Title: Exploring the role of global terrestrial ecosystems in the climate-carbon cycle interactions: An Integrated modeling and remote sensing approach Robert E. Dickinson (robted@eas.gatech.edu), Qing Liu, Yuhong Tian, Liming Zhou
1Exploring the role of global terrestrial
ecosystems in the climate-carbon cycle
interactions An Integrated modeling and remote
sensing approachRobert E. Dickinson
(robted_at_eas.gatech.edu), Qing Liu, Yuhong Tian,
Liming Zhou
?
School of Earth and Atmospheric Sciences, Georgia
Institute of Technology, 311 Ferst Drive NW,
Atlanta, GA 30332-0340
- KEY PAST PROGRESSES
- Some key advances were made in the following
areas in our past NASA projects in conjunction
with our proposed NASA work.
OVERVIEW OF SCIENTIFIC PLAN
- ABSTRACT
- Terrestrial ecosystems play important role in
the global climate system and carbon cycle.
Coupled climate-carbon models constrained by
remote sensing products provide important means
to address how the terrestrial ecosystem affects
and reacts to changing climate and global carbon
cycle. However, our current capability is still
limited because of inadequately developed
modeling and remote sensing data approaches. Our
current and past NASA projects have improved
satellite data representation of global
terrestrial properties and are developing new
model schemes for the merging of satellite data
and climate models. We propose to implement these
schemes to explore various terrestrial
biophysical coupling mechanisms that contribute
to the trajectory of atmospheric carbon, and to
develop a data assimilation system that allows
direct inference of vegetation dynamical
properties from remote sensed radiation fluxes in
weather and climate prediction. The proposed work
will be conducted using NASA MODIS products and
the modeling framework of the Community Climate
System Model (CCSM), and will significantly
improve the carbon and climate predictability of
the CCSM.
- Derived new land datasets from MODIS products for
use in climate models (1) a new land surface
dataset created from the latest MODIS high
quality products of LAI, vegetation continuous
field, IGBP land cover, and PFTs from the period
20002001 (2) a more realistic broadband
emissivity dataset created from MODIS thermal
infrared bands (3) a global monthly green
vegetation fraction dataset derived from MODIS
data and (4) an annual maximum FVC dataset
created from MODIS and AVHRR data. These datasets
significantly improved land surface climate and
energy balance simulations in the NCAR CLM2 and
Noah land model.
- Spatial pattern of percent cover difference
(new-old) in grass/ crop, tree, shrub, and bare
soil at the model spatial resolution as inferred
from MODIS land cover classification versus use
of AVHRR.
Objective 1 To improve and evaluate several
terrestrial processes in a coupled carbon climate
model that are important for determining future
concentrations of CO2 and CH4. Objective 2 To
develop and test an approach for the assimilation
of vegetation properties as a dynamic system in a
climate model from MODIS data.
- Spatial pattern of LAI difference between the
new and old land surface datasets (new-old as
derived respectively from MODIS and AVHRR land
products) in winter (DJF) and summer (JJA) .
- RESEARCH QUESTIONS TECHNICAL APPROACHES
- What determines the seasonal dynamics of leaf
level radiation fluxes, how can these be better
included in a climate model, and what are the
consequences of these fluxes for leaf level
carbon assimilation? - Approaches We continue to revise and test the
current CLM-DGVM (Dynamical Global Vegetation
Model) phenology scheme and the radiation scheme
for complex canopy. The phenology scheme will be
strongly constrained by MODIS LAI and EVI data
and validated using the MODIS derived Phenology
product. The canopy radiation scheme will be
evaluated against canopy light measurements from
BOREAS, LBA and the Safari 2000 MODIS validation
in southern Africa. We will also evaluate the
modeled sunlit leaf fraction against
climatological data of MODIS fPAR. - What determines soil moisture and water table
levels and consequently the distribution of
terrestrial seasonal and permanent wetlands?
Under what conditions will a wetland dry out?
