Visual Representations Of Non-Separable Spatiotemporal Covariance Models

Natural processes that relate to climatic variability (such as air circulation, air-water and air-soil energy exchanges) contain inherently stochastic components. Spatiotemporal random fields are frequently employed to model such processes and deal with the uncertainty involved. Covariance functions are statistical tools that are used to express correlations between process values across space and time. This work focuses on a review and visual representation of a series of useful covariance models that have been introduced in the Modern Spatiotemporal Geostatistics literature. Some of their important features are examined and their application can significantly improve the interpretation of space/time correlations that affect the long-term climatic evolution both on a local or a global scale.