This paper presents a new algorithm to retrieve the aerosol optical depth (AOD) from a Himawari-8 Advanced Himawari Imager (AHI). Six typical aerosol models that derived from the long-term ground-based observations of East Asia are used in AOD retrieval. To accurately determine the surface reflectance, improved channel relationships between red, blue, and shortwave infrared (SWIR) are built up according to the infrared Normalized Difference Vegetation Index (NDVI_SWIR). Based on the new derived aerosol models and improved channel relationships, AOD over East Asian is retrieved by using the AHI data. The results are compared with Moderate Resolution Imaging Spectroradiometer (MODIS) aerosol products (MOD04 and MYD04) and yielded a correlation coefficient lager than 0.8 (R = 0.87 and 0.92, respectively). In addition, the retrieved AOD values are also validated by ground-based measurements at 12 Aerosol Robotic Network (AERONET) locations and revealed a good agreement between them (R = 0.86).

Remote sensing images captured by satellites play a critical role in Earth observation (EO). With the advancement of satellite technology, the number and variety of remote sensing satellites have increased, which provide abundant data for precise environmental monitoring and effective resource management. However, existing satellite imagery often faces a trade-off between spatial and temporal resolutions. It is challenging for a single satellite to simultaneously capture images with high spatial and temporal resolutions. Consequently, spatiotemporal fusion techniques, which integrate images from different sensors, have garnered significant attention. Over the past decade, research on spatiotemporal fusion has achieved remarkable progress. Nevertheless, traditional fusion methods often encounter difficulties when dealing with complicated fusion scenarios. With the development of computer science, deep learning models, such as convolutional neural networks (CNNs), generative adversarial networks (GANs), Transformers, and diffusion models, have recently been introduced into the field of spatiotemporal fusion, resulting in efficient and accurate algorithms. These algorithms exhibit various strengths and limitations, which require further analysis and comparison. Therefore, this paper reviews the literature on deep learning-based spatiotemporal fusion methods, analyzes and compares existing deep learning-based fusion algorithms, summarizes current challenges in this field, and proposes possible directions for future studies.

Several essential improvements have been made in recent Ozone Monitoring Instrument (OMI) near‐ultraviolet (UV) aerosol retrieval algorithm version (OMAERUV version 1.4.2), but few regional validations for its aerosol optical depth (AOD) product are conducted. This paper assessed the OMAERUV AOD product over Central and East Asia. The OMAERUV Level 2.0 AOD product was compared with Aerosol Robotic Network (AERONET) Level 2.0 direct Sun AOD measurement over 10 years (2005–2014) at 27 selected AERONET sites. A combined comparison of OMAERUV‐AERONET AOD at 25 (2) sites was carried out and yielded correlation coefficient (ρ) of 0.63 (0.77), slope of 0.53 (0.57), y intercept of 0.18 (0.13), and 50.71% (57.24%) OMAERUV AOD fall within the expected uncertainty boundary (larger by 0.1 or ±30%) at 380 nm (440 nm). The more accurate (ρ > 0.70) OMAERUV retrievals are reported over eastern and northern China and South Korea. The two primary reasons for the underestimation of OMAERUV AOD over China are as follows: (1) the use of single‐channel (388 nm) retrieval method retrieves scattering AOD and not total AOD, and (2) the spectral dependence of the imaginary part of the refractive index in the near‐UV region assumed in the algorithm may not be representative of aerosols found over China. The comparisons for three predominant aerosol types indicate that smoke aerosol exhibits the best performance, followed by dust and nonabsorbing aerosol. It is consistent with the characteristic of near‐UV wavelength that it is more sensitive to absorbent particles. The comprehensive yearly (2005–2014) comparison at 25 sites and comparison between two periods (2005–2006 and 2009–2014) at selected four sites show no discernible decrease of temporal trend, which indicates that the OMAERUV algorithm successfully maintains its quality of aerosol product despite post‐2008 row anomaly instrument problem.

