Satellite Microwave Detection of Vegetation Phenology; Climate Constraints, Temporal Offsets
and Post-Fire Recovery as Compared to Optical-Infrared Phenology
Matthew O. Jones1,2, John S. Kimball1,2, Lucas A. Jones1,2, Kyle C. McDonald3,4
1The
3Jet
University of Montana Flathead Lake Biological Station, Polson, MT 2Numerical Terradynamic Simulation Group, The University of Montana, Missoula, MT
Propulsion Laboratory, California Institute of Technology, Pasadena, CA 4CUNY Environmental Crossroads Initiative and CREST Institute, City College of New York, New York, NY
Email Contact: matt.jones@ntsg.umt.edu
Websites: www.umt.edu/flbs & www.ntsg.umt.edu
Abstract
The Vegetation Optical Depth (VOD) parameter from satellite passive microwave remote sensing provides an alternative means for global phenology
monitoring that is sensitive to photosynthetic and non-photosynthetic vegetation canopy biomass and water content with minimal sensitivity to atmosphere
and solar illumination constraints. The VOD record from the Advanced Microwave Scanning Radiometer for EOS (AMSR-E) displayed North America
ecoregion start of season patterns and offsets (as compared to satellite optical-infrared remote sensing measures) that coincide with primary climate
constraints (temperature and water) to vegetation growth. The VOD start of season generally preceded optical-infrared NDVI and LAI greenup in cold
temperature constrained ecoregions and followed greenup in warmer, water limited ecoregions, with delays increasing for areas with greater woody
vegetation cover. The VOD also displayed alternate canopy recovery trajectories versus NDVI recovery following large scale fire disturbance in boreal
regions. VOD time series from the extreme 2004 fire year in Alaska and Canada showed an approximate two year recovery lag relative to NDVI; these results
are consistent with greater microwave sensitivity to woody biomass regeneration which is expected to show a slower response than canopy greenness
recovery. The AMSR-E VOD record provides new, independent phenological information for vegetation start of season measures as well as canopy post fire
recovery of both photosynthetic and non-photosynthetic biomass, complementing phenological information from NDVI and LAI measures. Synergistic
application of optical-infrared and microwave remote sensing data products are expanding the scope of observable vegetation phenology parameters and
advance regional disturbance, carbon, water and energy cycle studies.
This work was conducted at the University of Montana and Jet Propulsion Laboratory under contract to NASA (NNH07ZDA001N-TE).
VOD Start of Season Offset in relation to Climate Constraints on Net Primary Productivityb
AMSR-E Global Vegetation
Optical Depth (VOD) c,d
VOD Mean
2003-2010
VOD vs. NDVI Response to
Extreme 2004 Wildfires in
Alaska and Canadaa
The summer of 2004 was one of the hottest
and driest on record for interior Alaska
resulting in a panoply of large-scale fires
burning 2.7 million hectares of forest. MODIS
NDVI and AMSR-E VOD time series were
extracted over large scale fires (>1000km2)
and unburned (control) pixels. The VOD
record confirmed fairly rapid boreal
vegetation recovery consistent with previous
NDVI studies and showed an approximate 2
year recovery lag relative to NDVI. This lag is
consistent with greater VOD sensitivity to
photosynthetic and non-photosynthetic
woody biomass, which is expected to show a
slower response than canopy greenness
recovery alone.
VOD SOS earlier
VOD SOS later
No offset present.
VOD SOS offset by ecoregion
relative to NDVI Greenup date.
VOD SOS vs. MODIS-for-NACP NDVI Greenup Date by
ecoregion from 2003-07. Points are annotated by VOD
SOS offset matching regions on the accompanying map.
Daily 25 km resolution global EASE Grid brightness
temperatures from AMSR-E are used to derive daily
10.7 GHz frequency VOD retrievals over a global
domain. The VOD product integrates canopy
attenuation related to vegetation canopy biomass
and water content, while minimizing effects from subgrid scale open water variability and soil moisture.
