Insights into the Operational Use and Limitations of Spectral Vegetation Indices for Agriculture and

1
Insights into the Operational Use and Limitations
of Spectral Vegetation Indices for Agriculture
and Rangelands
Willem van Leeuwen University of Arizona,
OALS/ARSC GRD
2
What is RangeView?

• RangeView is an on-line decision aid for natural
  resource managers.
• The application provides spatially dynamic access
  to bi-weekly vegetation greenness images from
  1989 to date for any location in the United
  States using AVHRR and in the West using MODIS.
  Access to associated time-series -
• precipitation and climate surfaces

3
RangeView SVI Users

• End users include ranchers and other land owners
  as well as land and natural resource managers and
  researchers from a wide spectrum of local, state,
  tribal and federal agencies

4
RangeView Makes Satellite Imagery
OperationallyAccessible
http//rangeview.arizona.edu/
5
RangeView provides satellite time series data
every two weeks. This allows ranchers to compare
current conditions with past. Here, the Walker
Basin allotment is found to be greener than
average. This information could be helpful for
evaluating rangeland usage on USFS lands.
RangeViews decision support tool opens the door
to the information inside NASA satellite imagery
that we believe is a vital part of the future of
ranching in Arizona. John A. Scammon, Executive
Director, Arizona Cattle Growers Association
(12/15/03)
6
Acknowledgements

• Arizona Remote Sensing Center
• Drs. Stuart Marsh, Barron Orr, Charles F.
  Hutchinson
• Ann-Maree White, Wolfgang Grunberg, Jon Saints
• Laura Baker, Elisabeth vander Leeuw, Chih-lung
  Kuo,
• Grant Casady, Dan Tuttle, Colleen McDonald, Yu
  Yang
• Arid Lands Information Center
• Barbara Hutchinson and Anne Thwaits
• School of Renewable Natural Resources
• Drs. George Ruyle, Larry Howery and Paul Krausman
• Martin Pepper, Gwen Wolph
• Marketing Intelligence, LLC
• Dr. Kapil Jain and Chris Baker

7
The Extension goal is

• to bridge the gap between geospatial
  information research/technology and its use by
  every day people through
• seeds sown by NASA Earth Science, research and
  scholarship
• supported by Space Grant and extension
  facilitated by Land Grant.

8
New Initiatives in Alabama, Connecticut, New
Hampshire, North Dakota, Ohio, Oklahoma,
Nebraska, and Virginia
9
Our Programs
10
Diffusion Early Adopters Influence Others
11
Two-way Knowledge Exchange Training and Feedback
October 2002 Workshop Natural Resource Managers
from Government Agencies
12
(No Transcript)
13
(No Transcript)
14
(No Transcript)
15
(No Transcript)
16
(No Transcript)
17
(No Transcript)
18
(No Transcript)
19
(No Transcript)
20
(No Transcript)
21
RangeView next generation
Geographic Data Information Visualization
Analysis
22
A Sneak Preview

• GeoDIVA provides much more functionality than the
  current RangeView Dynamic Animation Tool
• 1 to 4 viewers (instead of 2)
• Ability to handle different spatial resolutions
  and time steps
• New spatial data MODIS products, climate
  surfaces, etc.
• New point data (rain gages) with the ability to
  dynamically graph rainfall through time as the
  animator allows you to visualize vegetation
  dynamics through time

23
(No Transcript)
24
(No Transcript)
25
(No Transcript)
26
(No Transcript)
27
User Quote

• Page or Tool MODIS NDVI 250m -Question or
  Comment
• Wondering why the latest MODIS images are more
  than 6 weeks old - we recently put cattle out on
  the North Kaibab - plan to use RS in our
  livestock operation in the future
• GIS Analyst - Grand Canyon Trust

28
Current use

• The depth of use (e.g. visualization for
  assessing pattern versus analysis of trends for
  critical decision making) depends on user
  confidence in the SVI values.
• Users rely on timely, spatially and temporally
  consistent and continuous data streams without
  having to worry about artifacts related to new
  and multiple sensors.
• Modelers and managers are moving from imagery to
  pixel based graphical representations and need
  tools to easily access, manipulate, integrate,
  analyze and correctly interpret these data.
• Accurate geo-registration of MODIS allows for
  pixel based and detailed time series analysis,
  making the current new generation of sensors much
  more attractive to modelers and land managers.

29
User insights

• Little advancement in connecting and integrating
  the associated uncertainties inherent to all
  steps of the processing and model chains (e.g.
  data capture, data input and SVI generation).
• Cross-comparison of uncertainty assessment is
  challenging to the end-product user because
  reporting of uncertainty tends to be research or
  data product-specific with limited emphasis on
  facilitating the interpretation of uncertainty
  associated with algorithm and processing quality
  for use by managers or decision makers.
• Consequently, the confidence in these data is
  often based on experience and visual confirmation
  of the spatial and temporal consistency in SVI
  imagery and time-series data.

