl. Opt.

Copernicus sentinel-2A calibration and products validation status

Rem. Sens.

Advancements in the Aerosol Robotic Network (AERONET) Version 3 database - automated near-real-time quality control algorithm with improved cloud screening for Sun photometer aerosol optical depth (AOD) measurements

Atmos. Meas. Tech.

A multi-temporal and multi-spectral method to estimate aerosol optical thickness over land, for the atmospheric correction of FormoSat-2, LandSat, VENμS and Sentinel-2 images

Rem. Sens.

Developing land surface directional reflectance and albedo products from geostationary GOES-R and Himawari data: theoretical basis, operational implementation, and validation

Rem. Sens.

An emerging ground-based aerosol climatology: aerosol optical depth from AERONET

J. Geophys. Res. Atmos.

Enhanced Deep Blue aerosol retrieval algorithm: the second generation

J. Geophys. Res. Atmos.

Aerosol properties over bright-reflecting source regions

IEEE Trans. Geosci. Rem. Sens.

High secondary aerosol contribution to particulate pollution during haze events in China

Nature