Assessment of numerical weather models with different spatial resolutions on tropospheric delay correction for InSAR

Abstract

Spatial variations in atmospheric parameters, including pressure, temperature, and humidity, significantly impact the precision of interferometric synthetic aperture radar (InSAR) measurements. This significantly limits the applicability of InSAR in the fields, such as terrain inversion and deformation monitoring. Fortunately, the evolving numerical weather models (NWMs) could offer a viable tropospheric correction solution. However, given the influence of complex terrain and areas with sparse atmospheric observation, the effectiveness of tropospheric correction with different NWMs' resolutions remains to be evaluated. This requires the examination of different NWMs tropospheric delay correction in detail with sufficient metrics. A total of 36 Sentinel-1 interferograms of 2023, which cover Berlin, Paris, and Milan, respectively, are used as examples. Tropospheric correction is carried out using ICOsahedral nonhydrostatic D2 (ICON-D2), the fifth-generation European Centre for Medium-Range Weather Forecast atmospheric reanalysis, and modern-era retrospective analysis for research and applications version 2. To assess the correction efficacy of various resolutions, standard deviation, semivariogram function, and phase-elevation correlation coefficient served as the evaluation method. Results show that the ICON-D2 model outperforms the other models in these metrics, especially in regions with significant topographic relief. Among them, the standard deviation of the corrected interferogram decreased by 21.6%-35.8%. NWMs have demonstrated effectiveness in mitigating altitude-related tropospheric delays without needing altitude assimilation. Overall, the present study underscores that despite potential uncalibrated atmospheric effects, high-resolution NWMs are anticipated to provide a more precise solution for InSAR, especially in regions exhibiting intricate and challenging terrain features.