【作者】 林芬芳；

【导师】 王珂；

【作者基本信息】 浙江大学 ， 农业遥感与信息技术， 2009， 博士

【摘要】 通过采样分析、资料收集和室外调研,在对工业化和城市化快速发展地区的土壤肥力因子和土壤重金属含量进行空间变异研究的基础上,系统评价了研究区的土壤质量状况,并将其与空间稳定性因子结合应用到基本农田保护、标准农田建设、耕地占补平衡的实施评价和管理上;同时,针对山区特定景观区域,开展了影响山区茶园土壤肥力质量的敏感因子及其空间预测方法研究。研究目的在于通过综合多种信息源,系统性和整体性地评价县(市)级尺度区域的土壤质量,从而能为大尺度区域的土壤质量评价提供技术途径和方法;在此基础上分析和揭示快速城市化进程对土壤质量和耕地质量的影响机制,为县(市)级土地(土壤)可持续利用管理决策提供依据;针对山区复杂地形,能建立一套准确预测山区土壤肥力质量的方法,以期为农业精准管理提供基础。论文的主要研究工作、认识和结论如下:(1)基本构建了工业化和城市化快速过程中县(市)级尺度区域的土壤质量评价方法的框架。通过对1:25000县(市)级区域的系统采样,首先,利用地统计学的空间结构分析功能、插值技术及互信息理论,分析了研究区土壤肥力因子和土壤重金属元素空间变异规律、空间分布格局、土壤污染状况及来源。通过建立基于岩性类型、土壤类型和土地利用方式的土壤质量评价的最小数据集(MDS),研究了富阳市土壤质量现状,分析了工业化和城市化对土壤质量的影响机制。根据全国耕地类型区、耕地地力等级划分标准(NY/T 309-1996),将基于MDS指标体系得到的土壤质量划分为4个等级。若不考虑土壤环境污染因子,那么研究区中Ⅰ级土壤资源占用全市总面积的31.36%,主要集中在研究区的西北部和中部地区,Ⅲ和Ⅳ级土壤资源主要分布在西北部往中部的方向,分别占总面积的22.51%和11.7 9%,而Ⅱ级土壤质量分布于南部和东部地区,占总面积的34.34%。若考虑研究区土壤重金属污染,研究发现各乡镇的土壤质量的高低排序与前述结果基本上一致,只是面积和土壤质量指数大小的变化。土壤Cd污染区的土壤质量相对比较高,土壤质量指数大部分在0.50-0.90之间,复合污染区的土壤质量相对更高,土壤质量指数平均值都在0.78以上,即Ⅰ级土壤质量,除灵桥镇外,这说明土壤质量较高的区域,土壤重金属污染较严重,这部分区域土壤的农业利用应以花卉苗木等为主。(2)提出了互信息理论结合决策树See5.0算法来定量研究自然条件和人为活动对土壤质量的影响。土壤质量受到各种因素的影响,不仅包括成土母质、土壤类型、地形等自然条件的影响,还受到人为因素的影响,尤其是城镇化进程的影响。通过互信息理论发现,众多连续型因子中影响土壤质量的主要因素包括到城镇的距离、到水域的距离、相对高程、到道路的距离和到工业类型的距离。采用决策树方法分析了这些因子是如何影响土壤质量过程的,结果表明利用基于互信息理论选取的因子的决策树结果明显优于利用全部因子的决策树结果,且无论是决策树还是决策规则,分类精度均达到80%以上。(3)以土壤质量评价成果和空问稳定性因子为基础,提出将耕地质量评价成果应用到基本农田保护、标准农田建设及耕地占补平衡。研究区现有基本农田中受到土壤Cd污染的面积占基本农田总面积的6.22%,基本农田中一等和二等土壤质量占50%以上,但另外45%左右的基本农田还是处于中低产田,需要及时地改造。同时,空间分析结果表明研究区有80%以上的一等基本农田分布在工矿企业和区级公路的1km以内,即一等基本农田存在的空间稳定性风险很大,尤其是分布在工矿企业旁边的基本农田,存在因工业化和城镇化发展而被建设用地占用的风险。研究区标准农田受到Cd污染占标准农田总面积的3.69%;一等土壤资源占19.82%,二等土壤资源占29.95%,三等土壤资源占27.14%,四等土壤资源占19.26%,即富阳市标准农田的土壤质量整体较高,但也存在一定比例的较差的标准农田。统计表明,研究区的标准农田的空间稳定性风险指数大于0.80的有74.24%,其中,距离工矿企业和区级公路1km以内的标准农田面积均达70%,离城区2km以内的占30%左右,可见,位于这些区域的标准农田易受工业污染,且转化为建设用地的可能性也较大。因此,研究区的标准农田建设成绩显著,但规划选址等决策存在缺陷。分析了1996-2004年问被占用耕地的土壤质量,发现富阳市1996-2004年间被占用耕地中一等土壤资源占48.05%;而补充耕地中,一等土壤资源仅占1996-2004年间补充耕地总面积的23.89%,二等土壤资源占32.44%,三等和四等土壤资源分别占24.16%和16.52%。可见,补充耕地中一等土壤资源与被占用耕地的一等相比,明显较少,仅为占用耕地的54.89%,而补充耕地中增加的耕地面积的土壤质量主要是二等、三等和四等。补充耕地中空间稳定性风险指数超过0.80以上的占总补充面积的61.75%。其中,离城区2km以内的补充耕地面积有28.97%,但是距工矿企业和公路1km以内的补充耕地面积均较多,都在60%以上。这些补充耕地有可能又面临建设占用的风险,同时也易受工业废物和废水等污染。(4)以浙江省富阳市丘陵山地的特色经济作物茶叶产地为研究对象,分析了影响茶园土壤肥力质量的敏感环境因子,比较了土壤肥力的空间预测方法,最后,以插值后的土壤肥力质量为变量,对茶园进行了精确农业管理分区,从而为准确预测山区土壤肥力质量和精准管理提供了依据。以不同地形因子、归一化植被指数和经纬度为自变量,分析比较了不同变量选择方法(相关系数法、主成分法、逐步回归法和互信息理论)和不同空间预测方法(逐步回归、普通克里格、协克里格、回归克里格、广义神经网络和BP神经网络),结果显示基于BP神经网络结合互信息方法的预测精度高于其它方法的预测精度,同时也表明了影响茶园土壤肥力质量的主要因子包括经纬度、相对高程和切线曲率。科学合理的土壤养分管理分区技术是实施精准农业变量施肥的高效手段。本研究采用模糊c-均值聚类对茶园土壤肥力进行管理分区,通过模糊性能指数(FPI)、分类熵(MPE)和分类距离(S)的检验发现最优模糊度和分类数分别为2.0和4。分区间差异显著性检验表明,除分区3与分区4外,不同分区之间土壤pH、有机质、交换性铝和全氮的均值均达到了显著差异;有效磷和速效钾在分区之间差异不显著;环境因子中相对高程在不同分区之间都达到了显著差异,这与前述关于土壤肥力质量指数的空间分布格局的分析结果较一致。更多还原

【Abstract】 Through sampling,materials collectiong and field investigation,based on the result of spatial variability of soil fertility factors and heavy metals in rapid industrialization and urbanization areas,this research systematically evaluated soil quality in the study area,and applied it and spatial stability fators to the implementation evaluation and management of basic farmland,high quality prime farmland and cropland requisition-compensation balance; Furthermore,in view of macroscopic soil quality evaluation result