由光達地面觀測可反演出時間連續且精確的氣膠光學厚度(AOD)，然而光達為單點量測，其空間覆蓋率有限。使用衛星影像反演二維的氣膠光學厚度產品則可提供氣膠空間分佈的資訊。在氣膠遙測反演法中的最佳逼近法(optimal approach)可能會因為地表反射特性造成在郊區及市區有差異。本研究中將檢視Landsat影像用對比法(contrast reduction method)及暗密植被法(dark target)者兩種方法反演台北地區AOD的分佈。針對像都市地表較亮的區域採用頻散係數法(dispersion coefficient method)並加入太陽光譜儀的資料以減少可能的誤差。相較之下，藍光(0.49um)、紅光(0.66um)及2.1um頻道和地表反射率有關係，dark target反演法較適用於濕土壤及植被類形的地區。利用MODIS AOD產品與AERONET(AErosol RObotic NETwork)驗證反演之AOD二維空間分佈的準確度。除此之外，本研究使用Angstrom Exponent及NGAI辨識氣膠種類，探討不同季節在台北地區的氣膠種類變化。;The aerosol optical depth (AOD) can be retrieved accurately with sequential ground based measurements of the direct and diffuse solar radiance. However, spatial coverage and location frequency causes the limitation. Hence, the use of satellite image data is a proper tool for retrieving aerosol optical depth products with more spatial information and the pattern of aerosol distribution. Currently, aerosol remote sensing may enhance our understanding of the optimal approach of AOD retrieval over urban and rural areas could different due to the characteristics of surface reflectivity. In this study, the concepts of contrast reduction and dark target approaches are examined with Landsat image and the observation of sun photometer (AERONET) for integrating AOD distribution over Taipei city in Taiwan. For the areas with bright surfaces such as urban area is applied by dispersion coefficient method with accompany sun photometer in order to reduce considerably errors in the product. In contrast, the dark target algorithm with the relationship of surface reflectance between blue (0.49 μm), red (0.66 μm), and infrared (2.1 μm) spectral bands, is suitable for moist soils and vegetation areas. The retrieval of AOD spatial distribution will be compared with MODIS AOD products and AERONET for verifying the accuracy of the result. Besides, the results suggest that discrimination of aerosol types based on Angstrom Exponent and NGAI method can be efficient to determine the variation of seasonal aerosols from AERONER as well as satellite data.
