Spatial interpolation of soil nutrients using algebra hyper-curve neural network

Spatial distribution of and interpolation methods for soil nutrients are the basis of soil nutrient management in precision agriculture.For study of application potential and characteristics of algebra hyper-curve neural network(AHCNN)in delineating spatial variability and interpolation of soil properties,956 soil samples were taken from a 50 hectare field with 20 m interval for alkaline hydrolytic nitrogen measurement.The test data set consisted of 100 random samples extracted from the 956 samples,and the training data set extracted from the remaining samples using 20,40,60,80,100 and 120 m grid intervals.Using the AHCNN model,three training plans were designed,including plan AHC1,using spatial coordinates as the only network input,plan AHC2,adding information of four neighboring points as network input,and plan AHC3,adding information of six neighboring points as network input.The interpolation precision of AHCNN method was compared with that of Kriging method.When the number of training samples was big,interpolation precisions of Kriging and AHCNN were similar.When the number of training samples was small,the precisions of both methods deteriorated.Since AHCNN method has no request on data distribution and it is non-linearization of neutron input variables,it is suitable for delineation of spatial distribution of nonlinear soil properties.In addition,AHCNN has an advantage of adaptive self-adjustment of model parameters,which makes it proper for soil nutrient spatial interpolation.After comparison of mean absolute error d,root mean squared error RMSE,and mean relative error%d,and the spatial distribution maps generated from different methods,it can be concluded that using spatial coordinates as the only network input cannot simulate the characteristics of soil nutrient spatial variability well,and the simulation results can be improved greatly after adding neighboring sample points’information and the distance effect as network input.When the number of samples was small,interpolation precision can be improved after properly increasing the number of neighboring sample points.It was also showed that evaluation of interpolation precision using conventional error statistic indexes was defective,and the spatial distribution map should be used as an important evaluation factor.

Optimised Spatial Sampling Scheme for Soil Electriclal Conductivity Based on Variance Quad-Tree (VQT) Method

The acquisition of precise soil data representative of the entire survey area, is a critical issue for many treatments such as irrigation or fertilization in precision agriculture. The aim of this study was to investigate the spatial variability of soil bulk electrical conductivity （ECb） in a coastal saline field and design an optimized spatial sampling scheme of ECb based on a sampling design algorithm, the variance quad-tree （VQT） method. Soil ECb data were collected from the field at 20 m interval in a regular grid scheme. The smooth contour map of the whole field was obtained by ordinary kriging interpolation, VQT algorithm was then used to split the smooth contour map into strata of different number desired, the sampling locations can be selected within each stratum in subsequent sampling. The result indicated that the probability of choosing representative sampling sites was increased significantly by using VQT method with the sampling number being greatly reduced compared to grid sampling design while retaining the same prediction accuracy. The advantage of the VQT method is that this scheme samples sparsely in fields where the spatial variability is relatively uniform and more intensive where the variability is large. Thus the sampling efficiency can be improved, hence facilitate an assessment methodology that can be applied in a rapid, practical and cost-effective manner.

Spatial Variability of Soil Carbon to Nitrogen Ratio and Its Driving Factors in Ili River Valley,Xinjiang,Northwest China

Soil carbon to nitrogen(C/N) ratio is one of the most important variables reflecting soil quality and ecological function,and an indicator for assessing carbon and nitrogen nutrition balance of soils.Its variation reflects the carbon and nitrogen cycling of soils.In order to explore the spatial variability of soil C/N ratio and its controlling factors of the Ili River valley in Xinjiang Uygur Autonomous Region,Northwest China,the traditional statistical methods,including correlation analysis,geostatistic alanalys and multiple regression analysis were used.The statistical results showed that the soil C/N ratio varied from 7.00 to 23.11,with a mean value of 10.92,and the coefficient of variation was 31.3%.Correlation analysis showed that longitude,altitude,precipitation,soil water,organic carbon,and total nitrogen were positively correlated with the soil C/N ratio(P < 0.01),whereas negative correlations were found between the soil C/N ratio and latitude,temperature,soil bulk density and soil p H.Ordinary Cokriging interpolation showed that r and ME were 0.73 and 0.57,respectively,indicating that the prediction accuracy was high.The spatial autocorrelation of the soil C/N ratio was 6.4 km,and the nugget effect of the soil C/N ratio was 10% with a patchy distribution,in which the area with high value(12.00–20.41) accounted for 22.6% of the total area.Land uses changed the soil C/N ratio with the order of cultivated land > grass land > forest land > garden.Multiple regression analysis showed that geographical and climatic factors,and soil physical and chemical properties could independently explain 26.8%and 55.4% of the spatial features of soil C/N ratio,while human activities could independently explain 5.4% of the spatial features only.The spatial distribution of soil C/N ratio in the study has important reference value for managing soil carbon and nitrogen,and for improving ecological function to similar regions.

