Get documents about "PowerPoint Presentation - ORNL DAAC
Document Sample
Bridging the Gap in Scales between
Flux Towers, Ecosystem Models and Remote Sensing
R.J. Olson1, R.B. Cook1, L.M. Olsen1, T. A. Boden1, J.T. Morisette2, and S.W. Running3
1Oak Ridge National Laboratory, Environmental Sciences Division, Oak Ridge, TN 37831-6038 United States*
2Goddard Space Flight Center, NASA, Code 923, Greenbelt, MD 20771 United States
3University of Montana, School of Forestry, Missoula, MT 59812 United States
B52A-02
Abstract
Many Scales:
Combining ground-based studies, ecosystem models, and remote sensing data
provides a broad basis for understanding the dynamics of ecosystem-atmosphere CO2 Need a Scientific Strategy and
exchange. Data from a global network of flux tower sites (FLUXNET) are being Information Management
The Array of Scales Represented
compared to ecosystem model outputs and to remote sensing products, such as the Strategy to Bridge Scales
photosynthesis product derived from the MODIS sensor on the Terra satellite. in the Various Sources of Data
Comparison of information from multiple types of local and regional studies is needed
Source Flux In situ Remote MODIS Models
prior to extrapolating site studies to larger-scale products derived from remote sensing
Towers Measure- Sensing (ETM, Products
or global-scale modeling.
Scale ments IKONOS)
This poster describes data being compiled by the ORNL DAAC to conduct the local Parameter NEE, respiration LAI, NPP, Land cover, fPAR, Vi, LAI, NPP, NEP,
and regional scale validation of remote sensing products. The FLUXNET database respiration NDVI, etc. PSN, NPP LAI
contains carbon, water vapor, sensible heat, momentum, and radiation flux
Spatial Varying footprint Points 30 m and finer 1x1 km Point
measurements with associated ancillary and value-added data products for a wide
(~1 km2) (<1 m2 )
range of ecosystems on five continents. The MODIS land products are 1-km resolution
data in the immediate vicinity of the flux tower. Modeling groups are using the data to Temporal 0.5 hr Periodic, Periodic 8-day Daily
run terrestrial biosphere models for 17 flux tower sites to compare model outputs, flux Annual composite
measurements, field measurements, and MODIS products. AmeriFlux, the network of
flux towers in the Americas, provides model initialization and driver data for this
exercise. Email Contact Information: R. J. Olson <olsonrj@ornl.gov>
*Managed by the University of Tennessee-Battelle LLC under contract DE-AC05-00OR22725 with the U.S. Department of Energy
Data for a typical site - Walker Branch Watershed (WBW), Oak Ridge, TN
Flux Tower In situ Remote Sensing
Measurements
EOS Core Validation Sites
Photo of tower located at Walker Branch Ecosystem
Watershed. Vegetation consists of mixed
species of broad-leaved deciduous forest.
