Spatial Interpolation
GEOS 5350 Demers, M.N., Geographic Information Systems Chang, Kang-tsung, Introduction to geographic information systems
Spatial interpolation is the estimation the value of properties at unsampled sites within the area covered by existing observations (control points). Calculates some property of the surface at a given point
Classifications of interpolations
1- Global & Local interpolation 2- Exact interpolation & Inexact Interpolation 3 - Deterministic and Stochastic
1 - Global and Local Interpolation
Global interpolation: uses all available control points Adequate for terraines that do not show abrupt variations Assumes good spatial autocorrelation on regional scales More generalized estimations Local interpolation: uses a sample of control points Adequate for terraines that show abrupt variations Assumes good spatial autocorrelation on local scales More local estimations
2 - Exact interpolation & Inexact Interpolation
Exact interpolation Predicts a value at control points that is the same as the observed values. The interpolation produces a surface that passes by the control points
Continue
Inexact Interpolation Predicts a value for the control points that differ from the observed value
3 - Deterministic and Stochastic
Deterministic interpolation No assessment of errors with the predicted values Stochastic interpolation Offers assessment of errors with predicted values. These methods assume random errors
Examples
I - Global (a) First-order trend surface
First-order trend surface (polynomial)
Example
(1) Set up 3 equations
(2) Re-write in matrix format (3) Calculate
X = 377 Y = 318 X2 = 29007 Y2 = 20714 XY = 23862 YZ = 4445.8 XZ = 5044
X 69 59 75 86 88 Y 76 64 52 73 53 Z 20.82 10.91 10.38 14.6 10.56 X2 4761 3481 5625 7396 7744 Y2 5776 4096 2704 5329 2809 XY 5244 3776 3900 6278 4664 XZ 1437 643.7 778.5 1256 929.3 YZ 1582.32 698.24 539.76 1065.8 559.68
377
318
67.27
29007
20714
23862
5044
4445.8
Continue
(4) Plug in values for 5 points
(5) Solve for b coefficients: Multiply inverse of left matrix by right matrix
(6) Use the b coefficients to calculate z for any point (X,Y) (69, 67)
(b) Higher-order trend surface
First order polynomials (inclined surface) can not represent the complex natural surfaces. A cubic or third order models can better represent such surfaces (e.g., hills, valleys)
Third-order trend surface (nine coefficients)
Example
II - Local
It is all about mechanisms for the selection of a suite of control points
Closest points
Points within a certain radius
(a) - Trends (polynomials) could be local
Local polynomials as opposed to global polynomials could be used as well for better representation of surfaces
(b) - Theissen Polygons
Also called proximal method Attempts to weight data points by area Commonly used for precipitation data
Continue
Triangles are drawn connecting control points (e.g., stations) using the Delaunay triangulation technique (also used for TIN) Lines are drawn perpendicular to sides of triangles at their midpoints Polygons are defined by intersections of these lines Values for control points are assigned to enclosing polygons
(c)- Density Estimation
Simple density functions: Number of points/cell size (e.g., 10,000 m2)(shown in shades of grey) Size of circle centered at center of cell size Other methods include Kernel density estimation Kernel density function is a commonly used alternative
(d)- Inverse Distance Weighted (IDW) Interpolation
Estimated value at point 0 Is the z value at control point i Distance between point I and point 0 The larger the k, the greater the influence of neighboring points. S number of used points
Continue
Zi 20.82 10.91 10.38 14.6 10.56 di 18 20.88 32.31 36.05 47.2 di2 324 435.97 1043.9 1299.6 2227.8 1/(di2) 0.0031 0.0023 0.0010 0.0008 0.0004 Zi x 1/(di2) 0.06426 0.02502 0.00994 0.01123 0.00474
SUM
0.0076
0.11520
zi 20.82 10.91 10.38 14.6 10.56
Between Distance points (di) 0,1 0,2 0,3 0,4 0,5 18 20.88 32.31 36.05 47.20
Example
Assuming k=2
Annual precipitation surface created by inverse distance squared
All values are within the maximum & minimum values of known points Small enclosed isolines are typical of this method
(e) - Radial Basis Functions (RBF)
Splines
A large group of interpolation methods Exact interpolators The difference between them is how the surface fits between the control points Each RBF also has a parameter that controls the smoothness of generated surface Differences between the outputs of these methods are small
Continue
Exact Function Good for large smoother surfaces Doesnt work well with abrupt changes
Continue
Creates a surface passing by control point and has the least possible change in slope at all points Unlike the IDW method, predicted values are not limited to the Max and Min dictated by data.
Kriging Next Lecture
Geostatistical Methods
Ordinary Kriging Simple Kriging Universal Kriging Indicator Kriging Probability Kriging Disjunctive Kriging Cokriging
Integrated Seismic Risk Map of Egypt
Generate Egypts first seismic risk map using a GIS approach
Red Sea-related Seismicity
Seismic Risks & Hazards
Seismic hazard --- Strength and frequency of shaking from earthquakes, Seismic risk ---- The chance of losing human life and/or property because of earthquake ground shaking.
Maps in the GIS
Seismic hazard map ---- Probability of occurrence of seismic ground shaking within a certain time frame Fault hazard map -- Surface & subsurface fault zones subject to re-activation, Amplification map -- Distribution of alluvium deposits that amplify ground shaking, Liquefaction map --- Areas susceptible (shallow groundwater, seismically active) to soil (sand/silt) liquefaction, Population density map --- Areas subject to increased risk of loss in human life and property.
Seismic Hazard Map
Probability of occurrence of seismic ground shaking of a certain intensity (peak ground acceleration PGA) in a specified time interval (250 years)
Distribution Map
Frequency Map # of earthquakes/cell
Hazard Fault Map
Faults proximal to earthquake epicenters & could be reactivated under current stress regimes
Hazard Fault Map
Intersection of the fault & distribution maps
Amplification Map
Earthquake ground motions are amplified by alluvium soil deposits
Soil Hazard Liquefaction Map
Intersection of earthquake coverage map with coverage map for soils that are saturated, porous, and have shallow water table (i.e., Nile deposits)
Population Risk Map
Highly populated areas affected by seismicity (low to high PGA)
