Objectives of lecture:
Student contribution: matrix algebra, decompositions and mathemagian work (Zach)
Projections Part I
1 Standardized projections: UTM & State Plane, etc.2 Numerical mechanics of projections
GCTP and other code for projections
3 Review of parameters required for metadataRegistration part II
1 Introduce basic geometric transformations: (rotation, translation, scale)
2 Application in orienting digitizers or material digitized (scanned/ sensed)
3 Least squares fit of "best" affine or projective
4 Influence of outliers ("blunders"); alternatives for fit
conventional choices of projection parameters; Overview Peter Dana Geographer's Craft site
Zone of standardized system (UTM
counts from Date Line eastward- we are in Zone 10)
State Plane zones have multiple numbering systems (alphabetic
state or FIPS codes...)
state plane zones are aggregates of counties, eg. central
Texas.
Most multi-zone states have a secret (usually unoffical) projection that covers the whole state. In Wisconsin, it is a TM zone positioned on the split between two zones in UTM (90 West), in Oregon, they have a special compromise Lambert rather openly adopted (including a set of error analysis, the area error image). Oregon law establishing coordinate systems (begin with 93.310)
Washington has the practice of extending South Zone to the North without any official sanction...
Official Coordinates for Washington State (RCW (the server) , Text (partially converted of Title 58 Chapter 20)
Other countries have similar choices, eg. Finland; UK; Ireland, (presumably M and N appear where expected...); France (3 Lambert zones);
Stephan Voser, great list of standard projections for Europe and elsewhere. (with ESRI parameters for most of them...)
Another summary with some specifics on European solutions
Bob Burth list of links on datums and coordinates (some links to specific country solutions)
the chatter it often takes to figure out that Switzerland is an oblique Mercator (1997 email, all links dead...)!
longitude scaled onto the X axis; Latitude on Y, issue is scaling of Y axis.
Centered over a Pole; R-Theta coordinate system from there; Parallels are concentric circles - the Radius of each circle, Angle scaled from longitude (at that parallel); then R-theta converted to X,Y.
Ellipsoid: Equat. radius, Polar radius, flattening (a,b,f)
eccentricity: e2 = 2f - f2 derived from flattening...
Horizontal and vertical datum(s?) [not data...] discussed lecture 08
Modification of "Normal Aspect":
Transverse (some meridian is new Equator);
Oblique; generally just recalculate lat/long based on new "pole" and "equator".
a location to be called (0,0) (given by latitude, longitude usually, and off in the far lower left of the study area...)
Registration involves transforming the coordinates of some collection
of data into a desired coordinate system. This is usually based
on having a key to the transformation by identifying some set
of points ("control") on both sources.
Some systems (eg. ARC/INFO) build this into the process immediately
(as suggested in Clarke page 69-70. I disagree, since I think
the original space is needed to perform quality control, then
transform later (with more ability to select how it happens)
How many points? ARC (see overheads) suggests 4, barely enough
to cross-check an affine! 16 or 20?
What points provide the best connection?
Ultimately, geodetically surveyed control points (see Geodetic
Control resources at National Geodetic Survey). Failing that,
often special control points (panelled on ground for photographs)
or simply points measured on maps (depends on which points you
use...)
US National Geodetic Survey
Data sheets for control pointsMonument data from WA DOT
GPS base station data (for differential corrections)
Affine: equation is simple, and general.
To translate, simply add (subtract) from X and Y; to scale multiply
(either by the same value or by differential scales (removes a
tilt from an airphoto, often, but not always...); rotation involves
cross products.
Combined through use of matrix notation, 3 X 3 matrices of three
components can be combined then applied to each coordinate.
alternatives to affine: projective (removes "barrel"
distortion based on distance from a focal point (deals with lens
distortions); higher order equations (usually impractical);
piecewise transformations (see Marvin White in his conflation
days).
Least squares: apply regression analysis to problem.
Compute transformation between the input X,Ys and the desired
"control" values that minimizes the squared deviations.
R squared (as in regression) is a measure of goodness of fit,
and the distances (variances) of the residuals are interesting
too.
Outliers: regression can be affected (sometimes heavily)
by some single bad value. Common method: examine outliers, remove
points, refit equation
"Robust" Alternative: Least-Median Squares (can be up
to 50% contaminated)
versions of code shown...
Speaking Truth to Power presentation