Measurement Frameworks by 01dGd1AR

VIEWS: 7 PAGES: 54

									Measuring Geographic Phenomena
   Data Models and Data Structures
   Measurement Frameworks

     Lecture 3 (b)
     ESTU 401
Terms
   Measurement Framework
       How the world is measured
       What do you control, what do you measure
   (GIS) Data Models (in simple terms)
       Raster or Vector
   (GS) Data Structures (in simple terms)
       Shapefile, geodatabase, coverage
       DEM, GRID, Imagine file, jpeg, tiff
       TIN
Measurement Frameworks…
   Provide a means (rules) for controlling certain
    components of a phenomena, allowing
    measurement of another component
   Geographic information has 3 components:
       Time
       Space
       Attribute
   Can’t measure all 3 simultaneously
       Control the less important component(s) in able to
        measure the desired component
Measurement Frameworks vs. Data
Models
   Measurement Frameworks are not the same
    as Data Models
   Often…:
       Attribute control = vector
       Space control = raster
   But, there are exceptions…
       Attribute controlled raster
       Spatially controlled vector
Data Models
   How the world is modeled
       In simple terms: Raster or Vector data
       Sometimes ‘Field’ vs. ‘Discrete Object’
   In not so simple terms, is composed of:
    A conceptual representation scheme…
    … using a Measurement Framework (or
     frameworks)
Vector & Raster Data Models
   Vector Data
       Vector data model measures the world as
        a set of discrete objects, representing
        features in the world
       Use points, lines, and polygons
   Raster Data
       Raster data model measures the world as
        a continuous field or surface
       Use tessellations
   These are conceptual
    distinctions
Data Models (ESRI side note)
   Note: ESRI also use the term ‘Data Model’ to
    refer to a series of industry-specific spatial
    database designs (or data templates) that they
    have helped to create (along with academic
    and industry professionals), e.g., for Hydrology,
    Fire Service, Transportation…
   See:
    http://support.esri.com/index.cfm?fa=downloads.dataModels.gateway
      Data Models (ESRI)




http://support.esri.com/index.cfm?fa=downloads.dataModels.gateway
Terms
   Measurement Framework
       How the world is measured
       What do you control, what do you measure
   (GIS) Data Models (in simple terms)
       Raster or Vector
   (GIS) Data Structures (in simple terms)
       Shapefile, geodatabase, coverage
       DEM, GRID, Imagine file, jpeg, tiff
       TIN
Data Structures
(or file structures or file formats)
   A specific implementation of a data model
   How the computer will represent your
    concepts
       Often related to specific software packages
   In simple terms:
       Shapefile, geodatabase, coverage
       DEM, GRID, Imagine file, jpeg, tiff
       TIN
Data Structures
   ESRI is primarily a vector based software

   ERDAS Imagine, or ENVI are raster based
    Vector Data
   Point, lines & polygons, based on x-y locations
   Points are discrete objects
   Lines are either discrete objects or are derived
    from points
   Polygons are either discrete objects (“whole
    polygon”) or are derived from lines (topological
    structure)…
Data Structures
   ESRI vector data structures
       ArcInfo Coverage*
       ArcView Shapefile**
       ArcGIS Geodatabase (.mdb, .gdb, .sde)
   ESRI raster data structures
       GRID

* Note that the coverage data structure was a connected coverage
   measurement framework (polygons)
** Note that the shapefile is an isolated object measurement framework
   (polygons)
    Shapefiles (ArcView 3.x)
   Multiple files make up a single ‘shapefile’




                 ArcCatalog
    Shapefiles (ArcView 3.x)
   “Whole polygon”
       ‘Spaghetti’ data
       Similar to CAD &
        drawing software
   Borders stored twice
   Each polygon is
    represented as a
    string of x,y points
   No topology is stored
       “Whole polygons”…                                                      Data File
                                                                                     C 3,0
                                                                      A 3,4
                                                                      A 4,4          C 3,2
      5
                                                                      A 4,2          D 4,2
                                                                      A 3,2          D 5,2
      4
                                                                      A 3,4          D 5,0
      3                                                                              D 4,0
                                                                      B 4,4
                               E                      A       B                      D 4,2
      2                                                               B 5,4
                                                                      B 5,2
      1
                                                                      B 4,2          E 1,5
                                                      C       D
                                                                      B 4,4          E 5,5
      0
                                                                                     E 5,4
                                                                      C 3,2          E 3,4
                     1             2            3         4       5   C 4,2          E 3,0
                                                                      C 4,0          E 1,0
                                                                                     E 1,5
http://www.utdallas.edu/~briggs/poec5319/struct.ppt
    Whole polygons…
   Spaghetti digitizing
    results in overlaps &
    slivers
       sliver = gaps
   Redundant borders (x,y
    points stored in both
    polygons)
   No topology
       No topological
        operations…
    Coverages (ArcInfo workstation)
   Multiple files stored in multiple file folders…




