【作者】 任红艳；

【导师】 庄大方； 潘剑君；

【作者基本信息】 南京农业大学 ， 土壤学， 2008， 博士

【摘要】 矿山开采为国民经济的快速发展提供了强大的能源和材料支撑,然而,不少矿山由于过度开采以及环保措施没有同步跟进而使得矿山开采面临着能否持续发展的问题,同时也对周边农田土壤-作物系统遭受严重的重金属污染——农田环境的健康和安全直接关系着人民的生命安全与和谐社会的构建。因为具有快速、简便以及无损的特点,遥感技术逐渐成为一种可供选择的、有效的农田环境质量监测手段和方法而备受人们的关注。本论文正是该领域一次探索,并得到了国家自然科学基金(40571130)和中国农科院农业部资源遥感与数字农业重点实验室的资助。宝山矿为一处具有多年开采历史的铅锌矿,伴生矿主要有铜、镉、砷,随着开采、冶炼生产的进行,铅、锌、铜、镉、砷等重金属元素进入矿区周边农田土壤-作物(水稻)系统。在土壤中,镉、砷、铅和锌在可见光-近红外区间均没有光谱响应特征,铜只有在高于4000 mg/kg浓度时才表现出比较明显的光谱响应,然而农田土壤中的重金属浓度难以达到这样的高污染水平,所以很难直接用土壤光谱反射率来预测土壤中重金属的浓度,但可以利用重金属元素与铁氧化物这种光谱特征成分之间的相关性实现对矿区农田土壤重金属浓度的预测;进入农田土壤的重金属被水稻吸收后将影响水稻的正常生长,对水稻叶片叶绿素合成以及光合作用造成破坏和抑制,这一变化将在水稻冠层光谱上得到响应,从而通过提取水稻冠层光谱特征来预测水稻冠层叶片重金属含量;同样,利用稻米中积累的重金属与蛋白质这一光谱特征物质之间较好的相关性实现稻米重金属含量的遥感监测。基于上述原理,本论文选择宝山矿区周边农田土壤-水稻系统为研究对象,分析了铅、锌、铜、镉、砷等重金属元素在土壤、水稻冠层叶片与稻米中的分布情况,同时,借助FieldSpec FR2500高光谱辐射计测量了土壤和稻米的室内高光谱,利用MSR-16R多光谱辐射计在野外测量了水稻冠层光谱,通过提取它们的光谱特征并对这些光谱特征参数与土壤、水稻冠层叶片和稻米重金属含量的关系进行了统计分析,对多/高光谱遥感技术在矿区农田土壤-水稻系统重金属污染的定量监测与评价研究中的应用展开了探讨,主要结果和结论如下:1,宝山矿区农田土壤中的Pb平均含量高达1767.42 mg/kg,远远超过水稻产地土壤环境质量要求(NY/T 847-2004,250 mg/kg);Zn的含量也远远超过该标准;土壤的Cu含量(81.91 mg/kg)比保障农业生产和人体健康的标准(50 mg/kg)高;土壤中的Cd含量的最低水平(2.63 mg/kg)就已经超过标准限定值(0.3 mg/kg)7倍之多,而土壤As含量的平均水平(140.16 mg/kg)接近二级标准限定值(30 mg/kg)的5倍,同时,该区域农田稻米食用安全性很低,其中稻米所遭受的Pb污染已经超过国家标准,需要引起国家相关部门的重视;2,利用这些重金属与具有光谱特征的Fe含量之间的关联,建立了利用土壤室内光谱反射率估算土壤重金属含量的预测模型,从而实现了对土壤重金属含量的间接测定,为土壤重金属污染状况的调查与监测提供了一种简单、快速的测定方法;同时,合适的光谱预处理方法可以提高模型的预测能力;3,为充分利用重金属污染下水稻的冠层光谱而提取了植被指数(Vegetationindex)、红边参数(Parameter of red edge position)、绿峰偏振斜率(Slope of grean peakpolarization)以及吸收三角形面积(Area of triangle absorption)等光谱特征,并通过这些光谱特征实现了水稻冠层叶片重金属含量的准确预测,并在红边参数提取的基础上提出了区别于重金属污染水稻生理临界值的光谱临界值的概念,其中,利用倒高斯拟合(Inversed Gaussian fitting)方法对红边参数的提取为多光谱遥感数据的充分利用提供了有益的参考;通过选择与CBERS(多光谱卫星)相匹配的波段预测水稻冠层叶片重金属含量的探索,为多光谱遥感卫星数据的农作物污染监测应用提供了有益的参考;4,通过利用高光谱仪器所获取的稻米室内近红外反射光谱实现了糙米蛋白质、重金属含量的快速预测,同时,多元散射校正(MSC)预处理后的光谱显著提高了回归模型的预测能力,结果表明可以通过糙米重金属与蛋白质含量之间的紧密相关和近红外光谱对蛋白质含量的响应实现对糙米重金属含量的预测;利用分蘖期的水稻冠层光谱特征研究了糙米蛋白质和重金属含量的预测,虽然其预测效果不如直接利用室内近红外反射光谱进行预测的效果,但是就对提前进行稻米品质和食用安全评价的意义而言,该方法不失为一种有益的补充。更多还原

