Study On The Spatial Variability And Scale Effect Of Soil Fertility Quality And Its Key Factor In Beijing

This paper took Beijing as a research area, rough set theory were introduced to to investigate the weight determination method in soil fertility evaluation. Traditional statistics and geostatistical method were used to analyze spatial variation and driving factors of soil fertility and soil organic carbon (SOC) and to discuss preliminarily the spatial variability response of SOC to sampling scale changes. The results can provide a theoretical basis for soil fertility quality assessment and nutrient management. The main results and conclusions are as follows:(1) The determination of weight using rough set theory for soil fertility evaluation involve several steps:data discretization, preliminary determination of soil fertility grade, attribute value reduction, equivalence partitioning, attribute significance calculation, and index weights calculation. Through the analysis of actual case, there was a significant linear correlation between crop yield and soil integrated fertility index (IFI) obtained by Delphi weighting method. The determination coefficient R1was0.77and root mean square error (RMSE) was1.25t hm-2. Again, a significant linear correlation between IFI (obtained by rough set theory weighting method) and crop yield was observed. The later method has higher accuracy, as indicated by higher values of R2(0.83) and lower value of RMSE (1.06t hm-2). Therefore, results of this study indicates that it is feasible to adopt rough set theory for determining the index weights of soil fertility.(2) Based on introducing soil microelements as soil evaluation indexes as well as soil macroelements, rough set weighting method and membership function were used for assessing soil fertility and its spatial variability was analyzed by using geostatistics method. The results showed that the contribution rates for integrated nutrient of soil conventional nutrient indexes and microelement indexes were0.66and0.34, respectively. The semi-variance analysis of the residuals showed medium degree of spatial autocorrelation. The overall spatial distribution trends of soil integrated nutrient in Yanqing Basin was mainly affected by organic matter and total N. But in the local area, the micronutrients such as available Cu and available Zn played a leading role. Both fertility levels for single available microelement and soil integrated nutrient of vegetable fields were higher than orchards and grain crop fields. Therefore, bringing microelement indexes into evaluation index system as same as conventional nutrient indexes is practicable and necessary, which can reveal satisfactorily the spatial variation pattern of soil fertility.(3) Spatial variation of SOC from1985to2009and its influencing factors were analysed. The results showed that the SOC contents in1985and2009were9.97g kg-1and10.46g kg-1, respectively. The long-term monitoring data showed the variation tendency of SOC was " increasing, decreasing and increasing". In both two periods, the SOC contents presented a2nd-order global trend from east to west and their spatial distribution was higher in north lower in south and higher in west lower in east. During 25years, the SOC of most area has increased with the largest increase for Fangshan and southwestern Mentougou area and the largest drop for Huairou and Miyun District. The terrain, soil textures, soil types, parent materials and land use were the significant influence factors for the spatial distribution of SOC. However, the land use changes directly affected the dynamic balance of SOC. Compared with maintained land, the SOC of the other lands transformed to dry lands was relatively small and to orchards was relatively increase. The SOC of dry lands transformed to other lands was relatively increase while orchards transformed to other lands was relatively decrease. This reason was extensive management for dry lands and high water and fertilizer inputs for orchards.(4) We designed four different sampling densities to investigate preliminarily the structural changes of the variogram and uncertainty of spatial prediction with the study scale changes. The results showed that the SOC was macroscopically related to terrain factor and low sampling density data were the most optimal for use in fitting the trend value. As sampling density increasing, the distribution of variogram of SOC and its residuals flattened out gradually, the random variation were growing strongly, and structural variation and uncertainty of spatial prediction decreased gradually. In addition, range of variogram may also affect the uncertainty of spatial prediction. Increasing sampling density and regression kiging method aided by terrain factors can improve prediction accuracy of SOC. Therefore, soil monitoring and management introducing auxiliary variable can to some extent cut the number of sampling points without reducing prediction accuracy.