论文部分内容阅读
准确预测土壤有机碳的空间分布,对于土壤资源开发和保护、应对气候变化和生态系统健康都具有重要意义.本文以塔里木盆地北缘盐土1300 m×1700 m样地为试验区,采集5~10 cm深度土壤样品144个,构建土壤有机碳含量的贝叶斯地统计空间预测模型,并以普通克里格、序惯高斯模拟和逆距离加权方法为对照,评价贝叶斯地统计对土壤有机碳含量的预测性能.结果表明:研究区土壤有机碳含量处于1.59~9.30 g·kg~(-1),平均值为4.36 g·kg~(-1),标准偏差为1.62 g·kg~(-1);半方差函数符合指数模型,空间结构比参数值为0.57;利用贝叶斯地统计方法,获得了土壤有机碳含量的空间分布图以及评价预测不确定性的预测方差、上95%分位数、下95%分位数分布图;与普通克里格、序惯高斯模拟和逆距离加权方法相比,贝叶斯地统计方法具有更高的土壤有机碳含量空间预测精度,显示出该方法对土壤有机碳含量预测的优越性.
Accurately predicting the spatial distribution of soil organic carbon plays an important role in the development and conservation of soil resources, climate change and ecosystem health.In this paper, a plot of 1300 m × 1700 m saline soil in the northern margin of Tarim Basin, cm depth of soil samples to build a Bayesian statistical spatial prediction model of soil organic carbon content. Using ordinary Kriging, sequential inertial Gaussian simulation and inverse distance weighting method as a control, Bayesian geostatistics was applied to evaluate soil organic The results showed that the soil organic carbon content in the study area was 1.59-9.30 g · kg -1 with the average value of 4.36 g · kg -1 and the standard deviation was 1.62 g · kg -1 -1). The semivariance function accords with the exponential model and the spatial structure ratio parameter is 0.57. The spatial distribution of soil organic carbon content and the prediction variance of evaluation uncertainty are obtained by Bayesian geostatistics. The upper 95% Quantile, and 95% quantile distribution. Bayesian geostatistical methods have higher spatial prediction accuracy of soil organic carbon than conventional Kriging, ordinal Gaussian and inverse distance weighting methods and show that The method of the soil there Carbon content prediction superiority.