The Sichuan Basin (Fig. 1), situated in southwestern China, ranks among the country’s four major basins, spanning an area exceeding 260,000 km². Bordered by the Qinghai-Tibet Plateau, Daba Mountains, Huaying Mountains, and Yunnan-Guizhou Plateau, the region exhibits a dramatic topographic gradient, with elevations ranging from 50 meters in low-lying areas to over 5,000 meters in peripheral highlands. The basin experiences a subtropical monsoon climate characterized by distinct thermal gradients: higher temperatures prevail in the eastern and southern sectors, with elevated thermal zones concentrated at the basin’s core and cooler conditions along its margins, resulting in concentric isothermal patterns. Seasonal temperature averages range from 24 ~ 28°C in summer to 4 ~ 8°C in winter. Annual precipitation totals 1,000 ~ 1,300 mm, with pronounced intra-annual variability—70 ~ 75% of rainfall occurs during the summer monsoon season (June–October) (Li et al. 2021; L. Zhang et al. 2019).

The basin hosts 17 major urban centers, including the megacities of Chengdu and Chongqing, each with populations surpassing 20 million. Cities in the southwestern sector exhibit particularly high population density and industrial activity. Rapid socioeconomic development in recent decades has coincided with frequent large-scale aerosol pollution events. Chengdu, the provincial capital, consistently ranks among China’s most air-polluted cities, with PM2.5 concentrations chronically exceeding WHO air quality guidelines. Air quality degradation remains a pressing concern across the basin’s urban centers, as documented in recent studies(Fang et al. 2021; Ning et al. 2018).

CALIPSO was the first satellite capable of monitoring the vertical distribution of clouds and aerosols(Hu et al. 2024). It provides global aerosol horizontal distribution characteristics, vertical profiles, and can automatically identify aerosol types(Hunt et al. 2009; Vaughan et al. 2004; Winker et al. 2009). The CALIPSO satellite is equipped with three instruments: a dual-wavelength (532nm, 1064nm) orthogonally polarized cloud-aerosol lidar (CALIOP), an Imaging Infrared Radiometer (IIR), and a Wide Field Camera (WFC) (Kou et al. 2023; Winker et al. 2007). The CALIOP lidar can collect remote sensing data both during the day and at night, with a revisit time of 16 days. After NASA processes the data, five levels of data products are available: Level 0, Level 1A, Level 1B, Level 2, and Level 3. This study primarily uses the Level 2 5km layer product, which provides 532nm aerosol optical depth information, capturing cloud and aerosol detection data over a 5km horizontal grid(Kim et al. 2018; Young et al. 2018). CALIPSO was officially launched in 2006 and served for nearly 17 years before being decommissioned in 2023.

MODIS, mounted on the Terra and Aqua satellites, is a key instrument in NASA's Earth Observing System (EOS) program for observing global biological and physical processes(Levy et al. 2013; Y. Zhang et al. 2024). MODIS features 36 spectral bands with moderate resolutions ranging from 0.25µm to 1µm, enabling observations of the Earth's surface every 1 to 2 days(Remer et al. 2005). It captures images of land and ocean temperatures, primary productivity, land surface cover, clouds, aerosols, water vapor, and fire events(Habib et al. 2019; Tripathi et al. 2005). In this study, we used MOD/MYD04_3K (MODIS Terra/Aqua Aerosol 5-Min L2 Swath 3km), a NASA Level 2 550nm aerosol product that provides global atmospheric aerosol optical properties (e.g., optical thickness and size distribution) and mass concentration over ocean and land environments. The product uses a lookup table (LUT) to retrieve reflectance and transmission fluxes, along with other quality control and auxiliary parameters, with a spatial resolution of 3 km.

