利用华中地区面向陆面模式的土壤质地数据集,开展了马尔可夫链模型模拟研究,假设检验表明,土壤质地垂向转移概率矩阵满足马氏性和平稳性,满足一维马尔可夫链要求,可以使用马尔可夫链模型模拟土壤质地的分布.第4层与第8层缺值模拟试验结果表明,模拟值与检验值相关系数均在0. 912之上,均方根误差为2. 249％ ～ 4. 215％,平均绝对误差为1. 573％ ～ 3. 514％.说明马尔可夫链模型能够较好地反映不同深度土壤质地的转移规律,应用于大尺度多层土壤质地部分缺值数据填补和重建有一定的可行性.%The soil texture data set for land models in central China was used for Markov-chain model simulation study. The hypothesis test results show that the vertical transition probability matrix of soil texture can meet the requirements for Markov property and stability,indicating that Markov-chain model can be used for the simulation of the distribution of soil texture as the minimum condition for it is meeting at least one dimension of the Markov-chain requirements. Missing value simulation test on the fourth layer and the eighth layer shows that the correlation coefficient between the fourth layer and the eighth layer is above 0. 912,the root mean square error is 2. 249 ～ 4. 215％,and the average absolute error is 1. 573 ～ 3. 514％. The results show that the Markov-chain model can well reflect the law of soil texture transfer at different depths,and it is feasible to fill and reconstruct the data of large scale multilayer soil texture.
展开▼
