A cruise observation was performed in the time period of 27 November to 25 December, 2001 by launching 84 radiosondes in the mid-Pacific for a wide latitude region of 25N to 48S with latitudinal intervals of about 1 degree. We extracted gravity wave components following the definition by SPARC gravity initiative to examine latitudinal variation of their dynamical characteristics. The gravity wave energy is largest around the equator just above the tropopause, but it seems that the energy peak spreads to higher latitudes at higher altitudes. The peak values are several times as large as the climatology in land regions shown by SPARC gravity wave initiative. The vertical wavenumber (m) spectra are steep and close to the shape proportional to m^-3 in the equatorial region while they become shallower at higher latitude regions. An interesting result obtained by the following detailed analysis is that the energy peak observed in the equatorial region is due to an inertia-gravity wave packet propagating from the Northern to Southern Hemispheres. In the meridional cross sections of winds and temperature, phase and energy propagation of a wave packet is observed in the latitude region from 7N to 20S, even though each profile was obtained at different time. In particular the phase structure is clear in the latitude region of 11S to 19S. Thus we performed hodograph analysis for each vertical profile in this region with a theoretical constraint of phase relation of an inertia-gravity wave between winds and temperature components. The temperature amplitude estimated from wind amplitudes and obtained wave parameters accords well with the observation, indicating that wave packets are due to an inertia-gravity wave. The inertia-gravity wave propagates southeastward with a horizontal wavelength of about 1500km and vertical wavelength of about 4.5km. These estimated wave parameters explain well the phase structure and energy propagation observed in the meridional cross sections. The reason why the wave packet propagation is nicely detected by this cruise observation is that the ship speed is almost the same as the meridional group velocity of the inertia-gravity wave. A satellite IR picture is examined when the wave packet is located at the tropopause (z=17km) when the ship is located at 7N on December 6, 2001. A clear structure of ITCZ was observed at this particular time and location. Thus it is likely that the observed gravity wave is excited by vigorous convection at ITCZ.