Simulation Modeling

1
Simulation Modeling

• What is a model?
• A computer software package (usual default
  definition)
• A yield table or site index curve
• A regression or other simple equation (e.g.,
  Olson's decay equation)
• A measurement (e.g., the height of mercury in a
  glass tube is a model oftemperature)
• All are an approximation of reality.

2

• Purposes of Modeling

• Prediction
• How do process and empirical models compare?
• Understanding
• To illustrate the consequences of our
  cumulative understanding of nutrient cycling
  processes (Johnson)
• To gain insights not possible from field studies
  and develop new hypotheses
• Help design field sampling protocols

3
Simulation Modeling

• Testing Models
• (Oreskes, Naomi, Kristin Shrader-Frechette, and
  Kenneth Blitz. 1994. Verification, validation,
  and confirmation of numerical models in the earth
  sciences. Science 263 641-646).
• Verification Implies "truth", which is
  unattainable. All must be know to achieve this. A
  typical false "verification involves
  "calibrating" the model for one-half of a data
  set, and running it for the second half. But what
  happens if the environment changes (the rules
  change)?

4
Simulation Modeling
5
Simulation Modeling

• Testing Models
• (Oreskes, Naomi, Kristin Shrader-Frechette, and
  Kenneth Blitz. 1994. Verification, validation,
  and confirmation of numerical models in the earth
  sciences. Science 263 641-646).
• "The goal of scientific theories (and, therefore,
  models) is not truth (because it is not
  attainable), but empirical adequacy (van Frassen
  1980). Empirical adequacy may be judged by
  comparing model output with actual data in a
  blind test. (This is very seldom done, although
  often possible).
• However, it will never be known with certainty
  whether or not the model obtains the right result
  for the wrong reasons (excessive "calibration").

6
Simulation Modeling

• Testing Models
• (Oreskes, Naomi, Kristin Shrader-Frechette, and
  Kenneth Blitz. 1994. Verification, validation,
  and confirmation of numerical models in the earth
  sciences. Science 263 641-646).
• "The goal of scientific theories (and, therefore,
  models) is not truth (because it is not
  attainable), but empirical adequacy (van Frassen
  1980). Empirical adequacy may be judged by
  comparing model output with actual data in a
  blind test. (This is very seldom done, although
  often possible).
• However, it will never be known with certainty
  whether or not the model obtains the right result
  for the wrong reasons (excessive "calibration").

7
Simulation Modeling

• What Good Are Models, Then?
• Management
• Predictive models (especially simple ones, like
  yield tables) are used all the time to good
  effect when not extrapolated beyond their
  appropriate range of environmental conditions.
  See Binkley for examples.
• Scientific
• "Models are most useful when they are used to
  challenge existing formulations, rather than to
  validate or verify them. Any scientist who is
  asked to use a model to verify or validate a
  predetermined result should be suspicious"
  (Orestes et al 1994).

8
Simulation Modeling

• Nutrient Cycling Models
• CENTURY (Parton et al., 1994)
• Developed for agro-ecosystems and grasslands
  recently applied to forests
• Simulates C, N, P, and S in soil-plant system
• Soil organic components divided into 3 fractions
  
• 1. Active (1.5 yr)
• 2. Protected (25 yr)
• 3. Resistant (1000 yr) Leaching bulked with other
  losses

9
CENTURY(Parton et al 1994)

• Simulates C, N, P, and S in soil-plant system
• Developed for agro-ecosystems and grasslands
  recently applied to forests
• Soil organic components divided into 3 fractions
  
• 1. Active (1.5 yr)
• 2. Protected (25 yr)
• 3. Resistant (1000 yr)
• Leaching bulked with other losses

10
Simulation Modeling

• FORTNITE (Aber et al)
• Modified a forest succession model to include N
  cycling.
• Heavy dependence upon litter quality/decomposition
  pathway.
• Used in many cases, included for management,
  without adequate testing for "empirical
  adequacy".

11
FORTNITE/LINKAGES(Aber and Melillo 1982 Pastor
and Post 1986)

• JABOWA forest development module
• Growth limited by N, light, moisture, and
  temperature
• N availability determined by decomposition and
  ligninN
• No Leaching

12
General Ecosystem Model (GEM)(Rastetter et al.,
1991)

• Process-based model of C-N interactions in
  terrestrial ecosystems. 
• Annual and daily time step versions
• Intended to be generally applicable to most
  terrestrial ecosystems
• Has been used in
• Temperate deciduous forests
• Tropical evergreen forests,
• Arctic tundra
• changes in CO2 concentration,
• To simulate changes in
• Temperature
• N inputs
• Irradiance
• Soil moisture

Rastetter, E. B., M. G. Ryan, G. R. Shaver, J. M.
Melillo, K. J. Nadelhoffer, J. E. Hobbie and J.
D. Aber.  1991.  A general biogeochemical model
describing the responses of the C and N cycles in
terrestrial ecosystems to changes in CO2, climate
and N deposition.  Tree Physiology 9101-126.
13
Pnet (Aber et al., 1992)

• Suite of three nested computer models
• Modular approach to simulating the carbon, water
  and nitrogen dynamics of forest ecosystems.
• PnET-Day is the instantaneous canopy flux module
• PnET-II adds nutrient allocation, a water balance
  and soil respiration to produce a monthly
  time-step carbon and water model which is driven
  by nitrogen availability.
• PnET-CN further extends the soil dynamics
  component and closes the N cycle by tracking
  nitrogen, along with carbon, throughout all
  compartments and fluxes

Aber, J.D. and C.A. Federer. 1992. A generalized,
lumped-parameter model of photosynthesis,
evapotranspiration and net primary production in
temperate and boreal forest ecosystems. Oecologia
92463-474
14
Simulation Modeling

• NuCM (Nutrient Cycling Model Liu et al)
• Comprehensive nutrient cycling model for growth,
  N, P, K, Ca, Mg, and S. Designed primarily to
  assess effects of atmospheric deposition. See
  attached documentation.

