Scale Effect Of Spatial Variability And Scaling Method Of Soil Attributes In A Complex Landscape Region

The aim of this paper was to reveal scale effects of spatial variability of soil attributes systematically,to identify dominant processes and factors of spatial variation at different scales accurately and to explore how multi-scale datas should be used more availably in soil attribute mapping,such as scaling procedure.That has important scientific significance and practical value understanding spatial variability of soil attributes in multi-scale,enhancing application of existing soil data to the research on soil science and soil resources utilization management practice.Dengfeng county is located in the southern foot of Songshan,charactered by complex topography,broad soil parent material and rich soil types.First of all,this paper analyzed statistical characteristic and spatial variation structure of soil attributes in different scales based on multi-scale samples.secondly,it studied the influence of parent material,soil types,local topographic factor and landuse factor on soil properties through correlation analysis,multiple comparison and so on in different scales and regions.Finally,taking prediction mapping of soil organic matter as an example,scale effect of modeling soil attributes,and a method scaling soil properties step by step along descending scales were researched.The results showed:(1)soil organic matter and gravel content represented multiple variation scales obviously in county-scale(2km to30km)under the complex landscape condition with mountains and hills,while soil total phosphorus not.(2)in region-scale,soil organic matter,gravel content and soil total phosphorus were effected remarkably by parent material and correlated with elevation significantly.However,in landscape-scale,the influencing factors of soil organic matter,gravel content and soil total phosphorus showed differentiation clearly.Soil organic matter correlated with elevation significantly only in piedmont alluvial sector,whereas gravel content and soil total phosphorus significantly still in three landscape area.The impact of soil type,landuse and aspect on soil properties was not general,played a role only in hilly diluvial parent material zone and piedmont alluvial sector.In addition,slope factor affected just soil total phosphorus of Purple Soil Zone.(3)the details of soil variations in small scale were expressed exactly could improved the fitting precision of large-scale data.In the meantime,it enhanced mapping accuracy of using small-scale data separately when joining large-scale data.(4)the accuracy was reduced after correcting landscape-scale mapping result compulsively with the law on regional-scale.Nevertheless,the law on landscape-scale player a positive role in improving the prediction of field-scale.Therefore,scale effect of soil attributes is obvious in a complex landscape region,and the main processes and factors of spatial variation among regional-scale,landscape-scale and field-scale display a marked difference.Moreover,the optimal way to predict soil attribute with regression kriging under nested sampling condition is that:fitting the trend item with large-scale data alone and describing the nested spatial variograms of residual item.Furthermore,the scaling effect on soil properties by"step by step along descending scales"method was not confirmed,from landscape to field it was valid but from region to landscape it was not.