广西甘蔗地不同耕作方式下土壤孔隙特征

为探究广西地区不同耕作方式对甘蔗地土壤孔隙结构的影响,选取经过3年耕作(CT)与免耕(NT)处理的两种不同耕作方式下的甘蔗地土壤作为研究对象,利用X射线CT扫描技术和ImageJ等软件获得表征土壤孔隙二维分布和三维特征的参数,并对土壤孔隙结构进行三维重构与可视化。结果表明:两种处理下土壤质地组成差异不大,但耕作使得土壤容重降低,改变了有机质的分布,使有机质含量显著增加。耕作主要影响土壤中的大孔隙(50μm)和中孔隙(0.2～50μm)的分布,使得0—30 cm土层大孔隙减少,中孔隙增加。耕作处理土壤孔隙的面密度和复杂度的波动大于免耕处理。耕作处理的土壤孔隙连通度均值(0.687)显著高于免耕处理(0.465),孔隙弯曲度均值(1.314)显著低于免耕处理(1.348),孔隙度和孔隙数均值(0.526%,8 174)显著小于免耕处理(1.181%,16 306)。免耕土壤孔隙在三维空间中分布较多且均匀,但孔隙孤立,连通较少,并且孔隙路径较弯曲。耕作土壤孔隙在三维空间中分布虽少,但分布呈网络型,交叉连通孔隙较多,孔隙通道连续且规则。因此,在广西地区甘蔗种植中,相较免耕而言,耕作增加土壤中生物性大孔隙,增加孔隙的连通度,降低孔隙通道的弯曲度,有利于土壤水分与溶质运移通道形成,并且促进作物根系生长和微生物活动,改善土壤的导水透气性,对甘蔗的生长发育具有积极作用。     

保护性耕作对喀斯特山区旱地土壤理化性质的影响

为揭示保护性耕作对喀斯特山区旱地土壤理化性质的影响,以常规耕作(CGGZ)为对照,在贵州省大方县典型喀斯特山区进行连续4年的对比试验,研究机械化保护性耕作(BHGZ)对喀斯特山区旱地土壤理化性状的影响。结果表明,在0-40cm土层,2种耕作方式的土壤孔隙度、毛管持水量、有机质、全氮、碱解氮、全磷、有效磷、全钾和速效钾含量均是耕层(0-20cm)高于下层,并随耕作年限的增加,这些指标逐渐增加,但土壤容重则相反。机械化保护性耕作土壤孔隙度、毛管持水量、有机质、全氮、全磷、有效磷、全钾和速效钾含量明显高于常规耕作的土壤,而容重和碱解氮的含量则略低于常规耕作土壤。4年机械化保护性耕作的耕层土壤有机质、全氮、全磷、有效磷、全钾和速效钾含量分别比常规耕作提高6.97%,9.13%,4.16%,9.16%,2.10%和3.09%,碱解氮降低4.46%;连续4年机械化保护性耕作20-40cm土壤有机质、全氮、全磷、有效磷、全钾和速效钾的含量分别比常规耕作提高12.58%,12.70%,7.71%,13.00%,2.90%和6.60%,碱解氮降低4.96%。     

利用基于偏微分方程的图像滤波技术研究土壤孔隙结构

利用自主研发的土壤切片数字图像处理分析系统软件,采用3种滤波方式,即BMO滤波(boundary mean oscillation有界平均振荡模型)、PM滤波(Perona-Malik偏微分方程模型)和中值滤波,对免耕、翻耕和旋耕3种耕作方式下不同深度土壤切片的数字图像进行去噪增强,并对处理后的图像进行了孔隙特征参数提取与统计。在此基础上,探讨了几种图像滤波技术的效果以及耕作方式对不同土层土壤孔隙形态结构的影响,以期找到适用于土壤切片数字图像的滤波技术,为后期土壤孔隙信息的提取与统计提供工具。结果表明,采用BMO滤波后,从土壤切片数字图像得到的土壤孔隙度与试验结果最为接近;其中,孔径5μm的孔隙度显著大于其他图像滤波技术的结果,而5~50μm与50μm孔径的孔隙度与其他图像滤波技术的结果没有显著差异;与免耕相比,翻耕与旋耕提高了表层土壤的孔隙度,增加了传导孔隙的比例。     

