Optimizing spatial distribution of parks based on the preference for park types: A case study with an improved two-step floating catchment area method

2SFCA (two-step float catchment area) has been commonly adopted in the analysis of accessibility to parks. However, the allocation of parks with partly different and partly substitutable functions has not been supported by the current method. This study incorporates the conception of preference of people groups for park types and the group-type matrix into the analysis of accessibility, and improves 2SFCA for the case study of Chengdu, China. As results firstly, age groups 0–17, 18–59 and ≥ 60 have significantly different preferences for country, theme and comprehensive parks, and the spatial distribution of these age groups is not even. Secondly, for each park type, a single-peaked spatial gradient is observed, wherein accessibility values are highest in a central sub-district and decrease towards the periphery. However, the locations of these central sub-districts vary across the three park types. Thirdly in the scheme for optimization, the satisfactory accessibility values in the samples could be taken as the target for the units requiring optimization. As two clusters at the urban sub-core area are identified as lacking parks, one cluster needs to build country park 56509 m², and the other one needs 20501 m² comprehensive park at the east part, country park 81991 m² and comprehensive park 73946 m² at the north part. In addition, the area at the urban periphery requires exercise equipment, playing and educational fields although it has sufficient open space. The policy implication and measures are further proposed correspondingly.