Accurate first-arrival picking from shot gathers is essential for traveltime tomography and near-surface velocity model building. In foothill areas, however, rugged topography and complicated near-surface conditions often result in extremely low signal-to-noise ratios, making first breaks, especially at far offsets, difficult to identify reliably. This limitation weakens ray coverage, degrades the stability of tomographic inversion, and ultimately reduces the quality of subsequent processing steps, such as static correction. To address this problem, we apply deconvolution-based supervirtual interferometry (DSVI) to enhance weak first arrivals in foothill seismic data. The workflow consists of two steps. First, deconvolution interferometry is used to generate virtual refraction responses while reducing source-wavelet imprint and sidelobe effects. Second, convolution interferometry is used to reconstruct supervirtual refractions through coherent stacking of phase-aligned energy. The enhanced gathers show clearer, more continuous far-offset first arrivals, allowing for more stable picking over a broader offset range. A field example from a foothill survey in Tarim Basin shows that DSVI markedly increases the number and spatial coverage of usable traveltimes, improves ray coverage in first-arrival tomography, and yields a more reliable near-surface velocity model. These improvements benefit tomography-based statics and also provide a better long-wavelength starting model for subsequent land full-waveform inversion.