Large Area Land Cover Monitoring: Current Status

Large Area Land Cover Monitoring:
Current Status and New Opportunities
Jim Vogelmann
USGS EROS
ICESat-2 Vegetation Tutorial
Ann Arbor, Michigan
May 7-8, 2014
Two Large Area Mapping National
Efforts at USGS EROS
• National Land Cover Data Base (NLCD)
– Land cover data sets for the US based on Landsat data
– 30-meter resolution products
– Supports wide variety of Federal, State, local, and nongovernmental
applications
– Created by the Multi-Resolution Land Characteristics Consortium
(MRLC)
• LANDFIRE
– Landscape Fire and Resource Management Planning Tools
– Provides over 20 national geo-spatial layers at 30 m resolution (e.g.,
vegetation, fuel, disturbance, fire regime) for fire applications
– Sponsored by Wildland Fire Leadership Council
– Principal Partners are US DOI, USDA Forest Service and Nature
Conservancy (includes partnership with MRLC and Monitoring Trends in
Burn Severity (MTBS) project)
In addition to CONUS
NLCD available for
Alaska, Hawaii and
Puerto Rico
Close-up of NLCD data near Research Triangle Park, North Carolina
Versions and Epochs of Available NLCD Data
Next major remapping for NLCD planned for 2016
NLCD Land Cover of Ann Arbor Region
LANDFIRE
• Lots of emphasis on:
–
–
–
–
Mapping vegetation “Existing Vegetation Type”
Mapping disturbance (keeping the data current)
Modeling canopy structure attributes (height, biomass)
Translating vegetation information to fuel type
• Uses a large reference database containing
hundreds of thousands of field observations
• Focusses a lot on natural vegetation (i.e., areas
that burn)
• Uses NLCD as baseline information from which to
build
Vegetation Disturbance in the Pacific Northwest; Red disturbance (fire) from MTBS
Description
LF
National
LF 2001
LF 2008
LF 2010
LF 2012
Original
products
Refresh
update:
Existing
vegetation
type, cover,
height
Refresh
update:
Disturbances*
succession,
fire, and fuels
Update:
Vegetation,
disturbances
*succession,
fire, fuels,
and Islands~
Update:
Vegetation,
disturbances
*succession,
fire, fuels,
and Islands~
w/ Improvements
for Conus ^:
Completed
2009
2011
2011
2014
2015 Estimated
Imagery
Date
19992003
1999-2003
1999-2003
*1999-2008
1999-2003
*1999-2010
1999-2003
*1999-2012
Extent
Conus,
AK, HI
Conus^,
AK, HI
Conus, AK,
HI
Conus, AK,
Conus, AK,
HI & ~
HI & ~
Insular Areas Insular Areas
Products
All
All
All
Majority
Majority
Veg. Cond.
Class (VCC) –
Ref.
Conditions
LANDSUM
(conus), Veg.
Dynamic Dev.
Tool {VDDT}
(AK/HI)
VDDT
VDDT
VCC
Summary
Units
ECOMAP
subsections &
Map Zones
Hydrologic Unit
Codes (HUCs)
(- Fire Regime,
users apply field
guide book method
to develop Veg
Cond. Class
{VCC}/FRCC)
(- Fire Regime,
users apply field
guide book method
to develop Veg
Cond. Class
{VCC}/FRCC)
HUCs
Recent Outreach Activities of Note
•
•
LANDFIRE worked with Association of Fish and Wildlife agencies partnering
to develop a Webinar series (2013)
– http://www.teaming.com/swap-revision-guidance-best-practices
– http://www.youtube.com/watch?v=ZPs6JCkdFuU&feature=youtu.be
– http://www.landfire.gov/documents_presentations.php
Other presentations and applications of LANDFIRE available at:
– http://www.youtube.com/watch?v=UY2dfj4vJgM&list=PL18E089F3B460
8422
– http://www.youtube.com/user/landfirevideo
http://www.youtube.com/watch?v=qrKdM2
GzlFQ&list=PL18E089F3B4608422
LANDFIRE Remap
• Notional Schedule
– Start: April-May 2015 (planning and prototyping)
– Major execution anticipated in Feb-March 2016
– Finish: 2-3 years after start of major activities
• Scale and Scope
– All US lands at 30 m pixels
– Appropriate for National/Regional/Landscape analyses
• New Data
– Best available (primarily Landsat 8)
– More ground plots and polygons
• Expanded LANDFIRE Reference Data Base (LFRDB)
• Point data from NRI, BLM, BIA, FIA, data calls
• Polygons for areas of change/disturbance when available
Three Top Problem Areas
• Need to map the changes closer to when the
changes occur
• Need to map and characterize vegetation
condition better (e.g., areas of growth, or
decline, or drought….what are the units of
change?)