15
NuCM(Liu et al 1991 Johnson et al 1993, 1995)

• Derivative of ILWAS model
• Stand-level cycling of N, P, S, K, Ca, and Mg
• Strong soil chemistry module
• Plant growth limited by water and nutrients
• Leaching explicitly simulated

16
The Nutrient Cycling Model (NuCM)
Deposition
Wet Dry
Translocation
Foliar Leaching
Throughfall
Mineralization, Weathering
Soil Solution
Soil, Unavailable
Soil, Exchangeable
Exchange, Adsorption/Desorption
Immobilization
Leaching
17
NuCMSites Calibrated

• Site Species
• Clingmans Dome, North Picea rubens
• Carolina Fagus grandifolia
• Coweeta, North Carolina Mixed Deciduous,
• Pinus strobus
• Walker Branch, Tennessee Mixed Deciduous
• Duke, North Carolina Pinus taeda
• Bradford, Florida Pinus elliotii
• Barton Flats, California Pinus ponderosa
• Little Valley, Nevada Pinus jeffreyii
• Nordmoen, Norway Picea abies
• Ås, Norway (Soil Columns) Pinus sylvestris

18
NuCMManipulations Simulated
Manipulation Site(s) References Atmospheric
Deposition Barton, Coweeta, Johnson et al.
(1993,1995a, (N, S, Base Cations) Clingmans,
Duke, 1996) Fenn et al. (1996) Nordmoen,
Ås Sogn and Abrahamsen (1997) Sogn
et al. (1997) Kvindesland
(1997) Harvesting Duke, Coweeta Johnson et al.
(1995a, 1998a) Liming Coweeta Johnson et al.
(1995b) Species Change Duke, Coweeta Johnson
et al. (1995a, 1997)
19
NuCMManipulations Simulated
Manipulation Site(s) References Precip.
Amount Walker Branch Johnson et al.
(1998b) Elevated CO2 Duke, Walker
Branch Johnson (in press) (via N cycle) Climate
change Clingmans Dome, NC Johnson et al
(2001) Walker Branch, TN Duke,
NC Coweeta, NC Bradford, FL Little
Valley, NV
20
NuCMSuccesses (General patterns mimicked)

• Response to liming (Coweeta, NC)
• Effects of N fertililzation in soil columns
  (Norway)
• Prediction of very high soil solution NO3- in
  ponderosa pine forests
• Harvest, species change patterns in Ca and Mg
  (Duke, Walker Branch)
• General patterns in soil solution (Nordmoen,
  Norway Red spruce, Clingmans Dome, NC)
• Soil solution SO42- patterns following NO3- pulse
  (Beech forest, Clingmans Dome, NC)
• Effects of precipitation on Ca2, Mg2 SO42-
  (Walker Branch, TN)

21
Figure 6. Field data and simulation of base
saturation 23 years after liming at Coweeta, NC
22
(No Transcript)
23
NuCMFailures (General patterns not mimicked)

• Chromatographic Mg2, Ca2, and Al2 leaching
  following NO3- pulse (Beech forest, Clingmans
  Dome, NC)
• Organic Al (Soil columns and Nordmoen, Norway)
• Effects of changing precipitation on Cl-, K
• (Walker Branch, TN)
• Overestimation of NO3- leaching during dormant
  season (Soil columns and Nordmoen, Norway)
• Sulfate patterns at Coweeta

24
(No Transcript)
25
(No Transcript)
26
(No Transcript)
27
(No Transcript)
28
(No Transcript)
29
(No Transcript)
30
(No Transcript)
31
(No Transcript)
32
Conclusions on the NuCM Model

• Considerable heuristic value
• Provides insights into the collective
  implications of what we think we understand about
  nutrient cycling processes
• Considerable heuristic value
• Provides insights into the collective
  implications of what we think we understand about
  nutrient cycling processe
• Questionable predictive value
• Mixed success in prediction
• Too many calibration knobs?
• No two users are likely to get the same
  predictions because of varying calibrations

33
General Comments on Prediction

• Empirical models (e.g., biomass regressions)
• Often good at prediction within their domain
• Usually limited heuristic value
• Process models (e.g., NuCM)
• Usually substantial heuristic value
• Questionable value for prediction
• Too many knobs, consequently
• Non-unique predictions
• What critical processes are left out?
• What critical processes are inaccurately
  portrayed?

34
Conclusions

• Process models are useful for exploring the
  collective implications of processes we think we
  understand (scientific/heuristic value)
• Process models should not be used to develop
  environmental guidelines because
• - They have too many knobs they cannot be
  relied upon to consistent predictions from user
  to user
• - They do not adequately simulate several
  important processes (like nitrate leaching, which
  is important for environmental standards).

35
Trans-scientific questions require a modeling
approach

• Some questions are trans-scientific Rastetter
  1996)
• They cannot be answered with experimentation
  (e.g. Global Change)
• They are beyond our current experience
• In these cases, process models may be the only
  alternative to no information at all.
• Are they better than no information at all?