森林土壤中量微量元素空间变异性研究

发布时间：2018-03-29 19:38

本文选题：森林土壤　切入点：中微量元素　出处：《华南农业大学》2016年硕士论文

【摘要】：森林不仅具有改善人类居住环境、涵养水源、防止水土流失的作用,还与森林土壤二者相互作用,影响土壤的发育与形成。因此,分析森林土壤中量与微量元素空间分布,对森林恢复与更新具有重要意义。本文以广东省云浮市云城区云安区为研究区域,运用BP人工神经网络预测模型,预测森林土壤中量及微量元素的空间分布,分析研究区内森林土壤中量与微量元素含量的空间分布特征,以期揭示当前环境条件下云城区云安区森林土壤中量与微量元素含量的分布现状,为该区森林的经营管理提供参考。本研究主要结论如下:1.对森林土壤中量及微量元素含量而言,整个研究区中量元素交换钙的含量相对较高,其次为全硫,交换镁含量最低;微量元素中,总铅的含量最高,其次为总锌,总镉含量最低。总镉、总锌的含量均在土壤环境质量标准一级标准的范围内,总铅的含量在二级标准的范围内。交换钙、交换镁、总镉、总锌的含量均小于全国背景值,总铅的含量大于全国背景值。2.通过土壤中量与微量元素变异系数的比较,云城区6个土壤中量及微量元素的变异系数分别在33.04%~98.19%之间,云安区的6个土壤中量及微量元素的变异系数在33.43~106.34%之间。其中云安区交换钙属于强变异,其余均属于中等变异,云城区与云安区交换钙的变异系数最大,分别为98.19%、106.34%。云城区总镉的变异系数最低,为33.04%。3.插值方法的比较,通过建模点与验证点的均方根误差与平均绝对误差的精度比较,可以看出泛克里格插值法比反距离加权插值法的精度高,进一步通过预测值与实测值的相关性显著度比较,发现泛克里格插值方法与反距离加权插值法建模点的预测值与实测值相关性都呈现出显著性,但是两种插值法的预测值与实测值相关系数差值不同,其中泛克里格的差值最小,但6个元素中泛克里格插值法对交换镁的预测性较好,对其他5个元素预测性欠佳。4.从模型的选取上来看,通过建模点与验证点实测值与预测值相关性显著度分析,结果发现交换钙、交换镁、全硫、总镉、总锌的预测,BP人工神经网络模型最好;总铅的预测,虽然泛克里格方法优于BP人工神经网络模型,仍需深入研究其最适模型。从空间预测分布图上来看,BP人工神经网络能更好的体现明显的地形之间的变化,与泛克里格插值法相比,更具有推广性,即BP人工神经网络最优。5.土壤中量与微量元素空间分布呈现出规律性,微量元素总镉、总铅、总锌的分布在整个研究区呈自西向东逐渐递减的空间分布趋势;交换钙总体呈自西向东逐渐增加再降低再增加的趋势,交换镁与全硫总体呈自西向东逐渐递减再增加的趋势,交换钙与交换镁的低值区呈一条不规则的带状由北向南从中部插整个研究区。中量元素受淋溶作用强,元素流失量多;微量元素受到人为干扰影响较大,包括施肥、农药、汽车尾气等影响。
[Abstract]:Forest not only has the function of improving human living environment, conserving water source and preventing soil erosion, but also interacts with forest soil, which affects the development and formation of soil. Therefore, the spatial distribution of trace elements and quantity in forest soil is analyzed. This paper takes Yunan District, Yunfu City, Guangdong Province as the research area, and uses BP artificial neural network to predict the spatial distribution of trace elements and quantity in forest soil. The spatial distribution characteristics of forest soil quantity and trace element content in Yunan district were analyzed in order to reveal the distribution status of forest soil quantity and trace element content in Yunan district under the current environmental conditions. The main conclusions of this study are as follows: 1. In terms of the amount of forest soil and the content of trace elements, the content of exchangeable calcium in the whole study area is relatively high, followed by total sulfur and the lowest content of exchange magnesium; Among trace elements, total lead content is the highest, followed by total zinc, total cadmium content is the lowest. Total cadmium and total zinc content are within the range of the first class standard of soil environmental quality standard, the total lead content is within the range of the second grade standard. The contents of total cadmium and zinc were all lower than the national background value, and the content of total lead was larger than the national background value. 2. By comparing the variation coefficient of soil quantity and trace elements, the variation coefficients of soil quantity and trace elements in six soils of Yuncheng District were 33.04 ~ 98.19%, respectively. The coefficient of variation of the six soils in Yunan area was 33.43 ~ 106.34%, in which the exchangeable calcium in Yunan area was a strong variation, while the others were medium variation, and the coefficient of variation between Yunan and Yunan was the largest. Respectively 98.19 and 106.34.The coefficient of variation of total cadmium in Yuncheng District is the lowest, 33.04k.3.Compared with the interpolation method, the accuracy of the root mean square error and the mean absolute error of the modeling point and the verification point are compared. It can be seen that the accuracy of the pan Kriging interpolation method is higher than that of the inverse distance weighted interpolation method, and the correlation significance between the predicted value and the measured value is further compared. It is found that the correlation between the predicted values and the measured values of the modeling points of the pankriging interpolation method and the inverse distance weighted interpolation method is significant, but the difference between the predicted values of the two interpolation methods and the measured values is different, in which the pan kriging method has the smallest difference. However, the Pan-Kriging interpolation method has a good predictability for the exchange magnesium, and the other five elements are not predictable. 4. From the model selection, the correlation significance between the measured values and the predicted values of the modeling points and the verification points is analyzed. The results show that the BP artificial neural network model is the best in predicting calcium exchange, magnesium exchange, total sulfur, total cadmium and total zinc, while the Pan-Kriging method is better than the BP artificial neural network model in the prediction of total lead. From the spatial prediction map, BP artificial neural network can better reflect the obvious terrain changes, compared with the pan Kriging interpolation method, it is more extensible. That is, BP artificial neural network is the best. 5. The spatial distribution of soil quantity and trace elements is regular. The distribution of trace elements, total cadmium, total lead and total zinc in the whole study area is gradually decreasing from west to east. The total exchangeable calcium increased gradually from west to east, then decreased and increased again from west to east, and the exchange magnesium and total sulfur decreased gradually from west to east, and then increased again. The low value region of exchangeable calcium and magnesium intercalated from north to south into the whole study area from north to south. The medium amount of elements was strongly eluted, the amount of elements lost was much, and the trace elements were affected greatly by human interference, including fertilization and pesticide. Automobile exhaust, etc.
【学位授予单位】：华南农业大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：S714.5

【参考文献】