Measuring soil Respiration
on Walker Branch Watershed
IKONOS Image of Models
Walker Branch Watershed. CO2
Model of
C,N,P and H2O in N2 Rubisco Leaf
MODIS ISIN or UTM? To reproject or not to reproject…? terrestrial
Stomata
Fire
Allocatio
ISIN UTM systems n Wood N2O, N2
Biological N Decomposition Denitrification
Fixation
P required Soil water
Flux Tower Flux Tower Source: Bob Scholes, Soil
Leaching
location location July 2001, Amsterdam
Participating Modelers
MOD12Q1- Quarterly Land Cover Lotec: ORNL – King
MOD12Q1- Quarterly Land Cover Difficult to collocate Biome BGC: UMT – Running
MOD09A1 - Surface Reflectance field measurements within PnET: UNH - Aber
remote sensing pixels
Making MODIS Products Available: 7x7-km ASCII Subsets Centered on Towers
MODIS Products Fraction of Photosynthetically Leaf Area Index Enhanced Vegetation Daily Photosynthesis
8-day composites Active Radiation (MOD15A2) (MOD15A2) Index (MOD13A2) (MOD17A2)
FPAR LAI EVI PSN
Top panels show spatial
pattern of 1x1-km pixels in
7x7-km subset centered on Sept 30 Sept 30 Sept 30 Sept 30
July 14 July 14 July 14
flux tower. EVI is a 16-day May 25
July 14
May 25 May 25
May 25
composite product. April 7
low high April 7 low high April 7 low high April 7
low high
FPAR - water Index (MOD13A2)
water
l water
Lower panels show
temporal patterns of pixels
with good QA flags, red line
is pixel containing tower,
dots are pixels near tower
ASCII Subsets available at:
//public.ornl.gov/fluxnet/modis.cfm
Strategies for Bridging Scales to Compare Diverse Observations B52A-02
Scientific Strategy: Information Management
Collocate data in time Strategy: Collocate data at
and space 1. 26 Core Sites in 6 biomes, sites central site or with
with active research programs Validation Internet links
Individual Site Data
2. Flux towers using eddy “Assessing by
covariance methods to sample independent means CDIAC/AmeriFlux
• Model initialization Modeling Groups
large footprint the uncertainties of • Estimating NPP
parameters
3. Gap-filling of flux data to create data products” • Weekly micromet data
and NEP for flux
complete temporal record towers
MODIS • Initial 0. 5 hr flux data
4. BigFoot project to develop • 25 sites
spatial scaling methods • 8 products
DAAC/FLUXNET
5. Fine-scale remote sensing to
• MODIS products – 7x7 cutouts
scale-up point measurements
• FLUXNET gap-filled data
6. Models to estimate parameters
• Links to field data via Mercury
that are difficult to measure and Comparison
to bridge scales Workshops
Additional validation data
First Steps to Bridge the Gap in Scales, Illustrated with Walker Branch Watershed Data
Gap-filled Flux Tower MODIS PSN for 2001 (red) Model Outputs (ensemble Biome BGC Model NEP for MODIS PSN for 2001 and
NEE for 1996-1998 and NEE for 1996-1998 including Biome BGC) 2000 and 2001 and NEE Biome BGC Model NEP
versus Measured NPP for 1996-1998 Outputs for 2000 and 2001
½ hr NEE data gap- Years differ, MODIS 8-day Models tended to be
filled and summed composite (best day higher than data at Daily model outputs MODIS PSN is mean
to 8-day periods in period) multiplied low NPP sites, lower summed to of 3x3-km subset, BGC
by 8 at high NPP sites 8-day periods NEP is for tower site
Comparing MODIS LAI MODIS LAI (2001) – red line
product to LAI field
Data Resources for Many Applications measurements
- 8-day composites with high quality QA flags
- average of 1x1 km pixels near flux tower
Field LAI measurements (1996 – 1999)
Available for EOS Core Sites - relative LAI with max LAI=1.0
- measured at a point near tower
- Source: Paul Hanson, ORNL
Comparisons
EOS
e
- MODIS LAI shows similar greening pattern in
rv
er
B F a nc ese
Core
nt
Ce
rk Bra t R
R le sh
spring as field data; however, fall pattern
-P d st
ar l ls h
as S A
ar es A
Pa r as
Site
B vi l n di
g.
