Please use this identifier to cite or link to this item: http://hdl.handle.net/11455/21978
標題: 遙測分析德基水庫之藻華現象
Analysis of Algal Bloom in Techi Reservoir with Remote Sensing
作者: 章國威
Chang, Kuo-Wei
關鍵字: Remote Sensing
遙測
Algal Bloom
藻華
出版社: 生命科學系
摘要: 水庫優養化會促使某些浮游生物產生爆發性繁殖,造成藻華現象。當藻華現象中含有產毒藻類時,則會影響到用水的安全。此外,由於海洋中的微藻是濾食性貝類及甲殼類的主要食物來源,當這些貝類及甲殼類濾食到產毒藻類時,其造成的負面影響將廣及全體食物鏈的成員。因此,有害藻華之監測與研究不但受到全世界湖沼及海洋學家的重視,也是全球共通的議題。 傳統的藻華監測模式,多採現場逢機取樣並將水樣攜回實驗室後再加以分析,故往往僅能獲取點狀式之資料而無法得到整體的水面資料。且此等傳統的監測模式,亦不易達到同時分析比對不同水體於同一時期之水質狀況,即使勉強達成也不易消除不同樣區於水樣分析時所產生之人為誤差。因此建立一套便捷有效又可同時比對不同地區、同一時期之水質狀況的統一監測模式,實有其必要性。衛星遙測資料(如Landsat、SPOT等)具有可定期取得大面積空間分佈資料及可追溯過去時空資料的特殊優點,因此,若能將衛星遙測技術和水質監測模式相結合,將可達到上述預期之目的。 本論文之研究,主要可分為三大部分。第一部分以經顯微鏡檢驗所得之二角多甲藻的數量為地真資料,並依此地真資料來選取訓練區,再以Mahalanobis監督式影像分類法,來分析德基水庫二角多甲藻之藻華區域。所得之藻華影像分析圖與地真調查之結果相近,其皆顯示藻華發生之區域多位於水庫上游區域,且藻華發生之區域及面積會隨季節變化而有所不同。究其原因發現,此乃因為水庫上游集水區多為梨山溫帶水果及高麗菜的主要產區,而這些產區配合農地施作之需要,常隨不同之季節而有不同之施肥作業。因此大量的營養鹽隨著山坡土壤的沖蝕而流入水庫上游,因而提供二角多甲藻大量繁殖之良好環境。此外,由於蔬果之施肥期及其施肥量會隨作物之生長期的不同而有所不同;故造成沖蝕流入水庫之營養鹽量會隨著季節的不同而有差異,因此使得藻華發生之區域面積會隨季節變化而有所不同。 以監督式影像分類法分析德基水庫的藻華現象,較傳統之船隻調查分析法的優點為,可獲取整體水面之藻華資訊而非傳統的點資料。但其缺點為須搭配地真資料來選取訓練區,再以此訓練區進行監督式影像分類;若無調查地點之地真資料庫,則較難進行監督式影像分類。因此,本研究將以Landsat TM資料建立德基水庫之藻華統計預測模式,來預測德基水庫之藻華現象。第二部分以Landsat TM資料利用統計方法,建立德基水庫藻華現象之預測模式。結果顯示,此預測模式預測德基水庫藻華現象發生的準確度可達74.07 %,而影響模式預測準確度的主要因子為光譜與空間解析度。如果我們可以獲取較高之光譜與空間解析度的遙測資料,我們將可得到較高之模式預測準確度。 隨著科技進步之日新月異,人造衛星上之酬載感測儀的空間解析度與波譜解析度也不斷地精進改良。1999年間陸續發射之高解析衛星(如SPIN、IKONOS、QuickBird、OrbView…等),其空間解析度可達約1 m之高解析度;而高光譜衛星亦不斷地在研發與發展中。這些高解析與高光譜衛星,不但可提供我們更清晰之影像,更可提升衛星影像的應用層次及其可行性。葉綠素a為所有藻類之共同色素,當藻華現象發生時,葉綠素a濃度亦會急遽升高,因此可借葉綠素a濃度的變化情形來作為藻華發生之指標。第三部分建立甲藻、矽藻及綠藻之生物量及其藻色素含量與高光譜資料之關連性,從而建立選取高光譜資料的模式,並進而以高光譜資料分析德基水庫之藻華現象。結果指出二角多甲藻、菱形藻與小球藻色素粗萃液之一次微分光譜曲線圖中,660~670 nm之一次微分值之變化與葉綠素a含量有關,且其一次微分值的大小也與葉綠素a濃度的高低相關。故研究中將660~670 nm之一次微分值與葉綠素a含量建立迴歸模式,進而以660~670 nm之一次微分值來估測葉綠素a含量。而該葉綠素a含量估測模式如下: Chlorophyll_a ( mg L-1 ) = 0.92691 * ( 660~670 nm first differential value ) - 0.39266 ........(1) 該葉綠素a含量估測模式之決定係數高達1,而其p值=0.00004遠低於0.005,故達顯著水準。本研究不但建立有效及極具應用價值之葉綠素a含量估測模式,並且可作為將來在應用高解析與高光譜衛星之先驅研究。 由本文所得之研究經驗中,我們認為遙測工具實可應用於水庫藻華監測的工作上。
Abstract The microscopic planktonic algae of the world's oceans are critical food for filter-feeding bivalve shellfish such as, oyster, mussels, scallops and clams as well as the larvae of commercially important crustaceans and finfishes. Some microscopic planktonic algae have the capacity to produce potent toxins that may be transfered through fish and shellfish to humans. On a global base, near 2000 cases of human food poisonings (15% mortality) from consumption of fish or shellfish consumption are reported each year. If not controlled, the economic damage through reduced local consumption and reduced seafood product exports can be considerable. Whales and porpoises can also become victims when they receive toxins through the food chain via contaminated zooplankton or fish. From water quality investigations in the past, dinoflagellate occur blooms are known frequently in the Techi reservoir in summer. Past algal bloom