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 error  •Map  derived  data  -­‐  map  scale,  digitizer    precision,  visual  acuity,    boredom  factor  •Data  out  of  date  Inadequate  spatial  data  model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