                             ArcCatalog
    Coverages (ArcInfo workstation)
   Topology is explicitly represented & stored
   “Point in polygon”
       Points used to make
        lines (can be reused)
       Lines used to make
        polygons (can be
        reused)
   No overlaps / slivers
       Also no islands or
        donuts…
                                                                              Points File
       “Point in polygons”…                                                   1    3,4
                                                                                          Polygons File
                                                                              2    4,4
                                                                              3    4,2    A 1, 2, 3, 4, 1
      5
                      11                                          12          4    3,2    B 2, 5, 6, 3, 2
      4                                                                       5    5,4    C 4, 3, 8, 9, 4
                                            1
                                                       2          5           6    5,2    D 3, 6, 7, 8, 3
      3                            13                 14          15          7    5,0    E 11, 12, 5, 1, 9,
                               E                      A           B           8    4,0       10, 11
      2
                                            4                      6          9    3,0
                                                       3
                                                      16          17          10   1,0
      1
                                                      C                       11   1,5
                                                                                          Label Points
                      10                    9
                                                                  D                         A 13
                                                          8           7       12    5,5
      0
                                                                                            B 14
                                                                              13 2,3        C 15
                     1             2            3             4           5
                                                                              14 3.5,3      D 16
                                                                              15 4.5,3      E 17
                                                                              16 3.5,1
http://www.utdallas.edu/~briggs/poec5319/struct.ppt                           17 4.5,1
    Geodatabases (ArcGIS)
                                   ArcCatalog
   Personal geodatabases (.mdb)
   File geodatabases (.gdb)
   ArcSDE geodatabases (.sde)
      Geodatabases (ArcGIS)
      Feature Classes
      Feature Datasets
      Relationships…
      Domains…




http://www.geog.ucsb.edu/~ta176/g176b/lab3/lab3_graphics/gdb_newobjects_9_ed.gif
    Geodatabases (ArcGIS)
   Topological rules are stored outside of the data
    layer itself
       Spatial relationships (coincident/adjacent geometry)
            Lines can/cannot cross or must be coincident
            Points must be within polygons
            Polygons must be adjacent
            Etc.
       Can exist between different layers
   Coverages are a topological data structure.
    Geodatabases can make use of topological
    relationships and rules…
      Geodatabases Topology Rules




http://webhelp.esri.com/arcgisdesktop/9.2/index.cfm?TopicName=Topology_rules
      Geodatabases Topology Rules (examples)
                                                                  No overlaps or
                                                                 gaps for polygons



                     Lines must                                                  Points must be
                    be coincident                                                 w/in a polygon



  Lines must connect
     (no dangles)
                                                                               Points must be on a line




http://webhelp.esri.com/arcgisdesktop/9.2/index.cfm?TopicName=Topology_rules
Raster data
   Thematic (or Discrete, or Categorical)
       Often derived from a attribute controlled measurement
        framework
       Often converted from
        vector data
   Continuous (or Field
    or Surface)
       Soils, geology,
        elevation
       Temperature
       Satellite imagery
       Distance
Raster data
   Imagery
       .jpg, .tif, .sid, .ecw…
       Air photo or satellite
       Cells with numbers representing color
   GRID’s
       Cells with numerical values – representing some
        measurement
      ESRI Raster data
             Imagery (jpg, .tif, .sid, .ecw)




                                                                               ArcCatalog
http://webhelp.esri.com/arcgisdesktop/9.2/index.cfm?TopicName=What_is_raster_data%3F
ESRI Raster data
   GRID (ArcInfo, ArcView, ArcGIS…)




                           ArcCatalog
Raster data (DEM)
   Digital Elevation Model (DEM)
   Technically, a USGS file format (.dem)
   Also (commonly) used as a generic name
    for any raster elevation data (e.g. an ESRI
    elevation GRID)
What’s the Most Important Decision
About Raster?
   Resolution (size
    of the grid cells)
       Often “inherited”
        from data…
       Sometimes an
        active decision
Raster: Cell Size
   Resolution (size of
    the grid cells)
       Larger or smaller
        cell size can
        produce very
        different results…


   High resolution (small cells) = large files
      (requires more storage space)
  Raster data
    Different structures for different purposes…



                                        Run-length Encoding



ASCII GRID Format
TIN’s
   Triangular Irregular Network
   Vector based surface model
       Irregular network of points, with an elevation
        value, connected by angled triangles
TIN’s