Sampling Strategies for Soil Available K and P at Field Scale

Field nutrient distribution maps obtained from the study on soil variations within fields are the basis of precision agriculture. The quality of these maps for management depends on the accuracy of the predicted values, which depends on the initial sampling. To produce reliable predictions efficiently the minimal sampling size and combination should be decided firstly, which could avoid the misspent funds for field sampling work. A 7.9 hectare silage field close to the Agricultural Research institute at Hillsborough, Northern Ireland, was selected for the study. Soil samples were collected from the field at 25 m intervals in a rectangular grid to provide a database of selected soil properties. Different data combinations were subsequently abstracted from this database for comparison purposes, and ordinary kriging used to produce interpolated soil maps. These predicted data groups were compared using least significant difference (LSD) test method. The results showed that the 62 sampling sizes of triangle arrangement for soil available K were sufficient to reach the required accuracy. The triangular sample combination proved to be superior to a rectangular one of similar sample size.

Multi-scale spatial relationships between soil total nitrogen and influencing factors in a basin landscape based on multivariate empirical mode decomposition

The relationships between soil total nitrogen(STN)and influencing factors are scale-dependent.The objective of this study was to identify the multi-scale spatial relationships of STN with selected environmental factors(elevation,slope and topographic wetness index),intrinsic soil factors(soil bulk density,sand content,silt content,and clay content)and combined environmental factors(including the first two principal components(PC1 and PC2)of the Vis-NIR soil spectra)along three sampling transects located at the upstream,midstream and downstream of Taiyuan Basin on the Chinese Loess Plateau.We separated the multivariate data series of STN and influencing factors at each transect into six intrinsic mode functions(IMFs)and one residue by multivariate empirical mode decomposition(MEMD).Meanwhile,we obtained the predicted equations of STN based on MEMD by stepwise multiple linear regression(SMLR).The results indicated that the dominant scales of explained variance in STN were at scale 995 m for transect 1,at scales 956 and 8852 m for transect 2,and at scales 972,5716 and 12,317 m for transect 3.Multi-scale correlation coefficients between STN and influencing factors were less significant in transect 3 than in transects 1 and 2.The goodness of fit root mean square error(RMSE),normalized root mean square error(NRMSE),and coefficient of determination(R2)indicated that the prediction of STN at the sampling scale by summing all of the predicted IMFs and residue was more accurate than that by SMLR directly.Therefore,the multi-scale method of MEMD has a good potential in characterizing the multi-scale spatial relationships between STN and influencing factors at the basin landscape scale.

Research on Influencing Factors of Spatial Variation of Soil Nutrients

Based on previous studies,the research methods and influencing factors of spatial variation of soil nutrients are summarized.It is concluded that the spatial variation of soil nutrients is studied generally by geostatistics methods,and the spatial distribution of nutrients is visually observed by using Kriging interpolation method.The influencing factors mainly include topography,sampling method,sampling spacing,sampling density and sampling scale.The influence of random sampling and grid sampling on interpolation is analyzed based on the specific conditions of the actual study area.The influence of sampling density and topography on the spatial variation of soil nutrients cannot be ignored,especially on available nutrients.When samples are collected in a large area(under a small and medium scale),the spatial variation of soil nutrients is large,and they have strong spatial autocorrelation;in a small area(namely under a large scale),the spatial variability of soil nutrients is small,and they have obvious spatial autocorrelation.This study can provide intuitive and convenient reference materials for the following researchers.

