Bootstrap Methods for Statistical Inference. Part II: Extreme-Value Analysis

Gilleland, Eric

doi:10.1175/jtech-d-20-0070.1

Cited by 20 publications

(9 citation statements)

References 23 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Quantile regression is a straightforward approach for practitioners since it shares similar theoretical background and implementation as traditional regression models. Other perspectives are through (1) bootstrapbased approaches (Gilleland, 2020) and (2) approaches based on the generalized extreme value (GEV) or threshold exceedance (e.g., generalized Pareto) models (Berrocal et al, 2014;Stein, 2017;Opitz et al, 2018). Bootstrap is a resampling procedure that can be used for estimating the sampling distribution about the trends and/or their uncertainty.…”

Section: Discussion Of Further Advanced Techniquesmentioning

confidence: 99%

Trend detection of atmospheric time series

Chang

Schultz

et al. 2021

Elementa: Science of the Anthropocene

View full text Add to dashboard Cite

This paper is aimed at atmospheric scientists without formal training in statistical theory. Its goal is to (1) provide a critical review of the rationale for trend analysis of the time series typically encountered in the field of atmospheric chemistry, (2) describe a range of trend-detection methods, and (3) demonstrate effective means of conveying the results to a general audience. Trend detections in atmospheric chemical composition data are often challenged by a variety of sources of uncertainty, which often behave differently to other environmental phenomena such as temperature, precipitation rate, or stream flow, and may require specific methods depending on the science questions to be addressed. Some sources of uncertainty can be explicitly included in the model specification, such as autocorrelation and seasonality, but some inherent uncertainties are difficult to quantify, such as data heterogeneity and measurement uncertainty due to the combined effect of short and long term natural variability, instrumental stability, and aggregation of data from sparse sampling frequency. Failure to account for these uncertainties might result in an inappropriate inference of the trends and their estimation errors. On the other hand, the variation in extreme events might be interesting for different scientific questions, for example, the frequency of extremely high surface ozone events and their relevance to human health. In this study we aim to (1) review trend detection methods for addressing different levels of data complexity in different chemical species, (2) demonstrate that the incorporation of scientifically interpretable covariates can outperform pure numerical curve fitting techniques in terms of uncertainty reduction and improved predictability, (3) illustrate the study of trends based on extreme quantiles that can provide insight beyond standard mean or median based trend estimates, and (4) present an advanced method of quantifying regional trends based on the inter-site correlations of multisite data. All demonstrations are based on time series of observed trace gases relevant to atmospheric chemistry, but the methods can be applied to other environmental data sets.

show abstract

Section: Discussion Of Further Advanced Techniquesmentioning

confidence: 99%

Trend detection of atmospheric time series

Chang

Schultz

et al. 2021

Elementa: Science of the Anthropocene

View full text Add to dashboard Cite

show abstract

“…(2022) used an inverse of standard error of the sample mean to represent the uncertainty associated with sample frequency and data variability, but this statistic is not necessarily representative of the uncertainty of the extreme percentiles (especially ozone data, which is often non‐normally distributed), and might not be appropriate for weighting extreme percentiles from different sources of data sets. Even though the uncertainty of extreme percentiles can be assessed through bootstrap‐based approaches (Gilleland, 2020; Kyselỳ, 2008), the weighting scheme for extreme values is currently absent and has not been thoroughly investigated. Therefore, in this study, we treat each data set equally when fusing the percentiles from different data sets.…”

Section: Methodsmentioning

confidence: 99%

Diverging Ozone Trends Above Western North America: Boundary Layer Decreases Versus Free Tropospheric Increases

et al. 2023

View full text Add to dashboard Cite

From a climate change perspective, ozone is the third most important greenhouse gas (after CO 2 and CH 4 ) and its radiative forcing is proportional to the global tropospheric ozone burden, which contributes to global warming on relatively long timescales of decades (see Chapter 7 of IPCC AR6, 2021). Therefore, long-term ozone trends at remote sites and in the free troposphere are typically based on observations from all months of the year, in addition to seasonal-based analyses (

show abstract

“…The best‐fit nonstationary GEV model is used to estimate design precipitation for selected return periods. The confidence interval on the design estimates (for both stationary and nonstationary models) is estimated using non‐parametric bootstrap resampling method explained by Gilleland (2020). In the procedure, non‐parametric resampling is performed on the fitted GEV distribution to obtain sampled set of extremes for making statistical inference.…”

Section: Methodsmentioning

confidence: 99%

An Integrative Framework for Assessment of Urban Flood Response to Changing Climate

Pal,

Saksena,

Dey

et al. 2023

Water Resources Research

View full text Add to dashboard Cite

Increasing frequency of extreme rainfall induced catastrophic urban flood events in recent decades demands proactive efforts to assess flood risk and vulnerability. Here, we develop an integrative framework for fine spatiotemporal scale assessment of urban flood response to historical and future projected changes in extreme precipitation. The framework includes three main components – nonstationary modelling of historical extreme precipitation, modelling of future precipitation, and urban flood simulations. It also provides robust estimates of uncertainty in design precipitation from statistical modelling, multiple climate models and stochastic uncertainty, estimated using machine learning techniques. We demonstrate the proposed framework for White Oak Bayou watershed in Houston, Texas, US. Two‐dimensional hydrologic‐hydraulic Interconnected Channel and Pond Routing (ICPR) model is used to simulate flood response from design precipitation for historical (1986–2020) and two future (2021–2050 & 2071–2100) periods in three (SSP1‐2.5, SSP2‐4.5 and SSP5‐8.5) future climate scenarios. Results show that nonstationary design estimates for historical precipitation are 14–25% higher than the stationary estimates for 100‐year event of 1‐hour to 24‐hour duration. Contrary to general global trends, we found a significant reduction in future design precipitation in all three emission scenarios. Additionally, stochastic uncertainty in future design precipitations is found to be larger than the modelling uncertainty in historical estimates and climate model uncertainty in future estimates. Flood response in terms of peak flood and total flood volume suggests that the difference in stationary vs. nonstationary historical and future design precipitation are substantial to cause considerable change in simulated flood.

show abstract

Bootstrap Methods for Statistical Inference. Part II: Extreme-Value Analysis

Cited by 20 publications

References 23 publications

Trend detection of atmospheric time series

Trend detection of atmospheric time series

Diverging Ozone Trends Above Western North America: Boundary Layer Decreases Versus Free Tropospheric Increases

An Integrative Framework for Assessment of Urban Flood Response to Changing Climate

Contact Info

Product

Resources

About