Ecological systems evolve in space and time. Until recently, however, research in ecology separately has focused either on the spatial domain (patterns) or on the temporal domain (processes). In this paper we describe novel approaches for progressing towards an integration of pattern and process, a goal long called for in ecology. First, we present a sequence of alternative stochastic models of spatially extended processes. Second, we advance two new methods for the estimation, or calibration, of model parameters from spatio-temporal processes observed in the field. Third, we provide tools for reducing the complexity of spatially extended ecological processes to manageable dynamical systems. Steps and techniques are illustrated in the context of data from a montane grassland community from the Czech Republic.