Font Size: a A A

Spatial Analysis Of Soil Nutrients And Comprehensive Evaluation Of Soil Nutrients Fertility In Beijing Arable Land

Posted on:2009-01-01Degree:MasterType:Thesis
Country:ChinaCandidate:R D ZhaoFull Text:PDF
GTID:2143360242487301Subject:Soil science
Abstract/Summary:PDF Full Text Request
Soil nutrient fertility in Beijing arable land is studied. By spatial analyses and comprehensive evaluation, spatial distribution laws of nutrient indexes and soil nutrient fertility grade in this area are mastered, which is helpful to some extent to rational fertilization and eco-environment protection.The spatial distribution and variability of soil nutrients, such as organic matter (O.M.), total nitrogen (total N), available phosphorus (avail.P) and available potassium (avail.K), are studied. Meanwhile, the nutrient fertility of Beijing arable area is evaluated by support vector machines (SVM). The mian conclusions are as following.At the measuring scale, results of spatial analysis are as following: the range of O.M. is 70.80km.The nugget effect of O.M. is 50%, which shows that the variation of O.M. is attributed to structural and random factors. The anisotropy of O.M. is significant with stronger anisotropic variability in NE9O°direction. Kiging optimum interpolation shows that the mean of O.M. in Beijing arable is 15.21g/kg, and most area is below 18.80g/kg. By analysing Kigring map, O.M. in Mentougou District is higher, and Daxing District, Yanqing District, Shunyi District are lower. In this study, interpolation accuracy is mainly affected by measuring scale and the regularity of sampling sites.The range of total N is 10.80km and the nugget effect of total N is 13.31%, which shows that the variation of total N is mainly affected by random factors. The anisotropy of total N is not significant. Kiging optimum interpolation shows that the mean of total N in Beijing arable is 1.01g/kg, and most area is below 0.91g/kg. By analysing Kigring map, total N in Mentougou District, Haidian District and Changping are higher, and Daxing District, Shunyi District are lower.The range of available P is 81.00km.The nugget effect of available P is 50%, which shows that the variation of available P is affected by structural and random factors. The anisotropy of available P is not significant. Kiging optimum interpolation shows that the mean of available P in Beijing arable is 18.41 mg/kg, and most area is below 19.26mg/kg. By analysing Kigring map, available P in Mentougou District, Haidian District and Pinggu are higher, and Daxing District, Yanqing District, Shunyi District are lower. The map of Kriging standard deviation of available P shows that area with low interpolation accuracy locates around research area, affected by measuring scale and the regularity of sampling sites.The range of available K is 4.68km, which is shorter than the distance of sampling step. Under this condition, the variation of available K is not reflected, therefore, anisotropy and interpolation are not made.As for three methods in the soil nutrient fertility evaluation, SVM(support vector machines) has obvious advantage compared with discriminant analysis, but not obvious compare with ANN (artificial neural networks), although it should be in theory. According to the result evaluated by SVM and nutrient fertility grade map with Kriging interpolation, soil nutrient fertility of Beijing arable is below the middle as a whole. Arable in Fanshan District, Haidian District, Fengtai District and Mentougou District are evaluated as high fertility, Huairou District and Pinggu District are in the middle, Tongzhou District, Changping District, Daxing District, Shunyi District and Yanqing District are low fertility area.
Keywords/Search Tags:nutrient fertility, spatial variability, geostatistics, support vector machines
PDF Full Text Request
Related items