Cold rain processes simulated with a nonhydrostatic cloud-resolving model developed by the Japan Meteorology Agency/Meteorological Research Institute and run at 1 km horizontal resolution (1-km-NHM) with a two-moment bulk parameterization scheme are validated using in situ aircraft observations for orographic snow clouds. To statistically validate cold rain processes simulated by the 1-km-NHM, aircraft observations collected during two winter seasons (March and December 2007, a total of 21 flights) over the Echigo Mountains are analyzed and compared with the model.
 For the cases where the differences in the cloud top heights and/or temperatures between the simulations and aircraft observations are small (i.e., the forecast errors are thought to be relatively small), the horizontal wind direction, wind speed, and vertical wind velocity for all three analysis height intervals (2.0-2.5 km, 2.5-3.0 km, and 3.0-3.5 km above sea level) and over all four analysis areas (Toukamachi, Senjyoji, Shimizu, and Naramata) exhibit reasonable agreement between the numerical simulations and aircraft observations, although the 1-km-NHM overestimated the horizontal wind speed in the cloud layer by 2-3 m s^-1. The simulated liquid water contents at every height interval and over every analysis area are significantly underestimated compared with the aircraft observations. Taking into account the underestimation of ice water content (IWC) measured with the Nevzorov TWC/LWC probe, the simulated IWCs in the upper and middle parts of the clouds are also slightly underestimated. The simulated total solid particle number concentrations (TNCs) in the upper and middle parts of the clouds were underestimated compared with the aircraft observations, while the simulated TNCs in the lower parts of the clouds are slightly overestimated. The ratios of simulated cloud ice number concentrations to snow number concentrations are less than unity and much smaller than the corresponding ratios obtained from the aircraft observations. This suggests that the overall conversion from cloud ice to snow in the 1-km-NHM, which occurs primarily through depositional growth, is faster than that in real clouds.