测绘通报 ›› 2020, Vol. 0 ›› Issue (9): 100-105.doi: 10.13474/j.cnki.11-2246.2020.0292

• 学术研究 • 上一篇    下一篇

基于深度置信网络的GNSS-IR土壤湿度反演

陈堃1,2,3, 沈飞1,2,3, 曹新运1,2,3, 朱逸凡1,2,3   

  1. 1. 南京师范大学地理科学学院, 江苏 南京 210023;
    2. 虚拟地理环境教育部重点实验室(南京师范大学), 江苏 南京 210023;
    3. 江苏省地理信息资源开发与利用协同创新中心, 江苏 南京 210023
  • 收稿日期:2019-11-27 出版日期:2020-09-25 发布日期:2020-09-28
  • 通讯作者: 沈飞。E-mail:shen.f@njnu.edu.cn E-mail:shen.f@njnu.edu.cn
  • 作者简介:陈堃(1996-),男,硕士生,研究方向为GNSS反射信号反演近地空间环境信息。E-mail:1013286921@qq.com
  • 基金资助:
    国家自然科学基金(41904018);江苏省自然科学基金(BK20190714);武汉大学地球空间环境与大地测量教育部重点实验室开放基金(18-01-04)

Retrieving GNSS-IR soil moisture based on deep belief network

CHEN Kun1,2,3, SHEN Fei1,2,3, CAO Xinyun1,2,3, ZHU Yifan1,2,3   

  1. 1. School of Geography, Nanjing Normal University, Nanjing 210023, China;
    2. Key Laboratory of Virtual Geographic Environment(Nanjing Normal University), Ministry of Education, Nanjing 210023, China;
    3. Jiangsu Center for Collaborative Innovation in Geographical Information Resource Development and Application, Nanjing 210023, China
  • Received:2019-11-27 Online:2020-09-25 Published:2020-09-28

摘要: 基于大地测量型GNSS接收机获取的反射信号反演土壤湿度是GNSS领域的研究热点。为克服常规线性回归和BP神经网络算法等的缺陷,本文提出了一种基于深度置信网络的GNSS-IR土壤湿度反演方法。试验结果表明,基于该方法得到的决定系数、土壤湿度平均绝对误差和均方根误差分别为0.909 8、0.017、0.021,与线性回归和BP神经网络算法相比,与实测数据吻合度更高,可有效提高土壤湿度反演精度,证明了方法的有效性和可靠性。

关键词: GNSS-IR, 土壤湿度, 信噪比, 深度置信网络, 反演

Abstract: Retrieving soil moisture based on reflected signals acquired by geodetic GNSS receivers is a research hotspot in the field of GNSS. In order to overcome the shortcomings of conventional linear regression and BP neural network algorithms, this paper proposes a GNSS-IR soil moisture retrieval method based on deep belief network. The results show that the coefficient of determination, the average absolute error and the root mean square error of soil moisture based on this method are 0.909 8, 0.017 and 0.021. Compared with the linear regression and BP neural network algorithm, they are more consistent with the measured data, and can effectively improve the accuracy of soil moisture inversion, which proves the validity and reliability of the method.

Key words: GNSS-IR, soil moisture, signal-to-noise ratio, deep belief network, retrieving

中图分类号: