【摘要】 目前,基于国内外发射的各种卫星传感器观测数据的陆地气溶胶光学厚度(AOD)反演方法与应用研究已相对成熟。然而,针对我国国产高空间分辨率(如高分一号)卫星数据的城市小区域尺度的反演与应用则相对较少。因此,选择使用深蓝算法对我国重要的工业型城市——徐州市进行国产GF-1 WFV数据AOD反演与空间特征分析,并利用地基AERONET同期数据进行反演结果验证。研究结果显示,2015—2017年6期的徐州市AOD整体空间格局稳定,均以人口聚集的城区呈高值分布为特征,不随季节变化,说明该地区空气污染源极可能以城市人为排放为主;与AERONET站点同期数据的相关性分析与时序分析结果表明,基于GF-1WFV高空间分辨率卫星数据的徐州市AOD反演结果质量良好。建议今后在大数据支撑下,应充分利用时空数据融合及机器学习等先进技术手段,弥补受云覆盖影响下单时间窗口GF-1卫星数据AOD反演的缺陷,以期达到国产高分卫星的更高时间、更高空间分辨率以及实时监测与预测的更高业务目标。更多还原

【Abstract】 At present, research on aerosol optical depth(AOD) retrieval over land and application is relatively mature, based on observational data from various sensors aboard domestic and foreign satellites. However, the studies of domestic satellite data with high spatial resolution(e.g. GF-1) in the urban or small area scale are few. Here, we selected Deep Blue algorithm for AOD retrieval of GF-1 WFV data and analyze AOD spatial characteristics of Xuzhou city which is one of important industrial city in China. Then, AERONET synchronous data were used for verifying retrieval results. The results showed that: the overall spatial pattern of AOD is stable in Xuzhou during six periods from 2015 to 2017 characterized by population in dense, urban areas with high values and does not change with seasons, which indicated that the main source of air pollution in this area is probably urban man-made discharge; Relevance analysis and time series analysis with AERONET data proved that AOD retrieval of GF-1 in Xuzhou city has good quality. Therefore, we suggested that some advanced technologies, such as spatio-temporal data fusion and machine learning, can supply the gap of AOD retrieval of GF-1 in single time window influenced by cloud cover, so that it can be expected to achieve the higher temporal and spatial resolution, real-time monitoring and prediction of domestic high-resolution satellites.更多还原