Annual 2012 Progress Report on the
Integrated Ecosystem Model for Alaska
and Canada Project
A collaborative project for the USGS/UAF Alaska Climate Science Center
And the Arctic, Western Alaska, and Northwest Boreal
Landscape Conservation Cooperatives
(Distributed 28 February 2013)
Helene Genet, Institute of Arctic Biology,
University of Alaska Fairbanks, Fairbanks,
AK 99775, hgenet@alaska.edu
Participants
Principal Investigators:
Stephanie McAfee, Scenarios Network for
Alaska and Arctic Planning, University of
Alaska Fairbanks, USGS Alaska Climate
Science Center, 4210 University Drive,
Anchorage, AK 99508,
smcafee4@alaska.edu
A. David McGuire, Professor, Alaska
Cooperative Fish and Wildlife Research
Unit, US Geological Survey, Institute of
Arctic Biology, 214 Irving I, University of
Alaska Fairbanks, Fairbanks, AK 99775,
admcguire@alaska.edu
Reginald Muskett, Geophysical Institute
Permafrost Lab, University of Alaska
Fairbanks, 903 Koyukuk Drive, Fairbanks,
AK 99775 rmuskett@gi.alaska.edu
T. Scott Rupp, Professor & Director,
Scenarios Network for Alaska & Arctic
Planning, University of Alaska Fairbanks,
3352 College Road, Fairbanks, Alaska
99709, tsrupp@alaska.edu
Yujin Zhang, Institute of Arctic Biology,
University of Alaska Fairbanks, Fairbanks,
AK 99775, yzhang27@alaska.edu
Co-Investigators:
Graduate Students
Eugenie Euskirchen, Research Assistant
Professor, Institute of Arctic Biology, 402
Irving I, University of Alaska Fairbanks,
Fairbanks, AK 99775,
seeuskirchen@alaska.edu
Elchin Javarov, Geophysical Institute
Permafrost Lab, University of Alaska
Fairbanks, 903 Koyukuk Drive, Fairbanks,
AK 99775 eejafarov@gi.alaska.edu
Sergei Marchenko, Research Associate
Professor, Geophysical Institute Permafrost
Lab, University of Alaska Fairbanks, 903
Koyukuk Drive, Fairbanks, AK 99775,
ssmarchenko@alaska.edu
Data Production and Management:
Tom Kurkowski, Operations Lead,
Scenarios Network for Alaska & Arctic
Planning, University of Alaska Fairbanks,
3352 College Road, Fairbanks, Alaska
99709, ttom.kurkowski@alaska.edu
Vladimir Romanovsky, Professor,
Geophysical Institute Permafrost Lab,
University of Alaska Fairbanks, 903
Koyukuk Drive, Fairbanks, AK 99775,
veromanovsky@alaska.edu
Michael Lindgren, Spatial Analyst,
Scenarios Network for Alaska & Arctic
Planning, University of Alaska Fairbanks
3352 College Road, Fairbanks, Alaska
99709, malindgren@alaska.edu
Postdoctoral Research Associates:
Amy Breen, Scenarios Network for Alaska
& Arctic Planning, University of Alaska
Fairbanks, 3352 College Road, Fairbanks,
Alaska 99709, albreen@alaska.edu
Programmers
Alec Bennett, Scenarios Network for Alaska
& Arctic Planning, University of Alaska
Fairbanks, 3352 College Road, Fairbanks,
Alaska 99709, apbennett@alaska.edu
2
Tobey Carman, Institute of Arctic Biology,
University of Alaska Fairbanks, Fairbanks,
AK 99775, tcarman2@alaska.edu
Hardware Systems Support
Dustin Rice, Scenarios Network for Alaska
& Arctic Planning, University of Alaska
Fairbanks, 3352 College Road, Fairbanks,
Alaska 99709, drrice@alaska.edu
Collaborators
W. Robert Bolton, International Arctic
Research Center, University of Alaska
Fairbanks, Fairbanks, AK 99775,
bbolton@iarc.uaf.edu
Paul Duffy, Neptune and Company Inc.,
8550 West 14th Avenue, Lakewood CO
80215, paul.duffy@neptuneinc.org
Zhaosheng Fan, Argonne National
Laboratory, zfan@anl.gov
Guido Grosse, Geophysical Institute,
University of Alaska Fairbanks, Fairbanks,
AK 99775, ggrosse@gi.alaska.edu
Mark Waldrop, U.S. Geological Survey
345 Middlefield Road, M.S. 962
Menlo Park, CA 94025,
mwaldrop@usgs.gov
Feng-Ming Yuan, Environmental Sciences
Division, Oak Ridge National Laboratory,
Building 2040, E274, P.O. Box 2008, MS6031, Oak Ridge, TN 37831,
yuanf@ornl.gov
3
1. Summary
In this project we are developing, testing, and applying the Integrated Ecosystem Model
(IEM) for Alaska and Northwest Canada to forecast how landscape structure and function might
change in response to how climate change influences interactions among disturbance regimes,
permafrost integrity, hydrology, vegetation succession, and vegetation migration. The IEM
framework couples (1) a model of disturbance dynamics and species establishment (the Alaska
Frame-Based Ecosystem Code, ALFRESCO), (2) a model of soil dynamics, hydrology,
vegetation succession, and ecosystem biogeochemistry (the dynamic organic soil/dynamic
vegetation model version of the Terrestrial Ecosystem Model, TEM), and (3) a model of
permafrost dynamics (the Geophysical Institute Permafrost Lab model, GIPL). This document
reports the first full year, i.e., 2012, of the IEM project. The data subgroup made great progress in
2012 in terms of producing 1 km driver data for the IEM framework over the entire IEM domain
for the historical period and for the Intergovernmental Panel on Climate Change Assessment
Report 4 (AR4) generation climate change scenarios. During 2012, the model coupling subgroup
made a lot of progress in synchronously coupling the component models of the IEM framework.
