Timothy G. Gregoire

Timothy G. Gregoire

J. P. Weyerhaeuser, Jr., Professor of Forest Management

Type

Professor Gregoire’s research is directed to the application and development of statistical methods for natural resources and environmental phenomena. One longstanding focus has been on probability sampling techniques and a second focus has been on statistical modeling of longitudinal and spatially correlated data. The results of his research have been published widely in the forestry, ecology, and statistical literature of both subject areas. His books include Sampling Methods for Multiresource Forest Inventory; Modeling Longitudinal and Spatially Correlated Data; and Sampling Strategies for Natural Resources and the Environment. Recent pursuits include the design and planning of the national forest inventory of Bhutan; the development of estimators of aboveground forest biomass following laser altimetry; the effect of measurement error when using ratio estimation; quantile regression modelling of tree diameter distributions; and the nature of statistical inference. Professionally, he has been a leader in organizations that promote the use of biometrics and environmental statistics. He is an elected Fellow of the American Statistical Association; a former regional president of the International Biometric Society; and the recipient of the Forest Science Award granted by the Society of American Foresters. He has served as a section editor of the multivolume Encyclopedia of Environmetrics, and currently serves as associate editor for Silva Fennica, for Environmetrics, and for Environmental and Ecological Statistics. He also serves on the board of directors of the Energy and Resources Institute–North America, as well as the Academic Council of TERI University in New Delhi, India.

My research is focused on 1) the development of statistical methodology with particular application in biological and natural resource contexts, and 2) the investigation of the statistical properties of estimators and methods of statistical modeling. Very recently I have been considering issues relating fundamentally to the basis of statistical inference. Publication of my research is distributed widely throughout the forestry, ecology, and statistical literature, and for which I received the Forest Science Award from the Society of American Foresters, and have been elected a Fellow of the American Statistical Association. In the area of probability sampling, my work has resulted in the development and explication of Importance Sampling and related Monte Carlo methods to estimate characteristics of individual trees; line intersect sampling with transect of multiple segments; development of inventory procedures in forestry and ecology to avoid edge-effect bias; examination of multiple sources of statistical error when using airborne lasers for purposes of forest inventory; authorship of a textbook of Sampling Methods for Multi-resource Forest Inventory (Wiley, 1993). Currently I have contracted with Chapman Hall/CRC to write and publish a textbook entitled Sampling Techniques for Natural and Environmental Resources. In the area of statistical modeling, I have had an abiding concern with fitting models to correlated data in a manner that explicitly accounts for the correlation structure. One product of this pursuit was my organization of a international conference in 1996 which resulted in the publication of Modelling Longitudinal and Spatially Correlated Data (Springer-Verlag, 1997). Other products includes numerous publications with colleagues in which we propose statistical models for forest plots and trees that have been measured repeatedly over time or space. Brief overviews of selected research topics follow.

  • A Realistic Analysis of the Variability of Carbon Estimates Using Airborne and Space LiDAR. This is a collaborative, international research project funded by NASA in response to the NASA Carbon Cycle Science program. Involving scientists from the U.S., Scandinavia, and Canada, the investigation aims to identify and reduce major uncertainties in the boreal forest carbon budget. The overarching objective of the study is to develop a realistic estimate of the variability of regional carbon estimates developed using space-based lidar measurements of forest structure.
  • Identification and monitoring effects of heavy vehicular traffic on the vegetative and physical characteristics of Alaskan tundra. A manipulative experiment was carried out on the tundra south of Prudhoe Bay, Alaska in order to observe and measure the impact of heavy equipment traveling over the tundra during winter at various degrees of tundra frozenness.
  • Line Intersect Sampling. With increased interest in the estimation of the abundance and distribution of coarse woody debris on the forest floor, the method of sampling has received renewed attention from practitioners. Techniques such as using ell-shaped transects, triangular transects, and Y-shaped transects have sprung up. We have developed unbiased estimators of CWD abundance when using segmented transects, and we have devised procedures to avoid edge-effect bias.

I teach two courses dealing with methods of statistical sampling. One is taught each year and is designed broadly for environmental and natural resource managers and students in our forest and environmental management degree programs. With this course I aim to acquaint students with the precepts of probability sampling for the estimation of population parameters. The other is taught less regularly and is more focused on techniques of advanced forest inventory.

I also teach a course each year entitled Statistics for Environmental Sciences which concentrates on applied regression modeling, and I have begun to co-teach a course on Applied Spatial Statistics. In alternate years I teach a course in Experimental Design, which  covers traditional aspects of statistical design of field and laboratory experiments.

Education

B.S., Princeton University; Ph.D., Yale University

This professor is accepting doctoral students