Goals and Objectives

1
Goals and Objectives

• CEOS Cal/Val Working Group on Global Land Cover
  Validation
• JRC, Ispra, 2728 March 2003

2
Primary Goal Best Practices Document

• Identify validation needs
• What validation is needed and why
• Design-based analysis
• Recommend sampling strategy
• Suggest ground truth strategy
• Chart the path for analysis techniques
• Account for costs
• Model-based analysis
• Quantifying the classification process
• Informal analysis
• Systematic examination
• Comparisons with other datasets

3
Tough Problems

• Coarse- vs. Fine-Resolution
• Legend mismatches
• Mixed pixels and mixture classes
• Geolocation errors
• Ground truth errors
• Imageryseasonality
• Reference materials
• Accuracy Metrics
• Agreement and Disagreement (Fuzzy)
• Spatial Autocorrelation
• For extended test sites

4
The Way Forward

• Milestones
• March, 2003 First draft outline with knowns and
  unknowns
• July, 2003 Outline with issues resolved
• October, 2003 Second meeting with wider group
• January, 2004 Draft document to circulate
• March, 2004 Final report
• BU NASA Proposal
• Global, Climate-Stratified, Land-Cover Validation
  Database
• Analysis and comparison of present and future
  global-scale land-cover products

5
Meeting Structure

• Presentations
• Issues in Accuracy Assessment
• Identify broad issues and problems that need to
  be addressed
• Validation Examples and User Perspectives
• More concrete or specific contributions or
  examples
• Discussions
• User needs, validation strategies
• Outline Sessions
• Convergence