基于辅助变量的森林半腐层厚度空间插值精度

基于地统计学方法,利用3种以海拔作为辅助变量的空间插值算法［局部平均的简单kriging法(simple kriging with locally varying mean,SKlm)、带有外部漂移的kriging法(kriging with an external drift,KED)和协kriging法(cokrging,COK)］计算了森林半腐层厚度的空间插值精度,并进行了交叉验证.结果表明：KED法既考虑了变量之间的空间变异,又考虑到影响局部空间变化的因素,与其他插值方法相比,其精度有很大提高;由于海拔与半腐层厚度之间的相关关系较弱,导致SKlm法的插值精度没有达到预期效果;COK法直接将海拔用于估计半腐层厚度,由于在边界地区缺乏采样点数据,因此边界地区的插值出现了多处突变区域.对比地统计学方法与距离反比权重法(inverse distance weighting, IDW)在本研究中的插值精度,除了KED方法的插值精度较高外,其余方法的插值精度均不及IDW,原因可能是利用辅助变量辅助地统计学插值时,主、辅助变量之间的相关关系在插值中起着重要作用.     

Mapping soil organic matter in low-relief areas based on land surface diurnal temperature difference and a vegetation index

Accurate estimates of the spatial variability of soil organic matter (SOM) are necessary to properly evaluate soil fertility and soil carbon sequestration potential. In plains and gently undulating terrains, soil spatial variability is not closely related to relief, and thus digital soil mapping (DSM) methods based on soil–landscape relationships often fail in these areas. Therefore, different predictors are needed for DSM in the plains. Time-series remotely sensed data, including thermal imagery and vegetation indices provide possibilities for mapping SOM in such areas. Two low-relief agricultural areas (Peixian County, 28 km × 28 km and Jiangyan County, 38 km × 50 km) in northwest and middle Jiangsu Province, east China, were chosen as case study areas. Land surface diurnal temperature difference (DTD) extracted from moderate resolution imaging spectroradiometer (MODIS) land surface temperature (LST), and soil-adjusted vegetation index (SAVI) at the peak of growing season calculated from Landsat ETM+ image were used as predictors. Regression kriging (RK) with a mixed linear model fitted by residual maximum likelihood (REML) and residuals interpolated by simple kriging (SK) were used to model and map SOM spatial distribution; ordinary kriging (OK) was used as a baseline comparison. The root mean squared error, mean error and mean absolute error calculated from leave-one-out cross-validation were used to assess prediction accuracy. Results showed that the proposed covariates provided added value to the observations. SAVI aggregated to MODIS resolution was able to identify local highs and lows not apparent from the DTD imagery alone. Despite the apparent similarity of the two areas, the spatial structure of residuals from the linear mixed models were quite different; ranges on the order of 3 km in Jiangyan but 16 km in Peixian, and accuracy of best models differed by a factor of two (3.3 g/kg and 6.3 g/kg SOM, respectively). This suggests that time-series remotely sensed data can provide useful auxiliary variable for mapping SOM in low-relief agricultural areas, with three important cautions: (1) image dates must be carefully chosen; (2) vegetation indices should supplement diurnal temperature differences, (3) model structure must be calibrated for each area.     

Combining geostatistical models and remotely sensed data to improve tropical tree richness mapping

Information on the spatial distribution and composition of biological communities is essential in designing effective strategies for biodiversity conservation and management. Reliable maps of species richness across the landscape can be useful tools for these purposes. Acquiring such information through traditional survey techniques is costly and logistically difficult. The kriging interpolation method has been widely used as an alternative to predict spatial distributions of species richness, as long as the data are spatially dependent. However, even when this requirement is met, researchers often have few sampled sites in relation to the area to be mapped. Remote sensing provides an inexpensive means to derive complete spatial coverage for large areas and can be extremely useful for estimating biodiversity. The aim of this study was to combine remotely sensed data with kriging estimates (hybrid procedures) to evaluate the possibility of improving the accuracy of tree species richness maps. We did this through the comparison of the predictive performance of three hybrid geostatistical procedures, based on tree species density recorded in 141 sampling quadrats: co-kriging (COK), kriging with external drift (KED), and regression kriging (RK). Reflectance values of spectral bands, computed NDVI and texture measurements of Landsat 7 TM imagery were used as ancillary variables in all methods. The R² values of the models increased from 0.35 for ordinary kriging to 0.41 for COK, and from 0.39 for simple regression estimates to 0.52 and 0.53 when using simple KED and RK, respectively. The R² values of the models also increased from 0.60 for multiple regression estimates to 0.62 and 0.66 when using multiple KED and RK, respectively. Overall, our results demonstrate that these procedures are capable of greatly improving estimation accuracy, with multivariate RK being clearly superior, because it produces the most accurate predictions, and because of its flexibility in modeling multivariate relationships between tree richness and remotely sensed data. We conclude that this is a valuable tool for guiding future efforts aimed at conservation and management of highly diverse tropical forests.     