常规耕作与短期免耕下葡萄园土壤孔隙结构差异的图像比较

本文利用数字图像分析方法,研究了新葡萄园和老葡萄园在常规耕作和短期免耕下土壤孔隙结构的差异。结果表明:常规耕作和短期免耕土壤在孔隙结构上具有较为明显的差异。与短期免耕相比,常规耕作土壤具有较多规则的小土粒,孔隙面积较大,其小孔隙比重较高。新葡萄园常规耕作和短期免耕土壤结构之间的上述差异相对弱于老葡萄园土壤。     

土壤孔隙质量分数维D_m二元图像分析及其影响因素研究

本文简要介绍了利用土壤切片的二元图像分析了土壤孔隙结构的质量分数维Dm 的方法 ,并研究比较了四类土壤类型的Dm 及其影响因素。结果表明Dm 可以较好地定量描述土壤孔隙的空间分布特征 ,Dm 与土壤孔隙面积、孔隙孔径分布、土壤深度以及土壤质地之间均具有一定的联系     

用于研究土壤孔隙三维结构的连续数字图像的制备

主要介绍了一种用于分析土壤孔隙三维结构的连续数字图像的制备方法，内容包括土样的处理、不同格式图像之间的转换、孔隙结构分析指标等；对一类具体土壤的研究结果表明，在2cm的间距内，土壤孔隙结构仍存在较为明显的异质性。     

利用土壤切片和数字图像方法土壤孔隙的垂直空间变异性

利用土壤切片和数字图像方法研究了两种土壤类型的也隙度在ｍｍ尺度上的垂直空间变异性。结果表明,土壤孔障度在ｍｍ尺度上存在较为明显的变异同的土壤变异性一定的差异;变异性与土壤孔隙呈反向关系,与工呈正向关系,在耕作期,变异性随呈现定一的波动趋势。     

耕作方式对土壤水动态变化及夏玉米产量的影响

一个连续２年的田间耕作试验在夏玉米生长期内完成,分析对比３种不同耕作方式对土壤水动态变化过程及对作物产量的影响。耕作扰动对土壤水动态变化的影响是明显的,夏玉米生长初期免耕下的表层土壤持有较高的水分,这归因于土壤非耕扰动、冬小麦残茬覆盖以及耕层土壤孔隙尺度分布的变化;另一方面,深松土壤受到耕作活动的强烈干扰,苗期耕层土壤蓄水明显小于传统耕作。耕作方式对土壤水差异的影响伴随着作物的生长发育过程显著减弱。深松耕作对作物根系生长发育状况及作物增产效果的作用是十分显见的。     

基于全卷积网络的土壤断层扫描图像中孔隙分割

针对土壤断层扫描图像中存在部分容积效应及因孔隙成分复杂、结构不规则等引起的分割精度低的问题,该文提出一种全卷积网络(fully convolutional network,FCN)土壤孔隙分割方法,为土壤科学研究提供技术支持。该文以黑土土壤断层扫描图像为研究对象,通过卷积和池化运算输出不同尺度的孔隙特征图;将孔隙的深层特征和浅层特征相融合,采用上采样算子对融合特征进行插值操作,从而输出孔隙的二值图。与大津法、分水岭法、区域生长法和模糊C均值聚类法(Fuzzy C-means,FCM)4种常用孔隙分割方法的对比结果表明,FCN法在低,中,高3种孔隙密度的土壤图像中优于其他4种方法。FCN法的平均分割正确率为98.1%,比4种常用方法分别高25.6%,48.3%,55.7%和9.5%;FCN法的平均过分割率和欠分割率分别为2.2%和1.3%,仅为次优方法(FCM法)的33.8%和23.6%。通过融合土壤孔隙结构的多重特征,FCN法能够实现土壤孔隙整体和局部信息的精准判断,为土壤学的研究提供了一种更加智能化的技术手段。     

