Spatial prediction of categorical variables with the Bayesian Maximum Entropy approach: the Ooypolder case study

Categorical variables such as water table status are often predicted using the indicator kriging (IK) formalism. However, this method is known to suffer from important limitations that are most frequently solved by ad hoc solutions and approximations. Recently, the Bayesian Maximum Entropy (BME) approach has proved its ability to predict categorical variables efficiently and in a flexible way. In this paper, we apply this approach to the Ooypolder data set for the prediction of the water table classes from a sample data set. BME is compared with IK using global as well as local criteria. The inconsistencies of the IK predictor are emphasized and it is shown how BME permits avoiding them.     

Pedometric mapping of soil organic matter using a soil map with quantified uncertainty

This paper compares three models that use soil type information from point observations and a soil map to map the topsoil organic matter content for the province of Drenthe in the Netherlands. The models differ in how the information on soil type is obtained: model 1 uses soil type as depicted on the soil map for calibration and prediction; model 2 uses soil type as observed in the field for calibration and soil type as depicted on the map for prediction; and model 3 uses observed soil type for calibration and a pedometric soil map with quantified uncertainty for prediction. Calibration of the trend on observed soil type resulted in a much stronger predictive relationship between soil organic matter content and soil type than calibration on mapped soil type. Validation with an independent probability sample showed that model 3 out‐performed models 1 and 2 in terms of the mean squared error. However, model 3 over‐estimated the prediction error variance and so was too pessimistic about prediction accuracy. Model 2 performed the worst: it had the largest mean squared error and the prediction error variance was strongly under‐estimated. Thus validation confirmed that calibration on observed soil type is only valid when the uncertainty about soil type at prediction sites is explicitly accounted for by the model. We conclude that whenever information about the uncertainty of the soil map is available and both soil property and soil type are observed at sampling sites, model 3 can be an improvement over the conventional model 1.     

Evaluating taxonomic adjacency as a source of soil map uncertainty

Soil types or map units are considered to be taxonomically adjacent if they differ in only one criterion, defined by an arbitrary threshold value. By treating soil types as nodes of a graph and taxonomic adjacency as the graph edges connecting nodes, algebraic graph theory can be used to produce a measurement of the uncertainty in a soil map associated with arbitrary classification boundaries between soil types. The largest eigenvalue of the adjacency matrix of a graph, the spectral radius, is an indication of network complexity. A larger spectral radius indicates a more complex network, and a greater degree of uncertainty or potential error associated with taxonomic adjacency. Benchmark values of spectral radius for cases of no taxonomic adjacency, including a single pair of adjacent soils, a chain or cycle‐type graph structure and a fully connected graph, are established so that taxonomic adjacency indices based on the spectral radius can be established. Examples are shown from two contrasting USA soil landscapes in the Ouachita Mountains, Arkansas, and the coastal plain of North Carolina, using both US Soil Taxonomy and the world reference base. The taxonomic adjacency indices are also useful in assessing soil richness and pedodiversity, with smaller values indicating a greater likelihood that identified soils represent distinct entities.     

Integrating soil map information in modelling the spatial variation of continuous soil properties

This paper presents two indicator algorithms that integrate soil map information into modelling the spatial variation of continuous soil properties: these are simple indicator kriging with varying means and the Markov–Bayes algorithm. Both methods are used to evaluate probabilities for copper and cobalt deficiencies in the Borders Region of Scotland. Results are compared with maps obtained by the polygonal method (Thiessen polygons) and an indicator kriging algorithm that does not use soil map information. Accounting for soil map information is shown to improve delineation of the deficient areas, especially where the sampling is sparse. Test locations are classified as deficient or not so as to minimize an expected cost of mis-classification that is derived from local probability distributions of copper or cobalt and functions measuring the cost of overestimating or underestimating metal concentrations. The comparison of classification results with actual copper and cobalt concentrations at test locations shows that the two proposed algorithms can decrease substantially the economic loss attached to misclassification.     