【Abstract】 Mining supports national economic development with large amount of energy sources and materials, however, over-mining and laggard environmental protection measures make many mines be faced with challenge of sustainable development and cause serious heavy metal pollution in the nearby soil-crop system. Security and health of agricultural environment is directly correlated to the subsistence security of people and constructing of harmonious society. As a quick, convenient and lossless method, remote sensing was gradually focused and has been used as an alternative and effective solution for monitoring environmental quality of farmlands. This dissertation was supported by the opening project funded by National Science Foundation of China (40571130) and Key Laboratory of Resources Remote Sensing and Digital Agriculture, Ministry of Agriculture.Pb, Zn, Cu, Cd and As entered into soil-paddy plant system when being mined from Baoshan Mine which is a Pb-Zn mine exploitated for many years. Pb, Zn, Cd and As have no spectral feature in the region of visible-near-infrared light, meanwhile, Cu shows spectral characteristics only if its concentration in soils exceeds 4000 mg/kg. However, heavy metal concentrations in agricultural soils are very little. It is hard to predict heavy metal concentrations in agricultural soils by soil spectral reflectance. Whereas, the close correlation between heavy metals and Fe that has significant spectral characteristics should be the basis on which heavy metal concentrations in agricultural soils can be predicted by soil spectral reflectance. Heavy metals absorbed by paddy plants would inhibit growth of root system, synthesis of chlorophyll and photosynthesis of leaves, which should be betrayed by canopy reflectance spectra of paddy plants. Therefore, heavy metal concentrations in rice canopy leaves should be predicted by characteristic spectra extracted from rice canopy reflectance spectra. Heavy metals accumulated in rice corns make passivation for rice protein and would threaten people’s health through food chain. Close correlation between heavy metals and protein that has significant spectral feature should also be the basis on which heavy metal concentrations in rice corns can be predicted by corn reflectance spectra.Based on above theories, heavy metal concentrations and spectra of soil-paddy plant system in the farmlands nearby Baoshan Mine were studied to research the feasibility of reflectance spectra of soils, rice canopy and rice com in predicting and assessing heavy metal pollution in soil-paddy plant system. This study can contribute to the usage of spaceborne or airborne multi/hyper spectral remote sensing in environmental pollution monitoring. The main results of this dissertation are listed below:1, Pb concentration in agricultural soils was 1767.42 mg/kg and exceeded much higher than the level required by Soil Environmental Quality Standard made by Ministry of Environmental Protection of P.R.C for rice producing area. Zn, Cu, Cd and As were similarily higher than the corresponding level required by Soil Environmental Quality Standard. At the same time, edible quality of rice is poor because Pb concentration in rice is much higher than the safety level, which should be seriously treated by relational government departments.2, On the basis of intercorrelation between heavy metals and Fe, spectral reflectance of soils was used to construct assessing models for the indirect estimation of heavy metal concentrations in polluted soils. Appropriate spectral pretreatments can promote cability of models. This reserach indicated that remote sensing can be an alternative solution of convenience and speed to investigate and monitor heavy metal pollution in agricultural soils.3, It is these spectral parameters such as vegetation indices, red edge position, slope of green peak polarization, and area of triangle absorption extracted from rice canopy spectra at the tillering stage that makes the best of canopy reflectance spectra of paddy plants polluted by heavy metals. By means of these spectral features, regression models were constructed for heavy metal concentrations in rice canopy leaves. Inversed Gaussian was used to fit reflectance spectra in the region from red light to near-infrared light for red edge position, which would be profitable for the sufficient usage of multispectral data. On the basis of extraction of red edge position, conception of spectral critical value was defined to discriminate physiological critical value for paddy plants polluted by heavy metal.4, Prediction for protein in rice by coarse rice reflectance spectra pretreated by multiplicative scatter correction was the best because the pretreatment promote predicting ability of regression models. It indicates that heavy metals in rice can also be indirectly predicted by coarse rice spectra on the basis of close correlation between heavy metals and rice protein.Prediction of protein content and heavy metal concentrations in coarse rice by rice canopy spectra at the tillering stage was not as good as that by coarse rice reflectance spectra. However, this method is one compensatory solution to predicting rice quality and assessing rice edibility.更多还原