×

INLA goes extreme: Bayesian tail regression for the estimation of high spatio-temporal quantiles. (English) Zbl 1407.62167

Summary: This work is motivated by the challenge organized for the 10th International Conference on Extreme-Value Analysis (EVA2017) to predict daily precipitation quantiles at the \(99.8\%\) level for each month at observed and unobserved locations. Our approach is based on a Bayesian generalized additive modeling framework that is designed to estimate complex trends in marginal extremes over space and time. First, we estimate a high non-stationary threshold using a gamma distribution for precipitation intensities that incorporates spatial and temporal random effects. Then, we use the Bernoulli and generalized Pareto (GP) distributions to model the rate and size of threshold exceedances, respectively, which we also assume to vary in space and time. The latent random effects are modeled additively using Gaussian process priors, which provide high flexibility and interpretability. We develop a penalized complexity (PC) prior specification for the tail index that shrinks the GP model towards the exponential distribution, thus preventing unrealistically heavy tails. Fast and accurate estimation of the posterior distributions is performed thanks to the integrated nested Laplace approximation (INLA). We illustrate this methodology by modeling the daily precipitation data provided by the EVA2017 challenge, which consist of observations from 40 stations in the Netherlands recorded during the period 1972–2016. Capitalizing on INLA’s fast computational capacity and powerful distributed computing resources, we conduct an extensive cross-validation study to select the model parameters that govern the smoothness of trends. Our results clearly outperform simple benchmarks and are comparable to the best-scoring approaches of the other teams.

MSC:

62G32 Statistics of extreme values; tail inference
62F15 Bayesian inference
62M30 Inference from spatial processes
62P12 Applications of statistics to environmental and related topics
PDFBibTeX XMLCite
Full Text: DOI arXiv

References:

