Objective: In modeling environment processes, multi- disciplinary methods are used to explain, explore and predict how the earth responds to natural human-induced environmental changes over time. Consequently, when analyzing spatial processes spatial domains, the spatial covariance of interest are always heterogeneous. However, this article proposed locally adaptive covariance for the spatial domain whose covariance is nonstationary in their spatial domain. The objectives of the study are to propose parametric, non-parametric and semi- parametric models for nonstationary spatial structure, continuous model for nonstationary spatial processes whose distance is far apart and to propose the adaptive weighting scheme approach that generates the optimal value for the nonparametric and semi-parametric models. Material and Methods: The spatial covariances are derived by applying the concept of adaptive weighting scheme approach on the covariance proposed in Nott and Dunsmuir (2002). Consequently, the local adaptive bandwidth for the nonstationary covariance was obtained for both the nonparametric and semi-parametric models. Simulations are conducted on the proposed model to examine the proposed model. Results and Conclusion: The results obtained are compared with existing models. The results indicate proposed spatial covariance are driven by the local bandwidths, penalty, weighted scheme, and tuning parameters. The adaptive models performed better in relation to existing covariances in terms of their mean square prediction errors (MSPE). The proposed models were further applied to real life Sulphate spatial data.
Keywords: Adaptive; locally, nonstationary; spatial covariance; variability
Amaç: Çevresel süreçleri modellerken, dünyanın zaman içinde doğal insan kaynaklı çevresel değişikliklere nasıl tepki verdiğini açıklamak, araştırmak ve öngörmek için multi-disipliner yöntemler kullanılır. Sonuç olarak, uzamsal süreçleri incelerken uzamsal alanlar, ilgilenilen uzamsal kovaryans her zaman heterojendir. Bununla birlikte, bu makalede kovaryansı uzamsal alanlarında durağan olmayan uzamsal alan için yerel olarak uyarlanabilir kovaryans önerilmiştir. Makalenin amaçları durağan olmayan uzamsal yapılar, mesafesi çok uzak olan durağan olmayan uzamsal süreçler için sürekli model için parametrik, non-parametrik ve yarı-parametrik modeller önermek ve, non-parametrik ve yarı-parametrik modeller için optimal değeri yaratan adaptif uyarlamalı ağırlıklandırma şeması önermektir. Gereç ve Yöntemler: Uzamsal kovaryanslar, Nott ve Dunsmuir (2002)'de önerilen uyarlamalı ağırlıklandırma şeması yaklaşımı kovaryans üzerine uygulanarak türetilmiştir. Sonuç olarak, hem non-parametrik hem de yarı parametrik modeller için durağan olmayan kovaryans için yerel adaptif band genişliği elde edildi. Önerilen modeli değerlendirmek için önerilen model üzerine simulasyonlar yapıldı. Bulgular ve Sonuç: Elde edilen sonuçlar mevcut modellerle karşılaştırıldı. Bulgular önerilen uzamsal kovaryansın, yerel bant genişlikleri, ceza, ağırlıklı şema ve ayar parametreleri tarafından yönlendirildiğini göstermektedir. Uyarlanabilir modeller, ortalama karesel öngörü hataları (MSPE) açısından mevcut kovaryanslara göre daha iyi performans göstermiştir. Önerilen modeller ayrıca gerçek hayattaki sülfat uzamsal verilerine de uygulanmıştır.
Anahtar Kelimeler: Uyarlanabilir; yerel, durağan olmayan; uzamsal kovaryans; değişkenlik
- Guttorp P, Sampson P. Methods for estimating heterogeneous spatial covariance functions with environmental applications, Hand-book of statistics Wiley, 1994; volume 12. [Crossref]
- Fuentes M. A new high frequency kriging approach for nonstationary environmental processes. Environmetrics, 2001; 12:469-483. [Crossref]
- Nott DJ, Dunsmuir WT M. Estimation of nonstationary spatial covariance structure. Biometrika Trust, 2002; 89(4):819-829. [Crossref]
- Shand L, Li B. Modeling nonstationarity in space and time. Biometrics, 2017; 1-10. [Crossref] [PubMed] [PMC]
- Sampson PD, Guttorp P. Nonparametric estimation of nonstationary spatial covariance structure. Journal of the American Statistical Association, 1992; 87:108-119. [Crossref]
- Schmidt AM, O'Hagan, A. Bayesian inference for non- stationary spatial covariance structure via spatial deformations. JRSSB, 2003; 65:743-758. [Crossref]
- Aberg S, Lindgren F, Malmberg A, Holst J, Holst U. An image warping approach to spatio-temporal modeling. Environmetrics, 2005; 16:833-848. [Crossref]
- Barry RP, Ver-Hoef J. Blackbox kriging: Spatial prediction without specifying variogram models. Journal of Agricultural, Biological, and Environmental Statistics, 1996; 1:297-322. [Crossref]
- Higdon D. A process convolution approach to modelling temperatures in the North Atlantic. Journal of Environmental Engineering and Science, 1998; 5:173-190. [Crossref]
- Higdon DM, Swall J, Kern J. Nonstationary spatial modeling. Bayesian Statistics, 1999; 6:761-768.