黄土高原不同生态治理小流域土壤有机质、容重及黏粒含量的对比

为探讨不同生态治理小流域土壤性质的差异,本研究分别从坡向、坡位、区段和土层4个方面分析了人工刺槐林流域杨家沟(YJG)与封禁荒草地流域董庄沟(DZG)土壤有机质(SOM)、土壤容重(BD)和黏粒含量(CC)的空间分异.结果表明: YJG与DZG的SOM、BD、CC分别为12.78 g·kg^-1、1.24 g·cm^-3、19.2%与11.13 g·kg^-1、1.21 g·cm^-3、18.2%,前者均略高,但差异不显著.各指标均为东坡大于西坡;SOM和CC顺坡向下有增加趋势,BD变异最小;SOM由上游至下游呈增大趋势,BD和CC不断减小;由土表向下至60 cm土深,BD和CC不断增大,SOM不断减小.各指标的空间敏感性依次为CC＞SOM＞BD,空间因素的影响效用依次为土层＞区段＞坡向＞坡位.上游CC、中游BD和CC在两流域间的差异显著,各指标对坡位、区段、土层的敏感性均为YJG＜DZG.     

利用辅助变量对污染土壤锌分布的克里格估值

采用协同克里格及与回归相结合的克里格方法,以36个下层(10～20 cm)土壤锌数据为目标变量、另36个下层土壤锌数据为校验数据、72个上层(0～10 cm)土壤锌为辅助变量,对沈阳市南郊某有色金属加工厂附近农田0.1 mol·L^-1 HCl浸提态土壤锌进行插值分析,并对这两种利用辅助变量的克里格方法在土壤锌空间分布研究中的适用性进行评价.结果表明,与回归相结合的克里格的估值效果明显优于协同克里格及普通克里格法.结合回归模型的变异函数理论模型决定系数较大、残差较小,估值精度比普通克里格法提高4%,且基于回归克里格的土壤锌分布图与利用72个样点普通克里格插值图具有高度的相似性.而协同克里格与普通克里格相比未表现出明显优势.借助辅助变量,采用基于回归模型的克里格方法是进行土壤重金属空间分布估值的一种有效方法.     

Ecological behavior of Quercus suber and Quercus ilex inferred by topographic wetness index (TWI)

The ecological behaviors of a network of pure evergreen oak stands (Quercus suber L. and Quercus ilex L.) in the Central-Western Mediterranean Basin were investigated toward climatic and edaphic factors implemented with the application of topographic wetness index (TWI). A Categorical Principal Component Analysis (Catpca) using climatic and soil physico-chemical parameters was performed on 23 cork oak and holm oak pure stands with the aim to understand better the effectiveness of TWI for characterizing soil ecology of the two species. Catpca pointed out that, although cork oak and holm oak are able to growth in similar Mediterranean conditions, they show different behaviors in terms of needs and tolerance to soil water content. TWI confirmed such results at local scale, allowing highlighting some interesting features of the species differential ecology. Although both species confirmed to be drought-tolerant, the heliophilous cork oak revealed to dominate the landscape on wettest soils with high TWI values—indicating the capacity to tolerate stresses due to periods of waterlogging—, while the shade-tolerant holm oak prevails for low-medium TWI values—drier and mesophilous sites. Despite the application of TWI to vegetation science and ecology is relatively recent, results are encouraging and suggest considering this user-friendly and synthetic index in ecological investigations and modeling.     