investigations at the Techi reservoir were conducted using boats and point sampling. Sometimes, the data can not represent truly the large areas due to time lacking and few samples per area are available. It is therefore a need of tool that can monitor algal blooms in large areas. Remote sensing data can acquire temporal, large spatial and vast spectral data and also track the past data. Remote sensing, which quantitatively measures the light reflected from the surface of the earth, is a powerful tool for studying regional-scale dynamic ecosystem of aquatic environment. Landsat TM data was used to monitor dinoflagellate blooms in the Techi reservoir with supervised classification in this study. The results afforded us comprehensive algal bloom information of the Techi reservoir instead of the conventional point data. This thesis is divided into three parts. First, Landsat TM data was used to monitor dinoflagellate blooms in the Techi reservoir with supervised classification in 1995 and the predicted accuracies for algal blooms reached higher than 87.5%. From the classification results, dinoflagellate blooms were predicted the most frequent in summer and the least in winter. The phenomenon was associated with the catchments management. The catchments are located on the upper stream of the Techi reservoir. The results afforded us the entire algal bloom information for the Techi reservoir instead of the conventional point data. The algal bloom areas and degree of potential seriousness were defined in this study. Second, we used ratios of logarithm transformed radiance values from Landsat TM data to establish statistical relationships to dinoflagellate densities. The procedure used a forward selection method to develop multiple linear regression models. The selected independent variables matched the dinoflagellate algal cell densities to build the bloom prediction model. The result showed that the bloom prediction model can predict the algal bloom phenomenon with 74.07 % accuracy in this study. The major limits were the spectral sensitivity and spatial resolution of the scanning device. If we can acquire greater spectral sensitivity and spatial resolution in the remote sensing data, we can attain higher accuracy of model analysis. Chlorophyll a is the common pigment in algae, so we can use the Chlorophyll a concentrations as the index of algal bloom phenomenon. Last, we use the hyperspectral data of algal suspensions, algal pigments crude extra and Chlorophyll a concentrations to established the prediction model of Chlorophyll a concentrations. So we can use the Chlorophyll a concentrations to make a judgment of algal bloom phenomenon happen or not. From the research of this thesis, we can find that the remote sensing is a powerful tool in monitoring the algal bloom phenomenon in the Techi reservoir.
URI: http://hdl.handle.net/11455/21978
Appears in Collections:生命科學系所

文件中的檔案:

取得全文請前往華藝線上圖書館



Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.