"Kernel Density Estimation Methods for a Geostatistical Approach in Seismic Risk Analysis: the Case Study of Potenza Hilltop Town (southern Italy)" Third International Workshop on "Geographical Analysis, Urban Modeling, Spatial Statistics"
Geostatistical Approach in Seismic Risk Danese Murgante Lazzari
1. Kernel Density Estimation Methods for a Geostatistical Approach in Seismic Risk Analysis: the Case Study of Potenza Hilltop Town (southern Italy) Maria Danese * , **, Maurizio Lazzari * , Beniamino Murgante ** International Conference on Computational Science and Its Applications (ICCSA 2008) - 30June -3July 2008 - Perugia, Italy * National Counsil of Research Archaeological and Monumental Heritage Institute, ** Università degli Studi della Basilicata, Dipartimento di Architettura, Pianificazione ed Infrastrutture di Trasporto
2. The problem: the seismic approach Program of prevision, prevention and protection Seismic event consequences evaluation and damages quantification Analysis of seismic damage scenarios
3. Analysis of seismic damage scenarios: instruments European Macroseismic Macroscale 1998 Vulnerability classes Damage levels Historical macroseismic scenarios Relationships between macroseismic intensity and damage levels
7. KDE: intensity and its measures First order effects (Absolute location) Second order effects ( Relative location ) Properties of a spatial distribution* *Gatrell et al. (1996)
12. KDE: a method for the right choose of bandwidth Nearest-Neighbor Index Nearest-Neighbor Expected Distance NNI > 1 observed distance is higher than the expected distance; events are more scattered than expected. NNI < 1 observed distance is smaller than expected distance Nearest-Neighbor Observed Distance
14. The case of study: Potenza hilltop town 1857 1930 1980 Over-consolidated clayey substratum Sandy-conglomerate deposit lays Narrow asymmetrical ridge
17. The case of study: parameters selection Intensity choice 6 5 D5 5 4 D4 3 3 D3 2 2 D2 1 1 D1 Intensity in northern sector Intensity in middle-southern sector Damage level
18. The case of study: parameters selection Kernel choice Cell size choice 0.1m
19. The case of study: parameters selection Bandwidth choice 6.8 Nearest neighbour mean calculated for whole point pattern Fixed for whole point pattern 1 KD map t (m) Methods used to estimate t Bandwidth approach Case
20. The case of study: parameters selection Bandwidth choice Sum of two resultant raster 3.9 6.8 D1-2-3 damage level: average of building’s minimum semi-dimension. D4-5 damage level: nearest neighbour mean calculated for whole point pattern. Two different fixed bandwidths 2 KD map t (m) Methods used to estimate t Bandwidth approach Case
21. The case of study: parameters selection Bandwidth choice Sum of two resultant raster 1.4÷9.9 6.8 4.1 D1-2-3 damage level: building’s minimum semi- Dimension D4-5 damage level: Building’s area ≤ mean + sd nearest neighbour mean calculated for whole point pattern Building’s area > mean + sd nearest neighbour mean calculated for whole point pattern multiplied by correction. One KDE with Fixed method, one with Adaptive method 3 KD map t (m) Methods used to estimate t Bandwidth approach Case