三峡库区柑橘土壤水分动态变化的自回归积分滑动平均模型构建
2021-07-16张明博朱士江徐文张涛刘彩虹彭玉强王斌李虎王浩贵树彪
张明博 朱士江 徐文 张涛 刘彩虹 彭玉强 王斌 李虎 王浩 贵树彪



摘要 采用时间序列分析法,针对三峡库区柑橘树根区60 cm深土壤水分建立ARIMA(1,1,1)模型。结果表明,拟合区与预测区相对误差分别为0.37%~7.67%、2.83%~5.56%,平均相对误差分别为4.33%、4.52%,均小于5%。该模型可以很好地模拟与预测柑橘树根区土壤水分的变化趋势,可以进一步应用在其他作物根区的土壤水分监测,为农作物优质生长、节水灌溉提供技术支撑。
关键词 ARIMA模型;构建;时间序列分析;柑橘;土壤水分;动态变化;三峡库区
中图分类号 S152.7 文献标识码 A 文章编号 0517-6611(2021)11-0001-04
doi:10.3969/j.issn.0517-6611.2021.11.001
开放科学(资源服务)标识码(OSID):
Construction of an Autoregressive Integral Moving Average Model for the Dynamic Changes of Citrus Soil Moisture in the Three Gorges Reservoir Area
ZHANG Ming-bo1,2,ZHU Shi-jiang1,2,3,XU Wen1,2 et al
(1.College of Water Resources and Environment, Three Gorges University, Yichang,Hubei 443002;2.Engineering Research Center of the Ministry of Education for the Ecological Environment of the Three Gorges Reservoir Area, Yichang,Hubei 443000;3. Key Laboratory of Efficient Utilization of Agricultural Water Resources, Ministry of Agriculture, Harbin,Heilongjiang 150000)
Abstract The time series analysis method was used to establish the ARIMA (1,1,1) model for 60 cm deep soil moisture in the citrus tree root area of the Three Gorges Reservoir area. The results showed that the relative error of the fitted area and the predicted area was 0.37%-7.67%, 2.83%-5.56%, respectively, the average relative errors was 4.33% and 4.52%, respectively, which were all less than 5%. The model could preferably simulate and predict the change trend of soil moisture in the root zone of citrus trees, and could be further applied to the monitoring of soil moisture in the root zone of other crops, providing technical support for high-quality crop growth and water-saving irrigation.
Key words ARIMA model;Construction;Time series analysis;Citrus;Soil moisture;Dynamic change;Three Gorges Reservoir Area
土壤水分作为影响作物优质生长的重要因素,受不同地区气候、土壤、灌溉等的影响,存在较大差异性。同时作物的产量与品质作为土壤水分差异性的体现形式,因此对土壤水分预报模型的研究持续被国内外学者广泛关注,并取得了大量成果。土壤水分预报模型可以分为确定性模型和随机性模型两大类[1],其中确定性模型包括水量平衡模型[2-3]、土壤-植物-大气连续体(SPAC)水分传输模型[4-5]、SPAC水热耦合传输模型[6-7]等,随机性模型包括数理统计模型(包括回归模型[8]、时间序列模型、人工神经网络模型[9-11]等)、随机水量平衡模型和随机土壤水动力学模型[12]等。其中一些精确度较高的模型,对于模型参数的要求存在一定門槛,该研究选取ARIMA时间序列模型对土壤水分进行研究,该模型对数据要求单一,只需要变量自身的历史数据,适用性较强[13]。一些学者……
