Inverse Distance Weighted Interpolation

Overview

  • explicitly makes the assumption that things are close to one another are more alike than those that are farther apart

  • to predict a value for any unmeasured location

    • IDW uses the measured values surrounding the prediction location

measured values closest to the prediction location have more influence on the predicted value that those farther away

  • it is assumed each measured point has a local influence that diminishes with distance

Definition

the expected result is a discrete assignment of the unknown function \(u\) in a study region:

\(u(x): x \rightarrow \mathbb{R}, x \in D \subset \mathbb{R}^n\)

where \(D\) is the study region

The set of \(N\) known data points can be described as a list of tuples

\([(x_1,u_1),(x_2,u_2),...,(x_N,u_N)]\)

the function is to be "smooth" (continuous and once differentiable), to be exact \((u(x_i)=u_i)\) and to meet the user's intuitive expectations about the phenomenom under investigation.

Power function

Weights are proportional to the inverse of the distance

  • raised to the power value \(p\)

  • as a result, as the distance increases, the weights decrease rapidly

    • the rate at which the weights decrease is dependent on the value of \(p\)
      • if \(p = 0\) , there is no decrease with distance
        • and because each weight \(\gamma_i\) is the same
          • the prediction will be the mean of all data values in the search neighborhood
  • as \(p\) increases, the weights for distant points decrease rapidly

    • if the \(p\) value is very high, only the immediate surrounding points will influence the prediction

when \(p=2\) the method is known as the inverse distance squared weighted interpolation

Search Neighborhood

Things close to one another are more alike than those that are farther away

  • as the locations get far away, the measure values will have little relationship to the value of the prediction location

    exclude more distant points that have little influence for speedup

    • it is common practice to limit the number of measure values by specifying a search neighborhood
      • restricting how far and where to look for the measured values to be used in the prediction

when to use IDW

  • a surface calculated using IDW depends on selection of power value \(p\) and the search neighborhood strategy

  • the output surface is sensitive to clustering and the presence of outliers

    • IDW assumes that the phenomenom being modeled is driven by local variation
      • which can be modeled by defining an adequate search neighborhood
  • since it does not produce prediction standard errors

    • justifying its use is problematic