本研究目的在於建構利用大地衛星八號(Landsat-8, L-8)搭載之影像儀(Operational Land Imager, OLI)所攝得之影像來對福爾摩沙衛星二號(簡稱福衛二號)上之遙測照相儀(Remote Sensing Instrument, RSI)進行相互輻射校正之方法，以取得正確之輻射轉換係數。福衛二號為我國第一顆光學遙測衛星，於西元2004年5月21日升空。其中搭載之遙測酬載為RSI。福衛二號升空及運作至今已超過十年，超出了其原訂僅為五年之任務壽命。因此，為維持RSI所觀測到之大氣層頂輻射率準確度，須對RSI進行絕對輻射校正。於校正場址之挑選，本研究採用國際上常用之非洲撒哈拉荒漠(Sahara Desert)、美洲索諾蘭荒漠(Sonoran Desert)與美國內華達與加州地區荒漠等場址。並參考頻段相近之OLI影像，將OLI影像提供之大氣層頂輻射率輸入輻射傳送模式中(6SV4.1)，進行大氣校正，以取得地表反射率，並將地表反射率輸入輻射傳送模式中，搭配當時福衛二號之觀測幾何、頻寬等條件，模擬RSI所觀測之大氣層頂輻射率。最後，將RSI影像之灰階值與模擬所得之RSI觀測輻射率進行統計迴歸，即可得出輻射轉換係數。研究結果顯示，於大部分之光學頻道，本文之校正結果與最新一期太空中心之校正結果和福衛二號發射前於實驗室內所測得之輻射轉換係數差異接近，可達±5%以內，代表本研究所建構之絕對輻射校正法具極高之可行性。;FORMOSAT-2 satellite (FS-2) was launched in May, 2004. It is the first Earth observation satellite operated by the National Space Organization (NSPO) of Taiwan. The main payload housed in FS-2 is Remote Sensing Instrument (RSI) with high spatial resolution. Landsat-8 (L-8) is an American Earth observing satellite launched in February 2013 quite recently. The main sensor on L-8 is Operational Land Imager (OLI). For any optical sensors, ensuring the accurate radiance observing is the most important issue for the applications to the scientific researches and environmental monitoring. Since RSI is operated more than 10 years, the optical characters may be altered. Therefore, the goal in this research is to construct a cross-calibration process and validate it by calibrating radiometric coefficients of FS-2 RSI sensor by using L-8 OLI image as a reference. For FS-2 RSI sensor, OLI is not only a new and well calibrated sensor but also use the similar spectral bands and bandwidth which can provide a credible data for calibrating RSI. The desert areas are selected for the cross-calibration in this study, including Sahara Desert in Africa, Sonoran Desert and deserts in Nevada and California in America. Those sites are usually used in other papers as a satellite sensor calibration site. The radiative transfer code, Second Simulation of the Satellite Signal in the Solar Spectrum version 4.1 (6SV4.1) is employed to drive land surface reflectance and the radiance RSI observed on the top of atmosphere. Eventually, the physical gains of RSI can be figured based on the relationship between observed radiance and the digital number. The results indicate that the differences of the radiometric conversion coefficients are quite similar to the dim file provided by NSPO. The differences can reach in ±5% in most spectral bands.
