Landscape Prioritization Models for Planning Dry Forest Restoration

1
Landscape Prioritization Models for Planning Dry
Forest Restoration
A Case Study from the Deschutes National Forest

• Alan Ager
• Pacific Northwest Research Station,
• Western Wildands Environmental Threat Assessment
  Center
• Prineville OR
• aager_at_fs.fed.us

2
Key questions

• Where do we treat?
• How much do we treat?
• When do we treat?
• How do we measure results?

3
Risk Science

• Risk science was developed to deal with uncertain
  events
• Used for analyzing potential impacts of natural
  disasters
• Definition
• Risk (likelihood) (consequence)

4
A Risk Based Framework for Planning

• Measure the relative risk among highly valued
  resources
• Measure the effects of management activities
• Prioritize treatment areas
• Choose treatment alternatives
• Determine effective mitigation methods (Risk
  assessment), i.e. how risk factors contribute to
  overall risk and how to change the outcome
• Likelihood
• Intensity
• Effects

Likelihood
Effects
Risk
Intensity
5
Wildfire Risk

• For a single resource value and considering only
  losses
• E(Loss) S(BPi) Lj)
• BPi is the burn probability at a given intensity
• Li loss at fire intensity i
• (Risk likelihood consequence)

6
Measuring Burn Probability

• Burn probability is a function of
• Ignition
• Weather and fuels conditions conducive to spread
• Escape initial attack (10 AM rule)
• Subsequent suppression fails
• Large fire event
• Probability of a point burning p(Fxyt) p(I)
  p(E) p(Sp) p(Su)
• Extensive literature for all these components

7
Umatilla National Forest Fire History 1970 - 2005
Large fire spread is the important factor
determining burn probability
8
(No Transcript)
9
Quantifying Losses from Wildfire

• Need loss functions to relate fire intensity to
  specific loss
• Structures
• Protected habitat
• Large trees
• Biomass
• Carbon
• Water quality
• Scenic and recreational values

10
Loss Functions
Fire intensity
2 ft 4 ft 6 ft 10 ft
11
Owl Risk Analysis for the Five Buttes Project
12
Deschutes Wildfire Risk Assessment

• Measure the relative wildfire risk among
  different resource values on the Forest
• Determine if conservation reserves are
  progenitors or victims of fire
• Measure the change in risk from proposed fuel
  treatments

13
Fire source
Fire source
Victims
Victims
14
Forest Management Plan

• 550,000 ha in protected areas
• Conservation reserves
• Recreation sites
• Research natural areas
• Wilderness
• 240,000 ha managed forest

15
(No Transcript)
16
Land Management Strata
17
The Experiment

• Simulate lots of extreme fire events with
  random ignition locations and estimate
• Burn probability
• BP F/N
• Marginal burn probability
• MBPi probability of a fire at the ith 0.5
  meter flame length category
• Conditional flame length
• CFL ?(MBPi /BP)(FLi)
• Fire size at each ignition point

18
Marginal burn probabilities
ignition
3 random ignitions encounter a pixel from
different directions
19
Burn probability

• Min 0.0
• Max 0.027
• Average 0.0074

20
(No Transcript)
21
Averages by Management Area
22
Fire size tagged on ignition pointMean fire
size 14,300 ha
23
Fire size tagged on ignition point
24
Progenitors versus victims

• Ratio (Firesize/BP)
• High values indicate that large fires are
  generated from an ignition at that point
• Low values mean that there is a relatively high
  likelihood of burning, but large fires are not
  generated from that point

25
(No Transcript)
26
(No Transcript)
27
Relative BP compared to managed forest
28

• Ignitions limited to RHCAs

29
Effect of land strata and burn probability on
fire size
30
Fuel Treatment Scenario

• 5 Treatment priorities based on restoration
  objectives (TNC) and Forest Plan restrictions
  (Green to Red)
• Scenario treats everything (63,000 ha) outside of
  reserves

31
(No Transcript)
32
Relative BP after treatment
33
Change in risk to old growth from treatments
34
Conclusions

• Wide variability in fire behavior within reserve
  systems
• There is good evidence that some of the reserves
  are progenitors of fire contributing risk to
  other reserves
• The outputs are a good starting point to address
  large fire threats to conservation reserves and
  restoration priorities
• Risk framework provides a quantitative measure of
  potential wildfire impacts and effects of
  treatments

35
Landscape Fuel Treatment Strategies

• SPLATS/SPOTS/TOM
• Maximize reduction of fire spread pre area
  treated (FlamMap)
• WUI (LHF)
• Treat the most accessible areas with minimal
  resource concerns and high values (GIS exercise)
• RNF - Restore Natural Fire
• Maximize area that can handle WFU per area
  treated (ArcFuels)

36
SPLATS/SPOTS/TOM Implementation problems

• Optimized treatment patterns are difficult to
  implement
• Reserves limit treatment placement
• Dry forests with patches of grassland fuels
  negate treatments
• National Fire Plan directs treatments to dry
  forest types to restore natural fire regimes
  (wildland fire use)
• And treat WUIs

37
RNF Algorithm

• Implemented in ArcFuels
• Searches landscape for best entry point to build
  a homogeneous area for future WFU
• User specifies area that can be treated and fire
  behavior threshold

38
(No Transcript)
39
Fire Modeling Resources

• www.wwetac.org/arcfuels
• www.fire.org
• aager_at_fs.fed.us

40
Future Work

• Refine treatment scenarios
• Estimate expected loss of large trees, and carbon
  emissions
• Economic analysis of fuel treatment effectiveness
• Annualize burn probabilities
• Time

41
Acknowledgements

• Dave Owens
• Dana Simon
• Leo Yanez
• Mike Simpson
• Helen Maffei

42
Figure 1.