本研究針對中華衛星一號所酬載的海洋水色照相儀(Ocean Color Imager, OCI)的頻道特性，建立一套處理OCI資料的大氣校正模式。在模式中以NDVI(Normalized Difference Vegetation Index)及OCI中670nm與865nm波段的反射率比值，估算出在晴空海域上這兩波段的海面反照率，進而依衛星觀測量及由LOWTRAN所建立的資料庫中選取適當的氣團特性值。再根據氣團特性與氣溶膠粒徑分佈及氣溶膠單次反照率的關係，推算出每一觀測點的氣溶膠光學厚度及氣溶膠散射量。再利用氣溶膠光學厚度與波長間的Angstron關係，求出每一觀測點上所有波段的氣溶膠散射量及離水面輻射量。 在研究過程中選取觀測時間相近的OCI及SeaWiFS(Sea-viewing Wide Field-of-view Sensor)資料，分別以本研究所建立的大氣校正模式(OCITRAN)及SeaDAS(SeaWiFS Data Analysis System)作業軟體進行大氣校正，並將兩者的演算結果進行比較。顯示OCI的大氣校正演算法所得的離水面輻射量和葉綠素濃度與SeaDAS所推算結果間有很好的一致性。另外從現場實測資料與OCITRAN所得結果的相對誤差量，與SeaDAS結果的誤差量也相近。 從OCI資料所反演出的氣溶膠光學厚度與AERONET(Aerosol Robotic NETwork)和OCI相近的4個波段的實測資料比較，結果顯示兩者間具有良好的相關性及一致性，其相關係數達0.79以上。在沙塵暴發生的期間，OCI資料所反演的氣溶膠光學厚度能忠實地呈現沙塵暴的跡象，所以OCI資料除了可取得來自海水表層的光學資訊，供生光模式推算海水表層的葉綠素濃度含量外，也可提供空氣品質的訊息。 OCITRAN, an atmospheric correction model, specifically designed for the data processing of the ROCSAT-1 Oceanic Color Imager (OCI) is developed in this research. In the model, the concept of the Normalized Difference Vegetation Index (NDVI) and the reflectance values of the OCI channels at 670 and 865nm are used to assess the ocean surface albedo under clear sky conditions. The air mass character can further be determined with the aid of a database constructed from a combination of Lowtran simulations and satellite observations. The aerosol optical depth and aerosol scattering radiance can also be estimated with the aerosol size distribution and single aerosol scattering albedo. Finally, the aerosol scattering and the water-leaving radiance for each OCI channel can be estimated pixel-by-pixel with the Angstrom relationship between the aerosol optical depth and channel wavelength. In order to evaluate the results, several pairs (each pair owned a similar observational time) of the OCI and SeaWiFS (Sea-viewing Wide Field-of-view Sensor) observational data were processed by their respective atmospheric correction model---OCITRAN and SeaDAS (SeaWiFS Data Analysis System), and compared. From the analysis, it showed that there existed a good consistency between the water-leaving radiance and chlorophyll concentration. Moreover, the sea surface truths showed that the margin of error of OCITRAN was similar to that of SeaDAS. Comparisons of four AERONET(Aerosol Robotic NETwork) channels which owned a similar bandwidth with four OCI respective channels were also made. The outcome showed that quite a nice correspondence existed, where the correlation coefficient reach 0.79. This model was also used to monitor the sandstorms that occurred during the year 2000 and 2001. The variations of the aerosol optical depth of the atmosphere during the events were clearly seen. It revealed that the OCI data not only can be employed to estimate the chlorophyll concentration in the oceans but also can provide the information of the air quality.
