Fig. 1. Dependence of spatio-temporal clustering on scale

Fig. 2. Spatial distributions of two different types of events

Fig. 3. Spurious spatial co-location patterns caused by the random interactions

Fig. 4. Illustration of human visual system

Fig. 5. Relationship between spatio-temporal clusters and data scale

Fig. 6. Meteorological division identified by the statistical method for spatio-temporal clustering

Fig. 7. Illustration of spatio-temporal cross outliers

Fig. 8. Significance tests on the cross outliers

Fig. 9. Induced spatial auto-correlations between different features

Fig. 10. Pattern reconstruction based on multi-modal summary characteristics

Fig. 11. Spatio-temporal co-location patterns determined by the statistical method

Fig. 12. Space-time support vector regression model considering both auto-correlation and heterogeneity