Kim Lowell - Using Digital Orthophotographs To Estimate Changes In Bushfire Threat To Built Structures For 1983 Ash Wednesday Fire Scars

Using Digital Orthophotographs to Estimate Changes in Bushfire Threat to Built Structures Prof. Kim Lowell, University of Melbourne/ Dept. of Primary Industries – Victoria

Ron Shamir1, Andreia Siqueira2, John White2, Alice O’Connor3, Gary Butcher3, Mark Garvey1, Michael Niven2 1Country

Fire Authority Victoria 2Victorian Department of Sustainability and the Environment 3Geomatic Technologies

1

Project Drivers


Has the bushfire threat to structures changed since Ash Wednesday 1983? –

–

Can all structures be located economically? (Cadastral records? Property sales?) Can bushfire threat for each be determined? (Structure characteristics vs. microsurroundings vs. macro surroundings)

2

Approach


Structure mapping: –




Human interpretation of 1982, 1983, 2006 digital orthophotographs

Bushfire threat: –

Landscape level weather, topography, and vegetation

3




Study Area Three Ash Wednesday (1983) Fire Scars + 500 m buffer

Macedon Ranges Dandenongs

Otways 4

Structure Data
Orthophotographs (0.35 m resolution) – – –

Houses, Sheds, Commercial, Other Type Confidence Presence Conf.

5

Validation


1 km2 samples – –

1. Independent visual re-checking for 2006 2. Ground-based verification if necessary

6




Weather – – –

Bushfire Threat Model(1)

Bureau of Meteorology Data Prevailing wind direction on high fire danger days Plus perceived shift in wind on days of high FD

7

Bushfire Threat Model (2)


Topography – –




Slope Aspect

Vegetation – – –

Digital Orthophotographs Classified Forest/Non-forest Processed with “TreeDen” algorithm

8




TreeDen algorithm Identifies density of Forest pixels –

Does not identify forest canopy density

1982

2006

Non--forest Non Dense Medium Sparse

“Real” TreeDen 9

Bushfire Threat Model (4)





Combine wind, topography, TreeDen layer for 1982 or 2006 Classify into LowHigh, Very High, Extreme

1982

2006

Decreased Risk No Change Increased Risk

Change 10

Results


Validation – – –

Errors of Omission/Commission: 12%/0.5% +10% “not seen” (22% Total EoO) Type agreement for Houses/Sheds: 91%/78%

Not seen Commission

True Error 11

Bushfire Structure Threat Dandenong

4000

80

3500

70

3000 2500

1982

2000

2006

1500 1000 500 0

60 50

1982

40

2006

30 20 10 0

Low-High Very High




Percentage of Structures

Number of Houses and Sheds

Dandenongs

Extreme

Low-High

Very High

Extreme

Plus… No relation between bushfire threat and Ash Wednesday bushfire damage 12

Bushfire Threat Change (Area) Decrease

No Change

Increase

Net Increase (Decrease)

Otways

8

82

10

2

Macedon Ranges

6

91

3

(3)

13

79

8

(5)

Dandenongs

13

Costs


$6.00/structure or $4.60/property – – – –




Data acquisition: 32% ($1.90/structure) Structure mapping: 39% ($2.30/structure) Threat modelling: 20% ($1.20/structure) Contingency: 10% ($0.60/structure)

Projected: $11.4 million for Victoria

14

Conclusions




22% of structures will not be mapped – mainly small sheds Improvement of the bushfire threat model requires ground-based data Since 1983… – –

Large increase in structures having LowHigh and Very High threats No substantial change in percentage of structures in each threat class 15