Many variable fertilization approaches based on ‘real-time’ crop N status were developed for making N fertilizer management in precision agriculture. Unfortunately, to date, only few papers reported the effect of variable fertilization algorithms on the spatial and temporal variability of crop parameters. Based on these problems, this study designed three different variable fertilization algorithms based on vegetation index (Y), SPAD (S) and crop growth model (Z), respectively, with uniform fertilization and no fertilization as controls. Results showed that wheat growth had strong spatial dependence, which become stronger after fertilization. Wheat yield also had strong spatial dependence. There were some similar spatial distribution between NDVIs, soil TN and yield, indicating that spatial variability of yield had strong relationship with crop growth status and soil TN content. The site-specific fertilization treatment based on crop growth model (Z) had the best adjustment capacity to promote crop growth and yield, and reduce their spatial variation, compared with other fertilization treatments.