基于深蓝算法的Sentinel-2数据气溶胶光学厚度反演

中高分辨率气溶胶信息对于高精度地表反射率反演以及城市空气环境质量监测具有重大意义,但在城市及稀疏植被等高亮地表区域,气溶胶光学厚度(AOD)的高精度反演一直是定量遥感领域的难点之一。以北京城市区和包头沙漠区为例,利用MODIS地表反射率产品构建先验知识约束条件,基于深蓝算法实现了13景Sentinel-2高亮地表的AOD反演。为验证算法精度,将反演结果与全球气溶胶自动观测网(AERONET)站点实测值、Sentinel-2官方插件Sen2Cor处理结果、Landsat-8反演值作对比。结果表明:①采用深蓝算法反演的AOD值与AERONET实测值具有显著的相关性(R^2>0.9,RMSE=0.056);②无论是沙漠高亮区还是植被较少的城市高亮区,Sen2Cor插件反演的AOD值整景均为固定值,无空间分布,不符合实际情况;③Sentinel-2深蓝算法反演结果与准同步过境的Landsat-8反演的AOD产品在空间分布上具有高度一致性,较好地反映了人类活动特征。相比于目前官方产品,深蓝算法适合Sentinel-2数据高亮区域的气溶胶反演,在绝对精度和空间分布趋势方面均具有明显优势。     

Himawari-8气溶胶产品的验证及应用

为准确分析中国地区气溶胶空间分布与时间变化特征规律,首先利用中国地区9个AERONET（Aerosol Robotic Network）地基站点观测资料对新一代静止气象卫星Himawari-8气溶胶光学厚度（Aerosol Optical Depth, AOD）产品数据进行一致性验证,并在此基础上选取2015年7月至2018年4月Himawari-8逐小时AOD数据分析了中国地区气溶胶光学厚度时空变化特征。结果表明：①Himawari-8 AOD与AERONET AOD之间相关性很高,9个站点的相关系数R在0.64 ~ 0.91之间,拟合曲线斜率k的范围为0.57 ~ 0.68。②Himawari AOD产品与AERONET AOD的相关性在中午时段较其他时段相对较低;北方地区Himawari-8 AOD冬季反演效果与夏季相比较差,南方地区则相反。③中国地区年平均AOD呈东高西低分布,春、夏两季AOD明显高于秋、冬两季,其中夏季最高,春季次之;地区间AOD月变化差异也较大;大部分地区AOD日变化呈现先下降后上升再下降的趋势,AOD最高值出现在午后14 ~ 16时,最低值出现在18时。研究结果为了解中国地区大气气溶胶的时空变化规律和全天时的大气污染监测方法提供新的参考。     

Himawari-8气溶胶产品的验证及应用

为准确分析中国地区气溶胶空间分布与时间变化特征规律，首先利用中国地区9个AERONET（Aerosol Robotic Network）地基站点观测资料对新一代静止气象卫星Himawari-8气溶胶光学厚度（Aerosol Optical Depth, AOD）产品数据进行一致性验证，并在此基础上选取2015年7月至2018年4月Himawari-8逐小时AOD数据分析了中国地区气溶胶光学厚度时空变化特征。结果表明：①Himawari-8 AOD与AERONET AOD之间相关性很高，9个站点的相关系数R在0.64 ~ 0.91之间，拟合曲线斜率k的范围为0.57 ~ 0.68。②Himawari AOD产品与AERONET AOD的相关性在中午时段较其他时段相对较低；北方地区Himawari-8 AOD冬季反演效果与夏季相比较差，南方地区则相反。③中国地区年平均AOD呈东高西低分布，春、夏两季AOD明显高于秋、冬两季，其中夏季最高，春季次之；地区间AOD月变化差异也较大；大部分地区AOD日变化呈现先下降后上升再下降的趋势，AOD最高值出现在午后14 ~ 16时，最低值出现在18时。研究结果为了解中国地区大气气溶胶的时空变化规律和全天时的大气污染监测方法提供新的参考。     

