In this paper, we examine the impact of GPS radio occultation (RO) data on the prediction of tropical cyclogenesis for ten typhoons over the Western Pacific in 2008-2010. The WRF (Weather Research and Forecast) model and its three-dimensional variational data assimilation system (3DVAR) are used to assimilate GPS RO soundings over a three-day period through continuous cycling, starting at five day prior to the actual cyclogenesis. The data assimilation cycle is followed by a two-day forecast, to assess the impact of GPS RO data on tropical cyclogenesis prediction. The results show that the assimilation of GPS RO data substantially increases the probability of detection for the tropical cyclogenesis. Moreover, the use of a nonlocal excess phase operator that takes into account the inhomogeneity of refractivity (due to moisture variability) in the lower troposphere improves the effectiveness of GPS RO data assimilation. The nonlocal excess phase operator is able to better extract the moisture information from the GPS RO data (as compared to the local refractivity operator) and produces an improved moisture analysis. The improved analysis associated with the use of nonlocal operator improves the model's ability to predict the tropical convection, which in turns improves the prediction of tropical cyclogenesis. Detailed diagnosis will be presented on the case of Typhoon Nuri to illustrate the impact of GPS RO data assimilation using the nonlocal excess phase operator on tropical cyclogenesis.