There is still code to be written and tested for completing the task of having the synchronously
coupled framework operational. During 2012, the tundra fire and treeline subgroup made
substantial progress on both ALFRESCO and TEM related issues. There is still some calibration
work to do in both models, and we found that some of the calibration issues are related to the new
1 km historical data sets. Supplement 1 is white paper developed by the thermokarst dynamics
subgroup to define a conceptual approach towards modeling thermokarst disturbance and
associated landscape transitions. In 2012, the wetlands dynamics subgroup made substantial
progress in developing a peatland version of TEM that could represent water table dynamics and
biogeochemistry in an open boreal fen. During 2013 we plan to extend this modeling effort by
focusing model development on an open boreal bog that has developed as a consequence of
thermokarst disturbance in a black spruce permafrost plateau forest. The parameterization and
testing of this model will rely substantially on the ongoing IEM wetland field studies being
conducted at the Alaska Peatland Experiment during 2013. In 2012, we recognized the need to
better engage management and other stakeholder needs through additional outreach activities. We
want to not only disseminate our research objectives and results, but also the management
implications and potential uses for IEM outputs. To accomplish this task, we specifically created
a position and hired a science communicator, Kristin Timm, to take on this role in 2013. Kristin
will participate in our monthly meetings and facilitate discussion among researchers, the
Landscape Conservation Cooperatives Science Coordinators, and the management community.
2. Preface
Ongoing climate change throughout Alaska and Northwest Canada has the potential to
affect terrestrial ecosystems and the services that they provide to the people living in the region.
These services include the provisioning of food and fiber by Alaskan ecosystems, the importance
of ecosystems to recreation, cultural, and spiritual activities of people in Alaska and Northwest
Canada, and the role Alaska ecosystems play in regulating the climate system. Assessments of the
effects of climate change on ecosystem services has in part been hindered by the lack of tools
capable of forecasting how landscape structure and function might change in response to climate
change. In Alaska and Northwest Canada, such tools need to consider how ecological processes
play out in both space and time. Landscapes may change substantially in time and space because
of shifting species composition (e.g., an increase of shrubs in tundra) and species migration (e.g.,
treeline advance). These shifts in landscape structure and function may be caused by changes in
4
disturbance regimes (e.g., fire, insects, wind throw), permafrost integrity, and hydrology across
the landscape. In this project we are developing, testing, and applying the IEM for Alaska and
Northwest Canada to forecast how landscape structure and function might change in response to
how climate change influences interactions among disturbance regimes, permafrost integrity,
hydrology, vegetation succession, and vegetation migration. This tool will provide scenarios of
changes in landscape structure and function that can be used by resource-specific impact models
to assess the effects of climate change on specific natural resources.
Our primary goal in this project is to develop the IEM modeling framework to integrate
the driving components for and the interactions among disturbance regimes, permafrost
dynamics, hydrology, and vegetation succession/migration for Alaska and Northwest Canada.
This framework couples (1) a model of disturbance dynamics and species establishment (the
ALFRESCO), (2) a model of soil dynamics, hydrology, vegetation succession, and ecosystem
biogeochemistry (the dynamic organic soil/dynamic vegetation model version of TEM), and (3) a
model of permafrost dynamics (the GIPL model). The IEM is an integrated framework that will
provide natural resource managers and decision makers an improved understanding of the
potential response of ecosystems due to a changing climate and to provide more accurate
projections of key ecological variables of interest (e.g., wildlife habitat conditions).
Our objectives in this project are to (1) synchronously couple the models, (2) develop
data sets for Alaska and adjacent areas of Canada, and (3) phase in additional capabilities that are
necessary to address effects of climate change on landscape structure and function. The
synchronous coupling of the models is both a technical activity that is necessary so that the
models can exchange data while they are running in parallel for the same climate scenario (see
Figure 1), and a scientific activity to evaluate that the temporal and spatial dynamics of the model
are operating properly. The consideration of Alaska and Northwest Canada allows us to deal with
landscape issues that do not necessarily stop at the Alaska-Canada border and will give the IEM
the capability to support assessments of trans-boundary resource responses to climate change.
With respect to current capabilities, the models have substantial expertise in addressing fire
disturbance dynamics, vegetation dynamics, and permafrost dynamics in interior Alaska,
particularly with respect to upland ecosystems. We have identified three priority issues that need
to be incorporated into the IEM so that it can more fully address issues throughout tundra and
boreal forest regions of Alaska and Northwest Canada: (1) tundra fire and treeline/tundra
succession dynamics, (2) landscape-level thermokarst dynamics, and (3) wetland dynamics. The
incorporation of tundra fire and treeline/tundra succession dynamics will allow us to better
forecast changes in landscape structure and function in northern and northwest Alaska and tundra
regions of Northwest Canada. Landscape-level thermokarst changes are important to incorporate
into the IEM because subsidence associated with the melting of previously frozen water in icerich permafrost can result in substantial changes in vegetation and habitat (e.g., turning a
graminoid tundra ecosystem into a wetland tundra ecosystem). Wetland dynamics are important
to represent because much of Alaska and Northwest Canada is covered by wetland complexes,
and changes in wetland structure and function has the potential to affect numerous animal species
that use wetlands (e.g., waterfowl).
This document reports the first full year, i.e., 2012, of the IEM project. We have arranged
the report as follows: (1) Activities and Progress, (2) Products, (3) Outreach Activities, (4)
Research Plans for 2013, (5) Outreach Plans for (2013), (6) References, (7) Tables, (8) and (9)
Figures. The sections on Activities and Progress and on Products are arranged by various
subgroups (data set development, model coupling, tundra fire and treeline dynamics, and
thermokarst and wetlands dynamics). See Table 1 for the proposed activities of these groups for
each year of the project. There is also a supplement to this report on our activities to define a
conceptual approach for the development of a thermokarst disturbance and transition model. That
document is a stand alone document and is not found at the end of this report.
5
3. Activities and Progress
3.1. Data Set Development
During 2012, the data set development sub-group met monthly in order to achieve several
objectives including defining the bounds of the study, taking an inventory of candidate data sets
to use, outlining several long term strategies, and creating the initial set of inputs for the coupled
model. Each of these objectives was discussed, investigated and debated. The proposed paths
forward were presented to the full research group and decisions were made (see below). One of
the key outcomes of our discussions is that a project of this spatial, temporal, and collaborative
scope must remain flexible to be ultimately successful.