This study utilized all CALIPSO level products during its operational period, covering a total of 16 years from 2007 to 2022 (data from 2006 and 2023 were excluded as they only cover half a year). Similarly, MODIS aerosol product data from the same 16 years were used. Since MODIS remote sensing data is limited to daytime, only daytime CALIPSO data were utilized. Due to differences in spatial resolution between CALIPSO and MODIS aerosol products, the first step was to resample the data. To ensure sufficient spatial coverage for analyzing the entire Sichuan Basin, a 1°×1° geographic grid was selected for nearest-neighbor resampling, and the data range was uniformly matched to 102°~110°E, 27°~33°N. The second step involved calculating annual, monthly, and seasonal averages, followed by analyzing annual and monthly changes in AOD across the entire basin and conducting probability distribution statistics. Scatter plots were then created by matching each grid's monthly average data from the Sichuan Basin with CALIPSO and MODIS monthly averages, followed by linear regression analysis. To further analyze the discrepancies, the differences between the corresponding monthly and seasonal averages of the two datasets were calculated, and the absolute differences were subjected to probability statistics and linear regression. Finally, the regional distribution of annual averages, seasonal averages, and seasonal discrepancies in AOD between the two datasets were visualized to more intuitively reflect distribution differences.

The annual average AOD values for the Sichuan Basin were calculated and linearly fitted, resulting in a trend graph of yearly average AOD values from 2007 to 2022 (Fig. 2). Overall, the AOD values from both MODIS and CALIPSO show a continuous decline. MODIS AOD values dropped from 0.53 in 2007 to 0.3 in 2022, while CALIPSO values decreased from 0.4 to 0.34. The AOD values of MODIS and CALIPSO converged in 2014. Before 2014, MODIS values were higher than CALIPSO's, but the trend reversed after 2014. The linear fit results indicate that the rate of decline in MODIS AOD is greater than that of CALIPSO AOD. The CALIPSO AOD values appear to follow a three-year cycle during the decline, where after three consecutive years of decrease, a rebound occurs. However, the overall trend remains downward.

Figure 3 shows the monthly average AOD values for the Sichuan Basin, with seasonal boundaries indicated by green dashed lines. The MODIS line shows a pattern of "high AOD in winter and spring, low in summer and autumn" in the Sichuan Basin. The peak value appears in March (0.52), and the lowest value in November (0.33). Conversely, the CALIPSO line follows a pattern of "high AOD in summer and autumn, low in winter and spring," with a peak in August (0.45) and a low in December (0.32). MODIS AOD values are higher than CALIPSO's in winter and spring, remaining above 0.4 from January to May, while CALIPSO values stay between 0.35 and 0.4. The largest difference, 0.15, occurs in March during the spring. After June, MODIS AOD values fall below CALIPSO's. Both show a similar trend in summer and autumn, with a rise after July, peaking in August, and then declining to a low at the end of the year. To further compare seasonal differences between MODIS and CALIPSO, we performed a statistical analysis of AOD in the Sichuan Basin at a sampling resolution of 1°×1°. This yielded scatter plots of MODIS and CALIPSO AOD (Fig. 4), which were then linearly fitted. The R² values for the linear fits in spring, summer, autumn, and winter are 0.14, 0.12, 0.12, and 0.27, respectively. This indicates that the overall correlation between MODIS and CALIPSO AOD values is weak, with the correlation being better in winter than in other seasons. Although the correlation is weak, some patterns can still be observed. The slopes of the linear fits are 0.7, 0.58, 0.44, and 0.68, all less than 1, indicating that as CALIPSO AOD values increase, MODIS tends to underestimate them. The intercepts are 0.22, 0.16, 0.19, and 0.16, all greater than 0 and around 0.2, suggesting the presence of systematic errors between CALIPSO and MODIS. These errors likely arise from sensors and algorithms. When AOD values are low, MODIS values tend to be higher than CALIPSO's.

Figure 5 shows the frequency distribution of all AOD pixels, revealing that the AOD values in the Sichuan Basin are mainly distributed below 0.8. In the ranges of 0.1–0.2 and above 0.6, MODIS has a higher frequency distribution than CALIPSO, indicating that MODIS is more sensitive to both low and high values compared to CALIPSO. Conversely, CALIPSO's AOD values are more prominently distributed in the 0.2–0.6 range, suggesting that MODIS performs better in extreme weather conditions than CALIPSO.

Figure 6 shows the absolute value of the difference between CALIPSO and MODIS AOD. In Fig. 6a, a scatter plot of CALIPSO AOD against the absolute deviation is presented, with a linear fit resulting in an R² of 0.007. The correlation is very weak, and the slope is nearly zero, indicating that the magnitude of AOD has little relationship with the difference between CALIPSO and MODIS. Figure 6b presents the frequency distribution of the absolute deviations, with over 60% of the deviations falling within the 0-0.2 range. The larger the deviation, the less frequently it occurs, indicating that the AOD deviation between CALIPSO and MODIS is relatively stable, with most differences being less than 0.2.