- Risser DM, Calder AC. Regression based covariance functions for nonstationary spatial modeling. Environmetrics, 2015; 26:284-297. [Crossref]
- Fuentes M. Spectral methods for nonstationary spatial processes. Biometrika Trust, 2002; 89:197-210. [Crossref]
- Paciorek CJ, Schervish MJ. Nonstationary covariance functions for Gaussian process regression. Advances in Neural Information Processing Systems, 2004; 16(16):273-280.
- Bornn L, Shaddick G, Zidek JV. Modeling nonstationary processes through dimension expansion. Journal of the American Statistical Association, 2012; 107:281-289. [Crossref]
- Schmidt MA, Guttorp P, O'Hagan A. Considering covariates in the covariance structure of spatial processes. Environmetrics, 2011; 22:487-500. [Crossref]
- Meiring W, Guttorp P, Sampson P. Space-time estimation of grid-cell hourly ozone levels for assessment of a deterministic model. Environmental and Ecological Statistics, 1998; 5:197-222. [Crossref]
- Le N, Sun L, Zidek J. Spatial prediction and temporal backcasting for environmental fields having monotone data patterns. The Canadian Journal of Statistics, 2001; 29:529-554. [Crossref]
- Damian D, Sampson P, Guttorp P. Bayesian estimation of semi-parametric nonstationary spatial covariance structures. Environmetrics, 2001; 12:161-178. [Crossref]
- Guttorp P, Meiring W, Sampson P. A space-time analysis of ground level ozone data. Environmetrics, 1994; 5(3):241-254. [Crossref]
- Ingebrigtsen R, Finn L, Ingelin S. Spatial models with explanatory variables in the dependence structure. Spatial Statistics, 2014; 8:20-38. [Crossref]
- Huang C, Hsing T, Cressie N. Nonparametric estimation of the variogram and its spectrum. Biometrika, 2011; 98:775-789. [Crossref]
- Genton GM, David JG. Non-parametric variogram and covariogram estimation with fourier-bassel matrices. Journal of Computational Statistics and Data Analysis, 2002; 41:47-57. [Crossref]
- Im H, Stein ML, Zhu Z. Semi-parametric estimation of spectral density with irregular observations. Journal of the American Statistical Association, 2007; 102:726-735. [Crossref]
- Cressie N, Johannesson G. Fixed rank kriging for very large spatial data sets. Journal of the Royal Statistical Society, Series B (Statistical Methodology), 2008; 70:209-226. [Crossref]
- Banerjee S, Gelfand AE, Finley AO, Sang H. Gaussian predictive models for large spatial datasets. Journal of the Royal Statsitical Society, Series B, 2008; 70:825-848. [Crossref] [PubMed] [PMC]
- Eidsvik J, Finley AO, Banerjee S, Rue H. Approximate bayesian inference for large spatial datasets using predictive process models. Computational Statistics and Data Analysis, 2012; 56:1362-1380. [Crossref]
- Ren Q, Banerjee S. Hierarchical factor models for large spatially misaligned datasets: a low-rank predictive process approach. Biometrics, 2013; 69:19-30. [Crossref] [PubMed] [PMC]
- Katzfuss, M. Bayesian nonstationary spatial modeling for very large datasets. Environmetrics, 2013; 24:189-200. [Crossref]
- Nychka D, Bandyopadhyay S, Hammerling D, Lindgren F, Sain S. A multiresolution gaussian process model for the analysis of large spatial datasets. Journal of Computational and Graphical Statistics, 2015; 24:579-599. [Crossref]
- Choi I, Li B, Wang X. Nonparametric estimation of spatial and space-time covariance function. Journal of Agricultural, Biological, and Environmental Statistics, 2013; 18(4):1-20. [Crossref]
- Cressie N. Fitting variogram models by weighted least squares. Mathematical Geology, 1985; 17:563-586. [Crossref]
- Gilmour A, Cullis B, Welham S, Gogel B, Thompson R. An efficient computing strategy for prediction in mixed linear models. Computational Statistics and Data Analysis, 2004; 44:571-586. [Crossref]
- Cherry S, Banfield J, Quimby W.F. An evaluation of a nonparametric method of estimating semi-variograms of isotropic spatial process. Journal of Applied Statistics, 1996, 23(4):435-449. [Crossref]
- Shapiro A, Botha JD. Variogram fitting with a general class of conditionally nonnegative definite functions. Computational Statistics and Data Analysis, 1991; 11:87-96. [Crossref]
.: Process List