1 Characterization of Land Surface Freeze/Thaw State, Temperature and Moisture Controls on Ecosystem Productivity: Carbon Cycle Science Addressed with NASA’s Proposed Soil Moisture Active/Passive (SMAP) Mission Kyle C. McDonald Department of Earth and Atmospheric Sciences The City College of New York, New York, NY, USA and Jet Propulsion Lab, California Institute of Technology Pasadena, California, USA John S. Kimball University of Montana Missoula, Montana, USA International Geoscience and Remote Sensing Symposium July 25-29, 2011, Vancouver, BC, Canada Portions of this work were carried out at the Jet Propulsion Laboratory, California Institute of Technology under contract to the National Aeronautics and Space Administration. This work has been undertaken in part within the framework of the JAXA ALOS Kyoto & Carbon Initiative. PALSAR data were provided by JAXA EORC.
Characterization of Land Surface Freeze/Thaw State, Temperature and Moisture Controls on Ecosystem Productivity: Carbon Cycle Science Addressed with NASA’s Proposed Soil Moisture Active/Passive (SMAP) Mission Kyle C. McDonald Department of Earth and Atmospheric Sciences - PowerPoint PPT Presentation
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
1
Characterization of Land Surface Freeze/Thaw State, Temperature and Moisture Controls on Ecosystem Productivity:
Kyle C. McDonaldDepartment of Earth and Atmospheric Sciences
The City College of New York, New York, NY, USAand
Jet Propulsion Lab, California Institute of TechnologyPasadena, California, USA
John S. KimballUniversity of Montana
Missoula, Montana, USA
International Geoscience and Remote Sensing SymposiumJuly 25-29, 2011, Vancouver, BC, Canada
Portions of this work were carried out at the Jet Propulsion Laboratory, California Institute of Technology under contract to the National Aeronautics and Space
Administration. This work has been undertaken in part within the framework of the JAXA ALOS Kyoto & Carbon Initiative. PALSAR data were provided by JAXA EORC.
2
SMAP Science ObjectivesSMAP Science Objectives
Primary Science Objectives:
• Global, high-resolution mapping of soil moisture and its freeze/thaw state to: Link terrestrial water, energy and carbon
cycle processes
Estimate global water and energy fluxes at the land surface
Quantify net carbon flux in boreal landscapes
Extend weather and climate forecast skill
Develop improved flood and drought prediction capability
Soil moisture and freeze/thaw state are primary surface controls on Evaporation and Net Primary Productivity
3
Conceptual relationship between landscape water content and associated environmental constraints to ecosystem processes including land-atmosphere carbon, water and energy exchange and vegetation productivity. The SMAP mission will provide a direct measure of changes in landscape water content and freeze/thaw status for monitoring terrestrial water mobility controls on ecosystem processes.
Terrestrial Water Mobility Constraints to Terrestrial Water Mobility Constraints to Ecosystem ProcessesEcosystem Processes
Landscape Water Content
Su
rfac
e R
esi
sta
nce
Thawed
Frozen
High
High
LowLow
Snow Accumulation
Increasing Biological Constraints
Freeze - Thawcycles
Landscape Water Content
Su
rfac
e R
esi
sta
nce
Thawed
Frozen
High
High
LowLow
Snow Accumulation
Increasing Biological Constraints
Freeze - Thawcycles
Freeze - Thawcycles
4
““Link Terrestrial Water, Energy and Carbon Link Terrestrial Water, Energy and Carbon Cycle Processes”Cycle Processes”
Do Climate Models Correctly Represent the Landsurface Control on Water and Energy Fluxes?
What Are the Regional Water Cycle Impacts of Climate Variability?
• L-band radiometer provides coarse-resolution (40 km) high absolute accuracy soil moisture measurements for climate modeling and prediction
SMAP Mission UniquenessSMAP Mission Uniqueness
SMAP is the first L-band combined active/passive mission providing both high-resolution and frequent revisit observations
• L-band radar provides high resolution (1-3 km) observations at spatial scales necessary to accurately measure freeze/thaw transitions in boreal landscapes
• Combined radar-radiometer soil moisture product at intermediate (10 km) resolution provides high resolution and high absolute accuracy for hydrometeorology and weather prediction
• Frequent global revisit (~3 days, 1-2 days for boreal regions) at high spatial resolution (1-10 km) enables several critical applications in water balance monitoring, basin-scale hydrologic prediction, flood monitoring and prediction, and human health
Comparison of SMAP coverage with other L-band missions
SMAP is the only microwave mission providing consistently high resolution and frequent revisits for the global land area
Range bars show the maximum and minimum parameters for the corresponding mission.
