Bayesian change-point analysis in hydrometeorological time series. Part 2. Comparison of change-point models and forecasting

Perreault, Luc; Bernier, Jacques; Bobée, Bernard; Parent, Éric

doi:10.1016/s0022-1694(00)00271-7

Cited by 95 publications

(62 citation statements)

References 17 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Several tests to detect the presence of abrupt changes in the mean of the distribution of the variable of interest have been proposed and developed (Potter, 1981;Buishand, 1984;Perreault et al, 2000;Rodionov, 2004;Reeves et al, 2007;Wang et al, 2007;Beaulieu et al, 2009), each based on different assumptions, such as normality of the data, prior knowledge or not of the year of the changepoint, single versus multiple change-points. In this study, the presence of change-points in the mean and variance of the annual maximum discharge is examined using the Pettitt test (Pettitt, 1979).…”

Section: Change-point and Trend Analysesmentioning

confidence: 99%

“…The test significance is computed using the limiting distribution for continuous distribution in Pettitt (1979). The vast majority of the change-point tests in the literature is designed to detect abrupt changes in the mean of the distribution of the variable of interest, and only a limited number can detect abrupt changes in variance (Perreault et al, 2000). However, abrupt changes in variance can have a large impact on the frequency of extreme events (Katz and Brown, 1992;Meehl et al, 2000;Ferro et al, 2005), making their detection an important step when examining the stationarity of a time series.…”

Section: Change-point and Trend Analysesmentioning

confidence: 99%

See 1 more Smart Citation

Analyses of extreme flooding in Austria over the period 1951–2006

Villarini

Smith

Serinaldi

et al. 2011

Intl Journal of Climatology

View full text Add to dashboard Cite

ABSTRACT:Analyses of extreme flooding in Austria is performed using daily discharge time series from 27 stations over the period . The main research questions revolve around: (1) temporal non-stationarities in the flood record, (2) upper tail and scaling properties of the flood peak records, and (3) relation between magnitude and frequency of flooding and the North Atlantic Oscillation (NAO). Two datasets are derived from the daily discharge time series: annual maximum daily discharge and peaks-over-threshold (POT) data. The validity of the stationarity assumption in the annual maximum discharge record is assessed by investigating the presence of abrupt and slowly varying changes using nonparametric tests. The time series are tested for abrupt changes both in the mean and variance of the flood peak distributions by means of the Pettitt test. The presence of monotonic trends is investigated by means of the Mann-Kendall and Spearman tests. Violations of the stationarity assumption are associated with abrupt rather than gradual changes. These step changes generally involve river regulation through construction of dams or other major engineering works. It is not possible to make conclusive statements about the presence of an anthropogenic climate change signal in the flood peak record. Similar conclusions are obtained when focussing on the frequency of POT floods. The Generalised Extreme Value distribution is used to study the upper tail and scaling properties of annual maximum daily discharge records. The location and scale parameters exhibit power-law behaviour as a function of drainage area. The shape parameters indicate that the flood peak distributions for Austria have a heavy tail. Non-stationary modelling of the annual maximum daily discharge and POT time series is used to explore the relation between flood magnitude and frequency and NAO. The results indicate that NAO is a significant covariate in explaining the magnitude and frequency of occurrence of flooding over a large part of Austria.

show abstract

Section: Change-point and Trend Analysesmentioning

confidence: 99%

Section: Change-point and Trend Analysesmentioning

confidence: 99%

Analyses of extreme flooding in Austria over the period 1951–2006

Villarini

Smith

Serinaldi

et al. 2011

Intl Journal of Climatology

View full text Add to dashboard Cite

show abstract

“…The statistical inferences on the unknown parameters with regards to change point location can then be made based on this posterior distribution. Bayesian approach in change identification problem for mean level in time series data have been used by previous researchers such as by Smith (1975), Lee and Heighinian (1977), Booth and Smith (1982), Perreault et al (2000a;2000b) and Kim et al (2009). The Bayesian approach in this study is based on a single shifting model and takes into consideration non-informative prior distributions on the unknown change point.…”

Section: Methodsmentioning

confidence: 99%

Bayesian Changepoint Analysis of the Extreme Rainfall Events

A.¹

2012

Journal of Mathematics and Statistics

View full text Add to dashboard Cite

Problem statement:This study assesses recent changes in extremes of annual rainfall in Peninsular Malaysia based on daily rainfall data for 50 rain-gauged stations over the period . Approach: Eight indices that represent the extreme events are defined and analyzed, which are extreme Dry Spell (XDS), extreme Rain-Sum (XRS), extreme wet-day intensities at 95% and 99% percentiles (I95 and I99), proportion of extreme rainfall amount to the total rainfall amount (R95& R99) and frequency of extreme wet-day at 95 and 99% percentiles (N95 and N99). Bayesian approach based on a single shifting model is used to investigate the change in the mean level of these extreme rainfall indices. The detection on whether the change has occurred or not is analyzed followed by the estimation of the location of change point. Results: The results of the analysis showed that half of the stations considered displayed significant changes. The analysis also found that in general, the changes occurred in the early 90s. More than 75% of the stations which recorded significant changes are situated on the west coast of the peninsula. Conclusion/Recommendations: The west coast of Peninsular Malaysia displays more significant changes in trend especially at stations located in urban areas compared to the east coast of the peninsula. In terms of the Bayesian methods used, the existence of any outlier in the data series may influence the result since the analysis is based on mean value which is very sensitive to any outlier.

show abstract

“…Taking inspiration from other research fields such as hydrology (Rao, 1996;Perreault et al, 2000aPerreault et al, , 2000b and avalanche science (Eckert et al, 2007(Eckert et al, , 2008 in which dealing with stochasticity is more common, this paper aims at defining a bouncing model explicitly distinguishing these different sources of variability.…”

Section: F Bourrier Et Al: Stochastic Bouncing Modelmentioning

confidence: 99%

Bayesian stochastic modeling of a spherical rock bouncing on a coarse soil

Bourrier¹,

Eckert²,

Nicot³

et al. 2009

Nat. Hazards Earth Syst. Sci.

View full text Add to dashboard Cite

Abstract. Trajectory analysis models are increasingly used for rockfall hazard mapping. However, classical approaches only partially account for the variability of the trajectories. In this paper, a general formulation using a Taylor series expansion is proposed for the quantification of the relative importance of the different processes that explain the variability of the reflected velocity vector after bouncing. A stochastic bouncing model is obtained using a statistical analysis of a large numerical data set. Estimation is performed using hierarchical Bayesian modeling schemes. The model introduces information on the coupling of the reflected and incident velocity vectors, which satisfactorily expresses the mechanisms associated with boulder bouncing.The approach proposed is detailed in the case of the impact of a spherical boulder on a coarse soil, with special focus on the influence of soil particles' geometrical configuration near the impact point and kinematic parameters of the rock before bouncing. The results show that a first-order expansion is sufficient for the case studied and emphasize the predominant role of the local soil properties on the reflected velocity vector's variability. The proposed model is compared with classical approaches and the interest for rockfall hazard assessment of reliable stochastic bouncing models in trajectory simulations is illustrated with a simple case study.

show abstract

Bayesian change-point analysis in hydrometeorological time series. Part 2. Comparison of change-point models and forecasting

Cited by 95 publications

References 17 publications

Analyses of extreme flooding in Austria over the period 1951–2006

Analyses of extreme flooding in Austria over the period 1951–2006

Bayesian Changepoint Analysis of the Extreme Rainfall Events

Bayesian stochastic modeling of a spherical rock bouncing on a coarse soil

Contact Info

Product

Resources

About