• 主题
高级搜索  >
历史搜索 清空历史
首页> 外文期刊>Journal of Hydroinformatics >An evaluation of nonlinear methods for estimating catchment-scale soil moisture patterns based on topographic attributes
【24h】

An evaluation of nonlinear methods for estimating catchment-scale soil moisture patterns based on topographic attributes

机译：基于地形属性的流域尺度土壤水分格局非线性估算方法的评价

获取原文
获取原文并翻译 | 示例
           
页面导航
  • 摘要
  • 著录项
  • 相似文献
  • 相关主题

摘要

Physical processes that impact soil moisture are typically expressed as nonlinear functions, but most previous research on the estimation of soil moisture has relied on linear techniques, in the present work, two machine learning techniques, a spatial artificial neural network (SANN) and a mixture model (MM), that can infer nonlinear relationships are compared to multiple linear regression (MLR) for estimating soil moisture patterns using topographic attributes as predictor variables. The methods are applied to time-domain reflectometry (TDR) soil moisture data collected at three catchments with varying characteristics (Tarrawarra, Satellite Station and Cache la Poudre) under different wetness conditions. The methods' performances with respect to the number of predictor attributes, the quantity of training data and the attributes employed are compared using the Nash-Sutcliffe coefficient of efficiency (NSCE) as the performance measure. The performances of the methods are dependent on the site studied, the average soil moisture and the quantity of training data provided. Although the methods often perform similarly, the best performing method overall is the SANN, which incorporates additional predictor variables more effectively than the other methods.
机译：影响土壤水分的物理过程通常表示为非线性函数，但是大多数先前对土壤水分估算的研究都依赖于线性技术，在本研究中，两种机器学习技术，空间人工神经网络（SANN）和混合方法可以推断非线性关系的模型（MM）与多重线性回归（MLR）进行比较，以便使用地形属性作为预测变量来估算土壤湿度模式。该方法适用于在不同湿度条件下在三个具有不同特征的流域（Tarrawarra，卫星站和Cache la Poudre）收集的时域反射计（TDR）土壤水分数据。使用Nash-Sutcliffe效率系数（NSCE）作为性能指标，比较了方法在预测变量属性数量，训练数据量和所使用属性方面的性能。这些方法的性能取决于所研究的地点，平均土壤湿度和所提供的训练数据的数量。尽管这些方法通常具有相似的性能，但总体上效果最好的方法是SANN，它比其他方法更有效地合并了其他预测变量。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱：kefu@zhangqiaokeyan.com

京公网安备：11010802029741号 ICP备案号：京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有

  • 客服微信

  • 服务号