土壤预测尺度效应Scaleeffectonsoilattributepredictioninacomplexlandscaperegion.pdfVIP

Scale effect on soil attribute prediction in a complex landscape region 1 1 1 1 Wu Zhen-fu , Zhao Yan-feng , Qi Li , Chen Jie 1 School of Environment and Water Conservancy, Zhenzhou University, Institute of Natural Resource and Eco-Environment, Zhengzhou University, Zhengzhou 450001, China Abstract. To research the scale effect on mapping soil organic material (O.M), total 283 top soil samples across landscape and region scale were collected in ?km2 area of Dengfeng county, Henan province, according to a nested sample strategy. Three scenarios about how to use these samples were designed to produce digital soil O.M map with regression kriging interpolation. I) just using macro scale samples in trend prediction. Ii) just using small scale samples in trend prediction. Iii) using all samples in trend prediction. (这个地 方你再理理，看到底是几种方案。) The precision of those maps were compared and it was found that the trend of soil O.M on the elevation factor was macroscopical and that could be fitted optimally using only large scale data. Though increasing the sample volume when add samples of small scale in simulation the precision of trend equation decreased actually. However, after the trend was extracted both scales of samples were contributive to the spatial simulation of residues. Small scale of samples enhanced the spatial structure of residual in large scale and revealed the spatial variability of residual in small scale. And due to that, in the residual simulation process, small scale data improved effectively the precision of prediction in large scale. Therefore, scenari