The University of Toledo

Assistant Professor

Research and Teaching Interests:

• Environmental and ecological statistics
• Bayesian hierarchical models
• Water quality and watershed modeling
• Stream ecosystem response to disturbance

(419) 530-4230 | song.qian@utoledo.edu

I am interested in the applications of modern statistical methods, especially Bayesian methods, in ecological and environmental studies, specifically:

Collaborating with USGS in studying the effects of urbanization on stream ecosystems. This research focuses on the analysis of a large data set from ecological studies conducted by the USGS National Water Quality Assessment (NAWQA) program. In the past 5 years, the collaboration resulted in a series of new methods for analyzing biological response data. The current focus is on the analytical methods of species compositional data and linking species composition change to environmental changes.

Bayesian hierarchical models for understanding drinking water pollution at a national scale. I developed Bayesian hierarchical models for assessing the US drinking water safety and compliance to the Safe Drinking Water Act. These models are capable of handling data with large amount of concentration values below method detection limits. Recently, I developed a series of models for analyzing data on the occurrence of pathogenic microorganisms in drinking water sources. These models are being used by the US EPA in its regulatory review of SDWA compliance. The models were also used for analyzing China's water quality in its drinking water sources. These models address a common feature in the data, that is, a large amount of fasle negatives.

Watershed modeling. Modeling the loading of nutrient and other pollutants to receiving waters is the first step of developing effective watershed water quality management strategies. A Bayesian SPARROW model is under study for modeling small-scale watersheds and for areas with inadequate data. The Bayesian updating process is included in the model so that not only information from similar watersheds can be used as a starting point, but also the resulting model can be used for developing an effective monitoring program.

Ecological threshold, its detection and quantification. Determining whether an ecological threshold exists along an environmental gradient is a contentious subject. The complexity of the problem is rooted at Hume's problem of induction. My interest lies in the use of statistical causal inference to determine the appropriate response model and the management implications. Currently, I am interested in testing a series of threshold models for different response patterns.

Phenological data analysis. Phonological data on plants and animals are important record on climate changes. Analysis of phenological data often follows the one-species-at-a-time approach. I am interested in developing Bayesian hierarchical models to integrate related data series (e.g., the lilac first bloom dates and cherry blossom dates in Washington D.C.)

2012

2011