论文部分内容阅读
国家尺度土壤属性数据是地球生物化学循环及水循环等领域研究的重要数据,目前,该尺度土壤属性数据的获取方法主要有两类:土壤属性-空间数据连接法和空间插值。为了确定哪一类方法更适合稀疏样点的国家尺度土壤属性制图,本文以中国吉林省的土壤有机质含量制图为例,采用8～32km格网样点和1∶100万土壤图,对这两类方法进行对比分析。独立样本验证结果表明,土壤属性-空间数据连接法的平均误差(ME)大于距离反比加权(IDW)插值,而平均绝对误差(MAE)和均方根误差(RMSE)都小于IDW插值。IDW插值获得的土壤属性图虽然能大致反映土壤属性空间分布的基本规律,但出现了类似“牛眼睛”的空间结构,且存在无样点区估计值不准确等问题;土壤属性-空间数据连接法尽管忽略了同种土壤类型内部的差异,保留了不同土壤类型边界处的属性值突变,但获得的土壤属性图更能反映土壤属性分布的基本规律,也具有比较详细的土壤属性空间结构。因此,在基于稀疏样点的国家尺度土壤属性制图中,土壤属性-空间数据连接法的制图效果要优于IDW空间插值法。
At the national scale, soil attribute data are important data in the field of biogeochemical cycle and water cycle. There are two main methods for obtaining soil attribute data at this scale: soil attribute-spatial data connection method and spatial interpolation. In order to determine which type of method is more suitable for the mapping of soil attributes at a sparse sampling point, taking the mapping of soil organic matter in Jilin Province of China as an example, using 8 ~ 32km grid samples and 1: Class method for comparative analysis. The independent sample validation results showed that the mean error (ME) of soil attribute-space data connection method was greater than the distance inverse-weighted (IDW) interpolation, while the mean absolute error (MAE) and root mean square error (RMSE) were less than IDW interpolation. Although the soil attribute map obtained by IDW interpolation generally reflects the basic law of spatial distribution of soil attributes, the spatial structure similar to that of “bull’s eye” appears, and there are some problems such as the inaccurate estimation of sample areas. The soil attribute-space Despite neglecting the internal differences of the same soil types, the data connection method retains the sudden change of attribute values at the boundaries of different soil types. However, the obtained soil attribute maps can better reflect the basic laws of soil attribute distribution and have more detailed soil attribute space structure. Therefore, in the mapping of soil attributes at the national scale based on the sparse sampling points, the soil attribute-spatial data connection method is better than the IDW spatial interpolation method.