not suitable for mountain areas,this study undertook soil fertility quality and its prediction in mountain area.The main objectives were:(1) systematically and globally evaluating soil quality in county(municipal) area by integrating multi-source data for providing the framework of soil quality evaluation in large scale districts;(2) on the basis of the soil quality result,analyzing and revealing changes of soil quality and cropland quality in urbanization to establish the foundation for land planning;(3) aiming at complicated terrain areas,obtaining the spatial prediction method of soil fertility for precision management.The main results,understandings and conclusions of the dissertation were summarized as follows:(1) Basically constructing of the framework of soil quality evaluation of county (municipal) area in fast industrialization and urbanizationFirstly,the research analyzed spatial variability of soil fertility factors and heavy metals, their spatial patterns,soil pollution and pollution sources by geostatistics and mutual information.Secondly,based on spatial distribution of soil quality indicators,the research evaluated soil qualtiy in study area and the mechanism of soil quality affected by industrialization and urbanization through the minimum data set related to geology,soil types and land use.Soil quality indices were divided into four grades according to the classification of region type and fertility of cultivated land in China.Without considering soil environmental factors,the first grade of soil resource accounted for 31.36%of the total area, and mainly located in the northwestern and central region of the study area;The second grade lied at the south and east of the area accounting for 34.34%;The third and fourth grade located in the region from northwest to middle,and accounted for 22.51%and 11.79%, respectively.If considering influence of soi heavy metals,the results showed that soil quality state in every town was the same as previous results and only changed in area.Soil quality index was high for 0.50-0.90 in polluted areas by soil Cd;Soil quality index was higher in combined pollution areas than that in single-factor pollution areas,and its average exceeded 0.78.(2) The method of mutual information combined of decision tree See 5.0 was found for quantitily studying the effect of natural and human factors on soil qualitySoil quality was affected by not only natural factors including parent material,soil type, terrain and etc,but also human activities,especially urbanlization.It was showed that factors affecting soil quality were mainly distance to town,distance to water,altitude,distance to road and distance to industry.In addition,in determining the process of the influence of these factors on soil quality by decision tree,it was suggested that the accuracy based on factors chosen by mutual information was higher than thatbased on all factors.