R
Title: Modeling Nitrates in the Des Moines River This seminar takes the form of a `progress report' on development of a simple statistical structure with which to model concentrations of nitrate/nitrite at 7 monitoring stations on the Des Moines River. The overall goal of the analysis is to develop a model with few parameters that is capable of describing nitrate concentrations in the enitre monitoring network using data (provided by the US Army Corps of Engineers) of roughly biweekly observations over the period 1982 to 1996. A subset of these data, consisting of 938 observations from 1982 to 1985 is used as the primary focus for model development. A conditionally specified Gaussian model is used, with dependence conceptualized as a temporally-varying spatial phenomenon. While any of a number of approaches might yield adequate model structures for this problem, the approach taken focuses on maintaining a simple description of large-scale structure over time and incorporating the majority of variation within the dependence structure.