An Enhanced Phenology Dataset for Global Drylands from 2001 to 2019

Abstract

Abstract Drylands dominate the interannual variability of global carbon sink, and phenology is a key driver of carbon sequestration. However, accurately retrieving dryland phenology from satellite data remains challenging due to sparse and heterogeneous vegetation. Existing land surface phenology (LSP) products exhibit low accuracy in drylands due to coarse spatiotemporal data sources and algorithms optimized for other ecosystems. Here we present the Global Dryland Phenology Dataset (GDPD) for 2001–2019, derived from daily 500-m two-band Enhanced Vegetation Index using MODIS NBAR data and an improved retrieval algorithm with dynamic, pixel-wise amplitude thresholds. GDPD covers 88.4% of global drylands, compensating for missing regions in other LSP products. GDPD shows strong agreements with in-situ phenology from PhenoCam GCC (SOS: r = 0.88; EOS: r = 0.72) and physiology from flux tower GPP (SOS: r = 0.96; EOS: r = 0.90). We highlight the importance of high-resolution data in improving dryland phenology retrieval. This dataset improves our understanding of how dryland ecosystems respond to climate change and supports the development of Earth system models.