• Canopy structure (especially important to
LANDFIRE, but also relevant to other
communities)
– Better biomass information necessary
– Better canopy density information necessary
SWIR/NIR Trends (1986-2010) at 3 sites
210
SWIR/NIR (X 100)
190
Fire Site
Sagebrush
PJ
170
y = 0.093x - 48.82
R2 = 0.0151
150
130
110
y = 1.3371x - 2567
R2 = 0.6894
90
70
1985
1990
1995
2000
2005
2010
Year
Trends; Time A
Trends; Time B
Trends; Time C
Crater Lake; Synthetic Data; Summer 2012
Which areas are actively growing and declining? 2003-2012
LANDFIRE Canopy Structure
• Canopy height derived using life-form information with
field plot information (largely FIA) integrated with
Landsat data and topographic data and an SRTM
product (Kellndorfer et al., RSE 2004) with regression
tree modeling…..
• Lower 48: broad classes (5 tree classes)
• Alaska: 2 classes of height
• Results better than one might expect, but need
improving
From
http://www.whrc.org/mapping/nbcd/
SRTM mission: February 2000
Is LIDAR the obvious solution for
deriving canopy structure information?
• It has been demonstrated by many that Lidar
data can provide great canopy structure
information
– Height information
– Canopy density information
– Both height and canopy density information important
for multiple applications, including fire, habitat,
carbon…
LIDAR Data Accessible through Earth Explorer
Credit: Birgit Peterson, EROS Center
Does GLAS give us what we need?
•
•
•
•
Large footprints widely spaced
Saturation of waveform
Lots of issues in using the data, like slope, clouds
But lots of data and one can do some good things with them
Mean Canopy Height distribution
comparisons pre-Katrina versus
post-Katrina in Louisiana forests
by wind zone and season of data
collection. From Dolan et al.,
RSE 2011
Recommendations for Linkages
between ICESAT-2 and Landsat
• We need good canopy structure information
– The third dimension is an important “missing link” in the land cover
arena
– We need accurate canopy density, height and biomass information
– These data layers will need regular updating
• While “wall-to-wall” 3-d would be nice, we should
be able to develop good canopy structure data
sets using Landsat-derived land cover and
sample-based data from ICESat-2
– This assumes that ICESat 2 can provide the requisite canopy
structure information for a given sample
Possible Collaborative Activities
• Can be used to refine vegetation structure data layers
– Lots of LANDFIRE field reference information: Can we develop quantitative
biophysical relationships between information in the data base and ICESat-2?
– Once these relationships are developed, we should be able to link this
information with Landsat-derived land cover patterns with the ICESat-2 data to
improve structure mapping.
• Interpretation of Landsat-based cover change
information (especially condition variables)
– Land cover trend information derived from Landsat time series clearly depicts
spatial patterns of changes taking place, but quantitative information of the
changes/trends (e.g. biomass change) is lacking
– Information from ICESat-2 could provide meaning to these changes from a
biophysical standpoint.
– This information would be especially important to the modeling community, who
needs information in easily understood and recognized biological units.
Thank you!