 ArcCatalog
       TIN’s




http://www.csiss.org/learning_resources/content/good_sa/tin.jpg

http://www.hbp.usm.my/thesis/heritageGIS/2Dto3DGIS_files/image007.jpg
Measuring & Counting the World…
   Examples
       Lakes
       Population
Measuring & Counting the World…
   Whatcom County Lakes
       Frames of reference?
       Measurement level?
       Measurement framework?
       Data model?
       Data structure?
Lakes?…
   We need to define “Lake”
       Is a small pond a ‘lake’?
       Does a ‘lake’ need to have water year-round?
       Is there a minimum depth for a ‘lake’?
       Are swampy areas ‘lakes’?
       Are wide spots in streams/rivers ‘lakes’?
Lakes?…
   Once we have defined ‘lake’ we can then
    measure the county
       Where are the lakes?
       How many lakes there are?
       What % of county is ‘lake’?
       What is the biggest lake?
       etc.
Lakes?…
   Frames of reference?
       Time                    2010 (winter vs. summer vs. aggregate?)
       Spatial                 UTM, State Plane
       Attribute               Lake classification system, depth…
   Measurement level?
       Nominal                 Lake/Not Lake, Pond-Lake-Swamp
       Ordinal                 Pond-Lake-Big Lake, or ‘lakeness’*
       Interval                Lake temperature
       Ratio                   Area, depth, perimeter…
              Lakeness scale:     0 = never holds water     3 = wet, clear, < 1 acre
                                  1 = occasional            4 = wet, clear, > 1 acre
                                  2 = always wet, (w/veg)
Lakes?…
   Measurement framework?
       Fix Time
                   Could control attribute (lakes, defined
       Control?     as being a certain size/type or
       Measure?     types) and then measure/locate
   Data Model?      any water body in the county
                     meeting that definition (probably
       Vector       using a vector data model)
       Raster     OR: Could control space, establish a
                     regular grid of cells (or sample
                     points) and for each cell measure
                     for lake presence, or lakeness, or
                     depth… (probably using a raster
                     data model – or sample points)
Lakes?…
   Vector model (attribute controlled), provides
    discrete objects (to be counted, located,
    mapped)…
   Raster model (space controlled), could include
    bathymetry (average, min, max depths) or could
    be part of a complete surface (or field) of land
    cover information… (percent of county that is
    lake, median lakeness category, etc.)
Measuring & Counting the World…
   Population of Whatcom County
       Frames of reference?
       Measurement level?
       Measurement framework?
       Data model?
       Data structure?
Population?…
   We need to define “Population”
       Human vs. other…
       Where people live vs. where they work…
            Daytime vs. nighttime population
       Permanent residents or anyone alive at certain point
        in time
       Adults vs. children
Population?…
   Once we have defined ‘population’ we can then
    measure the county
       Count how many people there are
       Where are they located?
       Calculate densities, clustering,
       Demographics (age, gender, income, etc.)
       etc.
Population?…
   Frames of reference?
       Time        2010
       Spatial     UTM, State Plane
       Attribute   Human population…
   Measurement level?
       Nominal
       Ordinal
       Interval
       Ratio       Number of people
Population?…
   Measurement framework?
       Fix Time
                   Could control attribute (looking for
       Control?     people) and then measure/locate
       Measure?     every person as a point (probably
   Data Model?      using a vector data model)
       Vector     OR: Could control space, establish a
                     regular grid of cells (or sample
       Raster       points) and for each cell measure
                     the number of people… (probably
                     using a raster data model)
Population?…
   Vector model (attribute controlled), provides
    discrete objects (to be counted, located,
    mapped)…
       Could be aggregated into groups (blocks, tracts)
   Raster model (space controlled), depicts entire
    area of county as a continuous population
    surface, with a given measurement (count,
    density, average income, etc.)
Measuring & Counting the World…
   Examples
       Lakes
       Population
   PLUS:
       Sampling methodology (random or regular)
       Spatial autocorrelation issues
       Scale
       Interpolation
       Repeatability
       etc…
Data Sampling and Measurement
   The world is infinitely complex…
   GIS (computers) are inherently finite
   Strive to make decisions to best
    model/measure the world – for a given
    aspect, task or question
   Data collected to fulfill the needs of one task
    may or may not be best suited or others…
Data Sampling and Measurement
   Can never completely measure the world
       30 m. satellite land cover imagery ignores minor
        variations in cover types smaller than 30 m.
       20 m. contour lines ignore smaller variations
        between contours
       10 year censuses ignore population fluctuations
        between census years
       Rain gauges cannot measure rainfall between
        gauge locations
       Habitat classes can never capture all of the finer
        detail of habitat variations from one location to
        another
Data Sampling and Measurement
   In order to fully evaluate the usefulness of a
    particular data set for a given project one
    must know the methods used for the data
    collection and processing, including the
    measurement framework used, the
    conceptual data model used to store the
    data…

    along with things like the sampling criteria,
    processing of the data…
Big Summary (to date)…
   GIS (messy, contextual and more than just software)
   Geographic information (Space, time & attribute)
   Frames of reference (Spatial, temporal, thematic)
   Measurement levels (Nominal, Ordinal, Interval, Ratio)
   Measurement framework (Attribute or space control)
   Data Model (conceptual model, raster/vector)
   Data Structure (specific implementation, file format)

								
To top