What will make boreal and tropical peatland
vulnerable to climate change? - Approaches We are using a simple
groundwater model developed in our current IDS
work as an efficient approach to represent
groundwater dynamics and will extend it with an
inter-grid horizontal water transport for use in
GCMs. Methane emissions are to be estimated using
models by Walter et al. (2001). The simulated
changes in wetland areas and in the net fluxes of
carbon and methane over these regions will then
be assessed. - What climate processes condition rapid losses
from and transitions between different forms of
terrestrial biomass carbon stores? - Approaches We are reformulating modules in the
CLM-DGVM by including will adding the current
CLM-DGVM fire module fire probability based on
the concept of fuel beds related to various
carbon stores. We will also design some rules to
characterize fractions of various carbon stores
removed on average by occurrence of fire at model
time step based on analyses of climatological and
remote sensing indices. The MODIS daily fire
product at 1-km will be used to constrain the
model simulated timing and spatial distribution
of fires. - Can we develop a data simulation framework that
could derive MODIS LAI in a forecast mode,
prototyped by use of MODIS albedos and a dynamic
vegetation model with atmospheric forcing for the
period of MODIS data? - Approaches We will develop and test a canopy
radiation model to relate the vegetation
dynamical properties with the albedo more
realistically for complex canopies. A filter-type
assimilation technique will be used to perform
data-model fusion. A statistical dynamic
vegetation model will be used to validate the
physical dynamical vegetation model and to test
the developed data assimilation system. - Can we establish proper quantifications of data
and model statistics for applications in the
global terrestrial ecosystem in which remote
sensing data and dynamic vegetation models can be
optimally integrated, i.e., What statistics of
the albedo do we use as input? How can we compute
matching statistics from the dynamic model? What
are the statistical properties of the dynamic
model we need to have for the model-data fusion
to be successful? How can we quantify the albedo
statistics of both the dynamic model and the
MODIS data? - Approaches The land surface model uncertainty
is largely attributed to uncertainties of model
parameters, and hence will be quantified through
Monte Carlo-type approach. We will use MODIS
pixel albedos to obtain an optimal estimate of
model grid level albedos and their data errors.
The albedo scaling-up can be as simple as
averaging the pixel values for homogenous grids
and more complex for highly heterogeneous grids.
Statistically-based criteria will be used to
determine the heterogeneity of each grid cell
before computing the optimal grid estimate.
- Developed spatially and temporally more realistic
and accurate MODIS land albedo and LAI products
(1) improved MODIS albedo and LAI retrievals
under snowy and cloudy conditions (2) more
realistic LAI seasonality (3) established global
bare soil albedo date set for climate models. - Improved parameterizations of the terrestrial
processes (1) a new snow cover fraction scheme
(2) investigating canopy effects on snow surface
energy budget (3) integration of the MODIS
BRDF/albedo data and land modeling (4) land
surface turbulence, bifurcation, and soil
moisture memory (5) parameterization of canopy
hydrological processes.
- Developed new canopy radiation model with
consideration of the plant canopy 3-D effects We
found that the modeling of the canopy light
environment as implemented in current CLM is
seriously deficient in its partitioning of light
between vegetation canopies and underlying
surfaces due to its unrealistic assumption of
plane parallel canopy geometry, in particular for
the semiarid systems with sparse shrubs, and
northern forest with winter snow-pack. This
consequently leads to the underestimation of the
light loading on the canopy, and hence its carbon
assimilation. We have been developing a more
realistic three dimensional canopy radiation
model for climate modeling to describe the canopy
geometric (shadow) effect to improve the model
climate simulations and energy partitioning
between canopy and its underlying surface.
- Revealed the cause of the notable differences
between the CLM modeled and MODIS observed
radiation flux We found that the FPAR value is
mainly determined by LAI in MODIS and both LAI
and stem area index (SAI) in CLM. The positive
FPAR bias is mainly attributed to CLM SAI of
deciduous canopy and higher LAI than MODIS for
evergreen canopy as well. The negative FPAR bias
results from several factors, including
differences in LAI and soil albedo between CLM
and MODIS or limitations of the geometric optics
scheme used in the model.
Collaborators Rong Fu, Wenhong Li, Qing Liu
(Georgia Institute of Technology) Gordon B.
Bonan (NCAR) Ruth DeFries, John Townsend,
Eugenia Kalnay, Shunlin Liang, Chris Justice
(University of Maryland) Mark Friedl, Yuri
Knyazikhin, Ranga Myneni, Nikolay Shabanov,
Crystal Schaaf, Alan Strahler (Boston
University) Nancy Kiang, Randal Koster, Jeffrey
Privette (GSFC, NASA) Wenge Ni-Meister (Hunter
College) Xubin Zeng (Univeristy of Arizona)
Zong-Liang Yang, Guo-Yue Niu (University of Texas
at Austin) Hanqin Tian (Auburn University)
Yong-Jiu Dai (Beijing Normal University, China)
David Erickson, John Drake (Oak Ridge National
Laboratory) Xiaowen Li, Qinhou Liu (CAS
Institute of Remote Sensing, China)
- FPAR differences (FPARag0.0-FPARag0.23)
simulated by the CLM and the MODIS FPAR
algorithm with the canopy underlying soil albedo
changing from 0 to 0.23 at SZA 30.
Please visit http//climate.eas.gatech.edu/dickins
on for more project information and published
papers.