(3) It was suggested that cultivated land quality should be applied to basic farmland, high quality prime farmland and cropland requisition-compensation balance-based on soil quality evaluating result and spatial stability factorsFor basic farmland,the area polluted by soil Cd was 6.22%of total basic farmland area. In soil quality,the first and second soil resources all exceeded 50%of basic farmland,but soil quality of remaining 45%in basic farmland was medium or low,which should be reconstructed.Furthermore,it was showed that more than 80%of the first basic farmland located in the region less than 1km from industry and road.This result suggested that basic farmland near industry are probably replaced by construction land.For high quality prime farmland,the area polluted by soil Cd was 3.69%of the total area. Soil quality of high quality prime farmland dominated by the second and third grade.It was statistically showed that spatial stability risk index was more than 0.80 in 74.24%of high quality prime farmland,the distance of 70%of which to industry and road was less than 1km, and the distance of 30%of which to town less than 2km.The result indicated that high quality prime farmland lying in these regions are easily polluted by soil heavy metal and easily transformed to construction land.For requisited cropland from the year 1996 to 2004,the area polluted by soil Cd was 4.14%of the total area.From view of soil quality,most of requisited cropland from the year 1996 to 2004 belonged to the first grade accounting for 48.05%;But the first grade only had 23.89%in compensated cropland from the year 1996 to 2004.In compensated cropland,the area of spatial stability risk index more than 0.80 accounted for 61.75%,the distance of 28.97%of which to town was in less than 2km,and the distance of more than 60%of which to industry and road was all less than 1km.Compensated cropland in these regions not only were replaced by construction land,but also polluted by industrial waste.(4) Based on tea in Fuyang county,this research chose optimally environmental factors affecting soil fertility quality in tea and the optimal estimation method of soil fertility quality. At last,management zones of tea soil were determined with the map of soil fertility quality interpolated.With topographical factors,NDVI,latitude and longitude as independent variables,this study analyzed and compared different variable selection methods(correlation coefficient method,principal component method,stepwise regression method and mutual information theory) and spatial interpolating methods(stepwise regression,ordinary kriging,co-kriging, regression kriging,generalized neural network and BP neural network ).The result showed that the method of BP neural network combined of mutual information was the best in prediction accuracy,and the factors affecting soil fertility quality in tea plantation mainly included latitude,longitude,altitude and tangential curvature.Scientific and reasonable soil management zoning is the effective means of variable rate fertilization in precision agriculture.Fuzzy c-means clustering algorithm was used to delineate management zones.It was found that through testing of the fuzzy performance index (FPI),classification entropy(MPE) and separate distance(S),optimal clustering exponent and number of classes were 2.0 and 4.0,respctively.To estimate the validity of zoning result, the general statistics analysis on the data was carried out.The zoning statistics showed that soil properties differed sharply between management zones.更多还原