基于环境变量筛选与机器学习的土壤养分含量空间插值研究

为了提高农田土壤养分含量空间插值精度,准确掌握土壤养分的空间分布特征,以新疆玛纳斯河流域绿洲为研究区域,测定土壤有机质含量、全氮含量、有效磷含量、速效钾含量、pH值和盐分含量,协同经度、纬度、地形、气象和植被指数因子作为环境变量,经过皮尔逊相关系数(Person correlation coefficient,PCC)、方差膨胀系数(Variance inflation factor,VIF)和极端梯度提升(Extreme gradient boosting,XGBoost)算法进行变量筛选,采用决策树(Decision tree,DT)、随机森林(Random forest,RF)、径向基函数神经网络(Radial basis function,RBF)和长短期记忆网络(Long short-term memory,LSTM)4种机器学习模型与普通克里格(Ordinary Kriging,OK)方法,对研究区农田土壤有机质、全氮、有效磷和速效钾含量进行空间插值。结果表明:研究区土壤有机质、全氮、有效磷、速效钾含量分别为0.226~32.275 g/kg、0.117~1.272 g/kg、3.159~53.884 mg/kg和81.510~488.422 mg/kg,变异系数为30.636%~43.648%,均属于中等程度变异。PCC、VIF和XGBoost变量筛选均表明,土壤有机质、全氮、有效磷和速效钾间具有一定的关联性,可用于目标属性空间插值的环境变量,但不同变量筛选方法对经度、纬度、地形、气象和植被指数因子筛选结果具有一定的差异性。XGBoost方法可以更有效地筛选出对空间插值结果重要的环境变量,且基于此方法筛选变量后建立的模型精度明显优于PCC和VIF筛选变量后建立的模型精度,而且协同环境变量的机器学习模型精度普遍优于未加入环境变量的OK模型精度,同一土壤养分含量空间插值模型精度从大到小依次为RF、LSTM、RBF、DT、OK,其中基于XGBoost筛选出的变量对土壤有机质、全氮、有效磷和速效钾含量构建的RF空间插值模型精度相较于未加入环境变量的OK模型有显著提高,决定系数分别提高43.02%、101.00%、86.04%和137.89%,均方根误差分别降低27.39%、42.78%、13.12%和28.39%,平均绝对误差分别降低29.01%、43.84%、11.20%和29.62%。利用RF模型对研究区农田土壤养分进行反演得到土壤有机质和全氮含量具有较强的空间分布一致性,含量较高的主要集中在研究区南部和东部区域,有效磷和速效钾含量具有一定的空间相似性,东南部、中北部区域含量较低。综上,XGBoost变量筛选方法结合RF模型可以更好地实现土壤养分空间插值,可作为土壤养分空间插值的有效方法。

顺平县平原区耕地土壤有机质预测精度研究——基于采样尺度和空间插值方法

为研究县域尺度土壤调查最适宜的采样尺度和插值方法,以河北省顺平县山前平原区为例,按照8点/km^(2)点位密度采集1532个土壤样品并测定土壤有机质。从样点中抽取250m×250m、500m×500m、1km×1km、2km×2km 4个采样尺度样本子集,结合地统计学和GIS对有机质空间变异特征进行分析,并对不同采样尺度和插值方法下有机质含量空间插值结果进行分析比较。结果表明:研究区有机质含量具有中等变异性,普通克里金法插值精度优于规则样条函数法和反距离权重法。为保证研究区有机质空间变异评价精度,较为适宜的采样尺度为1km×1km。

Optimal placement of two soil soundings for rectangular footings

This study uses a recently proposed algorithm for consideration of soil sounding locations in the bearing capacity estimations of spatially variable soil for rectangular footings.The objective of the study is to assess the possibility of indicating general guidelines for optimal soil sounding locations in the case of two soundings and rectangular footings.The possibility of proposing such general guidelines would be extremely valuable from the engineering practice point of view.Moreover,it would be promising for future studies concerning more complex foundation arrangements.For this reason,numerous scenarios are analyzed for a variety of vertical and horizontal fluctuation scales and a variety of rectangular foundation lengths.For generality of the results,two correlation structures are considered,i.e.the Gaussian and the Markovian ones.The optimal sounding location results are discussed.The observations indicate that,for a specified vertical fluctuation scale,all optimal borehole locations in dimensionless coordinates form a curve.This phenomenon can be utilized in practical applications.The potential applications of the obtained results and the directions for future studies in this area are also discussed.

