Regional evaluation of satellite-based methods for identifying leaf unfolding date

Abstract

Satellite-based methods have been widely used to identify the critical phenophases of terrestrial vegetation and detect responses of phenophases to climate change. However, few studies have been conducted to evaluate the performance of satellite-based methods for identifying the spatial and temporal changes of phenophases based on site-based phenophase observations. This study used the ground observations of leaf unfolding date (LUD) of woody and herbaceous plants over 31 sites in China to evaluate the accuracy of LUDs modeled by eight satellite-based methods using Moderate resolution imaging spectroradiometer (MODIS) vegetation index product. Our results showed an observed 2.62 and 2.45 days advance of LUD over the spatial scale with an increased mean annual temperature of 1 °C over the forest and grassland ecosystems respectively, but almost all eight methods underestimated the advance rates. In addition, the trends for the observed LUD varied from site to site, and the eight methods showed the poor performance in capturing the long-term trends, mostly because low temporal resolution of satellite data. Most methods tended to overestimate the trend of LUD over more than 40% forest sites. Our results highlight that there is a need for further improvements in the methods and satellite datasets used for identifying LUD.