Geostatistics Flashcards
Tobler’s first law
everything is related to everything else, but near things are more related than distant
spatial autocorellation (values of nearby things tend to be related)
Spatial interpolation
estimate of unknown values between sampled values within a convex hull
polygon enclosing all points in a dataset
convex hull
4 types of spatial interpolation
global, local, exact, inexact
global spatial interpolation
uses all input points for each cells calculation
local spatial interpolation
uses a localized subset of input
trend surface analysis
global
calculate a single function describing a surface that covers the entire map area
essentially a line of best fit but a surface
coefficients of a polynomial in trend surface analysis are determined through …
“least-squares” optimization
plane minimizes the sum of squared deviations between the sampled and estimated data
Euclidean allocation
local
each cell takes the value of the nearest input feature
geometry of output is determined by spatial distirbution of input
Thiessen/Voronoi polygons
value at an unknown location is equal to the value at the closest known location
unknown locations become the lines/borders of the polygons (as opposed to cell values changing in Euclidean allocation
how many values does a Thiessen/Voronoi polygon have
1, assigned from the contained point
area under polygon is closest to contained point than any other point in input
inverse distance weighting (IDW)
local
interpolates value at a POI as a weighted average of nearby points
unknown value to be interpolated by average of surrounding points
weighted - assumes spatial dependency
is IDW exact
Yes, interpolated results must match the value at a sampled point
Kernel density estimators
represent density change away from a known point
summit of kernel is point and pile around represents diminishing influence away from location
smaller search radius yields ____ density
peakier
no influence from outside the search radius
