Background

Due to the induced spatial auto-correlation among different features, they tend to be located together in close geographic proximity. Spatial co-location patterns are useful for understanding positive spatial interactions among different geographical phenomena. Existing methods for identifying spatial co-location patterns usually require users to specify two thresholds, i.e. the prevalence threshold and distance threshold. However, these two thresholds are difficult to determine in practice, and improper thresholds may lead to wrong decisions. On that account, the discovery of spatial co-location patterns is modelled as an nonparametric significant test problem by considering the distribution characteristics (e.g. auto-correlation) of each feature. In addition, the validity of the co-location patterns is evaluated based on scale-invariance principle of human perception and cognition.

Innovations

The spatial co-location patterns are interpreted as collections of interactive spatial features where the spatial auto-correlation of each feature is “induced” by underlying spatially auto-correlated variables, and the significance test is employed to determine significant patterns.

The null model of independency is constructed by reproducing several summary characteristics of the observed dataset of each spatial feature.

Applications

Case study 1: The proposed approach was further used to investigate the symbiotic relationships between species in a wetland located in the northeast of China based on the spatial co-location patterns discovered at multi-scales.

Case study 2: The nonparametric test was extended to discover network-constrained spatial co-location patterns among urban facilities in south Toronto. The precision and recall values of mining results detected by our method were higher than those of two state-of-the-art methods.