不同研究尺度下土壤有机质和全氮的空间变异特征

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。     

基于序贯高斯模拟的日最低气温空间不确定性评估

明确日最低气温对于评估低温对作物的危害、指导人们及时采取补救措施、保障粮食安全具有重要意义.克里格是近地面温度场插值的主流方法,但其平滑效应会导致低值区域被过高估计而高值区域被过低估计.对2011年12月12日冷空气影响下的海南岛日最低气温,采用交叉验证法评估了普通克里格和带漂移的克里格两种插值法的预测精度;并对克里格插值法和序贯高斯模拟法产生的当日海南岛最低气温的空间分布进行对比分析.结果表明：带漂移克里格法的预测精度（r=0.86）并不显著优于普通克里格法（r=0.86）;序贯高斯模拟能产生多个等概率的符合数据整体分布和方差函数的模拟结果,模拟结果克服了克里格插值的平滑效应,能够比克里格插值更加真实地反映当日最低气温的空间分布;在低温区域,气温变化小,序贯高斯模拟结果的条件方差小于普通克里格方差;潜在寒害区的空间不确定性能够通过多个序贯高斯模拟实现并加以量化.序贯高斯模拟在低温导致农业气象灾害的评估中具有较高的应用价值.
     

Spatial distribution of pH and organic matter in urban soils and its implications on site-specific land uses in Xuzhou,China

The spatial variation of soil pH and soil organic matter (SOM) in the urban area of Xuzhou, China, was investigated in this study. Conventional statistics, geostatistics, and a geographical information system (GIS) were used to produce spatial distribution maps and to provide information about land use types. A total of 172 soil samples were collected based on grid method in the study area. Soil pH ranged from 6.47 to 8.48, with an average of 7.62. SOM content was very variable, ranging from 3.51 g/kg to 17.12 g/kg, with an average of 8.26 g/kg. Soil pH followed a normal distribution, while SOM followed a log-normal distribution. The results of semi-variograms indicated that soil pH and SOM had strong (21%) and moderate (44%) spatial dependence, respectively. The variogram model was spherical for soil pH and exponential for SOM. The spatial distribution maps were achieved using kriging interpolation. The high pH and high SOM tended to occur in the mixed forest land cover areas such as those in the southwestern part of the urban area, while the low values were found in the eastern and the northern parts, probably due to the effect of industrial and human activities. In the central urban area, the soil pH was low, but the SOM content was high, which is mainly attributed to the disturbance of regional resident activities and urban transportation. Furthermore, anthropogenic organic particles are possible sources of organic matter after entering the soil ecosystem in urban areas. These maps provide useful information for urban planning and environmental management.     

基于局域模型的凉水国家自然保护区土壤全氮空间分布

以凉水国家自然保护区激光雷达数据为基础,建立数字高程模型,提取基本地形属性如坡度、坡向和复合地形属性湿度指数和相对径流强度指数等,在成土因素学说基础上,对全氮含量(TN)进行地理加权回归建模(GWR),同时运用反距离加权(IDW)、普通克里格(OK)和泛克里格(UK)对TN进行空间插值.结果表明： 对于研究区TN的预测,GWR模型预测精度(77.4%)高于其他3种空间插值方法,IDW预测精度(69.4%)高于OK(63.5%)和UK(60.6%)的预测精度.利用GWR模型预测研究区域土壤TN平均达到4.82 g·kg^-1;在高海拔、地形湿度大以及相对径流强的区域,土壤TN相对较高.对预测结果进行探讨发现,不同坡位、坡向的土壤TN也存在一定差异.因此,基于地形属性的局域模型是土壤属性空间分布预测的更为有效的方法.     