3.1.1. Definition of the bounds of the study
It was decided that the spatial domain of the study area will generally cover the full
extent mainland of Alaska, the Yukon, and that part of Northern British Columbia covered by the
Northwest Boreal LCC (Figure 2). The Aleutian/Bering Sea Islands LCC area was determined to
be beyond the scope of this study because of its focus on marine resource issues. The temporal
bounds of this study will be from 1901 - 2100, with annual and monthly outputs depending upon
the variable of interest. With such a large spatial extent, initial model runs may be limited to
smaller subdomains until we fully understand the computing resources needed to efficiently
operate the fully coupled model over the full domain. Additionally, some smaller areas within
this region will not be modeled because of limited data coverage in a 1km x 1km modeling
context.
3.1.2. Inventory of Climate Data
The data set development sub-group defined options and tradeoffs of various historical
and projected climate data to be used as inputs to the models. There is a long list of interpolated
historical climate data as well as various downscaled datasets available for use. For historical
data, the deciding factors included the need that temperature, precipitation, vapor pressure, and
radiation variables be available from an internally consistent dataset, from 1901 to approximately
current time, and that data sets be updated on a regular schedule. There are several reanalysis
datasets that come close to these requirements, but many either began in the 1970’s, have had a
significant methodological change in the middle of their series or have uncertain funding sources
for future updates. For these reasons, we decided to use the downscaled Climate Research Unit
(CRU) TS 3.1 (3.1.01 for precipitation only) data for historical climate inputs, as this release of
CRU data met all of these requirements to some degree.
To capture the largest range of projected fire variability, which is a major driver of
ecosystem dynamics, this study will use downscaled outputs used in the IPCC’s Fourth
Assessment Report (AR4) from the Canadian Centre for Climate Modelling and Analysis General
Circulation Model version 3.1 - t47 (cccma_cgcm31) and from the Max Planck Institute for
Meteorology European Centre Hamburg Model 5 (mpi_echam5) using the A1B emission
scenario.
3.1.3. Data Production, Storage, and Planning
The sub-group also established data set production strategies. Because many of the input
variables are only available at much coarser resolution than is required for this 1km resolution
modeling study, downscaling and resampling best practices were developed to obtain reasonable
input values. For instance, CRU TS 3.1 data are available as monthly outputs at 0.5 degree
resolution, but CRU also provides a CRU CL 2.0 climatology from 1961-1990 at 10 minute
resolution, but with slightly different, yet still related, variables (e.g., TS 3.1 cloudiness and vapor
6
pressure vs. CL 2.0 sunshine and relative humidity). To produce a higher quality product than
could be produced by simply taking a 0.5 degree resolution variable and resampling to 1km, we
decided to first convert the related TS 3.1 variables to the CL 2.0 climatology and then downscale
them via the delta method to 10 minute resolution, and then resample to 1km resolution. In some
cases, such as radiation, this required calculating the radiation at the top of the atmosphere and
taking into account the effects of clouds on radiation to relate that to the sunshine variable.
Details of all these methods will be outlined in the metadata for these datasets as well as a future
IEM data set publication, as the methods vary depending on the variable of interest.
A full review of current storage, backup, and processing infrastructure was completed to
support the data production, input variable storage, and ultimately the several versions of output
data expected to be produced in 2013. A new processing system with large storage capacity was
purchased and installed.
A Research Data Management Plan was produced to adhere to the Department of Interior
Climate Science Center Policy. This document outlines how the IEM team will facilitate full and
open access to data products produced by this study.
3.2. Model Coupling
During 2012, the model coupling subgroup conducted analyses of existing scientific
models, integration methodologies, and infrastructure needs for the development and operation of
the IEM.
3.2.1. Requirements Gathering, Model Analysis, Coupler Design, Infrastructure, and Tool
Selection
Prior to any start of code development, the model coupling subgroup began a
requirements gathering process. This process involved exploring the intent and expectations of
the project to maximize staff time and to make sure that the final product matches the
expectation. An initial analysis of each model component was undertaken to better understand the
necessary steps required to couple ALFRESCO, TEM, and GIPL. As part of this analysis, several
monthly meetings of the subgroup were dedicated to presentations of model architecture, data
sets, execution methods, and runtime requirements for ALFRESCO, TEM, and GIPL. Once the
requirements analysis was complete, coupling methods were explored that would meet various
requirements. Multiple methods were explored and refined to make a decision on the overall
architecture to be used for the coupled model. To facilitate communication and cooperation
among modeling groups, programmers, and scientists, an analysis of current infrastructure was
conducted, and the subgroup selected standardized tools and languages to facilitate efficiency
during the development process.
3.2.2. Progress
After completion of requirements gathering and model analysis, a coupling design based
on Loose Coupling methods was established that would allow each submodel to be maintained
independently by each modeling group, with minimal changes to individual code sets (Figure 3).
The coupler will consist of a central executable, and each submodel will be managed as a shared
library. Changes to each library may be made by each modeling group as necessary with minimal
impact on the other modeling groups. Variables and data will be shared as necessary through a
mediator object, allowing communication between models to be isolated to specific needs.
After analysis of individual tools and development methods, the modeling groups agreed
to standardize source code maintenance, operating platform, application build tools, and the
deployment process for the individual models. These decisions have led to a number changes in
ALFRESCO, TEM, and GIPL to standardize interactions among the models so that the models
will work coherently within the application framework.
7
3.3. Tundra Fire and Treeline Dynamics
The overall goal of the tundra fire and treeline subgroup during 2012 was to improve our
ability to forecast changes in landscape structure and function through incorporating tundra fire
and treeline dynamics into TEM and ALFRESCO. Our study region includes the area north of
and including the tundra-treeline ecotone in Alaska and western Canada, as well as the Seward
Peninsula and Yukon-Kuskokwim Delta (Figure 4). The modeling efforts will rely on
ALFRESCO and TEM passing information in a fully coupled framework. ALFRESCO will
provide TEM with information on burn severity and tree establishment in tundra regions and
TEM will model biomass and successional dynamics after fire.