The AMSR-E VOD and global land parameter database
is available online through the University of Montana
(http://freezethaw.ntsg.umt.edu) and the NASA
NSIDC DAAC (http://nsidc.org/data/nsidc-0451.html).
VOD and
NDVI Fire
Anomalies
Yearly summary
statistics of NDVI
and VOD fire
anomalies relative
to 2003(a) and
unburned control
pixels(b).
Fires. Background is IGBP MODIS land cover. White
box is zoomed region in yearly maximums figure.
VOD Yearly Maximums Relative to 2003
Ternary plot of relative
climate constraints on
vegetation NPP by
ecoregion; annotated
by VOD SOS offset.
Ecoregion level
climate constraints
on vegetation NPP
(Nemani et al.,
Science 2003)
Percent Tree Cover and SOS Offset
Differences in VOD and NDVI Post-fire Recovery and Tree Cover Loss
As an estimate of recovery, we determined post-fire years
(2005-2010) that were significantly (p<0.05) less than the prefire (2003) year. NDVI statistically recovered prior to VOD for
12 of the 14 fires. NDVI values returned to pre-fire levels
within 1 to 4 years (mean = 2.62 years) and 5 VOD values
returned to pre-fire levels within 3 to 5 years (mean = 4 years).
The ~2 year VOD recovery lag, relative to NDVI, is consistent
with greater VOD sensitivity to photosynthetic and nonphotosynthetic woody biomass and is related to percent tree
cover loss within fire perimeters. Plot (right) is normalized
post-fire percent tree cover loss (PTC2003 – PTC2005 / PTC2003)
versus the years to statistical recovery.
Summary & References
 VOD SOS Phenology offsets follow the distribution of low temperature and water constraints to NPP. VOD
provides an alternate land surface phenology data set that tracks seasonal changes in canopy water and biomass to
complement the greenness measures provided by optical-infrared data.
Tower Fluxes & VOD Start of Season
 VOD sensitivity to changes in canopy biomass provides a new methodology for tracking post-fire vegetation
recovery. The less steep (versus NDVI) post-fire trajectory of VOD tracks both photosynthetic and nonphotosynthetic vegetation biomass. Synergistic use of optical-infrared and microwave satellite data following
disturbance will advance our understanding of vegetation succession dynamics, aid in rehabilitation and restoration
efforts and inform carbon budget models.
aJones,
VOD SOS temporal offset vs. MODIS mean percent woody
tree cover by ecoregion. Ecoregions were classified as having
a dominant water constraint (>50%) for plant growth.
We used North America FLUXNET towers (n=33) from a range regional biomes. SOS
was derived from tower Gross Primary Productivity (GPP) and Ecosystem
Respiration (Reco) fluxes using TIMESAT and regressed against VOD ecoregion SOS.
Yearly VOD maximums were calculated relative to 2003
(YEARmax - 2003max). All fires within the region are included.
VOD pixels values show sensitivity to sub-pixel scale fires.
M.O., Kimball, J.S. & Jones, L.A. Satellite microwave detection of vegetation response to the extreme
2004 wildfires in Alaska and Canada, in prep, Global Change Biology.
bJones, M.O., Kimball, J.S, Jones, L.A., & McDonald, K.C. (2012). Satellite passive microwave detection of North
America start of season. Remote Sens. of Environ., 115
cJones, M.O., Jones, L.A., Kimball, J.S., McDonald, K.C. (2011). Satellite passive microwave remote sensing for
monitoring global land surface phenology. Remote Sens. of Environ, 123
dJones, L.A., Ferguson, C.R., Kimball, J.S., Zhang, K., Chan, S.K., McDonald, K.C., Njoku, E.G., & Wood, E.F. (2010).
Daily land surface air temperature minima and maxima from AMSRE IEEE J-STARS, 3, 111-123.
We thank the Ameriflux site principal investigators, R. Nemani, the Canadian National Fire Database and Alaska
Interagency Coordination Center for the data used in these studies.