SAR missions do not allow for complete global coverage.
7
Normal to late thaw& Carbon Source
[1995, 1996, 1997]
Source: Goulden et al. Science, 279.
Early thaw & Carbon Sink
[1998]
Spring thaw dates5/7 5/27 5/26 4/22
Primary thaw dates
Ecological Significance of the F/T SignalEcological Significance of the F/T Signal
Seasonal frozen temperatures constrain vegetation growth and land-atmosphere CO2 exchange for ~52% (66 million km2) of the global land area. Spring thaw signal coincides with growing season initiation and influences land boreal source/sink strength for atmospheric CO2.
8
R2 = 0.745; P < 0.0001
y = -4.0926x
-80
-60
-40
-20
0
20
40
60
80
-15 -10 -5 0 5 10 15
Spring thaw anomaly (days)
NP
P a
no
ma
ly (
g C
m-2
yr-1
)
Mean annual variability in springtime thaw is on the order of ±7 days, with corresponding impacts to annual net primary productivity (NPP) of approximately ±1% per day.
Day of Primary Thaw
45N
75N
180W120W
150W
60N
Day of Primary Thaw
45N
75N
180W120W
150W
60N
Spring Thaw vs Northern Vegetation Productivity AnomaliesSpring Thaw vs Northern Vegetation Productivity Anomalies
Source: Kimball et al., Earth Interactions 10 (21)
AK Regional Correspondence Between SSM/I Thaw Date and Annual NPP
9
-8
-6
-4
-2
0
2
4
6
8
1988 1990 1992 1994 1996 1998 2000
Th
aw
An
om
aly
(d
ay
s)
-2.5
-2
-1.5
-1
-0.5
0
0.5
1
1.5
2
2.5
Atm
. CO
2 a
no
ma
ly (
pp
m)
Spring Thaw Timing (SSM/I) Max. Annual CO2 drawdown
Freeze/thaw link to carbon source-sink activity: Early thaw years enhance growing season uptake (drawdown) of atmospheric CO2 by NPP; Later thaw years reduce NPP and CO2 drawdown.
NOAA CMDL Observatory at Barrow
Julian Day
Mean Thaw Date (SSM/I, 1988-2001)
R = 0.63, p = 0.015
Spring Thaw Regulates Boreal-Arctic Spring Thaw Regulates Boreal-Arctic Sequestration of Atmospheric COSequestration of Atmospheric CO22
Earlier thaw & larger CO2 drawdown (- sign)Later thaw & smaller CO2 drawdown (+ sign)
Source: McDonald et al., Earth Interactions 8(20)
10
Define F/T Affected RegionsDefine F/T Affected Regions
FT Affected Regions Defined by Cold Temperature Constraints Index & long-term reanalysis (GMAO) data
FT domain: Vegetated areas where CCI ≥ 5 d yr-1
11
Microwave Remote Sensing for F/T DetectionMicrowave Remote Sensing for F/T Detection
C-bandC-band
12
Algorithm Parameterizations:
– Seasonal frozen and thawed reference states• Varies with topography and landcover
– Threshold reference (T)• Selected based on difference in seasonal
frozen and thawed states
Approach for Assignment of Parameters:- Seasonal frozen and thawed reference states may be initially assigned using prototype
SAR datasets and radar backscatter modeling over representative test sites. - Ancillary landcover and topography information may be used to interpolate reference
states across the product domain.- The threshold reference (T) depends on landcover and topography.
Setting initial algorithm parameters is a key application of the algorithm testbed. - Final parameterization will be performed using the SMAP L2 radar data as part of
-1 L-band SAR landscape freeze-thaw classification
Backscatter (dB)
< -2
-4
-6
-8
-10
-13
< -18
Frozen
Water
ClassifiedState
17 Feb. (Day 48) 1 April (Day 91) 3 April (Day 93)
JERS -1 L- -
Backscatter (dB)
< -2
-4
-6
-8
-10
-13
< -18
Frozen
Water
ClassifiedState
Thawed
SMAP Freeze/Thaw Algorithm
14
Source: Kim et al. 2010. Developing a global record of daily landscape freeze/thaw status using satellite passive microwave remote sensing. IEEE TGARS, DOI: 10.1109/TGRS.2010.2070515.
