Using the spectral analysis method, this paper studied the young Quercus variabilis population that had obvious increase characteristics, and analyzed the population’s stability under disturbance. The result shows th...Using the spectral analysis method, this paper studied the young Quercus variabilis population that had obvious increase characteristics, and analyzed the population’s stability under disturbance. The result shows that Q.variabilis population has 20-47 year periodic fluctuations under the basic one, which is related to its regeneration and therefore can maintain its stability itself. But the internal periodic length of the basic one cannot be distinguished because the tree’s age is not old enough. The results also shows that human disturbance have affected the population’s fluctuation.展开更多
机器学习模型广泛应用于区域性滑坡易发性分析。模型的选择关系到评价结果的可信度、准确率和稳定性。现有滑坡易发性分析模型对比研究侧重模型的预测精度。模型的稳定性和数据量敏感性对机器学习模型的性能评估同样非常重要。本文以福...机器学习模型广泛应用于区域性滑坡易发性分析。模型的选择关系到评价结果的可信度、准确率和稳定性。现有滑坡易发性分析模型对比研究侧重模型的预测精度。模型的稳定性和数据量敏感性对机器学习模型的性能评估同样非常重要。本文以福建省南平市蔡源流域为研究区,以四川省绵阳市北川县为验证区,从预测精度、稳定性和数据量敏感性3个方面深入对比BP(Back Propagation)人工神经网络模型和CART(Classification and Regression Tree)决策树模型在滑坡易发性分析中的效果,主要结论如下:①在逐渐增加一定数量训练样本的过程中,BP人工神经网络模型预测精度的增长率更高。在蔡源流域内,当训练样本数量增加10000时,BP人工神经网络模型的预测精度上升5.22%,CART决策树模型的预测精度上升2.11%。②BP人工神经网络的预测精度高于CART决策树模型,且较为稳定。在100组数据集上,BP人工神经网络模型验证集预测精度的均值和验证集滑坡样本预测精度的均值分别为81.60%和84.86%,高于CART决策树模型的72.97%和76.59%。与此同时,BP人工神经网络模型对应预测精度的标准差分别是0.32%和0.37%,小于CART决策树模型的0.35%和0.67%。③BP人工神经网络模型分析的滑坡易发区相比CART决策树模型,更接近实际滑坡的空间分布。最后,北川县的验证实验也出现了相同的现象。展开更多
文摘Using the spectral analysis method, this paper studied the young Quercus variabilis population that had obvious increase characteristics, and analyzed the population’s stability under disturbance. The result shows that Q.variabilis population has 20-47 year periodic fluctuations under the basic one, which is related to its regeneration and therefore can maintain its stability itself. But the internal periodic length of the basic one cannot be distinguished because the tree’s age is not old enough. The results also shows that human disturbance have affected the population’s fluctuation.
文摘机器学习模型广泛应用于区域性滑坡易发性分析。模型的选择关系到评价结果的可信度、准确率和稳定性。现有滑坡易发性分析模型对比研究侧重模型的预测精度。模型的稳定性和数据量敏感性对机器学习模型的性能评估同样非常重要。本文以福建省南平市蔡源流域为研究区,以四川省绵阳市北川县为验证区,从预测精度、稳定性和数据量敏感性3个方面深入对比BP(Back Propagation)人工神经网络模型和CART(Classification and Regression Tree)决策树模型在滑坡易发性分析中的效果,主要结论如下:①在逐渐增加一定数量训练样本的过程中,BP人工神经网络模型预测精度的增长率更高。在蔡源流域内,当训练样本数量增加10000时,BP人工神经网络模型的预测精度上升5.22%,CART决策树模型的预测精度上升2.11%。②BP人工神经网络的预测精度高于CART决策树模型,且较为稳定。在100组数据集上,BP人工神经网络模型验证集预测精度的均值和验证集滑坡样本预测精度的均值分别为81.60%和84.86%,高于CART决策树模型的72.97%和76.59%。与此同时,BP人工神经网络模型对应预测精度的标准差分别是0.32%和0.37%,小于CART决策树模型的0.35%和0.67%。③BP人工神经网络模型分析的滑坡易发区相比CART决策树模型,更接近实际滑坡的空间分布。最后,北川县的验证实验也出现了相同的现象。
