Variance prediction for pseudosystematic sampling on the sphere (Q4804602)

From MaRDI portal

!

This is the item page for this Wikibase entity, intended for internal use and editing purposes. Please use the normal view instead:

scientific article; zbMATH DE number 1902380
Language Label Description Also known as
default for all languages
No label defined
    English
    Variance prediction for pseudosystematic sampling on the sphere
    scientific article; zbMATH DE number 1902380

      Statements

      Variance prediction for pseudosystematic sampling on the sphere (English)
      0 references
      0 references
      0 references
      26 August 2003
      0 references
      Bernoulli polynomial
      0 references
      covariogram
      0 references
      Fourier analysis
      0 references
      nucleator
      0 references
      periodic function
      0 references
      stereology
      0 references
      geometric sampling
      0 references
      Let \(f:\mathbb{S}^2\to\mathbb{R}^+\) be a nonrandom square integrable function on the sphere. The authors concentrate on the problem of predicting the precision of sampling of \(f\) by means of ray probes emanating from the origin, represented by sectioning planes which contain the vertical axis and that are free to rotate about it. For an efficient implementation, the sampling design should allow several rays to be sampled on a section plane. A natural design is based on a partition of \(\mathbb{S}^2\) into regions of equal area. This is a pseudosystematic sampling, since the regions are not of identical shape, and consequently the design is not rotation invariant in general. Let \(F(\varphi,\theta)\) be the average of \(f\) at \((\varphi,\theta)\) and its antipode, which yields a periodic measurable function. The parameter of interest is \(Q=\int_{\mathbb{S}^2}F\:d\omega\). An unbiased estimator \(\widehat{Q}_{n_1,n_2}\) of \(Q\) is first obtained by pseudosystematic sampling, where \(n_1\) and \(n_2\) are the sample sizes. Then, an exact representation of var\((\widehat{Q}_{n_1,n_2})\) is given in terms of the Fourier coefficients of the covariogram of \(F\). The results are applied with a suitable covariogram model, and the obtained modelled variance is estimated with the aid of a single sample. The paper ends with the description of an example and with some interesting comments.
      0 references

      Identifiers

      0 references
      0 references
      0 references
      0 references
      0 references