In recent years, ALFRESCO has been primarily used for modeling boreal forest
dynamics, although previous work had considered tundra dynamics as well. Since the tree
migration and tundra routines in ALFRESCO had not been used recently, a priority in this past
year has been to add this code back into the model. With this addition, both shrub and graminoid
tundra were included as vegetation types in ALFRESCO, with transitions occurring between the
vegetation types due to fire, succession, or tree colonization and establishment (Figure 5). This
new version of ALFRESCO has been simulated over our study region to year 2100 using future
climate inputs from the ECHAM5 and CCCMA global circulation model and the AIB emission
scenario (Figure 6). The model outputs pertaining to burn severity and tree establishment are
currently being provided to use as inputs to TEM in an asynchronous coupling, prior to
implementing the fully coupled framework.
We are currently adopting the current fire routine from the dynamic soil organic layer
version of TEM (DOSTEM) to the tundra, which includes parameterizing and calibrating the
model for the shrub, wet sedge, heath, and graminoid tundra types. Much of this effort has
involved parameterizing the model for soil carbon in these tundra ecosystems, taking into account
the fibric, amorphous, and mineral soil carbon to a depth of 5.4 meters. The information for this
parameterization is taken from the Alaska Soil Carbon Database (Johnson et al., 2011).
Currently, the parameterization and recalibration is in progress. After recalibration, the model
will be used at the plot scale to evaluate how the vegetation and soils change in tundra
simulations that include fire. Once this evaluation is complete, we will incorporate the burn
severity and tree establishment information from ALFRESCO, and perform simulations over the
study region with DOSTEM.
Two graduate students have been working with the treeline and tundra fire subgroup this
past year. As part of her PhD studies, graduate student Rebecca Hewitt is performing field
investigations of post-fire ectomycorrhizal limitation of seedling establishment across the boreal
forest-tundra boundary. She has also been working in the Tundra Fire and Treeline subgroup to
incorporate the knowledge from her field-based studies into ALFRESCO. Graduate student and
paleoecologist Ryan Kelly has been visiting UAF from the Univesity of Illinois during the fall
2012 semester. He has been interacting with the Tundra Fire and Treeline subgroup as he learns
how to use DOSTEM to model the effects of fire on carbon storage in Alaska during the
Holocene.
3.4. Thermokarst and Wetland Dynamics
The thermokarst and wetlands dynamics subgroups of the IEM team are working on
related issues. The thermokarst dynamics subgroup is charged with developing a model that will
simulate that timing and transitions of thermokarst disturbance across the landscape, which is
important for representing the spatial dynamics of wetlands across the landscape. The wetlands
dynamics subgroup is charged with (1) the development of a wetlands version of TEM for
incorporation into the AIEM and (2) providing information for designing, parameterizing, and
8
testing of a wetlands version of TEM. The wetlands version of TEM will simulate the internal
biogeochemical and vegetation dynamics of a wetland, but will not simulate changes in the area
of wetlands in response to climate change. Some initial progress on the development of a
wetlands version of TEM has already occurred with the development of peatland-DOSTEM (Fan
et al. 2013). The wetlands dynamics subgroup is also conducting field studies for providing
information for designing, parameterizing, and testing the wetlands version of TEM. Thus, the
efforts of the thermokarst and the wetlands dynamics subgroups are complementary as the groups
work to better represent the dynamics of wetland ecosystems in the IEM.
3.4.1. Thermokarst Dynamics
One of the priority issues for the development of the IEM was the need for the IEM to
consider landscape-level thermokarst dynamics. Landscape-level thermokarst changes are
important to incorporate into the IEM because subsidence associated with the melting of
previously frozen water in ice-rich permafrost can result in substantial changes in vegetation and
habitat (e.g., turning a graminoid tundra ecosystem into a wetland tundra ecosystem). The
representation of thermokarst disturbance in the IEM is also important for representing wetland
dynamics in the IEM, which are important to model as much of Alaska is covered by wetland
complexes, and changes in wetland structure and function has the potential to affect numerous
animal species that use wetlands (e.g., waterfowl).
A major goal of the thermokarst/wetland subgroup in 2012 was to develop a conceptual
design for a thermokarst disturbance model that would ultimately be incorporated into the IEM
alongside ALFRESCO, TEM, and GIPL. This conceptual design was to be provided in the form
of a white paper that would provide the vision for the development of this thermokarst
disturbance model. Please see Supplement 1 for this white paper.
3.4.2. Wetland Dynamics
3.4.2.1. Progress on Development of Wetlands Version of TEM
We made some initial progress in the development of a wetlands version of TEM through
the development of peatland DOSTEM (Figure 7). Our goal in developing the model was to
synthesize the results of a field water table manipulation experiment, APEX, conducted in a
boreal rich fen into a process-based model to understand how soil organic carbon and soil CO2
and CH4 fluxes might respond to projected climate change. Our approach was to calibrate the
model based on data from the control treatment of the manipulation experiment, and to validate
the model based on the data from two experimental treatments, including raised and lowered
water table manipulations. Because we measured gas fluxes for several years, we also captured a
large amount of interannual variation across all treatments, including a dry year and a 100-year
flood for the region. We conducted a scaling study to evaluate whether the model could be driven
by monthly instead of daily input climate data. The model was then used to simulate soil organic
carbon (SOC) dynamics (i.e., C inputs into the soil, CO2 and CH4 exchange with the atmosphere,
and changes in soil C stocks) of the control treatment under various CO2 emission scenarios
(high, midrange, and low emissions). We analyzed the results of these simulations to evaluate
how increases in atmospheric CO2, warming, and changes in precipitation influence SOC
dynamics in the rich fen. The full description of peatland DOSTEM and its application in this
study can be found in Fan et al. (2013). We are working in collaboration with Dr. Fan to further
develop peatland DOSTEM so that it can represent the biogeochemical and vegetation dynamics
after the transition of a forested peatland plateau to a collapse-scar bog located at the APEX study
site. This is the site that is being studied by the field component of the wetlands dynamics
subgroup.
