Assessing Land Cover Status and Change using Historical Landsat Data

1
Assessing Land Cover Status and Change using
Historical Landsat Data

• Jim Vogelmann USGS/EROS

2
To help set the stage.
USGS Science Strategy
U.S. Geological Survey, 2007, Facing tomorrows
challenges U.S. Geological Survey science in
the decade 20072017 U.S. Geological Survey
Circular 1309, x 70 p.
3
Six Science Directions

• Understanding Ecosystems and Predicting Ecosystem
  Change
• Climate Variability and Change
• Energy and Minerals for Americas Future
• A National Hazards, Risk, and Resilience
  Assessment Program
• The Role of Environment and Wildlife in Human
  Health
• A Water Census of the United States

Landsat plays a key role in these (especially
the first two)
4
Driving the need for Landsat

• Science understanding a changing planet
• Operational applications managing and
  monitoring resources for economic and
  environmental quality, public health and welfare,
  and national security
• Both require
• A global perspective
• A long-term record of observation
• Huge amounts of well-calibrated data

5
U.S. Landsat Archive Holdings(December 31, 2008)

• MSS Landsats 1-5
• 649,423 scenes
• 19TB Data
• TM Landsat 4 Landsat 5
• 734,627 scenes
• 368TB of Data
• Archive Grows by 40GB Daily
• ETM Landsat 7
• 786,700 scenes
• 730TB of Data
• Archive grows by 260GB Daily

6
This is a phenomenal amount of data! And it is
free!

• What CAN we to do with all of these data sets?
• And related to this.
• What SHOULD we be doing with all of these data
  sets?

7
Free data markedly changes how we go about our
land cover mapping and monitoring efforts
8
Some Case Studies How are we using Landsat
data? LANDFIRE Monitoring Trends in Burn
Severity Monitoring Trends in Vegetation Health
9
LANDFIRE in One Page

• Intended applications
• Fire hazards
• Fuel reduction
• Incident planning
• National strategic planning (e.g. FPA)
• Ecosystem restoration
• Other environmental/ resource management
  applications

• Objectives
• A national assessment of vegetation, fuel and
  ecosystem conditions
• Implementation of National wildland fire policies

• 24 primary data products, 30-meter nominal
  resolution nationwide
• Vegetation (potential and existing vegetation
  types and structure, succession classes)
• Fuels (surface and canopy)
• Fire regime conditions (reference conditions,
  landscape departure from reference conditions)

10
LANDFIRE Existing Vegetation Type (EVT)
Data(generated from multiple TM/ETM data sets)

• .

11
Close-up of the EVT for one zone

• a

12
LANDFIRE Updating Efforts

• A lot of concern was expressed (especially from
  southern US foresters) that the 2000 vintage
  LANDFIRE data were out of date even before they
  were developed
• Lots of changes due to timber operations
• Changes due to fire
• Started to prototype updating efforts in
  southeastern US
• Impetus for work was to keep data relevant

13
Remote Sensing of Landscape Change (LANDFIRE
updating)
Trends in Veg Health
14
Vegetation Change Tracker (VCT developed by C.
Huang at U of Maryland)
15
VCT Product with Causality Assigned (Okefenokee
Swamp)
16
Monitoring Trends in Burn Severity (MTBS)
17
The Monitoring Trends in Burn Severity (MTBS)
Project

• Collaborative effort between USGS/EROS and the
  Remote Sensing Application Center (RSAC) of the
  Forest Service.
• MTBS Project uses the Landsat archive to assess
  environmental impacts of wildland fires at a
  national level.
• Goal is to create a nationally consistent
  assessment of historical and current fires,
  including evaluation of burn severity of all
  large fires that have occurred in the US since
  1984
• http//mtbs.gov

18
12,932 fires (gt 1000 acres) recorded by MTBS from
1984-2007 (1.16 million ha burned)
19
The MTBS Project Burn Severity Products
Thematic Burn Severity Image Stats
Prefire
Postfire
Perimeter
RdNBR
dNBR
Metadata
20
Monitoring Trends in Vegetation Health
21
BackgroundSeveral different types of changes
that we can detect and monitor using remote
sensing
Abrupt changes
Gradual changes
22
Insects, Disease and Forest Health
WRS Path 34 Row 35 Landsat TM September 30, 2006
Santa Fe Natl Forest
Albuquerque
23
A tale of two pixels..
24
What we can do by analyzing Landsat spectral
trends Insect defoliation damage
B
A

• Trends of declining forest as measured
• by Landsat TM trend analysis (1995-2006).
• Red, blue and yellow indicate different rates
• of decline.

C
B. Trends of declining forest as measured
by Landsat TM trend analysis (1995-2006).
Red, blue and yellow indicate different rates of
decline.
C. Forest Health Monitoring Program
data (multiple defoliations caused by
western spruce budworm) as detected by
aerial sketch mapping (1998-2006)
25
Changes in plant vigor from mid 1990s to present
(on MODIS Image backdrop)
Red significant decrease in vigor Yellow
moderate decrease in vigor Blue increase In
vigor
26
Some Future Directions and Thoughts on
Operational Land Monitoring
27
Importance of Access to Ready-to-use Data

• Multiple sources of imagery
• Radiometrically corrected (to reflectance)
• Geometrically corrected
• Easy to access
• Same format
• Lots of ground information key to making this all
  work (need to link spectral data with meaningful
  field observations)
• Ground data are important for classification
  training
• Ground data are important for validation
• LANDFIRE reference data base might be a model for
  this
• Ancillary data (e.g. DEM, soils, climate, etc.)

28
Data Generation versus Data Analysis

• To date, we have largely been in data
  generation mode
• LANDFIRE, NLCD, MTBS exemplify this
• We will likely continue in this mode for a long
  time
• Essential Climate Variables (ECVs)
• Updated land cover data sets
• Continued need for providing of data to meet
  operational and management objectives

29
Data Generation versus Data Analysis

• Need to do more towards answering major science
  questions
• What are the patterns and rates and causes and
  consequences of land cover change?
• What do these changes mean in terms of
  important processes?
• Climate change issues
• Biodiversity issues
• Carbon sequestration issues
• Socio-economic issues

30
Technological Hurdles

• Volumes of data formidable
• What are the best algorithms to do the work?
• What types of tools can be developed to
  facilitate user analyses of the data?
• What are some of the best ways of integrated
  multiple sources of data?
• Clouds and image composites

31
Thank You!