[1] Bacro, J.N., Gaetan, C., Opitz, T., Toulemonde, G.: Hierarchical space-time modeling of exceedances with an application to rainfall data, arXiv:1708.02447 (2017)
[2] Balkema, AA; Haan, L., Residual life time at great age, Ann. Probab., 2, 792-804, (1974) · Zbl 0295.60014 · doi:10.1214/aop/1176996548
[3] Banerjee, S., Carlin, B.P., Gelfand, A.E.: Hierarchical Modeling and Analysis for Spatial Data. Chapman & Hall/CRC, Boca Raton (2004) · Zbl 1053.62105
[4] Bopp, Gregory P.; Shaby, Benjamin A., An exponential-gamma mixture model for extreme santa ana winds, Environmetrics, 28, e2476, (2017) · doi:10.1002/env.2476
[5] Casson, E.; Coles, S., Spatial regression models for extremes, Extremes, 1, 449-468, (1999) · Zbl 0935.62109 · doi:10.1023/A:1009931222386
[6] Castro Camilo, D., Huser, R.: Local likelihood estimation of complex tail dependence structures in high dimensions, applied to US precipitation extremes, arXiv:1710.00875 (2017)
[7] Chavez-Demoulin, V.; Davison, AC, Generalized additive modelling of sample extremes, J. R. Stat. Soc.: Ser. C: Appl. Stat., 54, 207-222, (2005) · Zbl 1490.62194 · doi:10.1111/j.1467-9876.2005.00479.x
[8] Coles, S.: An Introduction to Statistical Modeling of Extreme Values. Springer, London (2001) · Zbl 0980.62043 · doi:10.1007/978-1-4471-3675-0
[9] Coles, SG; Tawn, JA, Modelling extremes of the areal rainfall process, J. R. Stat. Soc.: Ser. B: Stat. Methodol., 58, 329-347, (1996) · Zbl 0863.60041
[10] Cooley, D.; Nychka, D.; Naveau, P., Bayesian spatial modeling of extreme precipitation return levels, J. Am. Stat. Assoc., 102, 824-840, (2007) · Zbl 1469.62389 · doi:10.1198/016214506000000780
[11] Daouia, A.; Girard, S.; Stupfler, G., Estimation of tail risk based on extreme expectiles, J. R. Stat. Soc.: Ser. B: Stat. Methodol., 80, 263-292, (2018) · Zbl 1383.62235 · doi:10.1111/rssb.12254
[12] Davis, RA; Klüppelberg, C.; Steinkohl, C., MAX-stable processes for modeling extremes observed in space and time, J. Kor. Stat. Soc., 42, 399-414, (2013) · Zbl 1294.62118 · doi:10.1016/j.jkss.2013.01.002
[13] Davison, AC; Huser, R., Statistics of extremes, Annu. Rev. Stat. Appl., 2, 203-235, (2015) · doi:10.1146/annurev-statistics-010814-020133
[14] Davison, AC; Ramesh, N., Local likelihood smoothing of sample extremes, J. R. Stat. Soc.: Ser. B: Stat. Methodol., 62, 191-208, (2000) · Zbl 0942.62058 · doi:10.1111/1467-9868.00228
[15] Davison, AC; Smith, RL, Models for exceedances over high thresholds (with discussion), J. R. Stat. Soc.: Ser. B: Stat. Methodol., 52, 393-442, (1990) · Zbl 0706.62039
[16] Hosking, JR; Wallis, JR, Parameter and quantile estimation for the generalized Pareto distribution, Technometrics, 29, 339-349, (1987) · Zbl 0628.62019 · doi:10.1080/00401706.1987.10488243
[17] Huser, R.; Davison, AC, Space-time modelling of extreme events, J. R. Stat. Soc.: Ser. B: Stat. Methodol., 76, 439-461, (2014) · doi:10.1111/rssb.12035
[18] Huser, R., Wadsworth, J.L.: Modeling spatial processes with unknown extremal dependence. J. Am. Stat. Assoc. To appear (2018)
[19] Huser, R.; Opitz, T.; Thibaud, E., Bridging asymptotic independence and dependence in spatial extremes using Gaussian scale mixtures, Spatial Stat., 21, 166-186, (2017) · doi:10.1016/j.spasta.2017.06.004
[20] Jonathan, P.; Randell, D.; Wu, Y.; Ewans, K., Return level estimation from non-stationary spatial data exhibiting multidimensional covariate effects, Ocean Eng., 88, 520-532, (2014) · doi:10.1016/j.oceaneng.2014.07.007
[21] Katz, RW; Parlange, M.; Naveau, P., Extremes in hydrology, Adv. Water Resour., 25, 1287-1304, (2002) · doi:10.1016/S0309-1708(02)00056-8
[22] Koenker, R.: Quantile Regression. Cambridge University Press, Cambridge (2005) · Zbl 1111.62037 · doi:10.1017/CBO9780511754098
[23] Lindgren, FK; Rue, H., A note on the second order random walk model for irregular locations, Scand. J. Stat., 35, 691-700, (2008) · Zbl 1199.60276 · doi:10.1111/j.1467-9469.2008.00610.x
[24] Lindgren, FK; Rue, H., Bayesian spatial modelling with R-INLA, J. Stat. Softw., 63, 1-25, (2015) · doi:10.18637/jss.v063.i19
