【作者】 周广柱；

【导师】 杨锋杰；

【作者基本信息】 山东科技大学 ， 大地测量学与测量工程， 2007， 博士

【摘要】 本论文针对德兴铜矿区实际，采用了遥感生物地球化学方法，对矿区部分优势植物包括对盐肤木、狼尾蒿、臭蒿、香薷、芒萁等植物进行了深入探讨，研究了矿区酸性水污染区和尾矿库土壤—植物系统中重金属分布规律。利用小波分析方法进行了噪声方差估计，并通过仿真试验和理论分析确定了波谱中存在的噪声的类型，针对噪声类型提出了最佳空域相关去噪方法。结合光谱反射率、反射峰最大反射率及其位置、吸收谷位置、面积、深度、宽度、红边参数和植被指数等共计44个光谱特征参数，对研究区采集的五种植物的光谱特征进行了分析，并针对叶部重金属含量及其富集特征的逐步多元回归模型开展了研究。研究结果表明：德兴矿区酸性水污染区和尾矿库土壤中元素总含量呈现尾矿库坝中>坝项，河床>污染区>废石场汇水水沟边的趋势。与土壤地球化学背景值相比，各采样点土壤中Cu含量远高于中国A层土壤及A层红壤背景值。Zn、Mn、Cd、Ni、Pb等的含量高于背景值，而Cr含量与背景值相近。对比中国土壤环境质量标准值，大多数土壤不满足国家标准二级标准要求，但几乎所有的土壤均满足三级标准要求。这些土壤基本能够满足农林业生产和植物正常生长的需要，但不能保障农业生产，维护人体健康。在研究区采集的几种植物在重金属含量较高的土壤环境中生存，明显受到了土壤重金属含量影响。重金属大多富集于根部，部分植物在叶部富集。几种植物均表现出了对铜元素的不同程度的富集特性。香薷器官中重金属含量很高，有明显的富集铜的特征。野外光谱噪声估计与降噪方法研究认为，利用第一尺度分解的小波高频系数估计噪声方差具有较高精度和稳定性，基于空域相关算法可以实现该噪声方差的估计。结果表明，野外波谱数据中存在的噪声主要集中在第一、二尺度，而噪声方差也随波长变化而变化。1400nm和1900nm附近水吸收波段噪声显著，野外波谱噪声属于除性复合噪声，采用对数变换结合小波变换可以令其白化，有利于噪声的去除。三类小波降噪方法中，空域相关去噪最适合于野外波谱数据处理，其次是模极大去噪，阈值去噪方法不适于野外波谱数据降噪。改进的自适应空域相关去噪法适用于波谱数据处理。该方法可以有效地去350nm～2500nm波段的噪声，其中包含1400nm附近水吸收波段的噪声。但1900nm附近噪声不能有效去除，原因可能在于光谱仪器系统的精度低。光谱特征研究结果指出，盐肤木、狼尾蒿、臭蒿、香薷、芒萁等五种植物绝大部分光谱特征随环境变化显著，各光谱特征值与叶部重金属含量的线性相关性各不相同，存在的一个共同的规律是：重金属的效应表现为它改变了植物的水分含量、色素含量、叶部组织结构等，促进或阻碍水的吸收；部分重金属阻碍色素合成，而Zn等则能促进色素合成；部分重金属可以破坏叶部组织结构，使红外反射率升高。一种重金属效应在多处光谱特征有所反映。不同重金属如Pb、Zn等元素，在同一光谱特征上(比如在水吸收波段)都有不同程度的反映。重金属在植物体内往往与蛋白质结合，作为生物催化剂——酶的组成，参与了不同的生化过程，可能导致了以上规律的存在。但由于生物对必需重金属元素如Zn的需求也是有一定限度的。当生长基质中这种重金属含量增达到一定程度时，就会产生毒化效应。而这种毒化效应体现在某些生物机能上而非全面影响。对于生物非必需元素如Pb在产生毒化效应时，则将影响多种生物机能如水吸收、色素合成等正常进行，从而导致了多处光谱特征响应。但不同植物之间也存在一些差异，这可能与不同物种对重金属元素的吸收、富集、耐受能力不同有关。叶部重金属含量及其富集特征的逐步多元回归模型研究结果认为，利用逐步回归法可以建立盐肤木、狼尾蒿、臭蒿、香薷、芒萁等五种植物的光谱特征与叶部化学物质之间的关系模型，从模型计算值与实测值比较结果来看，模型比较稳定，相对偏差大多在±10％以下。进入各元素逐步回归模型的光谱特征各不相同，进入同种元素各种植物逐步回归模型的光谱特征差异很大。研究中采用了一阶、二阶微分光谱对狼尾蒿、臭蒿、香薷三种一年生草本植物叶部重金属含量进行逐步回归建模。研究结果表明，除Pb、Mn较差外，微分光谱模型的估测效果均较好，相对偏差大多低于15％。一阶微分和二阶微分模型的估测精度有差异。采用一阶微分光谱模型计算叶部Cu、Ni的估测效果较好，而对于Cr、Mn、Pb、Cd、Zn，采用二阶微分光谱模型计算效果较好。研究中还采用了微分光谱对狼尾蒿、臭蒿、香薷三种一年生草本植物叶部重金属富集系数进行逐步回归建模，结果表明，微分光谱模型对Mn的富集系数估测结果较好；Cu、Cr、Ni、Pb、Cd、Zn其他相对偏差太大，估测价值不高，有待于进一步研究和探讨。更多还原