资源一号04星WFI数据气溶胶反演与验证——以北京市及北京周边地区为例

相比目前常用的传感器,资源一号04星(CBERS04)WFI传感器具有较优的时间分辨率、空间分辨率和幅宽优势,对区域气溶胶遥感估算具有较大的应用潜力。针对WFI数据,结合暗像元法和深蓝算法,构建动态查找表参与辐射传输计算,有效地反演出复杂地表上空的气溶胶光学厚度(aerosol optical depth,AOD)。结果表明,所提出的构建动态查找表并集成暗像元法和深蓝算法的方法,能较好地反映实际气溶胶分布情况;基于WFI反演的AOD空间分布与MODIS气溶胶产品基本一致,但具有更广的空间分布,更能反映区域AOD的分布细节;与AERONET站点AOD值间具有显著相关性(r0.94),且67%的反演结果位于误差区间以内,但低值AOD容易出现高估现象。     

Himawari-8气溶胶产品的验证及应用

为准确分析中国地区气溶胶空间分布与时间变化特征规律,首先利用中国地区9个AERONET(Aerosol Robotic Network)地基站点观测资料对新一代静止气象卫星Himawari-8气溶胶光学厚度(Aerosol Optical Depth,AOD)产品数据进行一致性验证,并在此基础上选取2015年7月至2018年4月Himawari-8逐小时AOD数据分析了中国地区气溶胶光学厚度时空变化特征。结果表明:①Himawari-8 AOD与AERONET AOD之间相关性很高,9个站点的相关系数R在0.64～0.91之间,拟合曲线斜率k的范围为0.57～0.68。②Himawari AOD产品与AERONET AOD的相关性在中午时段较其他时段相对较低;北方地区Himawari-8 AOD冬季反演效果与夏季相比较差,南方地区则相反。③中国地区年平均AOD呈东高西低分布,春、夏两季AOD明显高于秋、冬两季,其中夏季最高,春季次之;地区间AOD月变化差异也较大;大部分地区AOD日变化呈现先下降后上升再下降的趋势,AOD最高值出现在午后14～16时,最低值出现在18时。研究结果为了解中国地区大气气溶胶的时空变化规律和全天时的大气污染监测方法提供新的参考。     

从短波红外与红光波段反演华北地区气溶胶

针对MODIS数据监测华北地区陆地气溶胶存在的植被稀少和仪器老化的问题,开展了基于短波红外波段和红光波段地表反射率比值反演气溶胶。讨论了华北地区MODIS传感器红波段和短波红外波段地表反射率的时间变化特征,发现二者之间的比值较为稳定,据此去除地表反射贡献,建立了陆地气溶胶反演算法。利用2016年9月—2017年8月过境华北地区的MODIS数据进行了算法测试和验证,该算法能够同时在植被稠密和稀疏的地区获得反演结果,较好地反映了气溶胶的空间分布。与AERONET北京Radi站和香河站气溶胶产品对比表明,该算法对于城市和乡村区域的应用效果较好,与地面观测结果的相关系数高于0.9,但算法出现了整体低估;分季节验证结果表明,季节之间的差异较小。     

改进的浓密植被法反演Landsat-8 OLI气溶胶光学厚度

针对利用传统浓密植被法难以准确确定地表反射率的不足,分析了Landsat-8 OLI影像红蓝波段的地表反射率与归一化植被指数、散射角对短波红外波段(2.1μm)地表反射率的关系。分析表明,红、蓝波段与短波红外波段(2.1μm)地表反射率之间的比例关系随归一化植被指数和散射角的改变而不同,据此提出构建用于确定红、蓝波段地表反射率的关系模型,用于实现气溶胶光学厚度的反演。选取美国中东部云覆盖较小的地区进行气溶胶反演,使用AERONET站点数据进行反演结果的验证。结果表明,使用该模型得到反演结果与AERONET站点的实测值具有很好的一致性,拟合结果较好;大约70%的数据位于误差线内,反演结果满足精度要求。     

利用Landsat-8 OLI反演大气气溶胶的可见光谱段地表反射率估算

准确估算地表反射率的贡献一直是遥感反演大气气溶胶光学厚度过程中的重点和难点。为了促进Landsat-8OLI传感器在地表参数定量化特别是大气遥感领域的应用,本文提出一种利用OLI 1.6μm、2.2μm短波红外谱段数据估算遥感影像可见光地表反射率的方法。该方法依托于MOD04产品地表反射率估算模式,通过光谱归一化和构建新的短波红外植被指数等过程,建立OLI地表反射率估算模式,通过误差分析发现该模式能够有效地降低由于传感器光谱响应不同对估算结果的影响,对应用在OLI遥感影像的计算结果与同时间同区域MOD04产品地表反射率进行比较,表明其结果有较高的相关性和可靠性。     

