Abstract
The proper understanding of precipitation variability, seasonality, and predictability are important for effective environmental management. Precipitation and its associated extremes vary in magnitude and duration both spatially and temporally, making it one of the most challenging climate parameters to predict on the basis of global and regional climate models. Using information theory, an improved understanding of precipitation predictability in the conterminous United States over the period of 1949–2010 is sought based on a gridded monthly precipitation dataset. Predictability is defined as the recurrent likelihood of patterns described by the metrics of magnitude variability and seasonality. It is shown that monthly mean precipitation and duration-based dry and wet extremes are generally highly variable in the east compared to those in the west, while the reversed spatial pattern is observed for intensity-based wetness indices except along the Pacific Northwest coast. It is thus inferred that, over much of the U.S. landscape, variations of monthly mean precipitation are driven by the variations in precipitation occurrences rather than the intensity of infrequent heavy rainfall. It is further demonstrated that precipitation seasonality for means and extremes is homogeneously invariant within the United States, with the exceptions of the West Coast, Florida, and parts of the Midwest, where stronger seasonality is identified. A proportionally higher role of variability in regulating precipitation predictability is demonstrated. Seasonality surpasses variability only in parts of the West Coast. The quantified patterns of predictability for precipitation means and extremes have direct applications to those phenomena influenced by climate periodicity, such as biodiversity and ecosystem management.
Advances in Climate Processes, Feedbacks, Variability and Change for the West African Climate System
