Standards - GIS Courses...Scale_Factor_at_Central_Meridian: .09996 Longitude_of_Central_Meridian:...
Transcript of Standards - GIS Courses...Scale_Factor_at_Central_Meridian: .09996 Longitude_of_Central_Meridian:...
Standards An established or sanctioned measure, form, or method
Data standards: used to format, assess, document and delivery spatial data
Analysis standards: ensure most appropriate methods are used; get the best possible answers)
Professional or certification standards: establish the education, knowledge or experience of the analyst
Source: http://www.m-w.com/cgi-bin/dictionary
Data Standards
Media standards (CD-‐ROM, tape… ISO physical and device format)
Format standards (Data file components and structures; TIFF, IMG, Shapefiles…. .zip files)
Spatial data accuracy standards (quality of the positional and attribute values stored in the data set)
Documentation standards (how to describe spatial data's: origin/sources, transformations, manipulations, storage)
File Extensions http://www.geocomm.com/channel/esri/av_fileextensions.html
File Extensions http://www.geocomm.com/channel/esri/av_fileextensions.html
Data Accuracy
Accuracy measures how close an observation is to the true value
Precision refers to how repeatable a process or measurement may be (measure of predictable, consistency)
Data Accuracy
How close is the observation to the truth?
Expressed as
Percent (%) How often the value is wrong; “4% of the fields listed as row crops are perennial grasses”
Expressed as a number or probability distribution of the size of an error “the average positional error is 12.4 meters for power pole locations, or more than 90 percent of the digitized geodetic monuments are with 3.2 meters of their surveyed locations”
Data Accuracy
Positional accuracy should be quantified
Provide a physical measurement of the expected accuracy Most often statistics used are
•Mean Error •Error frequency threshold
Data Accuracy
Errors due to:
•How features are conceptualized
•Methods of data collection and analysis
•Field measurement error -‐ instrument and observer precision and bias, blunders
•Sampling density -‐ natural spatial variation
•Interpolation error •Map derived data -‐ map scale, digitizer precision, visual acuity, boredom factor •Data out of date Inadequate spatial data model
1936 1997Bellvue, Washington
Sources of Error Data always contain error-‐ take steps to minimize errors
Data Accuracy
Measuring and Documenting
Depends on the problem
Remember “truth” may not be completely known
•Must know the accuracy of our measure of the “truth” •It must be independent •It must have a higher order of measurement
Data Accuracy Measuring and Documenting
Four main ways to measure spatial data accuracy
Positional (geometric) Accuracy – Are coordinates correct? Attribute Accuracy – Are attributes correct Logical Consistency-‐ Are there overshoots, open polygons, duplicate or missing labels, are roads in rivers, etc. Completeness – how current is the data, what was included, how are the objects identified and defined, what was the minimum mapping unit?
(Understand and document these four within “Lineage” or what are the data sources and processing steps taken.)
18
NSSDA – National Standard for Spatial Data Accuracy defined in 1998
Defines test statistic, methods, reporting of positional accuracy
Five steps:
•Select test points •Define independent control data set •Collect measurements from both sources •Calculate positional accuracy statistic •Report the accuracy statistic in a standardized form included in the metadata
Well distributed samples
For each test point
For all points
Accuracy Assessment Summary Table
Note: RMSE is not the same as the average distance error not a typical distance error. It is a statistic that is useful in determining probability thresholds for error
RMSE x 1.7308 = 95% threshold
No established standards for accuracy of Linear features One common approach is to define a epsilon band (characterizing the uncertainty in line position)
(Another approach for straight lines:Using nodes and vertices to access accuracy)
Linear features (continued)
Attribute error Blunders Inappropriate model -‐ ordinal vs. continuous
How to characterize?
•Sampling. Visit points, precisely measure attribute at location, and query database. (Need to accurately locate yourself)
•Continuous variables, report mean, variance, etc., for measurements and database values
•For nominal variables, usually a proportion wrong. Can be summarized in a contingency table
Contingency or Confusion Table
Spatial_Reference_Information:Horizontal_Coordinate_System_Definition:Planar:Grid_Coordinate_System:Grid_Coordinate_System_Name: Universal Transverse MercatorUniversal_Transverse_Mercator:UTM_Zone_Number: 10-19Transverse_Mercator:Scale_Factor_at_Central_Meridian: .09996Longitude_of_Central_Meridian: -123 00 00Latitude_of_Projection_Origin: 0.0False_Easting: 500000False_Northing: 0.0Planar_Coordinate_Information:Planar_Coordinate_Encoding_Method: coordinate pairCoordinate_Representation:Abscissa_Resolution: 2.54Ordinate_Resolution: 2.54Planar_Distance_Units: metersGeodetic_Model:Horizontal_Datum_Name: North American Datum 1927Ellipsoid_Name: Clark 1866Semi-major_Axis: 6378206.4Denominator_of_Flattening_Ratio: 294.98Vertical_Coordinate_System_Definition:Altitude_System_Definition:Altitude_Datum_Name: National Geodetic Vertical Datum of 1929
Metadata(“Data about data”)
http://www.fgdc.gov/
http://www.fgdc.gov/
http://www.fgdc.gov/clearinghouse/clearinghouse.html
http://www.fgdc.gov/clearinghouse/clearinghouse.html