Uncovering regional climate patterns in Turkey using multivariate functional data analysis
Abstract
This study investigates regional climate patterns in Turkey through multivariate Functional Data Analysis (FDA) using daily observations recorded at 81 meteorological stations over 2012–2018 (T = 2557 days). Five coupled climate variables—mean temperature, total precipitation, temperature range, evaporation, and sunshine duration—are transformed into smooth functional trajectories using Fourier basis expansions with roughness control, where smoothing levels are selected via generalized cross-validation (GCV). To summarize joint seasonal variability and cross-variable dependence, we apply multivariate functional principal component analysis (MFPCA) and obtain low-dimensional score representations for each station. These scores are then used in a model-based clustering framework, with the number of climate regimes selected by the Bayesian Information Criterion (BIC). Finally, functional ANOVA (FANOVA) procedures are employed to evaluate whether the identified regimes differ significantly for each variable on the functional scale. The results reveal distinct data-driven climate regimes and indicate that regime separation is driven primarily by temperature range, evaporation, and sunshine duration, whereas differences in mean temperature and precipitation are comparatively weaker. Overall, the proposed FDA pipeline provides an interpretable and statistically principled framework for regional climate classification and functional inference, offering useful insights for climate monitoring and environmental modeling in Turkey.
Keywords
Climate patterns , Fourier basis , Functional clustering , Functional data analysis , Turkey
Kaynakça
- Abraham, C., Cornillon, P. A., Matzner-Løber, E., & Molinari, N. (2003). Unsupervised curve clustering using B-splines. Scandinavian Journal of Statistics, 30(3), 581–595.
- Boullé, M. (2012). Functional data clustering via piecewise constant nonparametric density estimation. Pattern Recognition, 45(12), 4389–4401.
- Bouveyron, C., & Jacques, J. (2011). Model-based clustering of time series in group specific functional subspaces. Advances in Data Analysis and Classification, 5(4), 281–300.
- Craven, P., & Wahba, G. (1979). Smoothing noisy data with spline functions. Numerische Mathematik, 31, 377–403.
- Ferraty, F., & Vieu, P. (2006). Nonparametric functional data analysis. Springer.
- Happ, C., & Greven, S. (2018). Multivariate functional principal component analysis for data observed on different (dimensional) domains. Journal of the American Statistical Association, 113(522), 649-659.
- Jacques, J., & Preda, C. (2014a). Functional data clustering: A survey. Advances in Data Analysis and Classification, 8(3), 231–255.
- Jacques, J., & Preda, C. (2014b). Model-based clustering for multivariate functional data. Computational Statistics & Data Analysis, 71, 92–106.
- Ramsay, J. O., & Silverman, B. W. (2002). Applied functional data analysis: Methods and case studies. Springer.
- Ramsay, J. O., & Silverman, B. W. (2005). Functional data analysis (2nd ed.). Springer.