Spatial variability of soil resistivity and rational sampling for corrosion assessment of carbon steel in Daqing area

Soil resistivity is one of the key indicators of the corrosive classification assessment on metal materials in soil environment. This paper presents variance characters of various quantity of soil resistivity samples data based on the semi-variance function methods of Geo-statistical Analysis by analyzing the regional soil resistivity sampling data in Daqing area. Furthermore, the variance of the soil resistivity as well as entire soil circumstance due to different sampling amounts are also analyzed and compared by means of using the characteristic parameters of the semi-variance function. In addition, this work also studied the rational sampling quantities according to various measurement errors required and evaluated the local soil corrosivity on carbon steel based on the actual measuring data in this area.

Sampling Size Requirements to Delineate Spatial Variability of Soil Properties for Site-Specific Nutrient Management in Rubber Tree Plantations

Site-specific nutrient management is an important strategy to promote sustainable production of rubber trees in order to obtain high yields of natural rubber. Making effective nutrient management decisions for rubber trees depend on knowing the spatial variations of soil fertility properties in advance. In this study the Kriging geostatistical method was used to examine the spatial variability of soil total nitrogen(TN), organic matter(OM), available phosphorus(AP) and available potassium(AK) in a typical hilly rubber tree plantation in Hainan, China. The spatial variability of the soils was small for the TN and OM and had medium variability for the AP and AK variables. Anisotropic semivariograms of all soil properties revealed that elevation and building contour ledge can profoundly affect the spatial variability of soil properties in the plantation, except for the AK variable. Soil samples had to be collected in alignment with the direction of elevation and perpendicular to the direction of building contour ledges, which was needed to obtain more reliable information within the study area in the rubber tree plantation. In formulating a sample scheme for AK, the distribution features of the soil’s parent material should be considered as the influence factor in the study field. The Kriging method used to guide the soil sampling for spatial variability dertermination of soil properties was about 2-5 times more efficient than the classic statistical method.

玉米南繁基地土壤速效氮空间变异性对采样间距的响应研究

【目的】探明玉米南繁基地土壤速效氮含量的空间变异性及其尺度效应,确定合理的采样间距,为精确土壤采样设计提供理论依据。【方法】以玉米南繁地块为试验样地,按10 m采样间距采集土样且为原始尺度,并以20、30、40 m距离依次抽样形成不同采样间距数据集,再利用空间自相关莫兰指数、半方差函数及分形维数分析不同采样间距条件下土壤速效氮含量的空间结构变化规律。【结果】试验地的土壤速效氮平均含量为90.24 mg/kg,属中等水平,变异系数为19.77%,属中等变异范围;在10、20、30、40 m等采样间距中,土壤速效氮样本数据结构变化不大,仅变异系数差异较大,其中以20m采样间距的变异系数最大、为29.37%,其他采样间距相差不大;随着采样间距增大,土壤速效氮含量的空间自相关莫兰指数呈减少趋势,其空间聚集性变弱,而块金值、块基比和分形维数呈增加趋势,表明随机因素所引起的空间变异逐渐增多,随机因素逐渐削弱结构因素引起的空间变异,使空间相关性逐渐变弱;20 m采样间距的土壤速效氮含量空间分布与10 m采样间距原始尺度一致,其空间预测的均方根误差值最小,空间预测精度为最高。【结论】随着采样间距增大,土壤速效氮含量的空间自相关性逐渐减弱,表征出随机因素所引起的空间变异性逐渐增强,结构因素引起的空间变异性逐渐削弱,其中20m采样间距的土壤速效氮含量空间变异特征和空间分布与10m采样间距原始尺度相似,且空间预测精度最高,是较为合理的采样间距,本研究可为玉米南繁基地的氮肥精准施用管理提供技术支撑。