Particulate matter (PM_2.5) is a significant public health concern in China, and the Chinese government has implemented a series of laws, policies, regulations, and standards to improve air quality. This study documents the changes in PM_2.5 and evaluates the effects of industrial transformation and clean air policies on PM_2.5 levels in urban and suburban areas of China’s three largest urban agglomerations, Beijing–Tianjin–Hebei (BTH), the Yangtze River Delta (YRD), and the Pearl River Delta (PRD) based on a new degree of urbanization classification method. We used high-resolution PM_2.5 concentration and population datasets to quantify the differences in PM_2.5 concentrations in urban and suburban areas of these three urban agglomerations. From 2000 to 2020, the urban areas have expanded while the suburban areas have shrunk. PM_2.5 concentrations in urban areas were approximately 32, 10, and 7 μg/m³ higher than those in suburban areas from 2000 to 2020 in BTH, YRD, and PRD, respectively. Since 2013, the PM_2.5 concentrations in the urban regions of BTH, YRD, and PRD have declined at average annual rates of 7.30, 5.50, and 5.03 μg/m³/year, respectively, while PM_2.5 concentrations in suburban areas have declined at average annual rates of 3.11, 4.23 and 4.69 μg/m³/year, respectively. By 2018, all of the urban and suburban areas of BTH, YRD, and PRD satisfied their specific targets in the Air Pollution and Control Action Plan. By 2020, the PM_2.5 declines of BTH, YRD, and PRD exceeded the targets by two, three, and four times, respectively. However, the PM_2.5 exposure risks in urban areas are 10–20 times higher than those in suburban areas. China will need to implement more robust air pollution mitigation policies to achieve the World Health Organization’s Air Quality Guideline (WHO-AQG) and reduce long-term PM_2.5 exposure health risks.

This study conducted the first comprehensive assessment of the aerosol optical depth (AOD) product retrieved from the observations by the Advanced Himawari Imager (AHI) onboard the Himawari-8 satellite. The AHI Level 3 AOD (Version 3.0) was evaluated using the collocated Aerosol Robotic Network (AERONET) level 2.0 direct sun AOD measurements over the last three years (May 2016–December 2018) at 58 selected AERONET sites. A comprehensive comparison between AHI and AERONET AOD was carried out, which yielded a correlation coefficient (R) of 0.82, a slope of 0.69, and a root mean square error (RMSE) of 0.16. The results indicate a good agreement between AHI and AERONET AOD, while revealing that the AHI aerosol retrieval algorithm tends to underestimate the atmospheric aerosol load. In addition, the expected uncertainty of AHI Level 3 AOD (Version 3.0) is ± (0.1 + 0.3 × AOD). Furthermore, the performance of the AHI aerosol retrieval algorithm exhibits regional variation. The best performance is reported over East Asia (R 0.86), followed by Southeast Asia (R 0.79) and Australia (R 0.35). The monthly and seasonal comparisons between AHI and AERONET show that the best performance is found in summer (R 0.93), followed by autumn (R 0.84), winter (R 0.82), and spring (R 0.76). The worst performance was observed in March (R 0.75), while the best performance appeared in June (R 0.94). The variation in the annual mean AHI AOD on the scale of hours demonstrates that AHI can perform continuous (no less than ten hours) aerosol monitoring.

Change detection in remote sensing (RS) images typically involves processing and analyzing RS images of the same geographic location captured at different times to identify changes. In recent years, deep learning has seen increased application in the field of RS, particularly in the context of change detection in RS images. Compared to conventional approaches, deep learning-based methods can automatically identify subtle changes in high-resolution RS images with greater accuracy and efficiency. This article provides an overview of deep learning-based change detection research from a variety of perspectives, including data samples, algorithms, computational power, and commercially driven developments. First, this paper compiles datasets for RS change detection, including SAR, multispectral, hyperspectral, and multimodal images. Second, it introduces the process and framework of deep learning-based change detection, along with mainstream deep learning networks employed for image feature extraction. In addition, change detection is described in terms of different detection granularity sizes and different detection methods. Furthermore, this article provides the first analysis of deep learning-based RS change detection from the perspective of computational power and commercially driven developments. This article also discusses the current major challenges and future directions in the field of deep learning change detection. Deep learning-based methods need to overcome the problem of undesirable samples, integrate multimodal RS images, focus on the extraction of multiple change information, and advance the development of foundation models for change detection.