Gas convection is observed in the solar photosphere as the granulation, i.e., having highly timedependent cellular patterns, consisting of numerous bright cells called granules and dark surroundingchannels called intergranular lanes. Many efforts have been made to characterize the granulation, which may be used as an energy source for various types of dynamical phenomena. Although the horizontal gas flow dynamics in intergranular lanes may play a vital role, but they are poorly understood. This is because the Doppler signals can be obtained only at the solar limb, where the signals are severely degraded by a foreshortening effect. To reduce such a degradation, we use Hinode's spectroscopic data, which are free from a seeing-induced image degradation, and improve its image quality by correcting for straylight in the instruments. The dataset continuously covers from the solar disk to the limb, providing a multidirectional line-of-sight (LOS) diagnosis against the granulation. The obtained LOS flow-field variation across the disk indicates a horizontal flow speed of 1.8-2.4 km/s. We also derive the spatial distribution of the horizontal flow speed, which is 1.6 km/s in granules and 1.8 km/s in intergranular lanes, and where the maximum speed is inside intergranular lanes. This result newly suggests the following sequence of horizontal flow: A hot rising gas parcel is strongly accelerated from the granular center, even beyond the transition from the granules to the intergranular lanes, resulting in the fastest speed inside the intergranular lanes, and the gas may also experience decelerations in the intergranular lane.Hinode is a Japanese mission developed and launched by ISAS/JAXA, collaborating with NAOJ as a domestic partner, NASA and STFC (UK) as international partners. Scientific operation of the Hinode mission is conducted by the Hinode science team organized at ISAS/JAXA. This team mainly consists of scientists from institutes in the partner countries. Support for the post-launch operation is provided by JAXA and NAOJ (Japan), STFC (U.K.), NASA, ESA, and NSC (Norway). We are grateful to the Hinode team for obtaining a series of datasets well suited to this analysis. We also thank S. K. Solanki for providing helpful comments and T. L. Riethmüller for instructions regarding how to treat numerical simulation code.