#### Research interests

I am interested in practical Bayesian modelling and computation for understanding and interpreting large and complex data sets. Such data sets may arise from systems that vary over space, time or both, and may be multi-variate, mis-measured and spatially mis-aligned and may also contain missing observations. The primary aim of my research is to reduce uncertainty in inferential statements by developing predictive Bayesian models for the quantities of interests.

Recently my research efforts have focused on modelling specific air pollution surfaces, for example, ozone and particulate matter concentration levels, and deposition levels of sulfate and nitrate compounds over the eastern United States. These model-based methods are being used to estimate the probability of compliance to the government standards regarding levels of exposure to air pollution at unmonitored locations based on temporally replicated observed data at sparsely located monitoring sites.

Methodological research interests include: development of data assimilation techniques for fusing sparsely located, point referenced data with grid-level computer model output; development of hierarchical models linking ozone levels and meteorology to estimate meteorologically-adjusted spatially varying trends; updating and forecasting of air pollution surfaces in a real-time dynamic environment. Some of these projects involve collaboration with Professor Alan Gelfand, Duke University and Dr David Holland, US Environmental Protection Agency.

I am currently involved in an EPSRC funded research project jointly with Dr Duncan Lee, Glasgow University and the Met Office. The project entitled "A rigorous statistical framework for estimating the long-term health effects of air pollution" aims to investigate health effects of air pollution. The project will develop methodology in both air pollution and health outcome data modelling and their integration.

I am also working as a co-investigator in the NERC funded research grant, Quantifying annual cycles of macronutrient fluxes and net effect of transformations in an estuary: their responses to stochastic storm-driven events. I am developing statistical models and model based statistical inference techniques to study nutrient dynamics in annual and seasonal periods.

Previously, I have worked on an EPSRC funded (2010-2012), cross-disciplinary research project jointly with Prof Paul Harper, Cardiff University, the Met Office, and the the Southampton University Hospital Trust. The project entitled "MetSim: A Hospital Simulation Support Tool" developed methods for improving planning and management of health services by accurately predicting demand by using meteorological information. The statistics arm of this project, led by myself, developed a Bayesian model for forecasting the number of hospital admissions and the model is being currently piloted by the Met Office for use by several UK hospitals.

#### Research group

Statistics

#### Research project(s)

**Professor Sujit Sahu**Building 54 Mathematical Sciences University of Southampton Highfield Southampton SO17 1BJ

**Room Number:** 54/9013

**Facsimile: **(023) 8059 5147

Professor Sujit Sahu's

personal home page