[25] Lindgren, FK; Rue, H.; Lindström, J., An explicit link between gaussian fields and Gaussian Markov random fields: the stochastic partial differential equation approach (with discussion), J. R. Stat. Soc.: Ser. B: Stat. Methodol., 73, 423-498, (2011) · Zbl 1274.62360 · doi:10.1111/j.1467-9868.2011.00777.x
[26] Martins, TG; Simpson, DP; Lindgren, FK; Rue, H., Bayesian computing with INLA: new features, Comput. Stat. Data Anal., 67, 68-83, (2013) · doi:10.1016/j.csda.2013.04.014
[27] Naveau, P.; Huser, R.; Ribereau, P.; Hannart, A., Modeling jointly low, moderate, and heavy rainfall intensities without a threshold selection, Water Resour. Res., 52, 2753-2769, (2016) · doi:10.1002/2015WR018552
[28] Northrop, PJ; Jonathan, P., Threshold modelling of spatially-dependent non-stationary extremes with application to hurricane-induced wave heights (with discussion), Environmetrics, 22, 799-809, (2011) · doi:10.1002/env.1106
[29] Opitz, T., Latent Gaussian modeling and INLA: a review with focus on space-time applications, J. French Statistical Society, 158, 62—85, (2017) · Zbl 1378.62095
[30] Papastathopoulos, I.; Tawn, JA, Extended generalised Pareto models for tail estimation, Journal of Statistical Planning and Inference, 143, 131-143, (2013) · Zbl 1251.62020 · doi:10.1016/j.jspi.2012.07.001
[31] Pickands, J., Statistical inference using extreme order statistics, Ann. Stat., 3, 119-131, (1975) · Zbl 0312.62038 · doi:10.1214/aos/1176343003
[32] Rue, H., Held, L.: Gaussian Markov Random Fields: Theory and Applications. Chapman & Hall/CRC, London (2005) · Zbl 1093.60003 · doi:10.1201/9780203492024
[33] Rue, H.; Martino, S.; Chopin, N., Approximate Bayesian inference for latent Gaussian models by using integrated nested Laplace approximations, J. R. Stat. Soc. (Ser. B), 71, 319-392, (2009) · Zbl 1248.62156 · doi:10.1111/j.1467-9868.2008.00700.x
[34] Rue, H.; Riebler, A.; Sørbye, SH; Illian, JB; Simpson, DP; Lindgren, FK, Bayesian computing with INLA: A review, Annu. Rev. Stat. Appl., 4, 395-421, (2017) · doi:10.1146/annurev-statistics-060116-054045
[35] Scarrott, C.; MacDonald, A., A review of extreme value threshold estimation and uncertainty quantification, REVSTAT, 10, 33-60, (2012) · Zbl 1297.62120
[36] Simpson, DP; Rue, H.; Riebler, A.; Martins, TG; Sørbye, SH, Penalising model component complexity: A principled, practical approach to constructing priors, Statistical Science, 32, 1-28, (2017) · Zbl 1442.62060 · doi:10.1214/16-STS576
[37] Stein, M.L.: Interpolation of Spatial Data: Some Theory for Kriging. Springer, New York (1999) · doi:10.1007/978-1-4612-1494-6
[38] Thibaud, E.; Opitz, T., Efficient inference and simulation for elliptical Pareto processes, Biometrika, 102, 855-870, (2015) · Zbl 1372.62011 · doi:10.1093/biomet/asv045
[39] Thibaud, E.; Mutzner, R.; Davison, AC, Threshold modeling of extreme spatial rainfall, Water Resources Research, 49, 4633-4644, (2013) · doi:10.1002/wrcr.20329
[40] Tierney, L.; Kadane, JB, Accurate approximations for posterior moments and marginal densities, J. Am. Stat. Assoc., 81, 82-86, (1986) · Zbl 0587.62067 · doi:10.1080/01621459.1986.10478240
[41] Turkman, KF; Turkman, MAA; Pereira, JM, Asymptotic models and inference for extremes of spatio-temporal data, Extremes, 13, 375-397, (2010) · Zbl 1226.60083 · doi:10.1007/s10687-009-0092-8
[42] Wadsworth, JL; Tawn, JA, Dependence modelling for spatial extremes, Biometrika, 99, 253-272, (2012) · Zbl 1318.62160 · doi:10.1093/biomet/asr080
[43] Wilks, D.: Statistical Methods in the Atmospheric Sciences. Elsevier, Oxford (2006)
[44] Wintenberger, O.: Editorial: special issue on the Extreme Value Analysis conference challenge “Prediction of extremal precipitation”. Extremes To appear (2018)
This reference list is based on information provided by the publisher or from digital mathematics libraries. Its items are heuristically matched to zbMATH identifiers and may contain data conversion errors. In some cases that data have been complemented/enhanced by data from zbMATH Open. This attempts to reflect the references listed in the original paper as accurately as possible without claiming completeness or a perfect matching.