[Paper Review] Using a Numerical Weather Model to Improve Geodesy

A. E. Niell, Leonid Petrov|ArXiv.org|Jan 23, 2004

GNSS positioning and interference10 references3 citations

TL;DR

This paper evaluates the use of numerical weather model (NCEP) data to improve tropospheric mapping functions in geodetic VLBI and GPS solutions. By replacing climatological models (NMF) with isobaric mapping functions (IMF) derived from NCEP's 2.5° grid data, the study shows significant improvements in baseline repeatability and reduced harmonic site position variations, especially for the hydrostatic component (IMFh), though the wet component (IMFw) shows only marginal gains due to coarse grid resolution.

ABSTRACT

The use of a Numerical Weather Model (NWM) to provide in situ atmosphere information for mapping functions of atmosphere delay has been evaluated using Very Long Baseline Interferometry (VLBI) data spanning eleven years. Parameters required by the IMF mapping functions (Niell 2000, 2001) have been calculated from the NWM of the National Centers for Environmental Prediction (NCEP) and incorporated in the CALC/SOLVE VLBI data analysis program. Compared with the use of the NMF mapping functions (Niell 1996) the application of IMF for global solutions demonstrates that the hydrostatic mapping function, IMFh, provides both significant improvement in baseline length repeatability and noticeable reduction in the amplitude of the residual harmonic site position variations at semidiurnal to long-period bands. For baseline length repeatability the reduction in the observed mean square deviations achieves 80 of the maximum that is expected for the change from NMF to IMF. On the other hand, the use of the wet mapping function, IMFw, as implemented using the NCEP data, results in a slight degradation of baseline length repeatability, probably due to the large grid spacing of the NWM that is used.

Motivation & Objective

• To assess whether numerical weather model (NWM) data can improve tropospheric mapping functions in geodetic positioning.
• To evaluate the impact of NWM-derived isobaric mapping functions (IMF) on baseline length repeatability and site position stability in VLBI solutions.
• To determine if IMF can allow inclusion of lower-elevation data without degrading vertical accuracy, compared to standard climatological models (NMF).
• To investigate the limitations of IMFw when applied with coarse-resolution NWM data, particularly in reducing residual harmonic site variations.

Proposed method

• The IMF mapping functions were derived from NCEP's 2.5° horizontal grid NWM data, providing vertical profiles of temperature, pressure, and water vapor for each site.
• Hydrostatic and wet mapping functions were calculated using a three-term continued fraction approximation (equation 1) fitted to raytraced delays from radiosonde data.
• The IMF parameters (a, b, c) were calibrated as linear functions of meteorological variables and geographic location to enable real-time use in geodetic processing.
• The IMF implementation was integrated into the CALC/SOLVE VLBI data analysis software for global solutions using 11 years of VLBI data.
• Baseline repeatability and residual harmonic site position variations were compared between NMF and IMF solutions across multiple stations.
• Height uncertainty was estimated by combining formal position errors with mapping function-induced errors, using sensitivity factors of 0.3–0.8 mm per mm of delay error at minimum elevation.

Experimental results

Research questions

• RQ1 Can NWM-derived atmospheric profiles improve the accuracy of tropospheric mapping functions in geodetic VLBI solutions?
• RQ2 To what extent does using IMF instead of NMF reduce baseline length repeatability and residual harmonic site position variations?
• RQ3 Why does the wet mapping function (IMFw) fail to improve performance despite theoretical potential?
• RQ4 At what minimum elevation can data be safely included when using IMF compared to NMF without degrading vertical accuracy?

Key findings

• The use of IMFh (hydrostatic component) reduced baseline length repeatability errors by 80% of the theoretical maximum improvement compared to NMF, indicating substantial performance gains.
• IMFh significantly reduced the amplitude of residual harmonic site position variations at semidiurnal to long-period bands, improving site stability.
• IMFw showed only marginal improvement over NMFw, with performance degradation likely due to the coarse 2.5° grid spacing of the NCEP NWM limiting wet delay accuracy.
• Formal height uncertainties were maintained at NMF levels down to 7° minimum elevation when using IMFh, whereas NMF caused significant uncertainty increases below 10°.
• For a 15° minimum elevation, atmosphere modeling errors contributed less than 2 mm to vertical uncertainty, but correlated errors over days may slow error reduction in weekly solutions.
• The results suggest that future geodetic solutions should incorporate temperature-dependent antenna height corrections and higher-resolution NWM data to further reduce systematic tropospheric errors.