Environmental Statistics

DOI10.1002/0470024747MaRDI QIDQ4828656zbMATH OpenOpenAlexFDO

Publication date 26 November 2004

Published in Wiley Series in Probability and Statistics (Search for Journal in Brave)

Full work available at URL https://doi.org/10.1002/0470024747

Many areas of statistical methodology and modelling find applications in environmental problems. Some notable emphases and needs are as follows: The study of extremes, outliers and robust inference methods is relevant to so many fields of inquiry, none more so than environmental issues. Particular modern sampling methods have special relevance and potential in many fields of environmental studies; they are important in monitoring and in standard-setting. For example, ranked-set sampling aims for high-efficiency inference, where observational data are expensive, by exploiting associated (concomitant, often `expert-opinion') information to spread sample coverage. Composite sampling seeks to identify rare conditions and form related inferences again where sampling is costly and where sensitivity issues arise, whilst adaptive sampling for elusive outcomes and rare events modifies the sampling scheme progressively as the sample is collected. Other topics such as size-biasing, transect sampling and capture-recapture also find wide applications in environmental studies. Linear and generalized linear models play a central role in statistical investigations across all areas of environmental applications. Further developments are needed, particularly in relation to multivariate correlated data, random dependent variables, extreme values, outliers, complex error structure, etc., with special relevance to environmental, economic and social issues. Risk evaluation and uncertainty analysis are modern thrusts which still need more careful definition and fuller investigation to formalize their pivotal roles and to elucidate distinctions with conventional statistical concepts and methods. Applications of special relevance occur across the environmental spectrum. Temporal and spatial models have clear and ubiquitous relevance; all processes vary in time and space. Time-series methods have been widely applied and developed for environmental problems but more research is needed on non-stationary and multivariate structures, on outliers and on non-parametric approaches. Spatial methods need further major research development, including concern for the highly correlated and multivariate base of most application fields. Conjoint spatial-temporal models and methods are not well developed and are ripe for major advances in research. In this book, the author reviews and represents the wide range of applications and statistical topics in environmental statistics. From the Contents: Extremal stresses: extremes, outliers, robustness. Ordering and extremes: applications, models, order-based inference, extremal processes, practical study and empirical models, generalized extreme-value distributions. Peaks over thresholds and the generalized Pareto distribution. Outliers and robustness. Discordancy and models for outlier generation. Multiple outliers: masking and swamping. Accommodation: outlier-robust methods. Outliers in linear models. Collecting environmental data: sampling and monitoring. Finite-population sampling. Estimating mean and variance. Ratios and ratio estimators. Stratified (simple) random sampling. Comparing the simple random sample mean and the stratified sample mean. Comparison of proportional allocation and optimum allocation. Optimum allocation for estimating proportions. Developments of survey sampling. Inaccessible and sensitive data. Length-biased or size-biased sampling and weighted distributions Weighted distribution methods. Composite sampling. Attribute sampling. Ranked-set sampling. Optimal estimation. Imperfect ordering. Sampling in the wild. Quadrate sampling. Recapture sampling. The Petersen and Chapman estimators. Capture-recapture methods in open populations. Transect sampling. Point transects or variable circular plots. Adaptive sampling. Examining environmental effects: stimulus-response relationships. Regression-type models. Linear models. Transformations. The generalized linear model. Special relationship models, including quantal response and repeated measures. Toxicology concerns. Quantal response. Bioassay. Repeated measures. Standards and regulations. Environmental standards. The statistically verifiable ideal standard. Guard point standards. Standards along the cause-effect chain. A many-dimensional environment: spatial and temporal processes. Time-series methods. Space and time effects. Estimating or eliminating trend. Periodicities. Stationary time series. Time-domain models. Frequency-domain models. Point processes. Spatial methods for environmental processes. Harmonic analysis of spatial processes. Spatial sampling and spatial design.

Recommendations

Cited in

(32)

This page was built for publication: Environmental Statistics

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4828656)