Figure 7 shows the spatial distribution of average AOD in the Sichuan Basin from 2007 to 2022, with city coordinates and boundaries marked on the map. Figure 7a shows the spatiotemporal distribution of CALIPSO AOD. High AOD values are concentrated in southwestern urban areas, with the highest value of 0.63 in Leshan City, followed by Zigong, Neijiang, and Yibin, where AOD values range from 0.5 to 0.6. In the western basin, due to the high altitude of the Qinghai-Tibet Plateau, AOD ranges from 0.2 to 0.3. In the eastern mountainous areas, AOD increases to around 0.3 to 0.4 due to altitude and population density. Figure 7b shows the spatiotemporal distribution of MODIS AOD. In high-value areas, MODIS AOD is higher than CALIPSO, with maximum values near Leshan and Meishan approaching 0.9. Zigong, Neijiang, and Yibin have values around 0.7 to 0.8, while Deyang and Mianyang also exhibit high values around 0.8. In low-value areas, MODIS AOD is lower than CALIPSO, with AOD in the western plateau generally below 0.2. Overall, the spatial distribution patterns of CALIPSO and MODIS are generally consistent, showing a trend of "high AOD in the basin's interior, low on the periphery; higher values in the western urban areas compared to the eastern, and lower values in the western mountains compared to the eastern."

The spatiotemporal distribution of AOD was analyzed in two-year intervals from 2007 to 2022. Figure 8 shows the spatiotemporal distribution of CALIPSO AOD. It can be observed that before 2018, there were some gaps in the Sichuan Basin due to the satellite's orbital path. This issue was improved after the satellite's orbit was lowered in 2018. High AOD values in the basin appeared in the Yibin area from 2007 to 2008 and in the Leshan area from 2013 to 2014, both approaching 0.8. Figure 9 shows the spatiotemporal distribution of MODIS AOD, with the highest values observed in 2011–2012. The AOD in the western urban clusters of the basin, including Meishan, Deyang, Mianyang, Zigong, and Leshan, were all close to 1.2. Starting from 2015, there was a significant decline in AOD values within the basin, with noticeable improvement in the western urban clusters.

Figure 10 shows the seasonal spatiotemporal distribution of AOD in the Sichuan Basin. For CALIPSO AOD, high values in spring are mainly concentrated in Zigong, Yibin, and Leshan, ranging from 0.6 to 0.7. In the northeastern Basin cities, AOD ranges from 0.4 to 0.5. In summer, AOD across the basin generally increases to between 0.5 and 0.6. In autumn, most AOD values in the basin remain around 0.5 to 0.6, with AOD increasing in the southwestern cities while beginning to decrease in the surrounding mountainous areas. In winter, overall AOD is higher than in other seasons, with AOD in the western urban clusters mostly exceeding 0.6. Looking at the seasonal distribution of MODIS AOD, it is noteworthy that spring AOD values in the basin are significantly higher than in other seasons, with the highest values in the western urban areas approaching 1. This is not as evident in CALIPSO's seasonal distribution. In summer, AOD decreases compared to spring, and further declines in autumn, especially around the periphery of the basin. In winter, MODIS AOD in the western cities of the basin is also relatively high, with Zigong, Neijiang, and Meishan reaching around 0.9.

To further analyze the differences between MODIS and CALIPSO AOD, we calculated the AOD difference based on the seasonal spatiotemporal distribution and created a plot (Fig. 11). The bias in the figure is calculated by subtracting CALIPSO AOD from MODIS AOD. The figure shows that the difference between the two is quite significant in spring, with the maximum bias reaching 0.8 in the Meishan and Chengdu areas. In other urban areas, the difference is also above 0.2. In summer, the bias improves considerably compared to spring, with deviations of about 0.3–0.4 in areas such as Deyang, Mianyang, western Chengdu, and Chongqing. In autumn, the bias is smaller, with urban areas showing deviations below 0.2. In winter, except for parts of Meishan and western Chengdu where the bias is around 0.4, the deviations in other urban areas are not particularly large. Additionally, we found that the bias in the plateau and mountainous areas surrounding the basin tends to be negative, indicating that CALIPSO tends to overestimate in low-value regions.