Two examples of space-time models for the atmosphere. Doug Nychka Geophysical Statistics Project National Center for Atmospheric Research This talk gives two examples of modeling a space/time process -- but not for the purpose of forecasting. One main point is that spatio-temporal processes are often important for other uses than predicting the next time period. The applications are from the areas of monitoring ozone pollution and quantifying the impacts of a changing climate. In monitoring environmental pollutants one would like to make inferences at locations where measurements are not available. For ozone we consider the spatial properties of the statistic related to the EPA standard: the fourth highest daily value observed over the summer. The approach is to model daily ozone and then aggregate to sample the posterior for the ozone standard. To study the impact of changing weather patterns on agriculture one uses daily meteorological fields, such as temperature and precipitation, as the inputs to numerical crop models. In this context, space-time models for meteorological variables are considered. These models, known as weather generators, involve an (interesting) mix of multivariate time series and spatial statistics techniques. As an example, a daily weather generator is presented that includes discrete components for precipitation occurrence. This weather generator is used to study the temporal and spatial distribution of yields in the Southeast US for corn.