The accuracy of surface deformation derived from Interferometric Synthetic Aperture Radar (InSAR) observations depends on the quality of the chosen interferogram subset. We present a method to select interferogram subsets based on unwrapping errors rather than temporal baseline thresholds. Using Sentinel-1 interferograms over the Tulare Basin (CA), we show that tropospheric noise dominates short temporal baseline subset solutions (with up to 2.9 cm/yr residuals at co-located GPS sites), while decorrelation leads to a systematic underestimation of true deformation rate in long temporal baseline subset solutions (with up to 5.5 cm/yr residuals). Our new workflow better mitigates these two noise sources at the same time. In the Eagle Ford (TX) region, our strategy revealed up to ~11 cm of cumulative line-of-sight (LOS) deformation over a ~900 km2 region. This deformation feature is associated with ongoing oil and gas activities and is reported for the first time here.