A Study On Aerosol Optical Depth Retrieval And Air Quality Forecasting Modeling

Atmospheric aerosol is a suspension of liquid and solid particles in air with radii varying from a few nanometers to more than 100 μm. Aerosols are a component of smog and air pollution and are spatially and temporally inhomogeneous and they may be found far from their sources.Traditionally, air quality has been monitored with networks of ground monitoring stations and the use of models that evaluate emissions and predict changes in air quality at discrete points. Ground-based, in-situ measurements can provide accurate point measurements. However, they are limited in space and incapable of describing transport patterns of air pollutants. Satellite aerosol and air-quality measurements can help to fill in spatial and information gaps where ground-based monitoring stations are not available. Furthermore, satellite measurements provide consistent, repeated monitoring that allow for comparison over time and between areas. The key parameter that relates satellite remote sensing data to near-surface respirable particulate matter(PM10) concentrations is aerosol optical depth(AOD). AOD is a critical physical quantity characterizing atmospheric turbidity, which reflects the degree of weakening to solar radiation by particulate matters in the whole atmosphere. Therefore, AOD is an indicator of particulate air pollution. Retrieval model of PM10 established by employing satellite retrieved AOD can be used to assess the temporal and spatial variation of near surface particulate pollution, overcome the shortcomings of traditional air quality models and assist regional air-quality forecasting.Based on the study and analysis of spectral characteristics of MODIS and currently mature AOD retrieval algorithms, the research focuses on high spatial resolution AOD retrieval algorithm over bright-reflecting surfaces, real-time retrieval method of PM10 concentrations and explores feasibility and reliability of satellite remote sensing in air quality forecasting applications and provides theoretical support in the operational PM10 monitoring and air quality forecasting. The major contents and findings of this research are listed as follows.(1) To overcome the drawbacks of existing MODIS dark target algorithm(bright surfaces and low spatial resolution), fine spatial resolution AOD retrieval algorithm from MODIS top-of-atmosphere(TOA) reflectacne over bright-reflecting surfaces and cloudless scenes is presented in this paper. Surface reflectance of urban areas is usually very bright in the red part of visible spectrum and in the near infrared, but is much darker in the blue spectral region(i.e., wavelength< 500 nm). The research retrieves 500 m × 500 m AOD from MODIS TOA reflectance based on a look-up table(LUT) approach. LUTs are constructed using the 6SV radiative transfer code for calculating the aerosol reflectance as a function of AOD under various sun-viewing geometries and relative humidity. Results show that there is a good agreement between AERONET AOD and 500 m AOD with high correlation and low uncertainty(r=0.923, SD=0.149, N=60). The retrieved AOD has rich spatial details and a continuous spatial coverage and is a key parameter of retrieval and forecasting models of PM10 concentrations.(2) Retrieval and forecasting models of PM10 concentrations using satellite remote sensing data are established in this research. Input to the retrieval model include retrieved 500 m AOD, effective radius derived through ?ngstr?m-α exponents, average extinction efficiency factor obtained by effective size parameter and planetary boundary layer height. Form of the air quality forecasting model is the same to the retrieval model with fine-tuned and corrected inputs. Retrieved and forecasted PM10 concentrations are validated by surface PM10 concentrations culculated from AQI. Comparisons with ground measurements show that the retrieved PM10 concentrations are highly correlated with surface PM10 observations(r=0.839, SD=47.462 μg/m3, N=260). However, accuracy of the focasting model is limited with large uncertainty(r=0.546, SD=39.25 μg/m3, N=260). Further studies will be necessary before satellite data can see more extensive applications in the operational air quality forecasting.