轮耕对土壤物理性状和冬小麦产量的影响

针对华北地区土壤连续单一耕作存在的主要问题,进行了土壤轮耕效应的研究。试验选择冬小麦夏玉米玉两熟区连续5 a免耕田,设置免耕、翻耕和旋耕3种轮耕处理（即免耕一免耕,免耕一翻耕和免耕一旋耕）,冬小麦播种前进行耕作处理。研究结果表明：多年免耕后进行土壤耕作（翻耕、旋耕）可以显著降低土壤体积质量;旋耕显著降低0～10 cm土壤体积质量,翻耕则降低0～20 cm体积质量;随时间变化各处理土壤体积质量差异逐渐降低。翻耕、旋耕均显著增加了0～10 cm土壤总孔隙,同时翻耕显著增加了10～20 cm土壤总孔隙;翻耕、旋耕显著提高了5～10 cm毛管孔隙。0～10 cm土壤饱和导水率表现为旋耕＞翻耕＞免耕,翻耕、旋耕在5%水平上显著高于免耕;10～20、20～30 cm土层均表现为翻耕＞旋耕＞免耕,且10～20 cm翻耕5%水平上显著高于免耕;饱和导水率与体积质量呈显著线性负相关。翻耕、旋耕有效穗数与免耕相比分别提高了24.1%、22.3%;冬小麦的实际产量表现为：旋耕＞翻耕＞免耕,翻耕、旋耕分别比免耕增产11.8%、16.9%。总之,长期免耕后进行土壤耕作有利于改善土壤物理性状,提高作物产量。     

土壤样本分形几何特征的图像分析方法

土壤水分的持留与水分和溶质的运移,在很大程度取决于土壤孔隙的几何性质,而用数字图像处理方法能够较准确地加以识别。本文采用面积法和周长法分别计算了土壤样本图像的孔隙分形维数和多重孔隙轮廓线分形维数,并分析了它们同土壤质地之间的关系。结果表明,土壤孔隙空间在一定尺度范围内才存在明显的分形特征,土壤质地越细(粘粒含量越高)分形维数越大。与周长法相比较,用面积法计算的孔隙分形维数对土壤质地更为敏感。     

免耕制度下耕作土壤结构演化的数字图像分析

利用数字图像方法 ,研究了室内模拟和自然田间免耕下的土壤结构演化状况。图像定性表明随免耕时间的推移土壤中团聚体由小变大 ,土体逐渐趋于紧实。图像定量分析结果表明随着免耕时间的推移 ,小团聚体和小孔隙减少而大团聚体和大孔隙增加。田间土样的结果表明了自然的阶段性降雨对土壤结构可能具有一定的调节作用 ,但从图像与团聚体粒径分布和孔隙孔径分布的变化来看 ,免耕是否有利于土壤性质和农业生产尚很难确定     

Three-dimensional analysis of a loamy-clay soil using pore and solid chord distributions

Examples of pore and solid chord distributions obtained for ideal porous media are presented, and the distributions of the porous and solid phases of a soil have been studied by pore and solid chord distributions. Serial sections, 100 μm apart, were cut in a soil core, impregnated with resin, and images were obtained of them. The 2D images from 160 sections were used to build a 3D reconstruction of the core. The initial 2D images, the 3D reconstructed medium and 2D computed images from the latter were studied. We found that the solid matrix of the particular soil is homogeneous and isotropic at the scale studied, and it could be characterized with a single 2D image. For pores ranging from 500 μm to 2 mm we also found similar pore chord distributions for the 2D images in the three orthogonal directions and the 3D medium. A single 2D image can be used to study these pores. For larger pores more than one 2D image is required, and we showed that eight 2D images are sufficient to describe these pores.     

土壤图像孔隙轮廓线分形特征及其应用

孔隙轮廓线分形维数是用来描述孔隙本身所具有的分形特征，它反映了土壤孔隙与固体颗粒接触界限的不规则性。该文采用数字图像处理技术研究了中国科学院封丘农业生态实验站内3种不同质地土壤样本图像的孔隙轮廓线分形特征。结果表明：土壤孔隙轮廓线分形特征与土壤质地之间存在着一定的相关关系，即土壤质地越细(黏粒含量越高)分形维数取值越大。同时，根据这3种土壤的孔隙轮廓线分形维数，结合不同分形方法分别预测了它们的水分特征曲线，并与实测的土壤水分特征曲线进行了比较。     

