【作者】 许冬梅；

【导师】 张世熔；

【作者基本信息】 四川农业大学 ， 土壤学， 2009， 硕士

【摘要】 本研究选取人口稠密的川中丘陵乡村区域为研究对象,以高分辨率卫星遥感图片与GPS、GIS技术相结合,首先对影响区域土壤氮密度的因素进行分析,结合相关影响因素,采用生态立地法、土地覆盖法、土壤类型法、普通克里格法、多元回归法和回归克里格法估算区域土壤氮密度;最后,采用蒙特卡罗模拟方法对结果进行不确定性的定量评价,获得了较为准确的土壤氮密度空间分布估算结果,找出了适合于村域景观土壤氮密度最佳的空间表达方式。主要结论分述如下:土壤氮密度受多因子综合作用。不同土壤类型之间土壤氮密度均达到极显著差异,可以作为土壤氮密度空间分布估算的参考分类标准,而地形属性中的海拔和坡度、土壤碳密度和粘粒含量均与土壤氮密度达到了极显著或显著相关关系,可以作为估算土壤氮密度空间分布的参数。根据各种估算方法对土壤氮密度常规统计特征、空间分布、分级面积表达效果的比较表明,生态立地法、土地覆盖法、土壤类型法、普通克里格法、多元回归法以及回归克里格法对研究区土壤氮密度的空间分布估算有一定差异,即没有一种估算方法在全部观察指标上完全占优势。同时,不能反映出这种估算结果的好坏程度,也即不能反映估算值与真实值之间的接近程度,需要对它们的估算结果进行不确定性评价。回归克里格法在不确定性评价中的不确定性最小。总体上来看,六种方法所表现出的土壤氮密度残差模拟值小于0.060 kg m^-2的概率的分布趋势是一致的,研究区残差模拟值小于0.060 kg m^-2的区域主要分布在中部两丘体之间的沟谷地带的高概率区,低概率区域主要集中在中部丘体顶部。对研究区残差模拟值小于0.060 kg m^-2的概率进行分级面积统计表明,回归克里格法估算的残差模拟值小于0.060 kg m^-2的不确定性最小。更多还原

【Abstract】 Basing on high-resolution satellite remote sensing image, GPS and GIS technologies, influence factors of soil nitrogen density of a densely populated village-scale district were analyzed and soil nitrogen density was estimated in Sichuan Hilly Region, using escotope method, land cover method, soil type method, ordinary kriging method, multiply regression method and regression kriging method, and besides, monte carlo simulation was used to assess the uncertainty. Finally, the best means was been found, and the content of soil nitrogen density was estimated in the region. The main results were as follows:Soil nitrogen density was influenced by many factors. The variation of soil nitrogen density reached significant at 0.01 probability level between different soil types. To discuss how topographical factors can impact the spatial distribution of soil nitrogen density, the results showed that natual logistic value of soil nitrogen density was notabilitily negatively correlated to elevation and slope at 0.01 probability level. The correlation of soil nitrogen density and soil silt content were notability positive at 0.01 probability level. There was remarkably positive correlation at 0.01 probability level between soil nitrogen density and soil carbon density. So, the factors above could be used to estimate soil nitrogen density in the region.Based on the descriptive statistics, spatial distribution character and area of grade for soil nitrogen density, the results indicated that the methods made the difference in expressing spatial variability of soil nitrogen density, and no one approaches were fit fully except for some aspects. Besides, spatial distribution estimation didnot reflect the smilarlity of estimation and measured value. So, uncertainty assessment needed to be used for the estimation results of soil nitrogen density.The trend of spatial distribution of simulation value of soil nitrogen density residual was the same. High probability region lowing the threshold of 0.060 kg/m~2 was in the valley between two hills in the middle of study aera. Low probability region was at the top of hills. Based on grading statistics of simulation value lowing the threshold of 0.060 kg/m~2, the results indicted that uncertainty of regression kriging method was the smallest in the process of uncertainty assessment.更多还原