用户名: 密码: 验证码:
珠江口海域浮游植物叶绿素-α浓度遥感反演模型研究
详细信息    本馆镜像全文|  推荐本文 |  |   获取CNKI官网全文
摘要
改革开放以来,经济飞速发展,珠江口海域环境与资源受到严重的影响,环境污染日益严重,工业与生活排污量日益增大,导致珠江口海域水质不断下降;浮游植物浓度不断升高,导致大面积的赤潮频频出现,不仅影响了附近海洋生物的生存,甚至威胁到当地居民的生活环境。叶绿素是浮游植物进行光合作用的主要色素,其浓度变化反映了水体中浮游植物的浓度、生物量及其分布状况,是反映海洋水质状况的一个重要的生物指标,同时也是海洋富营养化评价中最为重要的指标之一。
     本文以珠江口近岸水域中浮游植物的叶绿素浓度作为主要研究对象,基于多源遥感影像对浮游植物叶绿素浓度进行反演,采用BP人工神经网络算法,研究海洋藻类叶绿素浓度的遥感反演方法,为近岸水域生态环境监测提供依据。
     本文把光学遥感和雷达遥感影像相结合,建立了海洋藻类叶绿素浓度的遥感反演BP人工神经网络模型。研究工作包括:(1)基于CCD影像对浮游植物叶绿素的光谱特征进行分析,选取了CCD数据的CCD1、CCD2、CCD3波段作为特征波段;(2)对雷达影像中的后向散射系数以及利用Cloude-Pottier分解原理对经过预处理后的图像进行非相干目标分解,得到平均散射角α、散射熵H等参数,并与藻类叶绿素浓度进行相关分析,确定了HH、VV极化下的后向散射系数以及平均散射熵H作为预输入参数;(3)通过对以上六个参数进行不同的组合来建立各种模型进行分析,确定了3层BP网络模型作为本研究的最终模型,然后建立各参数组合与实测叶绿素浓度的线性模型,将得到的预测值与实测值进行拟合,与BP人工神经网络模型的精度进行比较发现BP人工神经网络模型精度最高。
     研究结果表明:(1)浮游植物叶绿素浓度与CCD各波段光谱反射率以及雷达影像的后向散射系数以及平均散射熵H之间有一定的相关性;(2)单独使用一种数据建立的线性模型以及BP人工神经网络模型的精度都不如将光学数据与雷达数据结合使用时的精度高。(3)BP人工神经网络模型的自适应组织能力能够很好的模拟叶绿素a浓度与遥感参数之间的复杂的非线性关系。
Since the reform and opening up, the economic developments quickly, which hadbeen seriously affected the environment and resources of the Pearl River Estuary,worsening environmental pollution, industrial and domestic sewage volume isincreasing, leading to the quality of Pearl River Estuary water is declining; theconcentrations of phytoplankton rising, resulting in big area of red tide frequentlyhappened, not only affected the nearby marine life, even became a threat to the livingenvironment of local residents. Chlorophyll a is the main pigment of phytoplanktonphotosynthesis, the concentration and dynamics of which reflected the concentration,biomass and distribution law of phytoplankton in water, which is an objectivebiological indicators reflect the situation of marine water quality, but also is one of themost important indicators of the eutrophication assessment of marine.
     In this paper phytoplankton chlorophyll concentration of the pearl river mouth asthe main research object, various types remote sensing image based on source ofphytoplankton chlorophyll-a concentration inversion, the BP artificial neural networkalgorithm, the research of Marine algae chlorophyll remote sensing of theconcentration of the inversion method and the offshore waters for ecologicalenvironment monitoring provides the basis.
     In this study, by combining optical remote sensing and radar remote sensingimage to extract remote sensing inversion of marine algae chlorophyll-a concentrationof BP artificial neural network model. Research work after a few stages:(1) based onCCD image of phytoplankton chlorophyll spectrum characteristics are analyzed, theselection of the CCD data CCD1, CCD2, CCD3band as the characteristic band;(2) ofthe radar image to the scattering coefficient and use after Cloude-Pottierdecomposition principle through the image preprocessing the incoherent targetdecomposition, get average scattering Angle alpha, scattering entropy H with algaeparameters such as chlorophyll concentration correlation analysis between, final HH,VV polarization of backscatter coefficient and average scattering entropy H;(3) basedon the six parameters are different combinations to establish various model isanalyzed and determined the three layers of BP neural network model as the finalmodel, then build up each parameter combination with actual chlorophyllconcentration of linear model, will get the predicted values and, by fitting and the BPneural network model accuracy of comparison BP artificial neural network model has the highest.
