Statistics and Computing

, Volume 21, Issue 1, pp 107–120 | Cite as

Estimating and testing zones of abrupt change for spatial data

Article

Abstract

We propose a method for detecting the zones where a variable irregularly sampled in the plane changes abruptly. Our general model is that under the null hypothesis the variable is the realisation of a stationary Gaussian process with constant expectation. The alternative is that the mean function presents abrupt changes. We define potential Zones of Abrupt Change (ZACs) by the points where the gradient, estimated under the null hypothesis, exceeds a determined threshold. We then design a global test to assess the global significance of the potential ZACs, an issue missing in all existing methods. The theory that links the threshold and the global level is based on asymptotic distributions of excursion sets of non-stationary χ2 fields for which we provide new results. The method is evaluated by a simulation study and applied to a soil data set in the context of precision agriculture.

Keywords

Excursion set Gaussian random fields Geostatistics Gradient estimation χ2 random fields Wombling 

Preview

Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

References

  1. Adler, R.: The Geometry of Random Fields. Wiley, New York (1981) MATHGoogle Scholar
  2. Adler, R.: On excursion sets, tube formulas and maxima of random fields. Ann. Appl. Probab. 10, 1–74 (2000) MathSciNetMATHGoogle Scholar
  3. Adler, R., Taylor, J.: Random Fields and Geometry. Springer, Boston (2007) MATHGoogle Scholar
  4. Aldous, D.: Probability Approximations via the Poisson Clustering Heuristics. Springer, New York (1989) Google Scholar
  5. Aronowich, M., Adler, R.: Sample path behaviour of χ 2 surfaces at extrema. Adv. Appl. Probab. 18, 901–920 (1988) CrossRefMathSciNetGoogle Scholar
  6. Barbujani, G., Oden, N.L., Sokal, R.R.: Detecting areas of abrupt change in maps of biological variables. Syst. Zool. 38, 376–389 (1989) CrossRefGoogle Scholar
  7. Banerjee, S., Gelfand, A.E.: Bayesian wombling: curvilinear gradient assessment under spatial process models. J. Am. Stat. Assoc. 101, 1487–1501 (2006) CrossRefMathSciNetMATHGoogle Scholar
  8. Banerjee, S., Gelfand, A.E., Sirmans, C.F.: Directional rates of change under spatial process models. J. Am. Stat. Assoc. 98, 946–954 (2003) CrossRefMathSciNetMATHGoogle Scholar
  9. Banerjee, S., Gelfand, A.E., Finley, A.O., Sang, H.: Gaussian predictive process models for large spatial datasets. J. R. Stat. Soc. Ser. B 70, 825–848 (2008) CrossRefMathSciNetMATHGoogle Scholar
  10. Cao, J.: The size of the connected components of excursion sets of χ 2, t and F fields. Adv. Appl. Probab. (SGSA) 31, 579–595 (1999) CrossRefMATHGoogle Scholar
  11. Chaudhuri, P., Marron, J.S.: SiZer for exploration of structure in curves. J. Am. Stat. Assoc. 94, 807–823 (1999) CrossRefMathSciNetMATHGoogle Scholar
  12. Chaudhuri, P., Marron, J.S.: Scale space view of curve estimation. Ann. Stat. 28, 408–428 (2000) CrossRefMathSciNetMATHGoogle Scholar
  13. Chilès, J.-P., Delfiner, P.: Geostatistics: Modeling Spatial Uncertainty. Wiley, New York (1999) MATHGoogle Scholar
  14. Cressie, N.: Statistics for Spatial Data. Wiley, New York (1993), revised edn. Google Scholar
  15. Fortin, M.-J.: Edge detection algorithms for two-dimensional ecological data. Ecology 75, 956–965 (1994) CrossRefGoogle Scholar
  16. Fortin, M.-J., Drapeau, P.: Delineation of Ecological boundaries: comparison of approaches and significance tests. Oikos 72, 323–332 (1995) CrossRefGoogle Scholar
  17. Gabriel, E.: Détection de zones de changement abrupt dans des données spatiales et applications à l’agriculture de précision. Ph.D. Dissertation, Université de Montpellier II (2004) Google Scholar
  18. Gabriel, E., Allard, D.: Evaluating the sampling pattern when detecting zones of abrupt change. Env. Ecol. Stat. 4, 469–489 (2008) CrossRefMathSciNetGoogle Scholar
  19. Gabriel, E., Allard, D., Guérif, M., Mary, B.: Detecting zones of abrupt change in soil data, with an application to an agricultural field. Eur. J. Soil Sci. 58, 1273–1284 (2007) CrossRefGoogle Scholar
  20. Godtliebsen, F., Marron, J.S., Pizer, S.M.: Significance in scale-space for clustering. In: Lawson, A.B., Denison, D. (eds.) Spatial Cluster Modelling, pp. 24–36. Chapman and Hall/CRC, Boca Raton (2002) Google Scholar
  21. Hall, P., Peng, L., Rau, C.: Local likelihood tracking of fault lines and boundaries. J. R. Stat. Soc. B 63, 569–582 (2001) MathSciNetMATHGoogle Scholar
  22. Liang, S., Banerjee, S., Carlin, B.P.: Bayesian wombling for spatial point processes. Biometrics (2009, in press) Google Scholar
  23. Mary, B., Beaudoin, N., Machet, J.M., Bruchou, C., Ariès, F.: Characterization and analysis of soil variability within two agricultural fields: the case of water and mineral N profiles. In: Grenier and Blackmore (eds.), Proc. 3rd European Conference on Precision Agriculture, pp. 431–436 (2001) Google Scholar
  24. Miller, K.S.: Multidimensional Gaussian Distribution. SIAM Series in Applied Mathematics (1964) Google Scholar
  25. Shen, X., Hsin-Cheng, H., Cressie, N.: Nonparametric hypothesis testing for a spatial signal. J. Am. Stat. Assoc. 97, 1122–1140 (2002) CrossRefMATHGoogle Scholar
  26. Stein, M.: Interpolation of Spatial Data: Some Theory for Kriging. Springer, Berlin (1999) MATHGoogle Scholar
  27. Womble, W.H.: Differential systematics. Science 114, 315–322 (1951) CrossRefGoogle Scholar
  28. Worsley, K.: Local maxima and the expected Euler characteristic of excursion sets of χ 2, F and t fields. Adv. Appl. Probab. 26, 13–42 (1994) CrossRefMathSciNetMATHGoogle Scholar
  29. Worsley, K.: Testing for signals with unknown location and scale in a χ 2 random field, with application to fMRI. Adv. Appl. Probab. (SGSA) 33, 773–793 (2001) CrossRefMathSciNetMATHGoogle Scholar
  30. Yaglom, A.M.: Correlation Theory of Stationary and Related Random Functions. Springer, Berlin (1986) Google Scholar

Copyright information

© Springer Science+Business Media, LLC 2009

Authors and Affiliations

  1. 1.Unité de Biostatistique et Processus SpatiauxInstitut National de la Recherche AgronomiqueAvignonFrance
  2. 2.I3M, UMR CNRS 5149Université Montpellier IIMontpellierFrance

Personalised recommendations