基于HJ-1卫星数据反演长江三角洲地区气溶胶光学厚度

探索利用我国HJ-1卫星CCD数据,运用深蓝算法开展长江三角洲地区气溶胶光学厚度反演的可行性,并将结果与其他气溶胶光学厚度产品进行比较。针对HJ-1A和HJ-1B数据,反演结果分别与MODIS气溶胶光学厚度产品以及AERONET地基观测数据进行对比验证。结果表明:深蓝算法得到A星、B星的反演结果与MODIS气溶胶产品呈显著相关,但在数值上普遍高于MODIS产品;反演结果与AERONET站点数据之间的误差介于0.008~0.364之间,研究时段内站点数据缺乏,未对误差进行严格的统计分析。基于深蓝算法的HJ-1卫星数据反演结果虽然在数值上与MODIS气溶胶光学厚度产品存在系统性偏差,但在空间上能够较好地反映长江三角洲地区大气气溶胶分布状况,且具有空间分辨率高的优势。     

基于MODIS影像的PM10反演——以北京市为例

提出了一种从MODIS影像上反演可吸入颗粒物浓度(PM10)的方法。该方法的基础为从MODIS影像上反演得到的3个可见光波段气溶胶光学厚度(AOD)计算的ngstrm-α。ngstrm-α与颗粒物粒径有关,根据ngstrm-α能够得到颗粒物有效半径,进而估算颗粒物浓度。反演的气溶胶光学厚度由AERONET北京站与香河站验证。PM10反演结果由北京市环保局发布的AQI反插得到的PM10(AQI)进行验证。结果表明:从MODIS影像上反演的3个可见光波段AOD与AERONET基站AOD具有良好的相关性,相关系数为0.923,均方根误差为0.149。该方法反演的PM10与PM10(AQI)相关系数为0.794,均方根误差为48.34(μg/m3)。     

Retrieval of Aerosol Optical Depth based on Himawari-8

To obtain Aerosol Optical Depth (AOD) products with high spatial and temporal resolution and high precision in China. We enhance the AOD retrieval algorithm by applying the MODIS red and blue surface re?ectance ratio database in the algorithm. The enhanced algorithm is able to retrieve AOD over both dark and bright surfaces., we retrieve the 10-minute AOD of China from April 2018 to April 2019. The AOD retrievals from the enhanced red-blue ratio algorithm (RB AOD) were validated by the Level 1.5 AERONET (Aerosol Robotic Network) sunphotometer measurements and MOD04_3K AOD , and the retrieval of AOD were compared with the latest AOD product (version 020) released by Himawari-8.The results show that :(1) The AOD retrievals from the enhanced algorithm agreed well with those from the AERONET. Except Baotou station, the correlation coefficient R of the other five stations is between 0.728-0.863, and the percentage of sample points within the range of EE (Expectation of Error) is between 47.7% and 68.6%, which has great advantages over Himawari-8 AOD products.(2) The RB AOD are basically consistent with AERONET AOD in time series trend. the RB AOD results are higher than those of AERONET AOD when AOD > 1.The spring and summer trend of Himawari-8 AOD is relatively consistent with that of AERONET AOD. However, due to the obvious diurnal change of Himawari-8 AOD in autumn and winter, and the diurnal change is relatively large, its trend deviates greatly from AERONET AOD.(3) The spatial distribution of RB AOD is basically consistent with that of MODIS AOD products, and the retrieved results are slightly lower than those of MODIS AOD.In winter, the inversion range of red-blue ratio method is wider than that of MOD04_3K AOD.(4) the red-blue ratio retrieval algorithm has great advantages over Himawari-8 AOD and MOD04_3K in precision and range of retrieval results of high-reflectance surface area in north China in winter and spring.     

