Modeling Impacts of Climate Change on Carbon

1
Changsheng Li1, Steve Frolking1, Xiangming Xiao2,
Carl Trettin3 and William Salas4 1
Institute for the Study of Earth, Oceans, and
Space, University of New Hampshire, Durham NH.
2 Center for Spatial Analysis, University of
Oklahoma, Norman OK. 3 Center for Forested
Wetlands Research, USDA Forest Service,
Cordesville, SC. 4 Applied Geosolutions,
LLC, Durham, NH.
Modeling Impacts of Climate Change on Carbon
Dynamics in Northern High-Latitude Wetlands
Modeling framework
1. DNDC predicts C and N transport and
transformation by converting primary drivers
(e.g., climate, topography, vegetation, soil, and
anthropogenic activity) into environmental
factors (e.g., temperature, moisture, pH, redox
potential and substrate concentration gradients),
which then determine the rates biochemical and
geochemical reactions. DNDC possesses a
relatively complete set of biogeochemical
processes for simulating C and N biogeochemistry
by tracking vegetation biomass, soil climate, and
soil microbial activities (e.g., decomposition,
nitrification, denitrification, fermentation).
DNDC has been linked to regional datasets
generated from statistics or remote sensing
analysis for North America, Asia, Europe and
Oceania for quantifying impacts of changes in
climate or management on C sequestration and
trace gas emissions at the regional scales 2.
NEST is a process-based model for Northern
Ecosystem Soil Temperature, developed to quantify
the transient response of ground thermal regimes
to climate change (Zhang et al., 2003). The model
explicitly considered the effects of vegetation,
snow, and different ground conditions, and
integrates climate, vegetation, snow, water, and
ground temperature based on energy and water
exchange. 3. ARC-VEG is a nutrient-based, plant
community and ecosystem model designed to
simulate the transient effects of temperature
change on the biomass and community composition
of a variety of arctic ecosystems (Epstein et
al., 2000). The model is currently parameterized
for upland, mesic ecosystems in high arctic, low
arctic, treeline, and boreal forest climate
zones.  
Preliminary DNDC Model Results
Field data and Wetland-DNDC model output (upper
left) inter-annual NEE variation in a wetland
spruce forest at Tver, (upper right) seasonal
variation of NEE fluxes for a SSA fen at
Saskachewan, Canada, (lower left) water table
fluctuation at a cypress wetland at Gator,
Florida, Russia, and (lower right) CH4 emissions
from a fen at Marcell, Minnesota.
Remote Sensing Analyses
larch forest, Hokkaido, Japan
Vegetation Photosynthesis Model (VPM) We will
also use the VPM model to estimate gross primary
production (GPP) of terrestrial ecosystems. The
VPM is based on the concept of light absorption
by chlorophyll and estimates daily GPP, using EVI
and LSWI , temperature and PAR data as model
inputs. Figure above shows a case study for
deciduous needeleaf forest (TMK) in Japan.
OPTICAL/NIR - We use MODIS time series data to
identify land surface phenology for individual
sites, including the starting and ending dates of
plant growing season in a year. NDVI Normalized
Difference Vegetation Index, EVI Enhanced
Vegetation Index LSWI Land Surface Water Index
MICROWAVE - Poyang Lake Wetland Example The
image below is a multi-temporal PALSAR image of
Poyang Lake in China. From this image one can see
seasonal flooding dynamics of natural wetland
(light green hues) and anthropogenic wetlands
(rice paddies in red). Flooding will be mapped
based on the enhanced backscatter periods due to
double bounce between water and vegetation. We
are creating maps of hydro-period for natural
wetland systems for each study site using a time
series of PALSAR Fine Beam (FBS, FBD, PLR).
MICROWAVE SAR Mapping of Study Site Wetlands
L-band SAR systems are the single best option for
fine spatial resolution remote sensing of wetland
extent and characteristics over large regions
because they operate regardless of cloud cover,
and can distinguish basic vegetation structure.
A dual-polarization L-band system such as ALOS
PALSAR will provide improved accuracy in
discriminating between rough water surfaces and
bare ground, and improved mapping of vegetation
structural characteristics. Our pre-processing of
SAR data will include incidence angle
normalization and multi-temporal filtering (see
figure above for PALSAR processing flow chart).