利用高程辅助进行土壤有机质的随机模拟

为了探讨在条件模拟计算环境下,是否可以利用高程数据辅助提高土壤有机质空间变化的预测精度及相应的预测不确定性模拟的准确性,该文在北京市平谷区内选取研究样区,以土壤有机质作为目标变量,一方面利用序贯高斯模拟法对土壤有机质的空间分布进行模拟,另一方面以高程作为辅助信息,利用序贯高斯协模拟法对土壤有机质的空间分布进行模拟,然后对两种方法的模拟结果进行对比分析。结果表明,在土壤有机质的空间预测精度、模拟预测结果的局部不确定性和模拟预测结果的空间不确定性三方面,通过将高程数据考虑进有机质条件模拟过程中,准确性都得到了提高。这对于农业可持续发展以及全球碳平衡研究都具有十分重要的意义。     

Spatial interpolation of orchard soil pH using soil type and planting duration as auxiliary information

Intensified field management in orcahrds has resulted in significant and widespread acidification in the soils. However, effectively mapping the spatial patterns of soil pH aiming to support ecological management is impeded by its large variotions across soil types and planting durations. Kriging methods were used to integrate soil type and planting duration information for effective mapping of orchard soil pH in a case study in orchards of the Northeast Jiaodong Peninsula, East China. A total of 1 472 surface soil samples were collected, and the planting duration of each sampled orchard was acquired to generate a planting duration map via Voronoi tessellations. The performance of five kriging methods was compared, namely, ordinary kriging (OK), OK combined with soil type (OK_ST), OK combined with planting duration (OK_PD), cokriging combined with soil type and planting duration (OCK_STPD), and OK combined with soil type and planting duration (OK_STPD). Results showed that soil pH declined significantly with increasing planting duration and exhibited moderate spatial variability over the study area. Soil type and planting duration both had significant influence on the spatial distribution of soil pH. The OCK_STPD and OK_STPD methods showed better prediction efficiency than OK, OK_ST, or OK_PD. With regard to the predicted maps of soil pH, the OCK_STPD and OK_STPD methods highly reflected local variations associated with soil type and planting duration, but the OK method was poorly representative. Categorical soil type and planting duration information may be used as ancillary information to improve the mapping quality of orchard soil pH. The OCK_STPD and OK_STPD methods were practical and efficient methods for interpolating orchard soil pH in the study area. The resultant high-quality soil pH maps can contribute to improved site-specific management in the orchards.     

Refining soil survey information for a Dutch soil series using land use history

Abstract. Differences in land-use history within soil series, although not influencing soil classification, lead to variability of non-diagnostic soil properties in soil databases. Regional studies that use soil databases are confronted with this considerable variability. This has, for example, been reported in regional studies focused on nitrate leaching from agricultural land. Such findings have a direct impact on regional assessments of nitrate leaching from dairy farms on sandy soils, a major environmental issue in the Netherlands. There is thus a need to deal with this variability in soil properties.
We were able to relate soil organic nitrogen, soil organic carbon and its dynamics to land use history for a Dutch sandy soil series. Within one soil series, three different land use histories were identified: old grassland, reseeded grassland and grassland converted from continuous cropping with silage maize. The addition of landscape characteristics significantly improved the regression models based on land-use only. Once established for any given soil series, such relationships can significantly improve soil survey input into dynamic models of soil behaviour such as regional nitrate leaching studies.     

