Atmospheric delay induced by refraction is a significant factor influencing the accuracy of satellite laser ranging (SLR). While ray tracing technique is recognized for its precision in estimating atmospheric delay, there's a gap in the research concerning the computation of integration step size and its impact on results. Therefore, various calculation methods for integration step size were compared and ray tracing with different step sizes were employed to calculate atmospheric delay at Zhuhai Tianqin Station (referred to as "TQ station"). Under different elevation angles, a comparative analysis was conducted to assess the differences in the results and these results were further compared with those from the Mendes-Pavlis model and VMF3o model. The exploration of the relationship between the calculation errors and meteorological factors as well as integration step size indicate that it is essential to select an appropriate integration step size and parameter combination before conducting ray tracing. To achieve millimeter-level accuracy in results, the initial integration step size should be below 50 m. Nevertheless, to avoid data errors, excessively small initial integration step sizes should be avoided. This study highlights the importance of choosing an optimal integration step size and provides valuable insights for improving the precision of atmospheric delay estimation in SLR.