Examining the spatial distribution of flower thrips in southern highbush blueberries by utilizing geostatistical methods

Flower thrips (Frankliniella spp.) are one of the key pests of southern highbush blueberries (Vaccinium corymbosum L. x V. darrowii Camp), a high-value crop in Florida. Thrips' feeding and oviposition injury to flowers can result in fruit scarring that renders the fruit unmarketable. Flower thrips often form areas of high population, termed "hot spots", in blueberry plantings. The objective of this study was to model thrips spatial distribution patterns with geostatistical techniques. Semivariogram models were used to determine optimum trap spacing and two commonly used interpolation methods, inverse distance weighting (IDW) and ordinary kriging (OK), were compared for their ability to model thrips spatial patterns. The experimental design consisted of a grid of 100 white sticky traps spaced at 15.24-m and 7.61-m intervals in 2008 and 2009, respectively. Thirty additional traps were placed randomly throughout the sampling area to collect information on distances shorter than the grid spacing. The semivariogram analysis indicated that, in most cases, spacing traps at least 28.8 m apart would result in spatially independent samples. Also, the 7.61-m grid spacing captured more of the thrips spatial variability than the 15.24-m grid spacing. IDW and OK produced maps with similar accuracy in both years, which indicates that thrips spatial distribution patterns, including "hot spots," can be modeled using either interpolation method. Future studies can use this information to determine if the formation of "hot spots" can be predicted using flower density, temperature, and other environmental factors. If so, this development would allow growers to spot treat the "hot spots" rather than their entire field.     

Estimation of soil organic matter by geostatistical methods: Use of auxiliary information in agricultural and environmental assessment

Mapping soil properties such as soil organic matter (SOM), or soil organic carbon (SOC) content represent a problem often arising in agronomic and environmental surveys, since point data must be spatialized with the best interpolation. Deterministic methods (Inverse Distance Weighting, Splines, ecc.) do not account for the error, thus probabilistic methods as geostatistics (e.g. Ordinary Kriging, OK) have been successfully applied for many years. Maps derived from this kind of stochastic interpolation, based only on a recognized autocorrelation among measured points, could not be suitable in representing the reality, since they usually show a smoothed pattern. Hybrid interpolation methods, such as Regression Kriging (RK), combine an interpolation based only on point data and an interpolation based on a regression of the target variable with other continuous variables, spatially related, well known on the whole area.In the Teramo province, central Italy, a set of 250 soil samples, collected from the surface horizon of agricultural soils is available. From these samples the estimation of soil texture, SOC, SOM related to texture and C/N both by OK and RK was performed. The following predictors were used for RK: (i) indexes derived from Landsat TM imagery, (ii) morphometric parameters derived from DEM, (iii) soil subsystems map 1:250,000. The maps obtained by both OK and RK in this survey show substantial agreement, without significant improvement in map accuracy using auxiliary information.     

基于神经网络模型和地统计学方法的土壤养分空间分布预测

采用径向基函数神经网络模型与普通克里格法相结合的方法,预测川中丘陵区县域尺度土壤养分(有机质和全氮)的空间分布,并与普通克里格法和回归克里格法进行比较.结果表明：各方法对研究区土壤养分的预测结果相似.与多元回归模型相比,神经网络模型对验证样点土壤有机质和全氮的预测值与样点实测值的相关系数分别提高了12.3%和16.5%,表明神经网络模型能更准确地捕捉土壤养分与定量环境因子间的复杂关系.对469个验证样点预测结果的误差分析表明,神经网络模型与普通克里格法相结合的方法对土壤有机质和全氮预测结果的平均绝对误差、平均相对误差、均方根误差较普通克里格法分别降低了6.9%、7.4%、5.1%和4.9%、6.1%、4.6%,降低幅度达到极显著水平(P<0.01);与回归克里格法相比则分别降低了2.4%、2.6%、1.8%和2.1%、2.8%、2.2%,降低幅度达显著水平(P<0.05).     

基于神经网络模型和地统计学方法的土壤养分空间分布预测

采用径向基函数神经网络模型与普通克里格法相结合的方法,预测川中丘陵区县域尺度土壤养分(有机质和全氮)的空间分布,并与普通克里格法和回归克里格法进行比较.结果表明：各方法对研究区土壤养分的预测结果相似.与多元回归模型相比,神经网络模型对验证样点土壤有机质和全氮的预测值与样点实测值的相关系数分别提高了12.3%和16.5%,表明神经网络模型能更准确地捕捉土壤养分与定量环境因子间的复杂关系.对469个验证样点预测结果的误差分析表明,神经网络模型与普通克里格法相结合的方法对土壤有机质和全氮预测结果的平均绝对误差、平均相对误差、均方根误差较普通克里格法分别降低了6.9%、7.4%、5.1%和4.9%、6.1%、4.6%,降低幅度达到极显著水平(P<0.01);与回归克里格法相比则分别降低了2.4%、2.6%、1.8%和2.1%、2.8%、2.2%,降低幅度达显著水平(P<0.05).     

