Lecture 13 Models II

1
Lecture 13Models II

• Principles of Landscape Ecology
• March 31, 2005

2
Landscape Models A Fortuitous Juxtaposition

• Theory and concepts from landscape ecology
• Broader scales in both space and time
• Understanding spatially explicit processes
• Relation of pattern to process
• Technology improvements
• Computers
• Remote sensing
• Geographic information sytems (GIS)

3
Landscape model limitations and assumptions

• Knowledge-based limitations
• Ecological, management questions have gotten more
  complex
• Data-based limitations
• Often, adequate data to parameterize large
  landscape do not exist.
• Problem
• Determining how much knowledge is needed AND at
  what scale to answer our questions

4
Landscape model limitations and assumptions
Scale
typical ecological study

• How do we scale up small scale processes to the
  landscape?
• Scaling up involves extrapolating data to
  larger spatial and/or temporal scales. Process
  rates are inferred for larger scales often beyond
  the range of the data collected for estimation.

Phenomena of interest
Spatial scale
5
Landscape model limitations and assumptions
Scale
How do Modelers Scale up?
Many Routes Linear (or Multiplicative)
Approach Simply apply fine-scale data to
broad-scale area. However, most data cannot be
scaled up linearly, since it assumes constant
processes across scales and a lack of
thresholds. Additive Approachs Use
type-specific predictions to account for spatial
variability in the study area (divide and
conquer). Dangerous but often necessary - most
empirical data is collected at a fine resolution.
6
Landscape model limitations and assumptions
Scale
Scale insensitivity assumption Derived from
hierarchical theory. To understand the outcome of
a process at a given scale, complete
representation of underlying processes not
necessary. Multi-modeling Use predictive
sub-models that operate at the smallest scale of
the larger model. Regardless of Method The
Input Data should be derived from the appropriate
scale whenever possible!
7
Landscape ModelGradients
8
Other Landscape Model Conceptual Gradients
REALISM
Traditional trade-offs
PRECISION
SCOPE/GENERALITY
Additional trade-offs
SITE SPECIFIC
GENERAL BEHAVIOR
Parameterization
RULE BASED
Process Representation
MECHANISTIC
gt 6 Months
Usability/Learning Curve
DAYS
9
Example forest models
Competition /Light availability
Individual tree models
Raster models
Disturbance spread/Seed dispersal
Spatially interactive
Patch models
Gap models
None
Mechanistic detail
10
Examples of Landscape Models
11
X SORTIE
Raster models
Individual tree models
Spatially dynamic
Patch models
Gap models
Mechanistic detail
12
SORTIE
SORTIE is a mechanistic, spatially explicit,
stochastic model of individual trees. Parameters
Tree and Crown Sizes Growth Function Mortality
Function Dispersal Shading
13
SORTIE
14
SORTIE
15
SORTIE
Year 100 - Clear cut
Year 500
Year 1000
16
Raster models
Individual tree models
Spatially dynamic
Patch models
X VDDT/TELSA
Gap models
Mechanistic detail
17
VDDT/TELSA

• A Markov Model with Patch Dynamics
• Assumptions
• Constant transition probabilities - succession
  does not change over time (stationary).
• Ecological processes and structures do not
  aggregate or disaggregate.
• Constant seed rain.
• Model Procedure
• 1. Create successional pathway diagrams with
  disturbance probabilities.
• 2. Apply the above to a real landscape.

18
(No Transcript)
19
Ponderosa Pine Forest Dynamics - Jemez Mountains
PP Forest ? TPA multi-age stands? even-aged
clumps? ? surface fuels ? CF risk
Burned PP Forest standing dead trees little/no
soil or grass cover
Rx Fire Only (large scale)
Thinning to 40-100 TPA at WUI
Insects
Thin Burn to 100 TPA (med. scale)
severe
PP Woodland 100 TPA multi-age stands? even-aged
clumps? understory grass or oaks dominant?
PP Forest dead live trees little grass cover
Crown Fire high-intensity spring/summer El
Nino?La Nina
moderate
Thinning to 60-100 TPA (small scale)
Succession Recruitment, 6 yrs
Surface Fire
Drought Insects
Repeated Thin Burn landscape scale
PP-MC Forest 1,000-1,300 TPA even-aged
stands many sm-diam trees little grass cover high
fuel loads
Active FS 20th c.
PP Forest ? sm-diam TPA dead live trees little
grass cover
Thinning to 40-60 TPA at WUI
OG (19th c.) ? surface fuel de facto FS
Active FS 21st c.
PP Savanna 40 TPA multi-age stands? even-aged
clumps? understory grass or oaks dominant?
Surface Fire low-intensity MRI 2-15
yr spring/summer
Thinning to 40 TPA at WUI
PP Woodland 100-? TPA multi-age stands even-aged
clumps grassy understory
Surface Fire low-intensity MRI 2-15
yr spring/summer
20
(No Transcript)
21
VDDT/TELSA
TELSA Output

• TELSA requires
• Forest cover maps.
• Maps of zones with management constraints.
• Mgmt information (management limits, stand ages,
  size classes of management units, roads).
• Disturbance size-class distributions,
  between-year variation and temporal trends.

Note A very static world.
22
Examples of Landscape ModelsHuman Behavior
23
The Urban Growth Model

• Predicts spatial extent of urban expansion
• Based on past urbanization patterns
• Roads, settlements, slope, new centers
• Cellular Automaton model
• Uniform application of growth rules (
  coefficients)
• Calibrate coefficients from past land use
  changes.
• Four types of growth (via decision rules)
• Five growth coefficients

Urban development in Baltimore/Washington region
24
Santa Monica Mountains Study Area
1989
1947
2000 predicted
1976