Phenologists study the periodic plant and animal life cycle events. The annual growth cycles of these systems are related to long-term climatic patterns. Onset of greenness, peak of greenness and end of senescence are a few of the variables phenologists are particularly interested in characterzing and forecasting. Understanding vegetation phenology and its spatio-temporal variation is required to reveal and predict ongoing changes in Earth system dynamics. Remote sensing technology has provided an impetus to phenological studies, and many efforts are directed towards predicting seasonal phenological patterns over vast areas. In particular, India provides a very rich and challenging phenological environment, as it has a diverse set of vegetation types, ranging from tropical evergreen to dry deciduous. In this work, we propose several models based on conditionally specified Markov Random fields to assess the spatio-temporal variation in the Multi-temporal Medium Resolution Imaging Spectrometer (MERIS) Terrestrial Chlorophyll Index (MTCI) data. These data are processed from satellite images available at a spatial resolution of approximately 4.6 km, composited at an 8-day interval for the years 2003 to 2007. We address several challenges characteristic of working with large spatial data sets: nonstationarity, computational demands, downscaling.

Keywords: Spatio-temporal model; Markov random field models for areal data; Massive spatial data sets; Phenology

Biography: Associate Professor of Statistics in the Department of Statistics at Iowa State University. She received her BS in Applied Mathematics from University of Bucharest, Romania and earned her MS and PhD from the University of North Carolina at Chapel Hill. Her areas of research are the methodology and applications of spatial statistics and time series to environmental, ecological, meteorological and agricultural applications. Recently, she has proposed an approximation method for estimating parameters of spatial models for large spatial data sets; proposed and alternative parametrization of one-parameter conditionally specified spatial models, which allow for interpretable covariate manipulations. Currently she is working on developing wind forecasting models that use meteorological model output, developing methodologies for analyzing satellite images for soil moisture and chlorophyll indexes, monitoring dynamic forest systems, and analyzing dispersal of genetically modified pollen flow.