     The results showed that:(1) between the phytoplankton chlorophyllconcentration and CCD each band spectrum reflectance and radar image to thescattering coefficient and average after scattering entropy H has certain correlation;(2)when using a combination of optical data and radar data building BP artificial neuralnetwork model, which’s precision higher than used alone one data set up linear model.(3) BP artificial neural network model of the adaptive ability of organization can verygood simulation chlorophyll concentration and remote sensing between theparameters of the complicated nonlinear relation.
引文
[1]Liu Mei-ling,Liu Xiang-nan,Li Mi.Neural-network model for estimatingleafchlorophyll concentration in rice under stress from heavy metals using fourspectral indices,biosystems engineering.2010,6:223-233
    [2] Jiang Wensheng, Pohlmann Thomas, Sun Jun, Starke Mdreas.and numericalmodeling.VolSPMtransport in the Bohai Sea: field experiments44,2004,No.3-4, pp175-188.
    [3]MarlinS.AnIntroduce to Ocean Remote Sensing[M].Beijing:Oceanic Press,2008.
    [4] CEOS.The Use of Earth Observing Satellites for Hazard Support:Assessments andScenarios[R].Final Report of DisasterManagement Support Group of CEOS.
    [5] NASAHeadquarter.NASA'sSmallSatelliteMissionsforEarthObservation[R].2004,9,Washington D.C.
    [6] NASA.http://earth.nasa.gov/history/index.html.
    [7] IsaksenL,Stoffelen A.ERS Scatter meter Wind Data Impacton ECWMFs TropicalCyclone Forecasts[J].IEEE Transaction on Geoscience andSensing,2000,38:1885-1892.
    [8] Atlas R,Bloom S C,Hoffman R N,et al.Geophysical Validation of NSCAT WindsUsing Atmospheric Data and Analyses[J].Journal of GeophysicsResearch,1999,104:11405-11424.
    [9] Chang C P,Lin S C,Liou C S,et al.An Experiment UsingNSCAT Winds inNumerical Predication of Tropical Mesoscale Rainfall Systems under theInfluence of Terrain[J].Geophysical Research Letters,1999,26:311-314.
    [10]Figa J,Stoffelen A.On the Assimilation of Ku-band Scatter meter Winds forWeather Analysis and Forecasting [J].IEEE Transaction on Geoscience andRemote Sensing,2000,38:1893-1902.
    [11]JIANG X W,TANG J W,ZHANG M W el al.Application of MODIS data inmonitoring suspended sediment of Taihu Lake, China[J].Chinese Journal ofOceanology and Limnology,2009,27(3):614-620.
    [12]KEINER L E,YAN X-H.A Neural Network Model for Estimating Sea SurfaceChlorophyll and Sediments from ThematicMapper Imagery[J].Remote Sensing ofEnvironment,1998,66:153-165.
    [13]WARRICK J A,MERTES L A K,SIEGEL D A.Estimating suspended sedimentconcentrations in turbid coastal waters of theSanta Barbara Channel withSeaWiFS[J].International Journal of Remote Sensing,2004,25:1995-2002.
    [14]Cracknell. Remote sensing techniques in estuaries and coastal zones—anupdate.Int J of Remote Sensing.1999,19(3):485-496
    [15]MILTON L A K,SMITH M O,ADAMS J B.Estimating suspended sedimentconcentrations in surface waters of the AmazonRiver wetlands fromLandsat[J].Remote Sensing of Environment,1993,43:281-301.
    [16]KHORRAMS,CHESHIREHM.RemotesensingofwaterqualityintheNeuseRiverEstuary,NorthCarolina[J]PhotogrammetrieEngineering&remoteSensing.1985,51(1):53-62.
