Investigation on systematic deviations of absolute and double differential partial wet delay between GNSS, PS-InSAR, and ERA-5 model observations

Interferometric Synthetic Aperture Radar (InSAR) data stacks offer a means to derive integrated water vapor (IWV) from phase delay observations along the satellite line of sight with high spatial resolution. Since interferometric observations are differential in both space and time, they capture relative IWV changes but lack absolute values, which is the variable integration constant. Existing approaches to obtain absolute IWV from InSAR are typically validated against GNSS observations, weather models, or other remote sensing datasets. However, these validations primarily assess the integration methods rather than the observed interferometric phase delay representing partial wet delay.

In this study we investigate the reverse approach by comparing native differential partial wet delay observations from Persistent Scatterer InSAR (PSI), mapped to zenith wet delay (ZWD), with forward-modeled double differential ZWD (DsDt ZWD) derived from GNSS and ERA-5 data. The analysis focuses on a region in Central Europe spanning the France-Germany-Switzerland border from March 2015 to July 2019.

The methodology incorporates data from 4.2 million persistent scatterer (PS) points, ERA-5 ZWD interpolated to these locations, and hourly tropospheric wet delay data from 16 GNSS stations. ... mehrTemporal and spatial differences were computed to generate synthetic DsDt ZWD data stacks, enabling direct comparison of GNSS and ERA-5 ZWD. Analyses were conducted in single differential (temporal) and double differential (temporal and spatial) domains, with evaluations performed at GNSS stations using statistical metrics such as the coefficient of determination (R²) and the Kling-Gupta Efficiency (KGE) index. Seasonal variability was also assessed. Additionally, the study examines how local distances around GNSS stations affect the correlation (R² and KGE) between GNSS-InSAR and GNSS-ERA data, evaluating their impact on measurement consistency.

Results reveal valuable insights into the performance of InSAR, GNSS, and ERA-5 ZWD. In the single differential domain, the variance of Dt ZPWD from InSAR aligns closely with GNSS and ERA-5 data. While scatter plots confirm a linear relationship between GNSS and ERA-5, GNSS vs. InSAR trends appear nonlinear. Applying double differences significantly enhances the correlation between GNSS and InSAR, surpassing that of GNSS and ERA-5. The KGE index highlights improved GNSS-InSAR performance, particularly in correlation (R) and variability ratio (Alpha). Seasonal analyses show that GNSS-InSAR excels during summer, with mean R² values twice those of GNSS-ERA, whereas GNSS-ERA performs better in winter. Regional variability is observed, with higher differences in R² and KGE values at stations in the Rhine Valley.

In conclusion, this study demonstrates the capability of PS-InSAR to provide high-resolution, accurate differential ZWD estimates, particularly during summer. PS-InSAR shows a stronger correlation with GNSS data in the double differential domain compared to ERA-5, underscoring the value of high-resolution ZWD data. Systematic variations in GNSS-InSAR correlation, identified as potential quality indicators for GNSS ZWD products, further highlight the importance of integrating multi-source geodetic data to enhance ZWD monitoring.