Tonglin Zhang

Written by: Allison Cummins, M.S. candidate in Statistics

Tonglin Zhang

Tonglin Zhang

Globally, there has been increased awareness regarding the state of our environment and the future of the world in which we live. Purdue University is no exception. Purdue President, France Córdova, is advocating a Black, Gold, and Green community. In the Department of Statistics, Professor Tonglin Zhang is currently conducting research in the field of spatial statistics while addressing both environmental and public health concerns.

Professor Zhang's research is both applied and theoretical, and based on environmental and health issues. Specifically in environmental science, he analyzes carbon dioxide and biomass distributions as related to local environmental variables, such as temperature, precipitation, and pollution. In public health, he is developing statistical models to study the geographical distribution of human diseases. Ultimately, his goal is to provide computationally efficient solutions in a spatial statistics framework that address important environmental and health problems.

Currently, in work funded by the National Science Foundation (NSF), Professor Zhang is taking advantage of three different data sources for the purpose of identifying cancer clusters in the United States. These clusters are also known as "Hotspots" and have a higher incidence or mortality rate of cancer than surrounding areas. Using data from the Center for Disease Control (CDC) with information from Geographic Information Systems (GIS) and the United States Census Bureau, Professor Zhang's work attempts to model the location, shape, and size of the cancer clusters. It turns out that estimating the shape and gradient of the cancer cluster is computationally challenging and is currently an open question. Current methods for detecting clusters also fail to take into account local ecological variables, dispersion parameter and random effects. The goal of Professor Zhang's research is to create generalized linear mixed model approaches with spatial independent or dependent random effects. In these models, the response variables are counts of disease while the independent variables are at risk population sizes and ecological covariates. The target of this model is to find the cluster of subareas with high incidence or mortality rate and determine possible links to local environmental problems. This model could be used for detecting not only cancer clusters, but also clusters of other infectious or non-infectious diseases such as influenza.

Aside from analyzing clusters in the area of public health, Professor Zhang is also currently working on a project that explores an important area of environmental science, namely the amount of carbon dioxide produced by a country. Professor Zhang is investigating the role of forests in the global atmospheric concentration of carbon dioxide. In the United States Professor Zhang is determining the national amount of carbon dioxide by employing statistical methods that rely on data collected from a few thousand trees every few years. From the tri-annual data he calculates the amount of carbon dioxide emitted from the trees and formulates matrices to represent the concentration of carbon dioxide in different regions of the United States. Since this is a 50 year project that has yet to be completed, data from a total of over 100,000 trees will be collected, thus providing a 100,000 by 100,000 square matrix. In order to compute the required likelihood function calculation the inverse of this large matrix is required. Despite advances in technology, it takes more than one year of computing time to calculate the inverse of the 100,000 by 100,000 matrix. Together with Professor Hao Zhang, Professor Zhang is developing a method of dimension reduction that avoids computing the inverse of such a large matrix by providing the inverse of a smaller matrix, thus reducing the computational time and making the inverse obtainable within minutes. They are currently completing this research and hope to publish their work on dimension reduction in the next few months.

In conjunction with finding the amount of carbon dioxide emitted from forests in the United States, Professor Zhang and Professor Qianlai Zhuang of Earth and Atmospheric Sciences, Purdue University are studying the changes in carbon dioxide levels of Chinese forest ecosystems. The flux of carbon dioxide, calculated by finding a net carbon gain or loss for the entire forest ecosystem, is able to explain most of the seasonal and annual variations in global atmospheric carbon dioxide concentration. Past research has shown that the amount of carbon used by ecosystems has increased substantially over the last 20 years; however, the cause of such an increase is not fully understood. Professor Zhang's research will aid environmental scientists' understanding of the causes of this increase as well as its impact on global climate change.

Professor Zhang is currently an Assistant Professor of Statistics at Purdue University. He received a Bachelor of Science in Mathematics and a Master of Science in Statistics from Peking University. Professor Zhang received his PhD in Statistics from the University of Michigan in 2002. He is currently teaching Advanced Statistical Methodology (STAT 526). Professor Zhang, his wife, and 3 year old daughter, enjoy Lafayette's small town atmosphere after spending time in Minneapolis, MN and Ann Arbor, MI. It is not uncommon to see Professor Zhang sitting out by Purdue's bell tower enjoying the fresh air and blue sky, something he doesn't feel he can do in a larger city. For more information about Professor Zhang, please visit his homepage.

November 2008