高光譜影像拍攝的波段多，因此資料可形成連續的光譜，提供許多 RGB影像及多光譜影像無法提供的資訊。透過機載高光譜得到地表的反射光譜後可對地表做許多的分類應用，然而在觀測目標上方若是有著其他非觀察對象的物體覆蓋便會無法得到所需資料，如在觀測土壤時有許多地區會有植被覆蓋，而非全為裸露地。
本研究選擇位於台南的中石化(台鹼)安順廠作為研究對象，安順廠因為早期作為鹼氯工廠及五氯酚工廠時的操作不當，對於四周環境造成汞及戴奧辛的污染。故本研究利用 CASI-1500 所拍攝的安順廠的機載高光譜影像以及土壤監測資料來分析土壤中的戴奧辛濃度與植物反射光譜之間的關聯性。
透過相關性分析、植生指數進行資料轉換，接著使用多變數線性迴歸模型分析。並搭配觀察不同區域的木麻黃及銀合歡的光譜特徵來探討結果後，可以觀察到植物的反射光譜受戴奧辛影響最為明顯的波段區間落在445.2~531.2nm，該波段範圍亦與葉綠素的吸收波段相符，推測是戴奧辛造成的生長逆境使得植物的光合作用能力變差，也因此也可以透過 PRI 看出光譜間的差異。
但利用該波段範圍的光譜作預測時，結果會受其他貧瘠土地的影響，如受鹽分逆境的海岸防風林雖未受戴奧辛影響但預測值仍偏高。本研究雖未考慮到其他逆境的條件，但若先經過簡單的土壤環境背景調查，本研究的方法應仍可作為傳統方法的輔助，用來更全面的了解污染物的濃度分布情形。

Through airborne hyperspectral images, one can get lots of information from the
surface via the reflectance spectrum. However, some barrier could be exist on top of the target, which could block the essential information for data analysis. For example, vegetation exists on contaminated land blocks the sensor from getting data from soil directly.
The purpose of this study is to find the relationship between reflectance spectrum of plants and the dioxin concentration of soil through airborne hyperspectral images, as a tool to locate places may be polluted.
This study chose CPDC An-Shun Site , located in Annan District, Tainan City, as
research objective for it severely contaminated by mercury and dioxins.
The hyperspectral image used in this study is took by airborne hyperspectral sensor CASI-1500.After atmospheric correction by empirical line, data transformation are done by analysis the correlation to the vegetation index. Multivariate linear regression (MLR) is applied to find out which information fits the situation of ground best.
This study also takes consideration in the spatial difference, finding the same species plants through the Google Street View, then compare the relationship between the spectrum variation of it and the distance from the plant to An-Shun Site.
The result of the MLR shows up the best result of distinguishing when using
445.2~531.2nm bandwidth data as analysis source, with a normalized root mean square(NRMSE) of 0.24. There is difference between same species of plant that were affect by dioxin pollution under different level, the difference is mainly reflected at 450-550nm data, which is wavelength reflecting the Chlorophyll absorption of plants.
In conclusion, the concentration of dioxin in soil mainly affects the reflectance
spectrum of plants at. It's possible to find difference between contaminated and
uncontaminated soil through spectrum, but it's still unable to know the exact amount of dioxin at each target region individually.

Baugh, W. M., & Groeneveld, D. P. (2008). Empirical proof of the empirical line. International Journal of Remote Sensing, 29(3), 665–672.
Dao, P. D., He, Y., & Lu, B. (2019). Maximizing the quantitative utility of airborne hyperspectral imagery for studying plant physiology: An optimal sensor exposure setting procedure and empirical line method for atmospheric correction. International Journal of Applied Earth Observation and Geoinformation, 77, 140–150.
Du, X. Y., Liu, X. Y., Shang, Z. Y., Han, W. S., & Zhang, H. (2021). Detection techniques for monitoring dioxin-like compounds: Latest techniques and the comparison. Journal of Physics: Conference Series, 2045(1).
Fiedler, H. (1996). Sources of PCDDIPCDF and Impact on the Environment. In Chemosphere (Vol. 32, Issue 1).
Gamon, A., Pe uelas, J., & Field, C. B. (1992). A Narrow-Waveband Spectral Index That Tracks Diurnal Changes in Photosynthetic Efficiency* (Vol. 41).
Gitelson, A., & Merzlyakb, M. N. (1994). Quantitative estimation of chlorophyll-u using reflectance spectra: experiments with autumn chestnut and maple leaves. In J. Photochem. Photobiol. B: Bioi (Vol. 22).
