Intercomparing multiple measures of the onset of spring in eastern North America

摘要

Measuring the onset of deciduous tree leaf flush and subsequent development during the spring season in temperate climates can be accomplished using multiple ground and satellite-based techniques. Although all these measurements are valid (i.e. record a real characteristic related to plant development), they typically are poorly interrelated due to incompatible levels of spatial representation and differing methodologies. Given recent and likely future impacts of climate change on spring leaf development, the need to reconstruct past patterns, and the lack of standardized vegetation change measurements around the world, more work is needed to determine the relationships among the various measures, and the degree to which they may serve as substitutes for each other. In this article, we use observations and measurements at two phenology 'super-sites' in eastern North America and four other supporting sites to evaluate the relationships among multiple spring leaf development measures, and explore strategies to standardise their intercomparison. The results show infrequent significant correlations among 10 satellite-derived 'start of season' (SOS) measures (which suggests they are often not detecting the same phenomena), along with more common significant correlations among six ground phenology measures. However, when ground phenology and satellite-derived SOS are compared, there are few significant correlations, even at sites with extensive native species phenology available. Modelled phenology, based on daily temperature data (Spring Indices First Bloom date) does as well as any of the direct native species measures, and is well suited to facilitate intercomparisons. In order to effectively compare ground-based and satellite-derived SOS measures, approaches that use limited numbers of individual plants face considerable challenges. Given that satellite-derived measures are areal and at a scale of 250 m and larger, we suggest collecting ground phenology data at the same areal scale in order to make effective comparisons.