紫色土土壤养分空间变异性研究

【作者】 朱益玲；

【导师】 刘洪斌；

【作者基本信息】 西南农业大学 ， 土壤学， 2002， 硕士

【副题名】地统计学方法

【摘要】 土壤并非一个匀质体，而是一个时空连续的变异体，具有高度的空间异质性。传统描述土壤空间异质性的方法基本上是定性地土壤分类，而不是定量地研究土壤空间变异。 70年代，地统计学方法被引入土壤科学研究领域，克服了应用经典的fisher统计理论在研究土壤性质空间变异性规律方面的不足，其理论依据是区域化变量理论。地统计学既可用来估测土壤性质的分布，也可用于确定土壤变异的空间尺度和形式，以提高采样的有效性；还可用于研究引起土壤变异的各种过程。目前，应用地统计学的原理和方法仍是国内外定量化研究土壤空间变异的趋势之一。随着地理信息系统(GIS)的广泛应用和地理学学科的发展，利用地统计学，结合GIS技术来研究土壤性质空间变异已成为目前的土壤学的研究热点之一。但现今土壤空间变异规律的研究主要集中在土壤物理性质方面，而土壤化学性质方面的研究还较少。而且，大部分的研究对象主要集中于北方平原地区，南方丘陵地区的土壤空间变异的研究才刚刚开始。 以下是运用GIS技术，采用地统计学方法，对丘陵地区紫色土表层(0～20cm)土壤中的氮、磷、钾等13种土壤养分的空间变异性规律进行研究的结果： 1．本次试验的采样区位于江津市，为低山丘陵地貌，土壤类型主要为紫色土和水稻土，土地利用现状主要为水田和旱地。土壤耕作多年，每个田块的土壤肥力不同导致生产力的不同，说明土壤养分含量具有高度的空间异质性。 2．养分的变异系数范围在12.9％～106％之间，速效磷的变异系数最高，为106％，有效铁的变异系数也较高，为98.5％，pH值的变异系数最低，为12.9％，其它大多数养分的变异系数属中等强度变异。养分全量的变异系数变化范围不大，说明养分全量在土壤中比较稳定，但速效养分变异程度较大。土壤中养分分布类型涉及到对土壤养分的正确评价。变异函数的计算一般要求数据符合正态分布，否则可能会使变异函数产生比例效应。在分布类型上，除全磷、速效磷、有效铁服从对数正态分布外，其余全为正态分布。 3．各向同性分析的结果表明：12种(除全氮)土壤养分在一定的范围内观测值之间存在着空间相关性，变异函数γ(h)随间距增大而增大，且有基台值，说明存在空间变异结构特征，在50m的采样间距内存在空间相关性。其变异函数可用球状模型、指数模型、高斯模型拟合，有效铁存在漂移现象，可用孔穴效应模型拟合。变程在270m～770m之间。12种土壤养分的块金值都较小，块金值与基台值之比大多在25％～75％之间，都具有中等强度的空间相关性，是结构性因素和随机性因素共同作用的结果。速效钾的块金值与基台值之比为81.2％，说明人为活动作用占主要因素。CEC、全钾、有效铜的块金值与基台值之比小于25％，说明由空间自相关引起的空间变异占主要部分。全氮的空间变异与间距无关，存在纯块金效应，块金值为0.2，对于全氮，采样间距应小于50m。各向异性分析的的结果表明，除全氮外的12种土壤养分含量在不同方向上具有明显的各向异性结构特征，有机质和速效磷为典型的几何异向性，其它则同时具有几何异向性和带状异向性。 4．在今后研究土壤养分空间变异结构时，对于采样点的设置，不仅要在大尺度上采样观测，还应同时在小尺度上进行套合取样，这样才能较准确地了解区域化变量在不同尺度上的变异特征。利用GIS的图形数据与属性数据相联的特性，进行随机采样，样点对之间的距离可很方便地测量出。这样，克服了用网格法采样时，样点数过多，工作量过大的弱点。土壤养分各向异性分析的结果，为今后采样点的设置也提供了一定的依据，在不同的方向上采样密度也应有所不同。在小尺度距离上相对要多一点，在大尺度距离上相对少一点，这样才能保证在变程a范围内的变异函数值能准确反映区域化变量的空间变异性。 5．根据各向异性结构分析后所得到的变异函数理论模型，采用普通克立格法和对数正态克立格法进行最优内插，绘制了各种土壤养分含量的空间分布格局（Krinng）图，可得到不同土壤养分含量的分布图、各级面积及比例，能对士壤养分的丰缺状况有整体性了解。Kropng插值的结果受变异函数模拟精度、样点的分布、邻近样点的选取数的影响。Kriging插值的结果表明：pH值由东到西逐渐增高；阳离于交换量和微量元素的空间变异规律与pH值的空间变异规律较为一致；有机质含量在采样区的东北部和南部较高，并和氮的空间变异规律较一致。全氮和碱解氮主要在东南部含量较高，全磷和速效磷在北部和中部偏南含量较高，全钾和速效钾则在北部和西南部含量较高，可以看出养分全量和速效养分在空间分布上有着一定的相关性。Kropng插值结果进一步表明土壤表层养分含量具有高度的空间异质性，决定了空间格局的存在，养分含量的斑块（patch）的大小、形状及空间分布等具有显著的差异，但是不论在什么方向，养分含量由低到高的分布梯度规律总是存在的。这种养分空间分布的特征与紫色土壤在不同空间位置上的各种土壤发生过程（物理、化学、生物等）有着重要的联系。 运用GIS技术，利用地统计学方法，研究西南丘陵地区紫色更多还原