基于时空变异的旱地土壤有机碳含量最优预测插值方法

以3.93×10^(4)km^(2)江苏北部旱地为试验区,基于1980年和2008年两期土壤样点数据,采用8种确定性和4种地统计学插值方法预测不同时期土壤有机碳含量,并筛选出时空演变下适宜的插值方法。结果表明:1980—2008年试验区土壤有机碳含量呈上升趋势,而空间变异性和自相关性有所减弱;1980年试验区土壤有机碳含量的最优预测插值方法为反高次曲面函数法,其相关系数和均方根误差分别为0.57和2.08 g·kg^(-1);2008年试验区土壤有机碳含量的最优预测插值方法为普通克里金法,其相关系数和均方根误差分别为0.60和2.17 g·kg^(-1)。总体来看,试验区土壤有机碳含量的时空变异性对不同插值方法的预测精度有很大影响。

Limited Spatial Transferability of the Relationships Between Kriging Variance and Soil Sampling Spacing in Some Grasslands of Ireland:Implications for Sampling Design

Sampling plays an important role in acquiring precise soil information required in modern agricultural production worldwide, which determines both the cost and quality of final soil mapping products. For sampling design, it has been proposed possibile to transfer the relationships between kriging variance and sampling grid spacing from an area with existing information to other areas with similar soil-forming environments. However, this approach is challenged in practice because of two problems: i) different population vaxiograms among similar areas and ii) sampling errors in estimated variograms. This study evaluated the effects of these two problems on the transferability of the relationships between kriging variance and sampling grid spacing, by using spatial data simulated with three variograms and soil samples collected from four grasslands in Ireland with similar soil-forming environments. Results showed that the variograms suggested by different samples collected with the same grid spacing in the same or similar areas were different, leading to a range of mean kriging variance (MKV) for each grid spacing. With increasing grid spacing, the variation of MKV for a specific grid spacing increased and deviated more from the MKV generated using the population variograms. As a result, the spatial transferability of the relationships between kriging variance and grid spacing for sampling design was limited.

不同尺度下土壤养分空间变异特征的研究

在不同取样尺度下采集了农田土壤样品并测定土壤有机质、全N和有效P含量,研究不同尺度下地统计学方法对土壤养分空间变异分析的适用性和空间插值质量.研究结果表明,大比例尺下土壤有机质含量、有效P含量、全N含量的分布呈正态分布,在中小比例尺下呈现轻微偏态分布.各养分空间变异系数存在明显空间变异(C.V.介于17%～27%之间),大比例尺下,空间变异较小;中小比例尺下,空间变异程度大(C.V和峰度值较小).在各种尺度下,土壤养分均存在典型的半方差结构,表明在各种取样比例尺下采用地统计学方法进行土壤养分空间变异分析都是可行的.在大中比例尺下,各养分块金系数介于0.28～0.38之间,表明土壤养分具有明显空间自相关;而在小比例尺下,块金系数介于0～0.17之间,表明土壤养分具有强烈空间自相关,其中有机质含量具有恒定的空间自相关(块金系数为0),有效P含量的空间自相关程度相对较弱;空间插值和交互检验结果表明,球状模型空间插值结果与实测结果符合性很好,交互检验均达到极显著水平.在小比例尺下其估计比大比例尺下精度更高.不同养分的空间预测结果质量不尽相同,土壤有机质含量空间预测符合度最高.

太湖流域土壤养分空间变异特征分析——以浙江省平湖市为例

以位于太湖流域的浙江省平湖市为例,采用GIS与地统计学相结合的方法对其耕层土壤(0～20cm)的有机质、全氮、速效磷、速效钾等4种养分要素的空间变异特征进行了分析。结果表明:有机质和全氮变异函数曲线的理论模型能很好地符合球状模型;速效磷经对数转换后也符合球状模型;而速效钾的变异函数曲线则符合指数模型。有机质、全氮和速效磷具有很强的空间相关性,说明受母质、地形、土壤类型等自然的结构性因素影响较大;速效钾具有中等的空间相关性,说明由施肥、作物、管理水平等随机因素引起的空间异质性起很大作用。整个研究区东南部土壤质地较轻,大量施肥在提高其作物产量的同时,也可能会带来农业非点源污染等问题。