应用Arcview制作土壤养分图技术

简要介绍了使用桌面地理信息系统软件Arcview制作土壤养分图的技术过程，包括野外样品采集、室内分析、建立土壤采样点图层及地块边界图层，使用Arcview空间分析拓展模块中Kriging插值法进行养分插值，从而制作养分等值线图。该方法同样可用于土壤其他属性的空间变异的研究。     

Rapid prediction of soil mineralogy using imaging spectroscopy

Hyperspectral images provide rich spectral and spatially continuous information that can be used for soil mineralogy discrimination. This paper proposes a method to evaluate the feasibility of Hyperion image in the rapid prediction of soil mineralogy. Four areas in Egypt were chosen for the current study. Preprocessing of the Hyperion data was done before applying the atmospheric correction. The minimum noise fraction transformation was used to segregate noise in the data. Various techniques were applied to the studied areas in which mixture tune matched filtering gave good results in a prediction of the end-members. Then, it employed to predict soil minerals in each cell using a spectral unmixing method. Illite, chlorite, calcite, dolomite, kaolinite, smectite, quartz, hematite, goethite, vermiculite, palygorskite and some feldspar were identified. In addition, sand and limestone, calcite and dolomite, and sand surface from similarly bright clouds can be distinguished easily based on the proposed method. The soil minerals obtained from X-ray diffraction analysis of the soil samples are in conformity with spectrally dominant mineralogy from Hyperion data. Different minerals can be identified using this method without any knowledge of field spectra or any a priori field data, thus configuring a “true” remote sensing method.     

Bayesian calibration as a tool for initialising the carbon pools of dynamic soil models

The most widely applied soil carbon models partition the soil organic carbon into two or more kinetically defined conceptual pools. The initial distribution of soil organic matter between these pools influences the simulations. Like many other soil organic carbon models, the DAYCENT model is initialised by assuming equilibrium at the beginning of the simulation. However, as we show here, the initial distribution of soil organic matter between the different pools has an appreciable influence on simulations, and the appropriate distribution is dependent on the climate and management at the site before the onset of a simulated experiment. If the soil is not in equilibrium, the only way to initialise the model is to simulate the pre-experimental period of the site. Most often, the site history, in terms of land use and land management is often poorly defined at site level, and entirely unknown at regional level. Our objective was to identify a method that can be applied to initialise a model when the soil is not in equilibrium and historic data are not available, and which quantifies the uncertainty associated with initial soil carbon distribution. We demonstrate a method that uses Bayesian calibration by means of the Accept-Reject algorithm, and use this method to calibrate the initial distribution of soil organic carbon pools against observed soil respiration measurements. It was shown that, even in short-term simulations, model initialisation can have a major influence on the simulated results. The Bayesian calibration method quantified and reduced the uncertainties in initial carbon distribution.     

基于序列信息的土壤CT图像超分辨率重建

受部分容积效应影响,土壤计算机断层扫描（Computed Tomography,CT）图像存在孔隙边界模糊现象,影响土壤孔隙结构研究的准确性。针对该问题,该研究提出基于序列信息的生成式对抗网络（Sequence information Generative Adversarial Network,SeqGAN）,实现土壤CT图像的超分辨率重建。针对土壤CT序列图像具有较高相似性的特点,SeqGAN法引入序列卷积块挖掘前后图像的序列信息,并将多重特征增强融合于目标图像中;利用多层残差块提取图像特征,构建残差块输入和输出的直接连接,以减少模型退化;利用对抗网络实现损失间接反馈,提高模型的特征学习能力。在序列相似性较高的土壤图像数据集验证了该方法性能。结果表明,SeqGAN法均方误差比次优方法GAN降低25%,峰值信噪比提升1.4 dB,结构相似性提升0.2%。重建的土壤图像具有较高准确率和清晰度,可为后续土壤物理学研究提供准确的数据基础。     

Hierarchical Bayesian models for soil CO2 flux using soil texture: a case study in central Hokkaido,Japan