Intercomparison of CALIOP,MODIS, and AERONET aerosol optical depth over China during the past decade

ABSTRACT

Aerosol optical depth (AOD) data from Cloud-Aerosol Lidar and Infrared Pathfinder Satellite Observations (CALIPSO) were intercompared and validated against ground-based measurements from Aerosol Robotic Network (AERONET) as well as space-based Moderate Resolution Imaging Spectroradiometer (MODIS) over China during June 2006 to December 2015. This article aims to evaluate CALIOP daytime AOD using MODIS and AERONET AODs. Comparing the AOD between CALIOP and AERONET in different regions over China using quality control flags to screen the AOD data, we find that CALIOP AOD is generally lower than AERONET AOD especially at optical depths over 0.4 likely due to differences in the cloud screening algorithms and general retrieval uncertainty. Comparison between CALIOP AOD and MODIS AOD results show that the overall spatio-temporal distribution of CALIOP AOD and MODIS AOD is basically consistent. As for the spatial distribution, both data sets show several high-value regions and low-value regions in China. CALIOP is systematically lower than MODIS over China, especially over high AOD value regions for all seasons. As for the temporal variation, both data sets show a significant seasonal variation: AOD is largest in spring, then less in summer, and smallest in winter and autumn. A long-term linear trend analysis based on the domain averaged monthly mean CALIOP and MODIS AOD shows agreement among CALIOP and MODIS for the trends over the 10-year period in four regions examined. The trends in AOD derived from CALIOP and MODIS indicate a decline in aerosol loading in China since 2006. It is found from frequency comparison that CALIOP and MODIS AOD generally exhibit a degree of correlation over China. Statistical frequency analysis shows that CALIOP AOD frequency distribution shows a higher peak than MODIS AOD when AOD < 0.4. For the most part, mean MODIS AOD is higher than mean CALIOP AOD. Evaluation of CALIOP AOD retrievals provides the prospect for application of CALIOP data. The intercomparison suggests that CALIOP has systematically underestimated daytime AOD retrievals, especially deteriorating with increasing AOD, and therefore, CALIOP daytime AOD retrievals should be treated with some degree of caution when the AOD is over 0.4.     

京津冀冬季气溶胶光学厚度遥感反演

针对传统的暗像元算法难以满足植被稀疏陆表气溶胶遥感监测需求的问题,提出了冬季植被稀疏的京津冀地区气溶胶光学厚度的遥感反演方法。以2016—2018年连续3年1—2月的AQUA/MODIS L1B数据为数据源,采用暗像元算法与深蓝算法结合的方法对冬季京津冀地区的气溶胶光学厚度进行了遥感监测。使用AERONET数据对结果进行了验证,并与MODIS MYD04_L2暗像元-深蓝气溶胶产品进行了对比。结果表明,该算法在冬季京津冀地区的气溶胶监测效果远好于暗像元算法,并与MODIS气溶胶产品表现出了显著的相关性,且有效监测范围更大、空间分辨率更高。根据连续监测结果,分析了京津冀地区冬季气溶胶光学厚度空间分布特征及其影响因素。     

Comparison of atmospheric aerosol climatologies over southwestern Spain derived from AERONET and MODIS

The southwestern area of Spain, by its geographical and climatological conditions, is a key location for the characterization of atmospheric aerosol properties. The present study is aimed at evaluating the reliability of satellite-based aerosol climatologies, as inferred from level 2 standard aerosol products such as the Terra-MODIS (Moderate Resolution Imaging Radiometer) MOD04 aerosol product, with an application over this region during the period 2000-2008.This evaluation is carried out by means of comparison with ground-based data from the AERONET station of El Arenosillo (Spain, 37.1N, 6.7W), which has been providing continuous data since 2000. The focus of this paper is the climatology of two aerosol optical parameters: the aerosol optical depth (AOD) and the Ångström exponent.AERONET ground-based measurements give an annual mean value of 0.16 ± 0.12 and a median of 0.12 for the AOD, and a mean value of 1.20 ± 0.47 for the Ångström exponent. The seasonal pattern is characterized by two maxima, the most important maximum occurs in summer months, and the other one in late-winter/early-spring. Lowest values appear in fall and winter, however, a local minimum is observed in July which is only detected with the long-term data series.The mean climatological AOD based on AERONET exhibits complex seasonal patterns (i.e. with multiple local extrema), which are not always captured by MODIS-based climatology. MODIS only reproduces low values of the AOD in winter and high values in summer, as well as the local minimum of July which is sharper when using over-land retrievals. The time series of the AOD retrieved from MODIS both over land and ocean are in relatively good agreement with the ground-based measurements, with a monthly overestimation of about 30% on average, and higher differences in spring. Seasonal patterns from MODIS are better reproduced over land than over ocean. The agreement between daily AERONET and MODIS, as assessed by linear regression, gives correlation coefficients above 80% and an intercept bias below 0.03.     

