Ecological Indicators (Sep 2024)
Comparing the performance of phenocam GCC, MODIS GCC, and MODIS EVI for retrieving vegetation phenology and estimating gross primary production
Abstract
Vegetation phenology serves as an important indicator for climate change and plays a crucial role in affecting the terrestrial water, energy, and carbon cycles. The green chromatic coordinate (GCC) obtained from digital repeat photographs has been widely applied in estimating phenology from the perspective of greenness, while the performance of satellite derived GCC is not well understood. We used flux tower GPP from seven deciduous broadleaf forest (DBF) and three grassland (GRA) sites over the Northern Hemisphere. The aim was to compare phenological events with GCC (obtained from digital repeat photographs and satellite remote sensing (GCCMODIS)) and the enhanced vegetation index (EVI). Meanwhile, we also explored the performance of these three indices in simulating GPP utilizing the light use efficiency (LUE) model at the DBF and GRA sites. Phenology retrieved by GCC, GCCMODIS, and EVI was all significantly correlated with GPP-estimated values at all sites (P < 0.001). It indicates the comparable performance of GCC, GCCMODIS, and EVI in estimating phenological events. The RMSE values between the GPP and three indices-estimated phenological events revealed that the three indices excelled in estimating the start of growing season (SOS) compared to the end of growing season (EOS) and the length of growing season (GSL). In terms of GPP estimation performance, the R2 values of GCCMODIS and EVI-estimated GPP increased by 2 % and 1 %, respectively, compared to GCC-simulated GPP. Meanwhile, the RMSE values for GCCMODIS and EVI reduced by 0.08, and the bias values were reduced by 0.06 and 0.12, respectively. This study showed that GCC obtained from satellite remote sensing data could be utilized as an effective tool in extracting phenology and has a great potential to estimate GPP, at least across the DBF and GRA regions.
