Absorption phenomenon attenuates seismic signal amplitudes, reduces the vertical resolution, and hassles the detection of thin layers. The key objective of this study is the vertical resolution enhancement of zero-offset vertical seismic profile data (ZVSP) by Gabor deconvolution. According to constant Q theory, it is reliable to apply a smoothing along hyperbolic trajectories in the time-frequency domain. In conventional hyperbolic smoothing, an empirical whitening factor is added to stabilize the process. Experiments on real seismic data show that the whitening factor can smear useful information or produce artifacts. To prevent these shortcomings, we apply a logarithmic magnitude spectrum (LMS) hyperbolic smoothing on ZVSP data. As a result, by replacing divisions with subtractions whitening factor is used. Smoothing the Gabor magnitude spectrum of seismic data along hyperbolic paths in the logarithmic spectrum can obtain the magnitude of the attenuation function, eliminate the effect of source wavelet, then, estimate the source wavelet amplitude spectrum. Applying different deconvolution methods on synthetic and real data we show that the performance of the Gabor deconvolution using the LMS is better than that of other methods including Wiener deconvolution and conventional hyperbolic smoothing method.