| 李红英,马国飞,金飞,段晓凤,王静,马力文,张晓煜.宁夏中部干旱带降水渗透深度预测模型研究[J].干旱地区农业研究,2016,34(6):138~142 |
| 宁夏中部干旱带降水渗透深度预测模型研究 |
| Study on a forecasting model of precipitation penetration depth in the middle arid zone of Ningxia |
| |
| DOI：10.7606/j.issn.1000-7601.2016.06.21 |
| 中文关键词: 降水 渗透深度 预测模型 土壤质地 天然草场 宁夏 |
| 英文关键词:precipitation penetration depth forecasting model soil texture natural grassland Ningxia |
| 基金项目:公益性行业（气象）科研专项项目“干旱致灾过程及机理”(GYHY201506001)；宁夏气象防灾减灾重点实验室基金项目“宁夏干旱区降水对土壤水分动态的影响” |
|
| 摘要点击次数: 1250 |
| 全文下载次数: 730 |
| 中文摘要: |
| 为掌握降水在宁夏中部干旱带天然草场土壤中的渗透情况，引入根据土壤含水量增量变化确定渗透深度的方法，并运用回归、逐步回归和相关分析等多种统计手段，建立了不同土壤质地下降水渗透深度预测统计模型。结果表明：(1) 运用回归方法建立的渗透深度预测模型（R2在0.60～0.67）比逐步回归预测模型(R2在0.49～0.58)显著性好，两种预测模型中降水量或降水日数的回归系数在置信度为0.05水平下均通过显著性检验。(2) 通过对两种预测模型预测结果与实测值的相关分析，回归预测结果、逐步回归预测结果和实测值之间的相关系数达0.70以上，特别是两种模型预测结果相关性显著（相关系数0.88～0.93），从模型简单可用的角度考虑，最终选用逐步回归预测模型。(3) 兴仁沙壤土条件下预测效果较好（81%～100%的样本相对误差均在30%左右或以下），同心壤土预测效果中等（55%～60%的样本相对误差小于30%），盐池粗砂土条件下效果一般（约50%左右样本相对误差在30%及以下）。(4) 同时，文中预测模型试验验证了不同土壤质地对降水渗透深度的影响，当相同的降水过程下，降水渗透深度大小顺序为粗砂土>沙壤土>壤土。 |
| 英文摘要: |
| In order to understand the penetration of precipitation in the soil at natural grassland of central arid zone in Ningxia, the penetration was calculated based on the increment of soil water content, and the statistical model for predicting the penetration depth of different soil texture was studied by statistical methods including regression, stepwise regression and correlation analysis. The results showed that the regression models for penetration depth prediction were more notable than the stepwise regression models. R2 of the regression models was 0.60～0.67, while R2 of the stepwise regression models was 0.49～0.58. The coefficients of precipitation or precipitation days in two models all had statistically significant relationships with penetration depth. Through the comparison between the predicted results and the measured values by correlation analysis method, the correlation coefficient between them was over 0.70. In particular, the prediction results by two models were significantly correlated (the correlation coefficient was 0.88～0.93). From the usage of the prediction model, the stepwise regression model was selected for penetration depth prediction. In addition, the prediction model performed better for the sand loam soil in Xingren than others and the relative errors of 81%～100% samples were about 30% or lower. The model for the loam soil in Tongxin was acceptable and the relative errors of 55%～60% samples were about 30 percent or lower. The model for the sand soil was not very good and the relative errors of half of samples were about 30 percent or lower. Furthermore, the influence of soil texture on the penetration depth was verified by the stepwise regression model. Under the same precipitation process, the penetration depth was in the following order: sand soil＞sandy loam soil＞loam soli. |
| 查看全文 查看/发表评论 下载PDF阅读器 |