Correcting MODIS aerosol optical depth products using a ridge regression model

Aerosol optical depth (AOD) is an important metric for the concentration of aerosols in the atmosphere. Dark target (DT) algorithm is a widely used physical model to retrieve AOD over land from Moderate Resolution Imaging Spectroradiometer (MODIS) data. However, due to the limitation of surface ‘dark-target’ in some regions and over certain surface types, it does not work very well. In this paper, we propose two hybrid frameworks based on ridge regression (RR) to improve the retrieval accuracy. They are serial and parallel approaches. In both frameworks, the DT algorithm is used as a baseline to derive an initial result, and the bias between the derived AOD and the ground-truth is corrected by the RR model. To validate the effectiveness of the proposed methods, we apply them on 3093 collocated MODIS and Aerosol Robotic Network (AERONET) observations, covering 10 stations at all available time in China. The obtained results demonstrate that the proposed methods can improve retrieval performance compared to the corresponding DT algorithm and the RR model.     

Improved estimation of aerosol optical depth from MODIS imagery over land surfaces

Estimation of aerosol loadings is of great importance to the studies on global climate changes. The current Moderate-Resolution Imaging Spectroradiometer (MODIS) aerosol estimation algorithm over land is based on the “dark-object” approach, which works only over densely vegetated (“dark”) surfaces. In this study, we develop a new aerosol estimation algorithm that uses the temporal signatures from a sequence of MODIS imagery over land surfaces, particularly “bright” surfaces. The estimated aerosol optical depth is validated by Aerosol Robotic Network (AERONET) measurements. Case studies indicate that this algorithm can retrieve aerosol optical depths reasonably well from the winter MODIS imagery at seven sites: four sites in the greater Washington, DC area, USA; Beijing City, China; Banizoumbou, Niger, Africa; and Bratts Lake, Canada. The MODIS aerosol estimation algorithm over land (MOD04), however, does not perform well over these non-vegetated surfaces. This new algorithm has the potential to be used for other satellite images that have similar temporal resolutions.     

Simultaneous non‐linear retrieval of atmospheric profiles and surface skin temperature from MODIS infrared radiances in Taiwan Strait

A non‐linear iterative physical algorithm that simultaneously retrieves atmospheric temperature, water vapour distribution and surface skin temperature from Moderate Resolution Imaging Spectroradiometer (MODIS) longwave infrared radiances is presented. The retrieval algorithm uses clear‐sky radiances measured by MODIS in Taiwan Strait for both day and night, and shows that it is capable of retrieving medium‐scale atmospheric temperature and water vapour. Sea surface temperature is retrieved with an accuracy similar to that achieved by MOD07 products. Evaluation of retrieval total precipitable water vapour (TPW) is performed by comparisons with retrievals from MOD07 products and data from a ground‐based sunphotometer. These show that MODIS retrieval of TPW, in general, agrees with other sounder retrievals of TPW. The total totals index (TTI) distribution retrieved from MODIS data is similar to that from MOD07 products.     

Retrieval of AOD from Sentinel-2 Data based on Deep Blue Algorithm

Medium-to-high resolution aerosol information is of great significance for surface reflectance inversion and urban ambient air quality monitoring. However, the high-precision aerosol optical thickness (AOD) retrieval in bright areas, such as cities and sparse vegetation areas, has long plagued the quantitative remote sensing applications. Taking Beijing urban area and Baotou desert area as examples, using MODIS surface reflectance products to construct prior knowledge constraints, the AOD inversion of 13 scenes Sentinel-2 images in bright areas was realized based on the deep blue algorithm. To verify the accuracy of the algorithm, the result were compared with the Sentinel-2 official algorithm processing result, the Landsat-8 official aerosol products and the ground-measured AOD data from the Global Aerosol Automated Observing Network (AERONET). The results indicate that the retrieved AOD values from deep blue algorithm is significantly correlated with the measured value of AERONET(R² > 0.90, RMSE = 0.056 0), and the AOD spatial distributions are also well consistent with those from Landsat-8, which reflects the characteristics of human activities. But, whether in desert bright area or urban bright area with less vegetation, the AOD values retrieved by Sen2Cor plug-in are fixed, no spatial distribution and do not conform to the actual situation. In general, compared with the current official products, the deep blue algorithm is suitable for aerosol retrieval in high-brightness areas of Sentinel-2 data，and has obvious advantages in terms of estimation accuracy and spatial distribution trend.