1. Geography and Environment 24/7 Population modelling for natural hazard assessment Alan Smith University of Southampton, UK Colloquium on Spatial Analysis, Copenhagen, 10 January 2013

2. Acknowledgments Developing the “Pop 24/7” methodologies: Professor David Martin Samantha Cockings Part of a PhD funded by the Economic and Social Research Council 2

3. Background Better population estimations are required for hazard risk assessment Censuses typically provide a decadal ‘night-time’ population estimation This does not take into account the large fluxes of temporary populations during the day Events of 2011/12 have focused global attention on natural hazards and their impacts 3

4. 4 Christchurch, NZ

5. 5 Miyako, NE Japan

6. 6 Southampton, New York, USA…

7. 7 …Southampton, UK Conventional density maps

8. 8 “Pop 24/7” overview Spatio-temporal gridded population modelling –Variable kernel density estimation (KDE) –Utilises population centroids –Redistributes resident populations according to a temporal profile –Population subgroups Removal of arbitrary administrative boundaries Allows locations of zero population density (Eg. Water)

9. 9 Methods Distribution between population centroids Creation of temporal profile using ancillary data sources Building new datasets to model (Eg. Retail)

10. 10 Data inputs

11. Large user and small business count UPCs Using commercial postcodes as georeferenced points to receive a known or estimated transient ‘daytime’ population (E.g. Shop locations for shoppers) 11

12. Temporal Profile A retail example of a temporal profile derived from the Time Use Survey 2000 indicating potential shopper numbers for a given time. P(Respondents Shopping) 12

13. 13 Results Variable grid size, currently using 250 metre resolution Visualization for public communication Population weighted to background mask Combination and analysis with hazard footprint data Application to a UK flooding scenario, using the Environment Agency’s ‘Flood Map’.

14. Example weekday population 14

15. Example weekday population 15

16. Example weekday population 16

17. Example weekday population 17

18. Example weekday population 18

19. Example weekday population 19

20. Example weekday population 20

21. Example weekday population 21

22. 22

23. 23

24. 24 Combing spatio-temporal population information and hazard footprints for analysis in GIS

25. Flood Map data Fluvial risk: 1% (1:100 years) Coastal risk: 0.5% (1:200 years) 25

26. 26 Next steps Further analysis of results Continued development of datasets and temporal profiles Application to a hazard scenario Demonstrate improved exposure estimations Advances in natural hazard risk management Many more potential applications…

27. Contact: Alan.Smith@soton.ac.uk Geography and Environment University of Southampton Southampton, UK SO17 1BJ