基于随机森林法的山核桃林地土壤速效养分含量空间分布特征研究

土壤氮磷钾是土壤肥力管理的重要元素,是植物生长的必要养分元素。对土壤氮磷钾的空间分布进行特征解译,可为精准管理临安山核桃产区林地土壤肥力,促进山核桃林产业可持续发展提供理论依据。研究以临安山核桃主产区为研究区域,利用随机森林（RF）、普通克里格（OK）和Shapley加性解释（SHAP）方法,结合地形因子、气候因子、土壤因子、遥感因子等环境变量,对山核桃林地土壤碱解氮（AN）、有效磷（AP）、速效钾（AK）的空间分布特征进行分析。研究结果表明：相比于OK模型,基于环境协变量所构建的RF模型对AN、AP和AK含量空间分布预测表现最佳,R²分别为0.68、0.60和0.64,均方根误差（RMSE）分别为20.005、10.287和22.426,平均绝对误差（MAE）分别为15.425、7.709 和21.628。RF模型SHAP分析显示,AN和AK含量分布主要受土壤有机质（SOM）的影响,并且SOM与AN和AK存在正相关性;AP主要受pH的影响,其次为色调指数,AP与pH和色调指数均具有负相关性;AK和AP同时受到海拔和坡向的影响。两种模型预测的氮磷钾空间分布趋势总体相似,不同速效养分存在明显的空间异质性。碱解氮高值区域主要分布于研究区东部;有效磷高值区域主要分布于研究区西部,但分散度高;速效钾高值区域则主要分布于研究区中部。总体而言,基于随机森林模型可以高精度模拟山核桃林地土壤氮磷钾含量空间分布特征,并依据主要环境协变量对土壤氮磷钾的影响关系,提出相应改良措施。在有效磷含量低值区域可以施用石灰来缓解土壤酸化,同时补追磷肥;碱解氮含量高值区域可以合理减少氮肥施用;速效钾含量低值区域合理施加钾肥;对于海拔较高及迎风坡多降雨的区域,可以构建林下高效水土保持植被,减轻水土流失;在林地施用有机肥料,改善土壤理化性质,增加土壤养分含量。     

LLWR Techniques for Quantifying Potential Soil Compaction Consequences of Crop Residue Removal

Harvesting crop residues for bioenergy or bio-product production may decrease soil organic matter (SOM) content, resulting in the degradation of soil physical properties and ultimately soil productivity. Using the least limiting water range (LLWR) to evaluate improvement or degradation of soil physical properties in response to SOM changes has generally been hampered by the extensive amount of data needed to parameterize limiting factor models for crop production. Our objective was to evaluate five pedotransfer functions to determine their effectiveness in predicting soil water holding capacity in response to different SOM levels. Similarly, two other pedotransfer functions were evaluated to determine the effects of SOM on cone index values. Predictions of field capacity and wilting point water content as well as the cone index–water content–bulk density relationship of soil strength using the pedotransfer functions were compared with field data from two tillage experiments near Akron, CO that had a range of SOM concentrations. Equations previously developed by da Silva and Kay gave the best estimates of LLWR for the pedotransfer functions we evaluated. These equations were then used to illustrate LLWR changes in response to different soil and crop management practices on a Duroc loam near Sidney, NE. The results showed that tillage and, possibly, soil erosion decreased the LLWR as tillage intensity increased. Therefore, we recommend that crop residue removal rates be limited to rates that maintain or increase SOM content to ensure soil physical conditions are not degraded.     

Using topographic wetness index in vegetation ecology: does the algorithm matter?

