The Wildland

1
The WildlandUrban Interface (WUI) From Simple
Models to Mitigating Fire Risk
Michael Ghil Ecole Normale Supérieure, Paris,
and University of California, Los Angeles V.
Spyratos, ENS ENGREF P. Bourgeron, INSTAAR,
CU, Boulder and C. Lampin M. Jappiot,
CEMAGREF, Aix-en-Provence
Please see these sites for further
details http//www.environnement.ens.fr/ ,
http//e2c2.ipsl.jussieu.fr/ http//www.atmos.ucla
.edu/tcd/, http//www.aix.cemagref.fr/htmlpub/divi
sions/afax/emax.htm
2
Motivation

• The WUI occupies about 10 of the surface and
  contains about 40 of the houses in the
  conterminous U.S.
• Fires have caused huge damages in the WUI in the
  Western U.S., as well as Southern Europe this
  past summer and fall.
• Further spreading of construction into
  (semi-)pristine vegetated areas.
• Global change seems to increase dryness in the
  areas of interest.
• Actual joint, interactive modeling of housing and
  vegetation seems to be novel and interesting.
• Simple, toy models can provide useful ideas.
• The hierarchical modeling approach allows one to
  go
• back-and-forth between toy (conceptual) and
  detailed (realistic) models, and between models
  and data.

3
Some spectacular WUI fires
A house is threatened by a wall of flames from a
wild fire west of the Missionary Ridge fire,
north of Durango, Colorado (AP/Charlie Riedel).
As Colorado burns in the summer of 2002, flames
from a new fire loom menacingly behind a mountain
home near Durango (Barry Gutierrez)
4
A toy model of the WUI
A lattice model with nearest-neighbor
interactions, already used in forest-fire
modeling novelty is including the houses.

• Panels (a, b, c) the
• central cell is ignited
• its a tree stand
• its a flammable house or
• Its a fireproofed house.

Panels (d, e) example of fire spreading
over the landscape red cells are on fire, black
cells are burnt green and yellow cells are
unburnt trees and houses.
5
Fire-spread probability

• Mean fire size S as a function of p, the
  vegetations
• probability of fire spread, for different house
  densities
• d d0 indicates fireproofed houses only, d1
  flammable
• houses only.

6
Fire proofing vs. Fuel treatment
Regime diagram of mean fire size S as a function
of p and d (d0 or d1) (a) note narrow
transition zone between widespread and limited
fire size and (b) higher efficiency of
fireproofing of houses.
7
Application to Mediterranean landscapes
Fire distribution over 2 days in Europe GIS tool
8

Motivations are similar to Western U.S.
Urbanization in a forested area, in the South of
France
Contexte (6/8)
9
Methodology Combine 2 criteria
Structure of vegetation
Structure of habitat
10
Density of fire outbreaks per 1000 ha over the
period 19972006
Source fire data ONF
C. Lampin, Ph. D. Thesis 2007
Site of Aix
11
Some conclusions /or questions

• What do we know?
• Its getting warmer and drier, which favors
  forest fires.
• Houses spread, which puts more people goods at
  risk.
• So, we should try to improve fire protection.
• Fire proofing entire communities seems to help,
  both the community and the vegetation in which it
  is immersed.

• What do we know less well?
• How, exactly, do houses interact with forest
  fires?
• How does climatic change affect this
  interaction?

• What to do?
• Better understand the system and its forcings.
• Use a full hierarchy of forest-fire and climate
  models to do so.

12
Some references

• Malamud, B.D., G. Morein, D. L. Turcotte, 1998
  Forest fires An example of self-organized
  critical behavior, Science, 281, 18401842.
• Hargrove, W.W., R.H. Gardner, M.G. Turner, W.H.
  Romme, D.G. Despain, 2000 Simulating fire
  pattern in heterogeneous landscapes, Ecological
  Modelling, 135, pp 243263.
• Spyratos, V., P. Bourgeron, and M. Ghil, 2007
  Development at the wildlandurban interface and
  the mitigation of forest-fire risks, Proc. Natl.
  Acad. Sci. USA, 104 1427214276 doi
  10.1073/pnas.0704488104.