3.4.2.2. Progress of the Field Component of the Wetlands Dynamics Subgroup
The goal of the field component of the wetland dynamics subgroup of the IEM project is
to conduct field studies that support the parameterization, validation, and verification of IEM
9
related to the effects of permafrost thaw and wetland development on vegetation composition and
productivity, net change in ecosystem C balance, CH4 fluxes from wetlands, and the controls on
these processes at APEX.
In 2010 we installed eddy covariance flux towers in the black spruce permafrost plateau
and the adjacent thermokarst bog to measure Net Ecosystem Exchange of CO2 (NEE), water and
energy fluxes, and a number of micrometeorological variables. These are managed by Eugénie
Euskirchen, faculty at UAF. In 2011 we installed two sets of 8 autochambers in the black spruce
and thermokarst sites that allowed us to measure Ecosystem Respiration (ER), NEE, and Gross
Primary Productivity (GPP) at a smaller scale and to isolate the effect of the forest understory
(see Figure 8). Autochambers were also made at different heights to increase atmospheric
temperatures and examine the temperature responses of different carbon flux pathways. In
addition, in 2011 we installed an eddy covariance flux tower at a moderately rich fen with no
near-surface permafrost.
In 2012 we began to focus on methane (CH4), soil redox chemistry, nitrogen availability
and mineralization, and decadal scale change in soil C balance. We installed a Picarro laser
spectrometer inline with the autochambers to measure CH4 fluxes and 13C-CH4 from permafrost
bogs. In September of 2012 we received a Licor open path CH4 analyzer to measure CH4 fluxes at
the thermokarst eddy covariance flux tower, and this was installed in October 2012. We installed
soil temperature and moisture probes at each of the chamber locations, and described the
vegetation within each flux collar. At the black spruce and thermokarst sites, we installed sap flux
sensors to better understand the relation between transpiration and photosynthesis. We also
collected 14C of ecosystem respiration to partition respiration between autotrophic and
heterotrophic sources. Affiliated scientists measured CH4 and CO2 ebullition fluxes. In 2012 we
collected soil cores from the permafrost thaw gradient to quantify the loss of forest soil carbon
and the accretion of sphagnum bog carbon in the decades following thaw. Recent studies have
shown that some of the greatest uncertainties surrounding loss of forest soil carbon occur in the
few decades following thermokarst development. In support of this line of research, we have been
conducting lab incubations and analysis of soil redox chemistry to understand the controls on
anaerobic decomposition of forest floor and permafrost carbon. In June 2012 we also installed sap
flux sensors on 10 trees in each of the black spruce and thermokarst bog ecosystems. The sap
flux measurements will be used to derive the tree GPP, which will be used to understand the
differences between the eddy covariance derived estimates of GPP (which take into account the
whole tree GPP) and the autochamber measurements of GPP (which do not take into account the
whole tree GPP).
Results from two years of flux data collection show that year to year variability in
ecosystem C fluxes can be large (Figure 9). In 2011 the bog was a net source of C to the
atmosphere of about 150 gC/m2, and in 2012 the bog was a net sink of about 70g C/m2 (Figure
9a). This was confirmed by analysis of autochamber fluxes (Figure 10). In 2012, the estimated
day of leaf-out occurred on May 10, while in 2011, the timing of leaf-out was one week later, on
May 17. This difference in leaf-out between the two years may partially explain why the timing
between the ecosystems switching to act as a source versus a sink of CO2 in the spring was earlier
in 2012 compared to 2011 at both the fen and thermokarst bog.
The black spruce ecosystem showed little interannual variability, acting as a sink of C of
~300 g C/m2 during both years. Autochamber fluxes and tower flux estimates correlated
reasonably well, but autochamber fluxes dramatically overestimate NEE in the black spruce
stand, because GPP from black spruce trees are not included in the autochamber NEE estimate.
Comparing tower and autochamber fluxes leads us to conclude that black spruce GPP may be up
to 400 g C/m2, a not unreasonable estimate. However, comparison of autochamber and tower
fluxes also leads us to conclude that understory GPP may be up to 50% of black spruce GPP. In
contrast to the bog, the boreal fen remained approximately carbon neutral during both years of
10
measurement, and the range of NEE was much lower than either the bog or black spruce
ecosystem.
Methane fluxes were estimated from the autochamber system throughout the summer. A
considerable amount of methods development was required, but it resulted in near continuous
monitoring of diurnal CH4 fluxes during the growing season. CH4 fluxes were remarkably stable
over the growing season and displayed little variation with atmospheric temperature or time of
day. Instead, fluxes were most affected by which chamber they were measured from, and by
implication they are likely influenced by plant community composition. Preliminary analysis
shows that the highest CH4 fluxes occurred in chambers that contained a large proportion of
Carex aquatalis, a common sedge with aerenchyma that can facilitate the flux of CH4 from deep
sources, and circumvent surface CH4 oxidation. Mean CH4 flux for the period of June 15 to
September 15 was 1.3 mg C-CH4/m2/d, a low estimate of CH4 flux for boreal bogs.
4. Products
4.1. Data Set Development
Input data sets at monthly time steps covering the full spatial, temporal, and modeling
bounds of the study were produced for the historical period (1901-2009, downscaled from CRU
TS 3.1) and the projected period (2010-2100, downscaled from cccma_cgcm31, mpi_echam5,
A1B Scenario) for climatic variables (temperature, precipitation, radiation, vapor pressure) and
static variables (initial land cover, elevation, aspect, slope). These data can be downloaded from
the following web site: http://www.snap.uaf.edu/data.php.
4.2. Model Coupling
A number of infrastructure pieces have been put in place. Source code for all submodels
and the coupler are being maintained through a common portal under the SNAP repositories on
github.com. Tools for source code compilation are being hosted by SNAP for the subgroups and
allow for automated building of applications and libraries using the Jenkins Continuous
Integration environment as well as the generation and serving of documentation relating to the
models.
