Assumptions

1
Assumptions Problem statements(real data has
real warts)

• Multibeam Quality Assessment Workshop
• January 4-5, 2006
• Dale Chayes ltdale_at_ldeo.columbia.edu

2
Assumptions

• Granularity per-file (independent of file size)
• File must be uniquely identifiable by name within
  the specified "repository"
• No subjective judgment calls", repeatable metrics
  
• Does not require access to an RDBMS to function
• Works seamlessly and scaleable to global holding
  and global tiles in a huge RDBMS
• Solutions, schemas, documentation are open source
  and freely available

3
Problem statements (1) Assuming that my
"repository" is "populated" with suitable
mbquality statistics and a query interface

• Show me the worst N-files in my holdings (based
  on each tool)
• Show me which files from a list have a metric
  greater, less than a.b or in the range of than
  w.x to y.z
• Show me all the files that have been assessed by
  tool X.Z
• Show me the statistics (metrics) from a named
  cruise, a specific ship, a time window, a
  geo-area, software/firmware version, personnel on
  board.

4
Problem statements (2)

• Look for symptoms sound speed errors
• Look for egregious outliers (bathymetry)
• Look for evidence navigation errors (offsets,
  crossovers)
• Does the just closed real-time data file seem
  reasonable (through multiple metrics)
• Show me the multi-year history of each vessels
  multibeam data using multiple metrics
• How do we assess biases (pitch, heading, roll)
  GOE adjustment method for doing patch tests by
  some Norwegians, commercial implementations
• Look for evidence that nav sensor offset errors

5
Problem statements(3)

• Create and maintain a public repository of patch
  test procedures, methods, LOCATIONS, and results
  (by platform.)
• How does this data file compare against a
  reference data set (global grid, tile, GEBCO,
  Sandwell Smith)
• Estimate RMS errors in positioning, heading,
  depth.