【Abstract】 According to distribution of barren and waste land in Dexing copper mine, remote sensing biochemistry methods were applied in the research on some of predominant plants, including Rhus Chinensis Mill, Sweet Wormwood Herb, Comnyza Canadensis (L.) Cronq. , Mosla Chinensis Maxim and Dicranopteris pedata. Heavy metals’ distributing pattern in soil-plant system at AMD contaminated land and tailings dam was analysed. Wavelet transformation was taken to estamate noise variance and denosing. Type of noise was made sure via theoretical analysis and emulational experiments. And spacially correlated filtration was chosen for field spectrum denoising. 44 indexes were employed to analyze the features of spectrum of these five plants. And heavy metals contained in leaves and its enrichment index were estimated by stepwise multianalytical regression model.The findings indicated that, total heavy metal concentration in acid mining drainage contaminated soil and in soil of the tailings storehouse was higher in the middle of storehouse than the top of dam, and showed the tendency as the river bed > the pollutes area > nearby the aqueduct of waste rock dam. Comparing with the soil geochemistry background value, the Zn, Mn, Cd, Ni, Pb, Cu content were higher than the their background value of the Chinese A-layer soil and A-layer laterite, especially Cu was far higher than its background value, while Cr is nearly equal to its background value. Comparing with the Chinese soil environmental quality standard value, the majority soils didn’t satisfy the standards requests of the second levels of the national standard, but nearly all soils satisfy the standards requests of the third levels. These soils can just satisfy production of the farming and forestry industry and just can meet the need for plant growth, but cannot safeguard the agricultural production, and the human body health.The five gathered plants survived in the study area where the heavy metal content is very high in soils, and was obviously influenced by the soil heavy metal. The heavy metals mostly concentrates in the root, while concentrate in the leaf of partial plants. The plants displayed varous enrichment characteristic to the copper element. In the organs of Mosla Chinensis Maxim the heavy metal content is very high, and has obviously enrichment characteristic of copper. Research result pointed that, the first scale decomposed details could be well employed to estimate the noise variance and this method has high accuracy and stability. This estimation could be realized based on the spacially correlation algorithm. The noise mainly distributed in the first and the second wavelet decomposed scale, and the noise variance also changes along with the wavelength change.Noise is remarkable nearby 1400nm and 1900nm where the water absorption exists. The emulational experiments and the theoretical analysis indicated that, the noise in the field vegetation spectrum belongs to be dividing noise, and the logarithm transformation and wavelet transformation can make it white. In three kind of wavelets noise reduction method, the spacially correlation filtration most suits to field vegetation spectral data processing, the next is the module maximum filtration. The threshold shrinking method is not suitable to spectral data noise reduction. The improved self-adapted spacially correlation filtration is suitable for spectral data processing. This method may effectively reduce the noise in the bands from 350nm to 2500nm, including bands nearby 1400nm where water absorption exists.But the noise nearby 1900nm cannot be effectively removed. The reason possibly lies in the precision insufficiency of the spectrum instrument system.The spectrum feature findings pointed out that, major spectrum indexes of the five kind of plants is remarkable different along with environmental variation. Linear relevance of spectrum indexes and heavy metal content in the leaf is various. An common rule is: The heavy metal effect often displayed as it changed plant’s moisture content, the pigment content, the leaf structure, and so on, promoting or counteract the water absorption. Some heavy metals hinder the pigment synthesis, but Zn can promote it. Some of the heavy metals may destroy the leaf structure and causes the higher valuesof the infrared indexes of reflection spectrum. One kind of heavy metal effect often has many response features in reflection spectrum. Different heavy metals like Pb, Zn, has the varying degree in the identical feature of the reflection spectrum, for instance in water absorption band. The reason lies in that the heavy metal in plant often combined with the protein, and acted as the biocatalyst, the enzyme, and participated in the different biochemistry process. But the biology also has certain limit of the demand of the essential heavy metals i.e. Zn. When essential heavy metal content achieved certain degree in the growth matrix, the poisoned effect presented. But this kind of poisoned effect influence on certain body functions but not on all of them. Regarding biological non-essential element like Pb, when the poisoned effect presented, many biological functions like water absorption, the pigment synthesis, and so on, were affected and could not be normmally carried on, thus it caused many spectral response. But between the different plant there were also some differences, it maybe owned to the different species ability of the heavy metal absorpting, concentrating and tolerating.The research on leaf heavy metal content and its enrichment indexes proved that, Stepwise multianalytical Regression model can be found. The model predicted value and the actual value comparison showed that the model is stable, and the relative deviation mostly below±10%. The first order and the second differential spectrum was employed to estamate the heavy metals contained in the leaves and their enrichment indexes for Sweet Wormwood Herb, Comnyza Canadensis (L.) Cronq. and Mosla Chinensis Maxim. The findings indicated that, except Pb and Mn, the differential spectrum model has good estimating effect, the relative deviation mostly is lower than 15%. The firstorder differential model and the second differential model has the difference on estamation of various heavy metal. The first order differential spectrum model can well estamate Cu and Ni in the leaves, while the second differential spectrum model can well estamate the Cr, Mn, Pb, Cd and Zn. The differential spectrum model can only well estimate the enrichment coefficient of Mn. As relative deviations of estimation on enrichment indexes of Cu, Cr, Ni, Pb, Cd and Zn are too high, the model can not be use to estimate the enrichment coefficients of these heavy metals. It looks forward for further studies and discussion.更多还原