The flow field around spur dike group is complex, noticeable, and widely encountered in the improving progress of urban riverways and coastlines. Detailed investigation on such flow phenomenon is necessary and of applied significance. In contrast to experimental study and field survey, the numerical simulation can provide much more details with relative low cost. Aiming to identify the appropriate Computational Fluid Dynamics (CFD) models for simulating flow around spur dike group, the flow fields around non-submerged and submerged spur dike groups, including eight spur dikes in staggered arrangement, were numerically investigated in this paper with selected CFD models and validated based on corresponding flume test, i.e., two sets of laboratory experiments and observed data collected. The numerical simulations were carried out using Finite Volume Method (FVM) and three turbulence models, i.e., standard k-e model, Reynolds Stress Model (RSM) and Large Eddy Simulation (LES) model. In each model, the free surface boundary was implemented respectively via two approaches, i.e., rigid-lid assumption (RLA) and volume of fluid (VOF) method. The comparisons between the CFD outputs and the observed data from flume experiments show that all three turbulence models are capable to simulate the three-dimensional flow around spur dike group to certain degree. It is noticed that, with comprehensive understanding of simulation accuracy and computational time, aiming to rapidly capture the field characteristics of main flow for non-submerged spur dike group, the standard k-e model under RLA method is recommended, while to achieve the fine simulation of spur dike field especially in backflow zone, LES model under VOF method is appropriate. For submerged spur dike group, comparing to simulation accuracy, simulation cost is not a major factor concerned and LES model based on VOF method is the most suitable one due to the overflow effect of spur dike crest and the backwater effect of spur dike body.