Scholarlink

<p>Multipath and diffraction effects can severely degrade the accuracy of GNSS precise positioning, particularly in applications such as control surveying and deformation monitoring. Considering the spatial autocorrelation of these errors offers the potential for enhancing residual-stacking error mitigation techniques. However, using spatial autocorrelation through covariance functions for all residuals faces significant challenges, including the complexity of solving equations with large covariance matrices and the negative impact of outliers on calibration accuracy. To address these issues, an autocorrelation-based approach rooted in kriging theory is proposed, incorporating three key strategies. First, the residuals are confined to local neighborhoods, significantly reducing the matrix size for equation solving. Second, the identification and removal of the outliers in the residuals are conducted using the leave one arc out validation method. Third, the determination of the variogram model relies only on the residuals within local neighborhoods, reducing the effect of the inherent nonstationarity on the covariances. This new approach is then applied to process GNSS observations from high mountainous areas where multipath and diffraction effects are evident. The results show that the approach significantly reduces the size of the covariance matrix, eliminating the need for partitioning in modeling even for datasets spanning more than 10&#xa0;days. Compared with the conventional grid method, the average variance reduction rate of the proposed approach improves from 26.0 to 45.6% for the residuals of station SCM2 on day-of-year (DOY) 248–257 of 2022 in precise point positioning mode, and from 60.0 to 79.6% for the residuals of baseline SCM3_SCM4 on DOY 042–051 of 2024 in relative positioning mode.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

An autocorrelation-based residual-stacking approach for GNSS multipath and diffraction mitigation and its application in high mountainous areas

  • 期刊论文

  • 发表日期：

    2025年6月7日

查看全文
  • Yumiao Tian,
  • Wenhao Xiong,
  • Yibing Liang,
  • Yu Tang,
  • Xi Mei,
  • Xuefeng Yang,
  • Miao Lin

摘要

Multipath and diffraction effects can severely degrade the accuracy of GNSS precise positioning, particularly in applications such as control surveying and deformation monitoring. Considering the spatial autocorrelation of these errors offers the potential for enhancing residual-stacking error mitigation techniques. However, using spatial autocorrelation through covariance functions for all residuals faces significant challenges, including the complexity of solving equations with large covariance matrices and the negative impact of outliers on calibration accuracy. To address these issues, an autocorrelation-based approach rooted in kriging theory is proposed, incorporating three key strategies. First, the residuals are confined to local neighborhoods, significantly reducing the matrix size for equation solving. Second, the identification and removal of the outliers in the residuals are conducted using the leave one arc out validation method. Third, the determination of the variogram model relies only on the residuals within local neighborhoods, reducing the effect of the inherent nonstationarity on the covariances. This new approach is then applied to process GNSS observations from high mountainous areas where multipath and diffraction effects are evident. The results show that the approach significantly reduces the size of the covariance matrix, eliminating the need for partitioning in modeling even for datasets spanning more than 10 days. Compared with the conventional grid method, the average variance reduction rate of the proposed approach improves from 26.0 to 45.6% for the residuals of station SCM2 on day-of-year (DOY) 248–257 of 2022 in precise point positioning mode, and from 60.0 to 79.6% for the residuals of baseline SCM3_SCM4 on DOY 042–051 of 2024 in relative positioning mode.