Hierarchical Bayesian (HB) methods are useful tools for modeling multifaceted, nonlinear phenomena such as those encountered in ecology, and have been increasingly applied in environmental sciences, e.g., to estimate soil gas flux from different soil textures or sites. We have developed a model of soil carbon dioxide (CO₂) flux based on soil temperature (T, 5 cm depth) and water-filled pore space (WFPS, 5 cm depth) using HB theory. The HB model was calibrated using a dataset of CO₂ flux measured from bare soils belonging to four texture classes in 14 upland field sites in a watershed in central Hokkaido, Japan, in the nonsnow-cover season from 2003 to 2011. The numerical software HYDRUS-1D was used to simulate daily WFPS, and the estimated values were significantly correlated with the measured WFPS (R² = 0.68, P < 0.001). Compared to a nonhierarchical Bayesian model (Bayesian pooled model), the CO₂ predictions with the HB model more accurately represented texture-specific observations. The simulation–observation fit of the CO₂ flux model was R² = 0.64 (P < 0.001). More than 90% of the observed daily data were within the 95% confidence interval. The HB model exhibited high uncertainty for high CO₂ flux values. The HB model calibration revealed differing sensitivity of CO₂ flux to T and WFPS in different soil texture classes. CO₂ flux increased with an increase in T, and it increased to a lesser degree with a finer texture, possibly because the clay and silt facilitated soil aggregation, thus reducing temperature fluctuations. WFPS values between 0.48 and 0.64 resulted in optimal conditions for CO₂ flux. The minimum WFPS value increased with an increase in clay content (P < 0.05). Although only a small number of soil types were studied in only one season in this study, the HB model may provide a method for predicting how the effects of soil temperature and moisture on CO₂ flux change with texture, and soil texture could be regarded as an upscaling factor in future research on regional extrapolation.     

Identifying potential management zones in a layered soil using several sources of ancillary information

Variation in soil texture has a profound effect on soil management, especially in texturally complex soils such as the polder soils of Belgium. The conventional point sampling approach requires high sampling intensity to take into account such spatial variation. In this study we investigated the use of two ancillary variables for the detailed mapping of soil texture and subsequent delineation of potential management zones for site‐specific management. In an 11.5 ha arable field in the polder area, the apparent electrical conductivity (EC_a) was measured with an EM38DD electromagnetic induction instrument. The geometric mean values of the EC_a measured in both vertical and horizontal orientations strongly correlated with the more heterogeneous subsoil clay content (r = 0.83), but the correlation was weaker with the homogenous topsoil clay content (r = 0.40). The gravimetric water content at wilting point (θg_{(?1.5 MPa)}) correlated very well (r = 0.96) with the topsoil clay content. Thus maps of topsoil and subsoil clay contents were obtained from 63 clay analyses supplemented with 117θg_{(?1.5 MPa)} and 4048EC_a measurements, respectively, using standardized ordinary cokriging. Three potential management zones were identified based on the spatial variation of both top and subsoil clay contents. The influence of subsoil textural variation on crop behaviour was illustrated by an aerial image, confirming the reliability of the results from the small number of primary samples.     

黑土土壤水分反射光谱特征定量分析与预测

选择单一土类黑土作为研究对象, 并准确调配其不同含水量,实验室测定土壤高光谱反射率,利用光谱分析与统计方法,定量描述了不同含水量黑土反射光谱特征,并建立了黑土含水量反射光谱预测模型,结果表明,随土壤含水量的增加,达到一定阈值（300 g kg-1）,反射率存在过饱和现象,但其倒数对数微分可以有效去除饱和问题;土壤反射率倒数对数微分对土壤含水量的响应表现出三个变化阶段,导致1 870 nm波段的倒数对数微分也表现为非线性变化,需要利用分段函数进行土壤含水量的光谱精确速测。     

