Spatiotemporal dimensionality and time-space characterization of vegetation phenology from multitemporal MODIS EVI

Abstract

Spatiotemporal dimensionality refers to the structure of the continuum of spatial and temporal patterns in an image time series. Time-Space characterization refers to an approach for representing this continuum as combinations of spatial and temporal components with a minimum of assumptions about the forms of the patterns. Patterns can be related to processes through modeling — both deterministic and statistical. By combining characterization and modeling, two complementary analytical tools can be used together so that each resolves a key limitation of the other. Empirical Orthogonal Function analysis, used in conjunction with Temporal Mixture Models, provide a way to 1) Represent the spatiotemporal dimensionality of an image time series, 2) Identify distinct temporal modes and their spatial distributions, and 3) Map the relative contributions of these modes to the observed image time series as spatially continuous fields. Some strengths and limitations of Time-Space characterization are illustrated using multitemporal MODIS EVI time series of vegetation dynamics on the Ganges-Brahmaputra delta.