Transformation grids for SIRGAS realization interoperability: An NTv2-based approach for frame and epoch unification across tectonically active Americas

Geodetic Reference System for the Americas (Sistema de Referencia Geodésico para las Américas, SIRGAS) comprises three geodetic realizations, SIRGAS95, SIRGAS2000, and SIRGAS-CON (multi-year solutions SIR17P01 and SIRGAS2022), aligned to successive International Terrestrial Reference Frame (ITRF) versions at distinct epochs. Conventional transformation workflows rely on ITRF parameters for frame changes and on velocity models or station-displacement rates for epoch reduction. However, in the tectonically active Americas, where station trajectories may include non-linear deformation, reference-frame transformations combined with generalized velocity models can produce decimeter-level residuals, limiting their suitability for high-precision geodetic applications. This study compares ITRF-based transformations using 7- and 14-parameter Helmert models combined with NUVEL-1A, DGF02P01, VEMOS2017, VEMOS2022, and direct-displacement strategies for epoch reduction with distortion models implemented using National Transformation version 2 (NTv2) grids for SIRGAS95-SIRGAS2000, SIRGAS2000-SIR17P01, and SIRGAS2000-SIRGAS2022. The analyses employed 51, 42, and 67 common stations, respectively, after excluding stations with coordinate inconsistencies and accounting for the tectonic context associated with the Maule 2010, Iquique 2014, and Illapel 2015 earthquakes. ITRF-based transformations produced larger residuals when epoch propagation relied on generalized velocity models, particularly for the longer SIRGAS2000-SIR17P01 and SIRGAS2000-SIRGAS2022 intervals. Horizontal residuals reached the decimeter level and locally exceeded one meter, highlighting the limited ability of generalized epoch-propagation models to reproduce spatially heterogeneous coseismic and postseismic deformation. Conversely, NTv2 grids generated using NTv2Creator showed smaller internal fitting discrepancies. NTv2Creator combines local seven-parameter Helmert transformations with Natural Neighbor interpolation of residual transformation differences. Maximum discrepancies at grid-construction stations were 6, 20, and 28 mm for the three transformation cases, respectively. These values characterize the internal fit of the modeled transformation fields and should not be interpreted as independent estimates of grid accuracy. The SIRGAS2000-SIRGAS2022 grid showed greater spatial heterogeneity, with larger discrepancies concentrated in tectonically active areas and regions with limited nearby control, indicating a less uniform representation of the realization-to-realization displacement field.