基于CARS算法的不同类型土壤有机质高光谱预测

不同土壤类型的理化性质和光谱性质存在差异,以往研究多以高光谱反射率或光谱吸收特征建立模型,输入变量类型结构单一,往往导致土壤有机质（Soil Organic Matter,SOM）预测模型的精度不高。为提高SOM高光谱预测模型精度,该研究以黑龙江省海伦市为研究区,将不同类型土壤分别以竞争自适应重加权采样（Competitive Adaptive Reweighted Sampling,CARS）筛选的特征波段、数字高程模型（Digital Elevation Model,DEM）数据和光谱指数作为输入变量,结合随机森林（Random Forest,RF）算法建立SOM预测模型。结果表明：1）通过CARS算法筛选后,各土壤类型特征波段压缩至全波段数目的16%以下,在很大程度上降低土壤高光谱变量维度和计算复杂程度,从而提高了模型的预测能力,说明CARS算法在提取特征关键波段变量、优化模型结构方面起到重要作用;2）不同类型土壤的SOM预测精度存在差异,沼泽土的预测精度最高为0.768,性能与四分位间隔距离的比率（Ratio of Performance to InterQuartile distance,RPIQ）为3.568;黑土次之,草甸土的预测精度最低,仅0.674,RPIQ为1.848。3类土壤的RPIQ均达到1.8以上,模型具有较好的预测能力;3）局部回归预测精度最优,验证集的调整后决定系数为0.777,均方根误差（Root Mean Square Error,RMSE）为0.581%,模型验证RPIQ为2.689,模型稳定性高。该试验筛选的预测因子通过RF模型可实现SOM含量的快速预测,简化了传统复杂的程序,可为中尺度区域不同类型土壤的SOM预测提供依据,为输入量的选择提供参考。     

数字土壤及其应用研究进展

  目的  为实现土壤剖面信息的快速采集,并保证数据的一致性与完整性。  方法  在土壤信息采集现状分析基础上,遵照MVC设计规范,采用SQLite数据库,基于iOS操作系统设计并开发了土壤剖面信息采集系统。  结果  土壤剖面信息采集系统设计了2个主表与83个查找表,方便用户快速录入剖面相关信息,增强交互界面友好性和数据规范化,开发了基于iOS系统的App程序,可以通过iPhone或iPad安装使用。  结论  通过系统测试及应用,表明利用该App进行土壤剖面信息采集系统切实可行,在土壤剖面信息采集方面具有很好的数据一致性,减少后期录入与整理的人工成本,可为科研工作者野外土壤剖面采样及第三次土壤普查等工作提供技术支撑,具有较强的应用价值。     

The soil cover pattern in Western Transbaikal region as displayed on a 1: 500 000 scale map

Specific features of the soil cover in the Western Transbaikal region are discussed. The soil cover has been studied during soil survey works on scales of 1: 25000 and 1: 100000 in the Kizhinga district. On this basis, a generalized map of the soil cover patterns has been compiled on a scale of 1: 500 000. Data of the large-and medium-scale soil mapping are not lost upon the generalization procedure due to the reflection of soil combinations on the small-scale map, which makes the latter very informative.     

Spatial prediction of soil nutrient in a hilly area using artificial neural network model combined with kriging

It is widely recognized that using correlated environmental factors as auxiliary variables can improve the prediction accuracy of soil properties. In this study, a radial basis function neural network (RBFNN) model combined with ordinary kriging (OK) was proposed to predict spatial distribution of four soil nutrients based on the same framework used by regression kriging (RK). In RBFNN_OK, RBFNN model was used to explain the spatial variability caused by the selected auxiliary factors, while OK was used to express the spatial autocorrelation in RBFNN prediction residuals. The results showed that both RBFNN_OK and RK presented prediction maps with more details. However, RK does not always obtain mean errors (MEs) which were closer to 0 and lower root mean square errors (RMSEs) and mean relative errors (MREs) than OK. Conversely, MREs of RBFNN_OK were much closer to 0 and its RMSEs and MREs were relatively lower than OK and RK. The results suggest that RBFNN_OK is a more unbiased method with more stable prediction performance as well as improvement of prediction accuracy, which also indicates that artificial neural network model is more appropriate than regression model to capture relationships between soil variables and environmental factors. Therefore, RBFNN_OK may provide a useful framework for predicting soil properties.