基于GIS和地统计学的土壤养分空间变异特征研究——以河北省遵化市为例

基于地理信息系统 (GIS)和地统计学研究了河北省遵化市土壤表层 ( 0～ 2 0cm )碱解氮、全氮、速效钾、速效磷和有机质等 5种养分要素的空间变异规律 .应用GIS能够将系统变量的属性数据同地理数据相结合 ,使大区域范围内进行地统计学分析变得较为方便 .研究表明 ,全氮、碱解氮、速效磷、有机质变异函数曲线的理论模型符合球状模型 ,速效钾的理论模型表现为指数模型和有基台值的线性模型的套合结构 ;碱解氮、全氮、速效磷、有机质的空间变异主要是由随机性因素引起的 ,但程度有所差异 ,全氮和有机质由随机性因素引起的空间异质性程度较高 ,碱解氮和速效磷较低 ;速效钾的空间变异则主要是由结构性因素引起 ;5种养分要素的空间自相关程度都属于中等的空间自相关 ,但空间变异的尺度范围不同 ,碱解氮和速效磷变异尺度基本相近 ,为 5和5 .5km ;全氮较大 ,为 14 .5km ;有机质为 8.5km ;速效钾的变异尺度有两个 ,0～ 3 .5km主要以指数模型为主 ,3 .5～ 2 5 .5km范围内以有基台值的线性模型为主 .5种养分要素的各向同性的范围不同 ,碱解氮和速效磷在整个范围 ( 0～ 2 8km )都表现出各向同性 ,全氮和有机质的其次 ,为 0～ 10km ;速效钾的较小 ,为 0～ 8km .

土壤水盐空间变异尺度效应的研究

水文学和土壤学中的尺度问题是目前水土科学研究的前沿课题之一。该文利用空间信息科学——地质统计学、根据设计的各种田间网格,在一维和二维(平面)空间中初步研究了黄河河套平原长胜试验区中小尺度的土壤水分和盐分随采样尺度所表现出的空间变异(结构性)的某些规律,表明采样尺度的划分和选取与水分和盐分的空间变异性大小有密切关系,尺度效应的研究对于指导农业技术研究中野外采样系统设计、节省外业调查的工作量及科学地进行内业计算、评估和揭示农业工程中具有地学特征的区域性自然规律有重要作用。

茂兰喀斯特原始森林土壤有机碳的空间变异性与代表性土样采集方法

茂兰原生林中20m×30m研究样地存在着7类小生境：石面、土面、石土面、石沟、石缝、石坑、石洞。样地内土壤有机碳的含量范围为40.1～203.5gkg^-1,样点与小生境间土壤有机碳含量的变异系数分别为43%和41%;同类型小生境间为22%～42%;小生境内为14%～57%;说明样地内土壤有机碳含量具有高度的空间变异性。为了促进岩溶山区土壤退化研究中数据之间的可比性,文中建议以土壤面积权重确定的小生境土样组成样地代表性土样的方法。选取样地内面积之和超过样地总土壤面积95%以上的几类小生境,由面积权重确定组成样地土壤代表样的各类小生境土样样品量,而各类小生境土样则分别由以面积权重确定的同类小生境样品量混合构成。利用该方法计算的研究样地土壤有机碳含量的面积权重值为92.1gkg^-1。

不同研究尺度下土壤有机质和全氮的空间变异特征——以北京市平谷区为例

在北京郊区面积为1075km^2的中尺度（平谷区）以400m×400m网格采样，共采集1076个样点，同时在平谷区内面积为28．8km^2的小尺度（马昌营镇）以100m×100m网格采样，共采集171个样点，测定其耕层土壤有机质和全氮的含量。应用传统统计学和地统计学方法，对两个研究尺度下的数据进行了分析，结果表明：有机质和全氮的变异系数范围为0．31～0．40，均属中等变异强度，随着研究尺度的缩小，土壤全氮的变异系数减少。半方差函数分析结果表明，两个研究尺度下，有机质和全氮均在一定范围内存在空间相关关系，区级尺度有机质和全氮的空间相关距离较大，分别为88．2km和4．9km，镇级尺度有机质和全氮的变程较小，均为0．7km，它们的空间异质性均主要由结构性因素引起。采用Kriging最优内插法对未测点进行了估值，绘制了等值线图，两个研究尺度下的有机质和全氮含量受地形、土壤类型、土地利用方式、施肥等因素的影响，均表现出明显不同的分布规律。