Seasonal Threshold Approach:
Annual Definition of SSM/I (37V GHz) Tb F/T Reference States
Frozen Non-Frozen
Pixel-wise Calibration using Tmx/Tmn from Global Reanalysis
6.9, 18.7 GHz); V (total col.); Global, daily coverage; Period of Record: 2002 – 2008. Product maturity: 3-7 (TRL) Available online (NSIDC & UMT) Reprocessing planned
20
Source: Kimball, J.S., L.A. Jones, et al., 2008. IEEE TGARS (in-press); 1Baldocchi, D., 2008. Aust. J. Botany 56, 1-26.
Satellite Mapping of Land-Atmosphere COSatellite Mapping of Land-Atmosphere CO22 Exchange using MODIS Exchange using MODIS
and AMSR-E: L4 Carbon Product Development for SMAPand AMSR-E: L4 Carbon Product Development for SMAP
• Application of MODIS - AMSR-E carbon model over boreal-Arctic tower sites indicates RMSE accuracies sufficient to determine NEE (net ecosystem exchange) to within ~31 g C m -2 yr-1, which is within 1estimated (30-100 gC m-2 yr-1) tower measurement accuracy.
• Sensitivity studies show SMAP will provide improved Ts and SM inputs, and resolve NEE to within ~13 g C m-2 over a ~100-day growing season.
Estimated Annual C Fluxes vs Site Ecosystem Model ResultsEstimated Annual C Fluxes vs Site Ecosystem Model Results
GPP (g C m- 2 yr- 1)
0
200
400
600
800
1000
1200
0 200 400 600 800 1000 1200
BI OME-BGC
MO
DIS
(M
OD
17A
2/3
)
I VO LTH OBS BRO OAS UPAD I ARC ATQ TLK
1:1RMSE = 25.3%MR = 7.1%
Rtot (g C m- 2 yr- 1)
0
200
400
600
800
1000
0 200 400 600 800 1000
BI OME-BGC
MO
DIS
-AM
SR-E
(C-M
odel
)
I VO LTH OBS BRO OAS UPAD I ARC ATQ TLK
1:1RMSE = 28.8%MR = 21.5%
• C-Model derived annual GPP and Rtot similar (RMSE<30%) to stand ecosystem process model results across latitudinal gradient of boreal-arctic tower sites.
• Uncertainty in residual NEE larger than component GPP/Rtot fluxes, especially for low productivity tundra sites.
Daily T and SM Time Series from AMSR-E and MERRADaily T and SM Time Series from AMSR-E and MERRA
WMO weather stations
USA Biophysical stations (SCAN, Ameriflux, …)
Source: Yi, Kimball, Jones, Reichle, McDonald, 2011. Journal of Climate
23
Prototype L4_C using MODIS-MERRA inputsPrototype L4_C using MODIS-MERRA inputs
Algorithm calibration and validation using FLUXNET tower CO2 (GPP, Reco, NEE) flux measurements across global range of land cover types.
L4_C and Tower Reco Comparison
FLUXNET Tower Eddy Covariance Measurement Network
24
Quantifying Land Source-Sink activity for CO2
Initial conditions (1ESRL)
Final optimized C-flux (1ESRL)
Initial conditions (L4_C)
Final optimized C-flux (L4_C)
1http://www.esrl.noaa.gov/gmd/ccgg/carbontracker
July 2003• The L4_C NEE (g C m-2 d-1) outputs provide initial conditions for 1CarbonTracker inversions of terrestrial CO2 source/sink activity;
• Differences in final optimized monthly C-fluxes relative to 1ESRL baseline are strongly dependent on these initial “first guess” C-fluxes (right);
• Atm. inversions provide additional verification of L4_C NEE against global flask network Obs. & other land models;
• Results link C source-sink activity to underlying vegetation productivity & moisture/temperature controls.
Soil Moisture Active and Passive (SMAP) Mission
26
Extra slides
27
Prototype L4_C Implementation using MODIS-MERRA inputsPrototype L4_C Implementation using MODIS-MERRA inputs
Latitudinal-zone average of NEE and GPP
Annual NEE was estimated at a 0.5 degree spatial resolution globally over a 7-year record using daily time series MERRA (SM, T) & MODIS (GPP) inputs. Estimated global carbon (NEE) source (+) & sink (-) variability is strongly affected by tropical (EBF) areas (above); large source activity in the tropics is driven by regional drought-induced GPP decline.
NEE (g C m-2 d-1)MODIS MOD17A2 Algorithm (Running et al. 2004)TCF Model (Kimball et al. 2008)SMAP L1/3 product streamsMicrowave RS based soil T (e.g. Jones et al. 07, Wigneron et al. 08)