C EA NS
Ji l an re
H ad SS
M oy i e
M alg sk
W nia C
M co p i
e
A
to RT
sn rar
R
o
i ob
ke Co
d r
ho
R _
c _
d e
ow F
an a
rg A
rn a
on a
O S
on B
pa a
uk a
B / CA
Jo ran
shows earlier leaf-off
K aP
K ada
U jo s
Vi A B
H ard
Ta
B EA
Imagery
Ta uz
Sk llet
La ow
Se A
SA gu
U ry
B n
LS
z
a
ke
d
RM
v
i
SD
r
vi
on
ar
al
ar
ra
ar
O
Data
WBW is a deciduous forest with annual
A
MODIS
- it is not clear why the MODIS LAI shows
SeaWiFS leaf-on and leaf-off phenology significant drop around day 151
ETM+ 2 1 7 3 1 1 7 1 1 5 5 2 1 8 1 2 9 5 1 10 2 1 1 2 5
IKONOS 1 3 1 5 5 1 2 4 4 1 1 1 2 3 5 2 2 1 3 1 1 3 5
ASTER 1 7 4 4 3 12
Data Collection and Flux-Model-MODIS
GLCTS
TM 90's era
MQUALS
AVRIS
CRESS
GLCF
EO-1
Comparisons are Ongoing
MISR planned
EOS
Ongoing Concerns
e
rv
er
B F a nc ese
Core
•
nt
Ce
rk Bra t R
Formal Comparisons
R le sh
Testing and refining of the
-P d st
ar l ls h
as S A
ar es A
Pa r as
Site in
B vi l n di
g.
C EA NS
Ji l an re
H ad SS
M oy i e
M alg sk
W nia C
M co p i
e
A
to RT
sn rar
R
o
i ob
ke Co
d r
ho
R _
c _
d e
ow F
•
an a
rg A
rn a
on a
O S
on B
pa a
uk a
B / CA
Jo ran
K aP
K ada
U jo s
Vi A B
H ard
Ta
comparison strategies are
B EA
situ data
Ta uz
Sk llet
La ow
Se A
SA gu
U ry
B n
Ecosystem Model-Data
LS
z
a
ke
d
RM
v
i
SD
r
vi
on
ar
al
ar
ra
ar
O
activities
A
EOS Val
needed Intercomparison (EMDI 3)
AERONET P • Spatial differences - scaling - 12 Modeling Groups
FLUXNET 3 1 1 1 3 1 3 3 1 1 P 3 1 1 1 3 2
LAI Net
of points to cells, comparing - NPP, NEE flux, MODIS products
one pixel vs. subsets
BIGFOOT
Networks / E, E,
• Temporal differences – - Kathy Hibbard, <kathyh@eos.sr.unh.edu>
Field
Studies A B B
H,
G E
H, F,
L P H H S
E,
E G S L C
E,
G E E satellite 8-day vs. field - Santa Barbara, CA; April 21-24,
URLs
Model
Runs observation on a specific day 2002
for EOS Val - sites with study funded by the EOS Validation NRA
• Data inconsistencies - • MODIS Vegetation Workshop
Data
FLUXNET - number indicates multiple towers in 200x200 km EOS Core Site area
Field studies - A-ARM, B-BOREAS, C-CIGNS, F-FIFE, E-LTER, G-GTNET, H-Landcover helicopter, L- methods, variables, formats - Products: fPAR, LAI, VI, PSN
LBA, P-PROVE, S-SAVE/S2K
Real-Time Validation - inter-comparison of MODIS, eddy covariance data, field data, and model outputs
• Sites - 26 core and ~170 - Steve Running, <swr@ntsg.umt.edu>
towers site; need for a wider - Missoula, MT, July 16-18, 2002
variety of ecosystem types - http://www.forestry.umt.edu/ntsg/MODISMTG/
Data Type Data Center URL
Flux data FLUXNET www.daac.ornl.gov/FLUXNET/fluxnet.html
Summary
Model Driver and AmeriFlux public.ornl.gov/ameriflux/Analysis/Model_Evaluation/index.html
Initial Parameters • Strategies for supporting validation have been designed and reviewed -
EOS Core Sites NASA GSFC modarch.gsfc.nasa.gov/MODIS/LAND/VAL/ tested and refinements are ongoing
MODIS Subsets ORNL DAAC public.ornl.gov/fluxnet/modis.cfm
• Distribution systems are in place and working
• Validation data are available and more data are being added
MODIS Images EDC DAAC edcdaac.usgs.gov/modis/dataprod.html
• Plans are in place to hold workshops and communicate validation results
In Situ data Mercury System mercury.ornl.gov/ornldaac/
via special journal issues