A new computing system has been procured as a dedicated resource for running large
scale simulations. This system is capable of running hundreds of processes simultaneously, and
has a large memory and storage infrastructure, which should allow for significant processing
using the IEM coupled model.
The coupling framework has been created, allowing the three submodels to run within the
same executable, to operate within synchronized time loops, and to read/write values from shared
memory. Data structures have been established to hold common data, and the source code for the
IEM coupler as well as each submodel is being hosted via github.com. Easily installable binary
executables are being produced for ALFRESCO, and this method will serve as a model for future
distribution.
To serve data products when simulations have completed, work has been done to stand
up a WMS server, capable of serving out geospatial data dynamically, allowing end users to view
the data easily through web interfaces.
4.3 Tundra Fire and Treeline Dynamics
4.3.1. Data Sets
Current data sets developed during the first year of the project include:
11
1. Spatial data from ALFRESCO simulations performed over the study region, to the year
2100, pertaining to fire occurrence, burn severity, and tree colonization and
establishment;
2. Information on the fibric, amorphous, and mineral soil carbon at depth for the heath,
shrub, graminoids, and wet sedge tundra ecosystems.
4.3.2. Publications
Alexeev, V., E.S. Euskirchen. Tundra burning in 2007: Did sea ice retreat matter? In
Review. Polar Geography.
4.3.3. Presentations
E.S. Euskirchen (presenter) and the Treeline and Tundra Fire Subgroup Participants.
Integrated Ecosystem Model for Alaska: Treeline and Tundra Fire Dynamics. Webinar.
Presented at the SNAP office to ~20 phone-in participants. April 2012.
4.4. Thermokarst and Wetland Dynamics
4.4.1 Thermokarst Dynamics
The thermokarst dynamics model white paper is available in Supplement 1. The thermokarst
dynamics subgroup also presented the following posters at the Fall 2012 AGU meeting:
Jafarov, E.E., H. Genet, V.E. Romanovsky, A.D. McGuire, and S.S. Marchenko. 2012. The
effects of forest fire on the frozen soil thermal state. Meeting of the American Geophysical
Union, San Francisco, California.
Zhang, Y. A.D. McGuire, H. Genet, W.R. Bolton, V.E. Romanovsky, G. Grosse, M.T.
Jorgenson. 2012. Modeling thermokarst dynamics in Alaska ecosystems. Meeting of the
American Geophysical Union, San Francisco, California.
4.4.2. Wetland Dynamics
4.4.2.1. Development of Wetlands Version of TEM
The initial progress in development of a wetlands version of TEM is represented by the
development of peatland DOSTEM and is described in the following publication:
Fan, Z., A.D. McGuire, M.R. Turetsky, J.W. Harden, J.M. Waddington, and E.S. Kane. 2013.
The response of soil organic carbon of a rich fen peatland in interior Alaska to projected climate
change. Global Change Biology. In press. doi:10.1111/gcb.12041.
4.4.2.1. Field Component of the Wetlands Subgroup
We have extensive flux datasets from the eddy covariance tower and flux autochambers
system. These datasets will be incorporated into the Bonanza Creek LTER’s online data
depository system. However, these datasets are still being QC’d and are not yet ready to be
made public. We have no publications to report. We have several presentations at the 2012 Fall
AGU Meeting this year concerning this research. These include:
Mark P. Waldrop; Jack McFarland; Eugenie S. Euskirchen; Merritt R. Turetsky; Jennifer W.
Harden; Kristen Manies; Miriam Jones; Anthony D. McGuire. Carbon Balance and
Greenhouse Gas Fluxes in a Thermokarst Bog in Interior Alaska: Positive and Negative
Feedbacks from Permafrost Thaw.
Mark P. Waldrop; Rachel Machelprang; Jenni Hultman; Kimberly P. Wickland. The tool of
microbial genomics research for interpreting the lability of permafrost carbon and
potential greenhouse gas feedbacks at different scales of resolution. (Invited)
Other presentations include:
Dustin Bronson, Xin Song, Michael Goulden, Kenneth L. Clark, Paul Bolstad, Tilden Meyers,
Jiquan Chen, Asko Noormets, Danilo Dragoni, David, Y. Hollinger, J. William Munger,
Stephen Wofsy, Timothy A. Martin, Russell K. Monson, Dennis D. BaldocchiAnkur R.
12
Desai, Eugenie EuskirchenWilliam J. Massman, Brent Helliker. Forest canopy
temperature: A comparison between an isotopic approach, and photosynthesis-weighted
air temperature. Ecological Society of America Annual Meeting, Portland, Oregon,
August 2012.
5. Outreach Activities
5.1. General Outreach Activities
5.1.1. Oral presentations
Breen, A. L., T. S. Rupp, D. McGuire, V. Romanovsky, E. Euskirchen & S. Marchenko. An
Integrated Ecosystem Model for Alaska. Alaska Center for Climate Assessment & Policy
Climate Webinar Series in Fairbanks, Alaska. April 2012.
Breen, A. L., T. S. Rupp, D. McGuire, V. Romanovsky, E. Euskirchen & S. Marchenko. The
Alaska Integrated Ecosystem Model. Alaska Cooperative Fish & Wildlife Research Unit
Annual Review. Fairbanks, Alaska. March 2012.
Breen, A. L., T. S. Rupp, D. McGuire, V. Romanovsky, E. Euskirchen & S. Marchenko. Alaska
Integrated Ecosystem Model: the pilot year. Webinar presentation to the steering committee of
the Arctic Landscape Conservation Cooperative in Fairbanks, Alaska. January 2012.
5.2.2. Other Outreach Activities
The Alaska IEM outreach activities are designed to advance understanding of the
modeling and study of ecosystems in Alaska. We accomplish this task through multiple efforts
including the integration of research and education, establishment of collaborations with federal
agencies, and dissemination of data and products to the public. We will next elaborate on each of
these efforts in turn below.