    [17]Zhang Xu-qin,Zhang Shi-kui,Wu Yongsen,et al.Progress inresearch onyellow-substance in seawater.Journal of Ocean-ography of Huanghai&BohaiSeas.2000,18(1):89-92
    [18]Aure′lio, Azevedo, Barreto-Neto.Application of Fuzzy Logic to the Evaluationof Runoff in a Tropical Watershed.Environmental Modeling&Software23.2007:244–253
    [19]TASSANS.Animproved in wateralgorithmforthedeterminationofthedeterminationofchlorophyllandsuspendedsediment concentrationfromThematicMapperDataincoastalwaters[J].IntRemotesensing,1993,14(6):1221–1229
    [20]GoulonHR, MorelA.RemoteAssessmentofoceanColorforInterpretationofsatelliteVisibleImagery[M].AreviewlectureNotesonCoastalandEstuarineStudies,New York,1983:114
    [21]Sathyendranath S.RemoteSensingofOceanColorinCoastalandOtherOpticall-yComplex waters[R].IOCCG Report Number3,2000,Canada
    [22]Xiangnan Liu,Nan Jiang.Application of Fuzzy Reasoning to Assessment of CropStress Level Based on MODIS Data:A Focus on Heavy Metal Pollution.TheInternational Archives of the Photogrammetry.Remote Sensing and SpatialInformation Science.2008:347—352
    [23]Morel A, Prieur L. Analysis of variations in ocean colour [J]. LimnolOceanogr,1977,22:7092722.
    [24]Lee Z P,Carder K L,Mobley C D,et al.Hyperspectral remotesensing for shallowwaters:2.Deriving bottom depths and waterproperties by optimization.AppliedOptics.1999,38:3831-3843
    [25]Buekton MongainE.The use of neural networks for the estimation of oceaniee onstituents based on MERIS instruct.In Remote Sense,1999,20(9):1841-1851
    [26]Keiner LE, Yan XH.Aneuaral network model for estimating sea surfacechlorophyll land sediment from The matie Maper imager.Remote SensingEnvironment.1998,66:153-165
    [27]Ohde T,Siegel H.Correction of bottom influence in ocean colour satellite imagesof shallow water areas of the Baltic Sea.International Journal of RemoteSensing.2001,22:297-313
    [28]Doerffer R,Schiller H.Determination of Casez water consituents using radiativetransfer simulation and its inversion by neural networks.Office of NavalResearch.Washington DC.1998
    [29] Schott J R., Salvaggio C., Volchok W J. Radio metric scene normalizationusingpseudo-invariant features[J].Remote Sensing of Environment,1988,26:1-26.
    [30]Han L,Donald C,Rundquitst D C.Comparion of NIR/RED ratio and firstderivative of reflectance in estimating chlorophyll concentration:a case study in aturbid reservoir.Remote Sensing of Environment,1997,62:253-261
    [31]N.Cristianini,T.J.Shawe.An Introduction to Support Vector Machine.CambridgePress,2000
    [32]Cherkassky V, MA Yunqian.Selection of meta-parametersfor support vectorregression[C].Proceedings of ICANN2002. Verlag BerlinHeidelberg:Springer,2002:687-693
    [33]Z.L.Wu,K.Y.N.Joseph.Location Estimation via Support Vector Regression. IEEETrans.on Mobile Computing,2007,6(3):311-321
    [34]Dekker A G,Peters S W.The use of the Thematic Mapper for analysis of entropiclakes:a case study in the Netherlands.Int J Remote Sensing,1993,14(5):799-821
    [35]M.Doumpos,C.Zopounidis.Additive Support Vector Machines for PatternClassification.IEEE Trans.on Systems,Man,and Cybernetics,Part B,2007,37(3):540-550
    [36] Vermote E F., Tanre D., Deuze J L., et al. Second simulation of the satellitesignalin the solar spectrum: An overview[J]. IEEE Transactions on GeoscienceandRemote Sensing,1997,35(3):675-686.