【Abstract】 The properties of soil are not homogeneous, whereas soil is a continuous spatio-temporal heterogeneity, with high spatial variability. The traditional methods of describing soil properties are on the whole qualitatively soil classification, not researching soil spatial variability quantitatively.In the 70s, geostatistics was introduced into soil science field, which overcame the problem of using classical fisher statistics theory to study the rules of soil properties, based on the regionalized variable theory. With the population applying of GIS and developing of geography, it has been a hotspot to use the combination of GIS and geostatistics to study the soil spatial variability today. Geostatistics can not only be applied to assess the distribution of soil properties, but also to estimate the spatial scales and forms of soil variation, in order to give rise to the validity of soil sampling, it also can be used to research the courses of soil variability. The theory and method of geostatistics are currently still one tide of researching spatial variability quantitatively both at home and abroad. But research about soil spatial variability still focus on soil physical characters, articles in soil chemical properties is rarely found. Furthermore, most objectives concentrate in northward, planning areas. Studies of soil spatial properties in southern hilly areas begin in recent years.With the help of GIS and geostatistics, the results of studying on the spatial variability of 13 soil fertilities, namely soil pH, CEC, organic matter, total nitrogen, available nitrogen, total phosphate, available phosphate, total potassium, available potassium, available copper, available zinc, available iron, available manganese in purple hilly areas can be showed.1. The sampling area lies in Jiangjin city, where is undulating topography, purple soil and paddy field soil account for most of its soil types, the status of land use are paddy and dry land. After year’s cultivation, the different soil fertility of each site resulted in diversities of soil productivity, which show the high spatial variability of soil fertility.2. Soil properties varied sharply, with available potassium showing the highest CV (106%), available iron the relative higher (98.5%), and soil pH the lowest (12.9%), with others the medium values. CV of total nutrients did not change greatly, which meant total nutrients are stable. On the other hand, CV of available nutrients did. The distribution types of soil nutrient may impact on the evaluation of soil fertility. Generally, the calculation of semi-variogram calls for data to conform to normal distribution, or it will cause proportion effects. The distribution patterns of total phosphate, available phosphate, available iron follow lognormal and others normal distribution.3. The analyst of isotropy indicated that the sample spots of 12 soil nutrient contents (except total nitrogen) were correlated in given spatial range. The semi-variogram values increased when the distances were enlarged, and nugget lied in, which meant the existence of spatial variability structure, and sample spots were correlated in the 50-meter of sampling distance. Spherical models, exponential models, gaussian models can be used to fit the semi-variogram. Draft phenomena of available iron showed hole effect model. Within the range of 270m~770m, macro-element (total phosphate, total potassium, available potassium), trace elements, CEC and soil pH were correlated in 50m with the rage of 780m. The nugget of 12 soil nutrient contents were small, the nugget-to-sill ratio was from 25% to 75%, with mild spatial relativity which due to the interaction of structural features and randomicity features. That of available potassium showed the interruption of human activities with the highest value (81.2%), and of CEC, total potassium, available copper showed that spatial auto-correlation was master with the value less than 25%. There was no relationship between spatial variability and distance, but with the nugget of 0.2, therefore, the dis更多还原