论文部分内容阅读
以江西省余江县中部地区为例,基于规则网格采样点,利用普通克里格(OK)和3种图斑连接方法(采样点分别与土壤图斑(PSt)、土地利用图斑(PLu)及土壤-土地利用复合图斑连接(PSt Lu))对土壤有机碳(SOC)进行点面拓展,并通过验证样点对比各方法的预测精度。结果表明,土壤类型和土地利用方式对红壤区SOC含量有显著影响,其中土地利用的影响大于土壤类型。OK方法未考虑土壤和土地利用对SOC的影响,空间预测误差最大;PSt和PLu分别考虑了土壤和土地利用类型间的差异,误差较OK分别降低了35.8%和46.0%;而PSt Lu综合考虑了土壤和土地利用的影响,误差较OK降低了48.9%。此外,PSt、PLu和PSt Lu方法得到的SOC分布分别与土壤类型、土地利用方式和土壤-土地利用组合的空间分布一致,均较OK能更好的反映现实情况。
Taking the central part of Yujiang County in Jiangxi Province as an example, based on the regular grid sampling points and the common kriging (OK) and three kinds of patch connections (the sampling points were respectively related to the soil patch (PSt), the land use patch PLu and PSt Lu) were used to expand the soil organic carbon (SOC). The accuracy of each method was verified by verifying the samples. The results showed that soil types and land use types had significant effects on SOC content in red soil area, of which land use was more affected than soil types. OK method did not consider the impact of soil and land use on SOC, and the spatial prediction error was the highest. PSt and PLu considered the difference between soil and land use types respectively, the errors were reduced by 35.8% and 46.0% respectively compared with OK; The impact of soil and land use, the error was reduced by 48.9% OK. In addition, the SOC distributions obtained by the PSt, PLu and PSt Lu methods are consistent with the spatial distributions of soil types, land use patterns and soil-land use combinations, respectively, which are better than OK to better reflect the actual situation.