    [37]B.Scholkopf,J.Platt,J.Shawe-Taylor.Estimating the support of a high-dimensionaldistribution.Neural Computation,2001,13(7):1443-1471
    [38]Wang Shujun,Guan Dongsheng.Remote Sensing Method ofForest BiomassEstimation by Artificial Neural Network Models. Ecology andEnvironment.2007,16(1):108-111
    [39]李宝华,荒川久辛.南黄海叶绿素a与初级生产力之间的相关分析.黄渤海海洋.1998,16(2):48-53
    [40]疏小舟,尹球,匡定波.内陆水体藻类叶绿素浓度与反射光谱特征的关系.遥感学报,2000年2月,4(1):41-45
    [41]丛爽.面向MATLAB工具箱的神经网络理论与应用.中国科学技术大学出版社.合肥.1998
    [42]丛王福,牛铮,蒙继华,李向军.1998~2003年卫星反演的中国陆架海叶绿素a浓度变化分析[J],海洋环境科学,200625(l):30一33。
    [43]曹仕,刘湘南,曹珊.成熟期水稻砷污染胁迫光谱诊断空间模型研究.中国生态农业学报.2010,18(4):234-238
    [44]周伟华,霍文毅,袁翔城等.东海赤潮高发区春季叶绿素a和初级生产力的分布特征.应用生态学报.2003,14(7):1055-1059
    [45]朱明远,毛兴华,吕瑞华.黄海海区的叶绿素a和初级生产力.黄渤海海洋.1993,11(3):38-51
    [46]疏小舟,尹球,匡定波。内陆水体藻类叶绿素浓度与反射光谱特征的关系[J],遥感学报,2000,Vol.4,No.1,41-45.
    [47]王书舟.支持向量机方法及其应用研究:[博士学位论文].哈尔滨:哈尔滨工业大学,2009
    [48]吕恒,李新国等.基于反射光谱和模拟MERIS数据的太湖悬浮物遥感定量模型[J].湖泊科学200517(2):104-109.
    [49]顾万花.基于遥感的淀山湖水体叶绿素a浓度估算研究.复旦大学硕士生论文.2011
    [50]吕恒,江南,李新国.内陆湖泊的水质遥感监测研究[J]地球科学进展.2005,20(2):185-191.
    [51]李素菊,吴倩,王学军等.巢湖浮游植物叶绿素含量与反射光谱特征的关系.湖泊科学.2002年9月,14(3):228-234
    [52]唐军武,王晓梅,宋庆君等.黄、东海二类水体水色要素的统计反演模式.海洋科学进展.2004,22:1-7
    [53]韩秀珍,吴朝阳,郑伟等.基于水面实测光谱的太湖蓝藻卫星遥感研究.应用气象学报.2010年12月,21(6):724-731
    [54]陈利博,李桐基,黄妙芬,等.两种水体后向散射系数测量方法的比较研究[J].海洋技术,2011,30(1),58-60
    [55]刘良明.卫星海洋遥感导论.武汉.武汉大学出版社.2005:120-142,145-153,216-220
    [56]崔廷伟.渤海生物光学性质与水色遥感反演[博士学位论文].青岛.中国海洋大学,2006
    [57]张卡,盛业华,张书毕.遥感新技术的若干进展及其应用[J].遥感信息,2004(2)
    [58]马跃良,王云鹏,贾桂梅.珠江广州河段水体污染的遥感监测应用研究〔J〕.重庆环境科学,2003,25(3):13一16.
