基于混沌理论的非饱和土含水率预测
Prediction of unsaturated soil moisture content based on chaos theory
朱悦璐 1吴奇俞1
作者信息
- 1. 南昌工程学院 水利与生态工程学院,江西 南昌 330099
- 折叠
摘要
针对无资料区土体含水率数据难以获取的问题,提出了一种基于卫星反演-相空间重构-非饱和入渗计算的组合方案,以研究区110 d土体表层含水率为基础,预测未来100 d无资料时段土体表层及内部含水率分布规律.计算结果表明:研究区含水率时间序列具备混沌特征,可由一维时间序列拓扑为一个嵌入维数m =5,迟滞τ =10 的相空间,由该相空间预测的土体表层含水率在验证期最小相对误差为0.7%,最大相对误差为2.4%,在预测期最小相对误差为2.2%,最大相对误差为8.3%,均满足工程需求,因此将其用于后续非饱和入渗计算的边界条件是真实有效的.该方案具有动力学特性和物理力学意义,可为无资料地区土体含水率估计借鉴.
Abstract
To address the problem of difficulty in obtaining soil moisture content data in no data areas,we proposed a combina-tion scheme based on satellite inversion,phase space reconstruction,and unsaturated infiltration calculation.Based on the soil moisture content of the 110-day surface layer in the study area,the distribution law of soil moisture content of the surface layer and internal part in the next 100-day no data period was predicted.The calculation results show that the time series of soil mois-ture content in the study area has chaotic characteristics,which can be topologically stretched to a phase space with embedding di-mension m of 5 and delay τ of 10.The minimum relative error of the soil surface moisture content predicted by this phase space during the validation period is0.7%,the maximum relative error is2.4%,the minimum relative error during the prediction peri-od is 2.2%,and the maximum relative error is 8.3%,all of which meet the engineering requirements.Therefore,it is used as the boundary condition for subsequent unsaturated infiltration calculation,and the result is realistic and effective.Compared with tradi-tional studies,the prediction scheme proposed in this paper has dynamical characteristics and physical-mechanical meanings,which can provide a new reference for soil moisture content estimation in no data areas.
关键词
土体含水率/非饱和入渗/相空间重构/混沌理论/Richards方程/非饱和土Key words
soil moisture content/unsaturated infiltration/phase space reconstruction/chaos theory/Richards equation/unsaturated soil引用本文复制引用
基金项目
江西省重点研发计划(20192BBHL80003)
国家自然科学基金(52069014)
出版年
2024
国家科技期刊平台
NSTL
维普
万方数据