The integration of research and education is accomplished through training of graduate
and postdoctoral students. The Alaska IEM project supports one graduate and five postdoctoral
scientists. These students and scientists are mentored by project PIs and participate in monthly
project meetings. The graduate student, Elchin Jafarov, successfully defended his PhD in late
November. His dissertation is titled, “The effects of changes in climate and other environmental
factors on permafrost evolution.” In addition, Elchin Jafarov and two of the postdoctoral
scientists funded by the project present their work at the American Geophysical Union’s meeting
in December 2012.
To integrate our research into broader programs and activities of national interest, we
collaborated with the Department of the Interior’s Landscape Conservation Cooperatives (LCCs)
in Alaska. To date, these include the Arctic, Western Alaska and Northwest Boreal LCCs. This
collaboration facilitates the synthesis of our research and results in formats understandable and
useful for state and federal agencies, resource managers, local communities, and other interested
members of the public. In 2012, we worked with these partners to create a project fact sheet and
are currently working together to create a list of FAQs and project deliverables to share with state
and federal agencies and the broader public. We are also working together to identify potential
groups with interests in developing resource-specific impact models that could be driven by
outputs from the larger Alaska IEM.
To disseminate our research to the public, we gave several presentations throughout the
year and are working toward making data generated by the project available online via the SNAP
website (http://www.snap.uaf.edu/data.php). The venue for presentations varied from the Alaska
Cooperative Fish & Wildlife Research Unit’s Annual Review in March 2012 to the Murie
Science and Learning Center’s Evening Speaker Series in June 2012. We had a large presence at
the American Geophysical Union’s annual meeting in December 2012 as we presented seven
13
posters and two talks related to the Alaska IEM project. A project-wide poster was presented at
the LCC poster session of the meeting.
5.2. Outreach Activities of the Wetland Dynamics Group
These field data being collected by the wetlands dynamics group are being used by
several modeling groups to formulate and verify models of wetland biogeochemistry in a
changing climate. First, the Yukon Flats data from 2011 is currently being incorporated into
TEM, with the aim of at understanding the effects of lake change on terrestrial biogeochemistry
and vegetation succession. Second, the IEM group will be incorporating the data and
understanding from the APEX project into their models of CO2 and CH4 fluxes from boreal
wetlands, and in relation to permafrost thaw.
E. Euskirchen has been working with Dr. Dustin Bronson and Dr. Brent Helliker at the
University of Pennsylvania on a project investigating the effects of canopy temperature on gross
primary productivity, providing them with the data from the black spruce site and installing a
canopy temperature sensor in the black spruce as part of a cross-site analysis undertaken by the
University of Pennsylvania Group. This has resulted in one presentation at the Ecological Society
of America annual meeting in 2012.
Our APEX site is now an affiliated project with the North American Carbon Program, the
central objective of which is to understand the sources and sinks of CO2 and CH4 from
ecosystems of North America. This work is being conducted in collaboration with science staff
and graduate students at UAF, U. Guelph, UC Irvine, USFS, and USGS through funding from the
YRB project, NRP, and the USGS FOCAL (FOCCSY) project.
6. Research Plans for 2013
The data subgroup made great progress in 2012 in terms of producing 1 km driver data
for the IEM framework over the entire IEM domain for the historical period and for the AR4
generation climate change scenarios. The major effort of the data subgroup in 2013 will be to
produce 1 km driver data for the IEM framework over the entire IEM domain using the IPCC
Fifth Assessment Report (AR5) models across representative concentration pathways (RCPs).
During 2012, the model coupling subgroup made a lot of progress in synchronously coupling the
component models of the IEM framework. There is still code to be written and tested for
completing the task of having the synchronously coupled framework operational. We expect this
work to be completed in the first half of 2013. During 2012, the tundra fire and treeline subgroup
made substantial progress on both ALFRESCO and TEM related issues. There is still some
calibration work to do in both models, and we found that some of the calibration issues are related
to the new 1 km historical data sets. We are currently addressing these calibration issues and
expect to have completed a first set of production runs during summer 2013. We anticipate
completing our proof of concept study using the synchronously coupled AIEM by the end of
2013. Now that the thermokarst dynamics subgroup has developed a conceptual approach
towards modeling thermokarst disturbance and associated landscape transitions, we will be
developing that model in 2013. We plan to focus on model development, testing, and analysis in
the Tanana Flats and in the Barrow areas with respect to the boreal forest and tundra landscape
unit thermokarst transition sets, respectively. By the end of 2013 we anticipate having a
thermokarst disturbance and transition model that can be incorporated into the IEM framework
during 2014. In 2012, the wetlands dynamics subgroup made substantial progress in developing a
peatland version of TEM that could represent water table dynamics and biogeochemistry in an
open boreal fen. During 2013 we plan to extend this modeling effort by focusing model
development on an open boreal bog that has developed as a consequence of thermokarst
disturbance in a black spruce permafrost plateau forest. The parameterization and testing of this
14
model will rely substantially on the ongoing IEM wetland field studies being conducted at APEX
during 2013.
7. Outreach Plans for 2013
In 2012, we recognized the need to better engage management and other stakeholder
needs through additional outreach activities. We want to not only disseminate our research
objectives and results, but also the management implications and potential uses for IEM outputs.
To accomplish this task, we specifically created a position and hired a science
communicator, Kristin Timm, to take on this role in 2013. Kristin will participate in our monthly
meetings and facilitate discussion among researchers, the LCC Science Coordinators, and the
management community. She has already begun work to develop content for the IEM Project
page on the Alaska Climate Science Center’s website, including highlighting the IEM project on
their homepage with a short video interview with the project PIs. She will also create additional
factsheets on various project activities such as model development, simulating tundra fire and
vegetation dynamics, and thermokarst and wetland dynamics. So far, we have only introduced
and summarized these activities in our outreach efforts. This will give us an opportunity to delve
deeper into these activities and introduce more content as we and others move forward toward
developing resource-specific impact models that make use of IEM outputs. The content and
methods of IEM project outreach efforts will continue to be driven by manager and other
stakeholder needs, so that project results and implications are disseminated far beyond our
research group.