Questions: How important is the choice of flow routing algorithm with respect to application of topographic wetness index (TWI) in vegetation ecology? Which flow routing algorithms are preferable for application in vegetation ecology? Location: Forests in three different regions of the Czech Republic. Methods: We used vegetation data from 521 georeferenced plots, recently sampled in a wide range of forest communities. From a digital elevation model, we calculated 11 variations of TWI for each plot with 11 different flow routing algorithms. We evaluated the performance of differently calculated TWI by (1) Spearman rank correlation with average Ellenberg indicator values for soil moisture, (2) Mantel correlation coefficient between dissimilarities of species composition and dissimilarities of TWI and (3) the amount of variation in species composition explained by canonical correspondence analysis. Results: The choice of flow routing algorithm had a considerable effect on the performance of TWI. Correlation with Ellenberg indicator values for soil moisture, Mantel correlation coefficient and explained variation doubled when the appropriate algorithm was used. In all regions, multiple flow routing algorithms performed best, while single flow routing algorithms performed worst. Conclusions: We recommend the multiple flow routing algorithms of Quinn et al. and Freeman for application in vegetation ecology.     

华南桉树人工林树液流通量及蒸散作用(英文)

通过对广东省湛江市雷州半岛桉树(Eucalyptus urophylla S.T.Blake)人工林的树液流通量、环境因子、土壤蒸发、林冠截留和林分特性相关指标一年多的连续观测,以及通过一个理论公式对日蒸散量的计算,得出了如下结论:(1)土壤特性及由此决定的土壤水势对树液流通量,以及树液流通量密度(SFD)与气温的关系有一定影响;(2)林冠层的VPD(空气饱和差)对SFD有显著影响;(3)由测定和计算得来的蒸散量在河头和纪家分别有5.26％和6.14％的偏差,可以认为这两种方法有较好的一致性;(4)河头和纪家的林分蒸腾量占总蒸散量百分比分别为62.2％和51.3％;(5)树种单位叶面积水平上的SFD是评价该树种水分利用的重要指标。     

基于辅助环境变量的土壤有机碳空间插值——以黄土丘陵区小流域为例

利用普通克里格法(OK)、反距离加权法(IDW)、径向基函数法(RBF)、基于土地利用类型修正的普通克里格法(OK_LU)4种插值方法,对黄土丘陵羊圈沟小流域的土壤有机碳含量进行空间插值。预测结果的准确性通过Pearson相关系数(R),平均绝对误差(MAE),均方根误差(RMSE),准确度(AC)来评价。研究结果表明:(1)在前3种常规空间插值方法中,OK对刻画区域土壤有机碳的空间分布趋势效果最佳,其预测MAE值和RMSE值均为最小,Pearson相关系数(R)和准确度(AC)最大,说明其预测结果的准确性最好、预测的极端误差也最小;其次为RBF;IDW预测的效果最差。(2)OK_LU在空间特征表达方面能够更好地反映复杂地形区的局部变异,其插值结果的精度相比OK有一定程度的提高,其平均绝对误差(MAE)从0.900%降到了0.567%,均方根误差(RMSE)从1.101%降到了0.777%,Pearson相关系数(R)从0.4026提高到0.5589,准确度(AC)从0.9081提高到0.9505。综合比较,在黄土丘陵地区,OK_LU能使插值结果的精度有较大提高,是土壤有机碳空间制图的有效途径。     

典型喀斯特小流域不同植被类型间土壤养分的差异性及其空间预测方法

植被类型制约着土壤结构和元素的异质化过程,致使土壤养分空间分布存在差异性.本文研究了典型喀斯特小流域不同植被类型间土壤养分(全氮TN、全磷TP、全钾TK、有机质SOM)含量分布的差异性,分析比较了普通克里金、回归模型、基于植被类型的回归模型对土壤养分预测的精度.结果表明: TN、TK、SOM与植被类型显著相关(P<0.05),TP与植被类型无显著相关(P=0.390),且TN和SOM在灌木林与耕地之间的差异性显著,TK在乔木林与灌草丛、灌木林与耕地、灌草丛与耕地间的含量差异皆显著;非连续的典型喀斯特小流域地形因子空间异质性较高,基于各样点间真实地形因子的多元线性回归预测模型精度优于基于已知点和预测点位置信息的普通克里金预测方法,且基于植被类型的回归预测模型提高了TN的预测精度.