预测土壤水力性质的形态学网络模型应用研究

土壤水力性质是研究非饱和带中水分和溶质运移的重要参数 ,可以用孔隙网络模型进行预测。通常采用的网络模型中的参数是任意指定的 ,无法真实反映土壤孔隙空间的形态特征。本文采用了一种基于孔隙形态学的网络模型来预测土壤的水力性质 ,即通过图像分析来直接测定孔隙的大小分布及其连通性 ,并将其结合到网络模型中 ,最大程度地再现了三维的土壤孔隙结构。本文根据河南封丘地区采集的砂壤土样本图像分析结果 ,采用形态学网络模型预测了其水力性质 ,同时也进一步评价了这种模型的优缺点及其应用前景。     

The relation between silty soil structures and their mercury porosimetry curve counterparts: fractals and percolation

Mercury porosimetry data can be interpreted in terms of soil structure using ideas drawn from (i) network modelling and percolation theory and (ii) fractal geometry. We linked mercury intrusion to soil structure quantified by image analysis within a relevant common pore radius scale. We compared (i) three independent methods for computing fractal dimensions of the matrix and of the solid–pore interface, namely fitted square boxes method and pore chord distribution on scanning electron microscope images of soil thin sections, and mercury porosimetry, and (ii) two independent methods for characterizing pore connectivity (image analysis) and percolation process (pressure threshold from mercury porosimetry). The results from analyses of the pore size distribution by mercury porosimetry differed from those from the image analysis. Mercury intrusion is controlled by both the connectivity of the pore space network and locally by pore throats leading to larger pore bodies. By contrast, image analysis is unaffected by pore connectivity and measures pore bodies. On the other hand, the chord length method might not adequately capture the scaling properties of the solid–pore interface, whereas the mercury porosimetry data were also difficult to interpret in terms of fractal geometry because of the effects of pore connectivity. However, fractal dimension values of both the solid phase and the solid–pore interface increased as a function of clay content, whereas both percolation probability values and throat radius values at the mercury percolation threshold decreased. The results show the merit of applying both fractals and percolation theory for determining structural parameters relevant to mercury and water transport in soil.     

孔隙结构图像分析中不同试验因素对分析结果的影响

研究了不同试验因素 (包括图像分辨率、土壤切片定向性、分析区域大小等 )对孔隙结构图像分析结果的影响以及试验误差 ,并简要介绍了用于分析土壤孔隙结构的土壤切片及数字图像的制备技术。结果表明 :不同试验因素对分析结果均会产生一定的影响 ,试验误差约在 1 0 %。这表明同一研究中应保持试验条件的相对一致以保证分析结果的可靠性和可比性     

Structural differences between bulk and rhizosphere soil

The physical characteristics of the soil at the root–soil interface are crucial because they determine both physical aspects of root function such as water and nutrient uptake and the microbial activity that is most relevant to root growth. Because of this we have studied how root activity modifies the structure and water retention characteristic of soil adjacent to the root for maize, wheat and barley. These plants were grown in pots for a 6‐week growth period, then the soil adjacent to the root (rhizosphere soil) and bulk soil aggregates were harvested. These soil aggregates were then saturated and equilibrated at matric potentials between ?600 kPa and saturation, and the water retention characteristics were measured. From subsamples of these aggregates, thin sections were made and the porosity and pore‐size distributions were studied with image analysis. Both image analysis and estimates of aggregated density showed that the rhizosphere soil and bulk soil had similar porosities. Growing different plants had a small but significant effect on the porosity of the soil aggregates. Image analysis showed that for all the plant species the structure of the rhizosphere soil was different to that of the bulk soil. The rhizosphere soil contained more larger pores. For maize and barley, water retention characteristics indicated that the rhizosphere soil tended to be drier at a given matric potential than bulk soil. This effect was particularly marked at greater matric potentials. The difference between the water retention characteristics of the bulk and rhizosphere soil for wheat was small. We compare the water retention characteristics with the data on pore‐size distribution from image analysis. We suggest that differences in wetting angle and pore connectivity might partly explain the differences in water retention characteristic that we observed. The impact of differences between the water retention properties of the rhizosphere and bulk soil is discussed in terms of the likely impact on root growth.