Limitations and Challenges of MODIS-Derived Phenological Metrics Across Different Landscapes in Pan-Arctic Regions

• Published in: Remote Sensing
• Main Authors: Wang, Siyu, Lu, Xinchen, Cheng, Xiao, Li, Xianglan, Peichl, Matthias, Mammarella, Ivan
• Other Authors: INAR Physics, Micrometeorology and biogeochemical cycles, Institute for Atmospheric and Earth System Research (INAR)
• Format: Article in Journal/Newspaper
• Language: English
• Published: MDPI 2018
• Subjects: land surface phenology; remote sensing; pan-Arctic; vegetation index; solar-induced fluorescence; LAND-SURFACE PHENOLOGY; INDUCED CHLOROPHYLL FLUORESCENCE; DECIDUOUS BROADLEAF FOREST; SUN-INDUCED FLUORESCENCE; NDVI TIME-SERIES; GROWING-SEASON; CANOPY PHOTOSYNTHESIS; SPRING PHENOLOGY; SNOW-COVER; 1172 Environmental sciences; Arctic; Climate change
• Online Access: http://hdl.handle.net/10138/278423

Recent efforts have been made to monitor the seasonal metrics of plant canopy variations globally from space, using optical remote sensing. However, phenological estimations based on vegetation indices (VIs) in high-latitude regions such as the pan-Arctic remain challenging and are rarely validated. Nevertheless, pan-Arctic ecosystems are vulnerable and also crucial in the context of climate change. We reported the limitations and challenges of using MODerate-resolution Imaging Spectroradiometer (MODIS) measurements, a widely exploited set of satellite measurements, to estimate phenological transition dates in pan-Arctic regions. Four indices including normalized vegetation difference index (NDVI), enhanced vegetation index (EVI), phenology index (PI), plant phenological index (PPI) and a MODIS Land Cover Dynamics Product MCD12Q2, were evaluated and compared against eddy covariance (EC) estimates at 11 flux sites of 102 site-years during the period from 2000 to 2014. All the indices were influenced by snow cover and soil moisture during the transition dates. While relationships existed between VI-based and EC-estimated phenological transition dates, the R-2 values were generally low (0.01-0.68). Among the VIs, PPI-estimated metrics showed an inter-annual pattern that was mostly closely related to the EC-based estimations. Thus, further studies are needed to develop region-specific indices to provide more reliable estimates of phenological transition dates. Peer reviewed