Validation Data Support Activities at the ORNL DAAC

1
Validation Data Support Activities at the ORNL
DAAC
Bob Cook, Steve Margle, and Lisa Olsen ORNL
DAAC Jeff Morisette, NASA GSFC John Dwyer, EROS
Data Center Faith Ann Heinsch, University of
Montana

• Land Cover Validation Workshop
• Boston, Massachusetts
• February 2, 2004

2
ORNL DAAC?

• Oak Ridge National Laboratorys (ORNL)
• Distributed Active Archive Center (DAAC)
• www.daac.ornl.gov

• Archives and distributes data for terrestrial
  biogeochemistry and ecosystem dynamics, including
  in situ data for validation

3
Compiling data to assist in validating MODIS land
products

• Good science requires understanding product
  accuracy / uncertainty
• Explicit statements of uncertainty foster an
  informed user community and improved use of data
• Use of global products produced by CEOS members
  requires characterization of each products
  uncertainty

4
Data Support for Validation

• Data needs and schedules defined by working with
  scientists funded to validate MODIS Products
• Field investigators have responsibility to
  process, document, QA, and provide access to
  their data
• ORNL DAAC provides Mercury system for search and
  retrieval, for both in situ and remote sensing
  data
• MODIS ASCII Subsets of key products prepared in a
  format convenient for comparison with field data
• When finalized move Land Validation data to
  long-term archive (ORNL DAAC)

5
MercuryMetadata Search and Data Access System

• Designed to provide access over the Internet to
  distributed research data
• Low requirements for the validation investigator
  (data provider)
• software tool available to help produce metadata
  and documentation
• Point and click download function
• Currently used for a number of activities
• EOS Validation, southern Africa (SAFARI 2000),
  and in Amazonia (LBA)

6

• Land Validation Core Sites
• 26 sites stratified by six biomes
• Based on existing stations, resources, experts
• FLUXNET, AERONET, BigFoot, LTER, etc.
• Long-term monitoring
• Jointly nominated by Instrument and Validation PIs

• Remote Sensing Data
• Landsat 5, ETM, IKONOS,
• ASTER, airborne sensors,
• subsetted over the core sites
• in situ Site Data
• Collected by various groups
• at core sites, available via Mercury
• Networks
• AERONET, FLUXNET, BSRN, etc.
• Ancillary Site Data/GIS Layers
• - site variables
• - elevation
• - land cover
• - reference layers (e.g., political
• boundaries, airports, water
• bodies)

7
Global FLUXNET Networkfor MODIS NPP Product
Validation
8
FLUXNET Activities

• 252 towers registered from 11 networks
• Gap-filled flux data available for 176 site-years
  from 53 sites
• Database includes
• Over 4,800 site ancillary values
• Over 2,000 instrument descriptors
• Over 5,000 literature citations

9
MODIS ASCII Subsets FY 2004

• Preparing and posting 9 Terra (C4) and 3 Aqua
  (C3) MODIS Products for 274 sites
• Developing web-based data visualization tools

Product Terra (C4) Aqua (C3)
Surface Reflectance MOD09A1 MYD09A1
Surface Temperature MOD11A2 MYD11A2
Land Cover MOD12Q1 (C3) -
NDVI / EVI MOD13A2 MYD13A2
LAI / fPAR MOD15A2 -
GPP, PSN MOD17A2 -
Annual NPP MOD17A3 -
Albedo MOD43B1 -
NBAR MOD43B3 -
10
MODIS ASCII Subsets ( 7 x 7 km)

• MODAPS at GSFC (NASA) provides subsets of
  selected products in HDF and SIN Projection for
  274 sites
• ORNL DAAC subsets to 7 x 7 km and reformats from
  HDF image into ASCII file of pixel values and QA
  flags, expanding as data are collected
• ASCII File posted on DAACs ftp site

MOD12Q1 (C3) for Walker Branch, Tennessee
MOD15A2 (C4) for Walker Branch, Tennessee
11
MODIS Land Cover Grids

• 7- x 7- km grids of MODIS Land Cover (MOD12Q1)
• Based on C3 processing for MODIS Land Cover (UMD
  Classification)
• Data available for other classifications and
  their uncertainty

12
Demonstrations

• WebMap Server
• Enables users to search for and select sites that
  have MODIS ASCII Subsets
• MODIS ASCII Subset Data Visualization
• Work in progress, currently only available for
  LAI/ fPAR (MOD15 C4)
• Grid Visualization
• Time Series

13
MODIS Validation What works?

• Growing buy-in to the core sites concept where
  the field, airborne, satellite data suite is
  helping to address scaling issues
• Exploitation of in situ networks (e.g. flux
  towers, AERONET, LTER/ILTER, BSRN) and field
  investigations (e.g., LBA, SAFARI 2000)
• Simplified data access with dedicated archives
  and easy to use data format (e.g., ASCII
  subsets)
• Frequent communication between Product PIs,
  Validation PIs, and data facilitators
• Best validation sites are those where
  investigators taking ground measurements are also
  interested in satellite products

14
MODIS Validation Whats in the works?

• Data for validation of MODIS Land products
  continue to be compiled and made available
• the number of sites is growing through
  coordination with science networks and CEOS
  members
• Improving formal documentation and making
  validation results more readily available to
  users
• Users can provide important feedback over and
  above formal validation investigators (e.g.,
  through DAAC surveys and user services)

15
Validation on-going needs?

• Promote stronger interface to the user community
  (e.g., through Validation Workshops)
• Develop stronger connection with the scientific
  community to establish how accurate the products
  need to be and how to relay products uncertainty
  so this information can be incorporated into how
  they are used

16
(No Transcript)
17
Best Practices for Preparing Ecological and
Ground-Based Data Sets to Share and Archive

• Best Practices include
• 1. Assign Descriptive File Names
• 2. Use Consistent and Stable File Formats
  
• 3. Define the Parameters
• 4. Use Consistent Data Organization
• 5. Perform Basic Quality Assurance
• 6. Assign Descriptive Data Set Titles
• 7. Provide Documentation
• Provided this document to
  SAFARI 2000, EOS Land Validation,
  BigFoot, and LBA projects
• Available on-line, in booklets, Ecol. Bull. 2001
• (http//www.daac.ornl.gov/DAAC/PI/bestprac.html)

18
in situ Data for EOS Core Sites
19
Leaf Area Index (MOD15A2)Walker Branch, Tennessee
20
Leaf Area Index (MOD15A2)
All pixels in 5 x 5 km area included Blue
good Red mean of 5 x 5 km area around
tower Green rejected based on clouds
Walker Branch, Tennessee