8. Acknowledgements
We thank Philip Martin and Jennifer Jenkins with the Arctic Landscape
Conservation Cooperative, Karen Murphy and Joel Reynolds with the Western Alaska
Landscape Conservation Cooperative, Amanda Robertson with the Northwest Boreal
Landscape Conservation Cooperative, Mark Shasby with the USGS Alaska Science
Center and Stephen Gray with the USGS Alaska Climate Science Center for constructive
comments and feedback during the past year.
15
9. References
Fan, Z., A.D. McGuire, M.R. Turetsky, J.W. Harden, J.M. Waddington, and E.S. Kane. 2013.
The response of soil organic carbon of a rich fen peatland in interior Alaska to projected
climate change. Global Change Biology. In press. doi:10.1111/gcb.12041.
Johnson, K.D., J. Harden, A.D. McGuire, N.B. Bliss, J.G. Bockheim, M. Clark, T. NettletonHollingsworth, M.T. Jorgenson, E.S. Kane, M. Mack, J. O’Donnell, C.-Lu Ping, E.A.G.
Schuur, M.R. Turetsky, and D.W. Valentine. 2011. Soil carbon distribution in Alaska in
relation to soil-forming factors. Geoderma 167-168:71-84.
16
Table 1. Timeline of activities for the project by year.
Year
Model Coupling
Data Sets
1
Synchronous coupling
and evaluation of AIEM
Prepare future
scenarios for
Western Arctic
Downscaled to
1 km resolution
2
Assessment of Changes
in Landscape Structure
and Function over the
Western Arctic
3
Assessment of Tundra
Fire and Treeline
Dynamics Responses
4
Assessment of
Thermokarst Dynamics
Responses over the
Western Arctic
5
Assessment of Wetland
Dynamics Responses
over the Western Arctic
Prepare
additional data
sets needed for
driving tundra
fire and
treeline
dynamics in
AIEM
Prepare
additional data
sets needed for
driving
thermokarst
dynamics
module in
AIEM
Prepare
additional data
sets needed for
driving
wetland
dynamics in
AIEM
Prepare
additional data
sets needed for
impact studies
Tundra Fire
and Treeline
Dynamics
Model testing
and evaluation
of tundra fire
and treeline
dynamics in
ALFRESCO
and TEM
Incorporation
of tundra fire
and treeline
dynamics into
the AIEM and
proof of
concept study
Thermokarst
Dynamics
Wetland
Dynamics
Conceptual
development of
approach to
representing
thermokarst
dynamics at a
landscape scale
Model
development
and testing of
landscape-scale
thermokarst
dynamics
module
Design wetland
dynamics field
program to
support AIEM
Work on
assessment of
tundra fire and
treeline
dynamics
Incorporation
of thermokarst
dynamics
module into the
AIEM and
proof of
concept study
Model testing
and evaluation
of wetland
dynamics
component of
TEM
Work on
resource
impact studies
involving
tundra fire and
treeline
dynamics
Work on
impact studies
involving
tundra fire and
treeline
dynamics
Work on
assessment of
thermokarst
dynamics
Incorporation
of wetland
dynamics into
the AIEM and
proof of
concept study
Work on
resource
impact studies
involving
thermokarst
dynamics
Work on
assessment of
wetland
dynamics
17
Initial
development of
wetland
dynamics
component of
TEM
Figure 1. Modeling framework for the synchronous coupling among ALFRESCO, TEM and
GIPL-1 in the Integrated Ecosystem Model for Alaska and Northwest Canada.
18
Figure 2. The spatial domain of the Integrated Ecosystem Model for Alaska and Northwest
Canada.
19
Figure 3. The Loose Coupling approach to synchronous coupling to allow each model
(ALFRESCO, TEM, and GIPL) in the Integrated Ecosystem Model for Alaska and Northwest
Canada to be maintained independently.
20
Figure 4. The initial land cover input data set for the Integrated Ecosystem Model for Alaska
and Northwest Canada.
21
Figure 5. Conceptual diagram of the ALFRESCO model. Since the model has been primarily
used in recent years to model boreal forest dynamics, as part of the treeline and tundra fire
dynamics effort, ALFRESCO has been modified to include shrub and graminoid tundra.
Through colonization, shrub or graminoid tundra may become white spruce, while following the
occurrence of tundra fire, the shrub tundra may revert to graminoid tundra.
22
(a)
(b)
Figure 6. In (a), ALFRESCO output showing tundra converted to spruce
forest. Simulated colonization of tundra by spruce is driven by historical (1900-2008) and
projected (2009-2100) climate scenarios. The colors indicate areas most likely to convert
to spruce forest at different time steps (blue=1900, pink=1950, orange=2000, yellow=2050
and green=2100). In (b), the colored areas show where white spruce has established in a
given grid cell by 2100, although the entire grid cell has not yet fully converted to white
spruce. The model was driven by air temperature and precipitation from the CCCMA
general circulation model with the A1B scenario.
23
Figure 7. Schematic of the peatland organic carbon module in peatland DOS-TEM.
24
Figure 8. Autochambers in the thermokarst site to measure carbon fluxes at smaller scales than
the eddy covariance tower at the site.
25
-2
Net Ecosystem Exchange (g C m )
(a
)
Bog 2011
Bog 2012
Spruce 2011
Spruce 2012
Feb
Mar
Apr
May
Jun Jul
Aug
Sep
Oct
-2
Net Ecosystem Exchange (g C m )
(b)
Fen 2011
Fen 2012
May
Jun
Jul
Aug
Sep
Oct
Figure 9. Cumulative net ecosystem exchange at the thermokarst bog and black spruce
ecosystems from February – October during 2011 and 2012 (a), and at the fen from May –
October during 2011 and 2012. A positive value of NEE denotes a source of C and a
negative value denotes a sink.
26
Figure 10. Net Ecosystem Exchange (NEE), Ecosystem Respiration (ER), and
Gross Primary Productivity (GPP) calculated from autochambers in Black Spruce
and Thermokarst Bog ecosystems. ‘Large’ and ‘small’ indicate the size of the
chamber. Chamber size was modified to try and increase atmospheric temperatures
in the chambers, but the effect was negligible (0.3 C).
27