30
Synthesize uncertainty information for the
end-user community.

• Measurement error for parameters and variables
  are often not provided with SVI products as they
  vary spatially and seasonally.
• In the best cases (e.g. MODIS products) it is
  difficult to translate the current quality
  assurance flags into an uncertainty,
• Implementing a feedback mechanism between SVI
  developers and end-users would also generate new
  perspectives on the utility and visualization of
  SVI products and their associated uncertainty
  fields.
• It is rare for a research paper to address all
  major factors that affect the uncertainty of
  spectral indices in a comprehensive fashion

31
User Community Requests Direct Quality
Assurance Fields with Spectral Vegetation Indices

• Provide more uncertainty data or some ranking
  system could be assigned to the many different
  flags (e.g. MODIS products that currently come
  with aerosol, shadow, cloud, bad data flags)
• Provide more information by visual coincident SVI
  and cloud and snow properties for each pixel

32
Future products

• Global modelers, vegetation science researchers,
  natural resource managers and decision makers
  alike would benefit from spatially-explicit
  uncertainty products that can be visually and
  quantitatively related to the SVI products
  themselves.
• Future research and product development should
  focus on improved characterization of uncertainty
  for all classes of end users.

33
PECAD - Production Estimate Crop Assessment
Division
PECADs Mission StatementTo produce the most
objective and accurate assessment of the global
agricultural production outlook and the
conditions affecting food security in the world.
Crop Explorer Regions
34
Global Crop Production Information and Analysis

• PECAD reports estimates of area, yield and
  production for 80 countries. (384 country X
  commodity pairs)
• Information is updated monthly and can be found
  at
• http//www.pecad.fas.usda.gov/wap.cfm

35
Within this framework/mission, PECAD has five
major objectives

• Provide accurate monthly production numbers by
  commodity and country.
• Provide the supporting evidence at monthly
  lock-up (when official estimates are developed
  by the World Agricultural Outlook Board).
• Provide bi-weekly web-updates for
  countries/regions during the growing season.
• Provide automated analysis products to external
  customers.
• Provide an ad hoc analysis capability for
  unanticipated issues as they arise.

36
Convergence of Evidence Methodology
Multi-source Data Processing and Expert Analysis
Global Ag. Production
37
Assimilation of NASA Products
Continuity NPP, NPOESS, GPM, LDCM
TBD -To Be Developed
38
PECADs Integrated System Solutions and
Enhancements (1)
Goal Improve the existing system and work on
continuity of products and processes that work

• Integrate tools and/or data (Landsat-and
  MODIS-like imagery) for comparative RS based crop
  analysis
• Access to multi-date field-level and national
  level multi-spectral observations and information
  (Landsat- and MODIS-like)
• Develop image processing and/or analysis tools
  for Landsat- and MODIS-like data (e.g. web
  services, ease of use, GIS tools Limitations of
  hardware, software and network capacity)

39
PECADs Information System Solutions and
Enhancements (2)

• Improve timeliness and/or quality of Landsat- and
  MODIS-like data (AWIFS/5d)
• Improve accuracy, timeliness, reliability and
  usefulness of precipitation/ weather data and
  analysis tools
• Improve crop stage and yield models and soil
  moisture models (country- crop-specific)
• Improve historical country- and
  commodity-specific area and yield data base, and
  contextual reference data (e.g. crop regions,
  counties, provinces)

40
PECAD Enhancement Goals (a few examples)

• Deliver MODIS Land Product images over major
  agricultural regions
• Rapid response, EVI and NDVI images
• False-color and shortwave composites
• Deliver time-series graphs over major
  agricultural regions
• Data drill images at original resolution.

41
PECAD operational requirements part of IEOS
requirements!

• Solicitation of requirements from the USDA/FAS
• based on global priorities, research agenda and
  needs, and the alignment of international
  collaboration of many agencies, institutions and
  local user communities.
• Requirements include aspects related to
• sensor characteristics (JACIE - Joint Agency
  Commercial Imagery Evaluation - NGA, NASA, USGS
  users)
• temporal and spatial scales of data acquisition
• data/product latency
• spectral considerations
• development of new models (e.g. crop, moisture)
• integrated tools and data
• continuity for fulfilling needs
• interoperability

42
PECAD input requirement strategies (GEOSS, MARS,
FAO)
43
Multiple SVI products

• Use strength of each sensor - latency vs
  spatial resolution vs consistency (time series)
  (RangeView)
• PECAD latency, redundancy, complementing data
  (confidence building in DSS)

44
Contact InformationWim van Leeuwenleeuw_at_ag.ari
zona.edu520-626-0058