    [59]陈清莲,王项南.海洋水色遥感及Lnadsat-5TM数据在海南岛东部海域水色分析中的应用.海洋技术.1995,14(3):47-60
    [60]邬国锋,崔丽娟,纪伟涛.基于时间序列MODIS影像的鄱阳湖丰水期悬浮泥沙浓度反演及变化[J].湖泊科学,2009,21(2):288-297
    [61]詹海刚,施平,陈楚群.利用神经网络反演海水叶绿素浓度.科学通报.2000,45(17):1879-1884
    [62]杜培军.Radarsat图像滤波的研究.中国矿业大学学报.2002,31(2):132-134
    [63]赵英时等.遥感应用分析原理与方法.北京.科学出版社.2003
    [64]金铭,刘湘南.基于冠层多维光谱的水稻镉污染胁迫诊断模型研究.中国环境科学.2011,31(1):733-737
    [65]任敬萍,赵进平.二类水体水色遥感的主要进展与发展前景.地球科学进展,2002,17(3):363-371
    [66]徐希孺.遥感物理.北京大学出版社.北京.2005:231-233
    [67]孙德勇,李云梅,乐成峰等.应用水表面下辐照度比估测太湖夏季水体叶绿素a浓度.湖泊科学.2007,19(6):744-752
    [68]唐军武,丁静,王其茂.大气散射对采用归一化植被指数进行赤潮遥感监测的影响研究.海洋学报.2004,26(3):136-142
    [69]疏小舟,汪骏发,沈鸣明.航空成像光谱水质遥感研究.红外和毫米波.2000,19(4):273-276
    [70]徐希孺.遥感物理.北京大学出版社.北京.2005:231-233
    [71]刘强,柳钦火,肖青.机载多角度遥感图像的几何校正方法研究.中国科学D辑.2002,32(4):299-306
    [72]毛志华,黄海清,朱乾坤.我国海区SaewiFS资料大气校正.海洋与湖沼.2001,32(6):581-587
    [73]杨建洪,王锦,赵冬至.海洋水色遥感大气校正算法研究进展.海洋环境科学.2008年2月,27(1):97-100
    [74]张亭禄,贺明霞.基于人工神经网络的一类水域叶绿素a浓度反演方法.遥感学报.2002,6(1):40-44
    [75]马荣华,戴锦芳.结合Landsat ETM与实测光谱估测太湖叶绿素及悬浮物含量.湖泊科学,2005,17(2):97-103
    [76]巩彩兰,樊伟.海洋水色卫星遥感二类水体反演算法的国际研究进展.海洋通报.2002.21(2):77-83
    [77]陈述彭主编.遥感大辞典.北京:科学出版社,1990
    [78]唐军武,王晓梅,宋庆君等.黄、东海二类水体水色要素的统计反演模式.海洋科学进展.2004,22:1-7
    [79]中国科学院空间科学技术中心.中国地球资源光谱信息资料汇编.北京.能源出版社.1987
    [80]吕恒,江南,罗潋葱.基于TM数据的太湖叶绿素A浓度定量反演.地理科学.2006,26(4):472-476
    [81]朱敏慧.SAR的海洋遥感探测技术综述.现代雷达.2010,32(2):1-5
    [82]郭华东等.雷达对地观测与应用.北京.科学出版社.2000
    [83]刘一良.微波遥感的发展与应用.沈阳工程学院学报(自然科学版).2008,4(2):171-173
    [84]张卡,盛业华,张书毕.遥感新技术的若干进展及其应用[J].遥感信息,2004(2).
    [85]张德海,郑震藩,姜景山.“神舟”四号中的微波遥感[J].物理,2004,33(1).
    [86]张雪虎,David McLaughlin,Elisabeth M T warog.机载真实孔径雷达在中小尺度区域海洋遥感中的应用.遥感技术与应用.2005年2月,20(1):127-132
    [87]张澄波.综合孔径雷达.北京:科学出版社,1989
    [88]李密.南海近岸水域叶绿素浓度遥感反演模型.硕士毕业论文.中国地质大学(北京)2011
    [89]韩秀珍,吴朝阳,郑伟等.基于水面实测光谱的太湖蓝藻卫星遥感研究.应用气象学报.2010年12月,21(6):724-731
    [90]陈雪洋.基于环境星CCD数据的冬小麦叶面积指数遥感监测模型研究.硕士毕业论文.2009
    [91]李传荣,贾媛媛,胡坚等.HJ一1光学卫星遥感应用前景分析[J].国土资源遥感,2005,77(3):1-3.
    [92]吴北婴.大气辐射传输实用算法[M].气象出版社,1998:21-40.
    [93]彭鹏,朱友志,张平.多视复图像的特征提取和分类方法[J].系统工程与电子技术,2008,30(8):1395-1398.
    [94]王超,张红,陈曦,等.全极化合成孔径雷达图像处理[M].北京:科学出版社,2008.
    [95]张钧屏,方艾里,万志龙,等.对地观测与对空监视[M].北京:科学出版社,2001
    [96]董长虹.Matlab神经网络与应用[M].北京:国防工业出版社,2005:91

© 2004-2018 中国地质图书馆版权所有 京ICP备05064691号 京公网安备11010802017129号

地址:北京市海淀区学院路29号 邮编:100083

电话:办公室:(+86 10)66554848;文献借阅、咨询服务、科技查新:66554700