李珂,刘朝顺,焦鹏龙.基于NPP/VIIRS夜间灯光遥感数据估算上海地区PM2.5浓度[J].环境科学学报,2019,39(6):1913-1922
基于NPP/VIIRS夜间灯光遥感数据估算上海地区PM2.5浓度
- Estimation of nighttime PM2.5 concentration in Shanghai based on NPP/VIIRS Day_Night Band data
- 基金项目：上海市自然科学基金（No.17ZR1408600）；国家理科基地科研训练及科研能力提高项目（No.J1310028）
- 李珂
- 华东师范大学地理科学学院, 上海 200241
- 刘朝顺
- 1. 华东师范大学地理科学学院, 上海 200241;2. 华东师范大学地理信息科学教育部重点实验室, 上海 200241
- 焦鹏龙
- 华东师范大学地理科学学院, 上海 200241
- 摘要：随着工业化和城市化的迅速发展，中国面临着越来越严重的大气污染危机，以细颗粒物（PM2.5）为标志的区域性大气复合污染问题已经成为我国大气环境污染治理的核心工作.目前，传统的地面监测站监测PM2.5浓度的方法空间分辨率较差且耗费人力物力，而卫星遥感数据监测PM2.5浓度的方法大部分只能应用于白天.由于PM2.5浓度在时间上存在周期性变化，因此利用卫星遥感数据监测的夜间PM2.5浓度可以作为表征PM2.5日变化规律的重要补充.本研究采用Suomi National Polar-orbiting Partnership（Suomi NPP）卫星搭载的可见红外成像辐射计套件（Visible Infrared Imaging Radiometer Suite，VⅡRS）中Day/Night波段（DNB）夜间灯光影像数据，依据大气气溶胶消光原理和站点气象数据，建立PM2.5浓度的预测模型，基于多元线性回归分析对上海地区9个空气质量监测站在2014-2018年冬季无月无云的晴朗夜间的PM2.5浓度值进行估计，并对PM2.5浓度的空间分布进行模拟.结果表明，在研究时段上海地区PM2.5实际浓度与模型估算PM2.5浓度之间R2=0.767，均方根误差（RMSE）为19.210 μg·m-3，验证了VⅡRS/DNB夜间灯光影像数据在估算PM2.5浓度方面有巨大潜力.
- Abstract：With rapid development of industrialization and urbanization, China has currently suffered from increasingly serious air pollution, in particular the fine particulate matter (PM2.5) related haze pollution episodes. Ground-level PM2.5 concentrations are routinely measured via sparsely distributed ground monitors, which is always costly and labor-intensive. Although satellite-based PM2.5 mapping is capable of providing PM2.5 measurements over large spatial area, the output data sets are largely limited to daytime concentrations. To better characterize the daily variability of PM2.5 concentration, this study estimated nighttime PM2.5 concentrations over Shanghai for the period of 2014-2018 in winter using Day/Night band image data (DNB) collected from the Visible Infrared Imaging Radiometer Suite (VⅡRS) on board Suomi National Polar-orbiting Partnership satellite (Suomi-NPP). The results showed that the predicted PM2.5 data agreed well with the co-located ground-based PM2.5 measurements, with a correlation coefficient of 0.767 and root mean square error of 19.210 μg·m-3, thereby indicating large potential of VⅡRS/DNB data in nighttime PM2.5 mapping.