Higashio, H., Ippoushi, K., Ito, H., & Azuma, K. (2011). Absorption and Transfer of Dioxins from Soil in Several Vegetables. Horticultural Research (Japan), 10(4), 467–473.
Liu, J., Zhang, Y., Wang, H., & Du, Y. (2018). Study on the prediction of soil heavy metal elements content based on visible near-infrared spectroscopy. Spectrochimica Acta - Part A: Molecular and Biomolecular Spectroscopy, 199, 43–49.
Liu, W., Li, M., Zhang, M., Long, S., Guo, Z., Wang, H., Li, W., Wang, D., Hu, Y., Wei, Y., & Yang, S. (2020). Hyperspectral inversion of mercury in reed leaves under different levels of soil mercury contamination. Environmental Science and Pollution Research, 27(18), 22935–22945.
Manolakis, D., Marden, D., & Shaw, G. A. (2003). Hyperspectral Image Processing for Automatic Target Detection Applications. In LINCOLN LABORATORY JOURNAL (Vol. 14, Issue 1).
Mao, Y., Pankasem, S., & Thomas, J. K. (1993). Photoinduced Oxidative Reactions of Dioxin and Its Chlorinated Derivative on Laponite Surfaces. In Langmuir (Vol. 9).
Newete, S. W., Erasmus, B. F. N., Weiersbye, I. M., Cho, M. A., & Byrne, M. J. (2014). Hyperspectral reflectance features of water hyacinth growing under feeding stresses of Neochetina spp. and different heavy metal pollutants. International Journal of Remote Sensing, 35(3), 799–817.
Schott, J. R., Salvaggio, C., & Volchok, W. J. (1988). Radiometric Scene Normalization Using Pseudoinvariant Features. In REMOTE SENSING OF ENVIRONMENT (Vol. 26).
SENSOR TYPE. (2010). CASI-1500. www.itres.com
Stazi, S. R., Antonucci, F., Pallottino, F., Costa, C., Marabottini, R., Petruccioli, M., & Menesatti, P. (2014). Hyperspectral Visible–Near Infrared Determination of Arsenic Concentration in Soil. Communications in Soil Science and Plant Analysis, 45(22), 2911–2920.
Thomas M. Lillesand, Ralph W. Kiefer, & Jonathan W. Chipman. (2015). REMOTE SENSING AND IMAGE INTERPRETATION.
Xia, J., Chanussot, J., Du, P., & He, X. (2016). Rotation-based support vector machine ensemble in classification of hyperspectral data with limited training samples. IEEE Transactions on Geoscience and Remote Sensing, 54(3), 1519–1531.
Zhao, L., Hu, Y. M., Zhou, W., Liu, Z. H., Pan, Y. C., Shi, Z., Wang, L., & Wang, G. X. (2018). Estimation methods for soil mercury content using hyperspectral remote sensing. Sustainability (Switzerland), 10(7).
中國石油化學工業開發股份有限公司，(2015)，安順廠及二等九號道路東側草叢區 土壤污染整治場址 污染整治第二次變更計畫 （定稿本）.
吳裕民，(2013)，以植生復育技術處理受戴奧辛及汞污染土壤之研究，國立中山大學海洋環境及工程學系研究所博士論文
施介嵐，(2004)，以光譜混合分析法進行台灣地區Master影像之研究，國立交通大學土木工程所碩士論文
曹晉銘，(2013)，以機載高光譜影像偵測小花蔓澤蘭分佈，國立成功大學資源工程所碩士論文
李焱沐、王紹強，(2017)，亞熱帶針闊混交林光化學植被指數與光能利用效率關係研究，RESEARCH, 36(11), 2239–2250.
林群勛，(2003)，都市垃圾焚化廠周界環境中空氣、植物及土壤所含多氯戴奧辛/呋喃之調查研究，國立成功大學環境醫學研究所碩士論文
草佳那子, 森泉美穂子, 土屋一成,上垣隆一. (2007). ダイオキシン類濃度が大きく異なる土壌で栽培したコムギ茎葉中のダイオキシソ類濃度の比較. 日本土壌肥料學雜誌, 78, 61–67
許博行，(2006)，海岸木麻黃林分易衰老原因之探討，台灣林業, 32, 40–44
中石化(台鹼)安順廠整治場址(2022)，查核及驗證數值，https://epb2.tnepb.gov.tw/cpdc/ch/default.asp
農業知識入口網(2022)，植物圖鑑，https://kmweb.coa.gov.tw/index.php