Network RTK generates spatial corrections by using differenced measurements from reference stations in the network, and the corrections are then provided to a rover. The rover, generally, uses linear interpolation, which assumes that the corrections at each reference station are spatially correlated, to obtain a precise correction of its location. However, an irregularity of the tropospheric delay is a real-world factor that violates this assumption. Tropospheric delay is a result of weather conditions, such as humidity, temperature and pressure, and it can cause spatial decorrelation when there are changes in the local climate. In this paper, we have defined the non-linear characteristics of the tropospheric delay between reference stations or user within a region as the "irregularity of tropospheric delay". Such an irregularity can negatively impact the network RTK performance. Therefore, we analyze the influence of the irregularity of tropospheric delay in network RTK based on meteorological data.