Snow is very of importance to regional water cycle and energy budget. Over the arid region with scarce in-situ measurements for snow study is still inadequate. In the present study, the spatiotemporal distribution of vertical snowfall rate (VSR) and radar reflectivity (Z) are investigated during 2006-2021 by W-band (94 GHz) cloud profiling radar (CPR) onboard CloudSat satellite and ERA5 reanalysis data over Xinjiang, Northwest China. The robust exponential relationships (R2 = 0.98) between VSR and Z from CPR in all seasons are established. The maximum of VSR and Z in summer are found owing to the warmer and moister conditions favor aggregation and riming process of hydrometers, so increasing the density and fall speed of snow particles. In addition, surface snowfall rate (SSR) from ERA5 is compared by CPR, indicating the qualitatively similar SSR during four seasons over the study region, but the obviously quantitative difference for both data source at high mountain areas (HMA) in summer. More than 5% of low confidence rate for SSR retrievals (such as the near-surface Z is toughly identified in steep-varying areas contributing to the unclear whether eliminating the effects of ground clutter on retrieval of snow events, increasing the probability of SSR overestimation) are calculated over HMA in summer can be possibly for the variability. This has implications when developing snowfall climatology from CloudSat products. In future, the linkage between cloud ice water path and SSR should be thoroughly analyzed in the study region, which can effectively avoid the interference of ground clutter for SSR retrievals.