Satellite- versus temperature-derived green wave indices for predicting the timing of spring migration of avian herbivores

M.S. Najafabadi, R. Darvishzadeh, A.K. Skidmore, Andrea Kölzsch, A. Vrieling, Bart A. Nolet, Klaus-Michael Exo, Nirvana Meratnia, Paul J.M. Havinga, Julia Stahl, A.G. Toxopeus

Research output: Contribution to journalArticleAcademicpeer-review

24 Citations (Scopus)
53 Downloads (Pure)


According to the green wave hypothesis, herbivores follow the flush of spring growth of forage plants during their spring migration to northern breeding grounds. In this study we compared two green wave indices for predicting the timing of the spring migration of avian herbivores: the satellite-derived green wave index (GWI), and an index of the rate of acceleration in temperature (GDDjerk). The GWI was calculated from MODIS normalized difference vegetation index (NDVI) satellite imagery and GDDjerk from gridded temperature data using products from the global land data assimilation system (GLDAS). To predict the timing of arrival at stopover and breeding sites, we used four years (2008–2011) of tracking data from 12 GPS-tagged barnacle geese, a long-distance herbivorous migrant, wintering in the Netherlands,breeding in the Russian Arctic. The stopover and breeding sites for these birds were identified and there lations between date of arrival with the date of 50% GWI and date of peak GDDjerk at each site were analyzed using mixed effect linear regression. A cross-validation method was used to compare the predictive accuracy of the GWI and GDDjerk indices. Significant relationships were found between the arrival date sat the stopover and breeding sites for the dates of 50% GWI as well as the peak GDDjerk (p < 0.01). The goose arrival dates at both stopover and breeding sites were predicted more accurately using GWI (R2cv= 0.68,RMSDcv= 5.9 and R2cv= 0.71, RMSDcv= 3.9 for stopover and breeding sites, respectively) than GDDjerk.The GDDjerk returned a lower accuracy for prediction of goose arrival dates at stopover ( R2cv= 0.45,RMSDcv= 7.79) and breeding sites (R2cv= 0.55, RMSDcv= 4.93). The positive correlation between the absolute residual values of the GDDjerk model and distance to the breeding sites showed that this index is highly sensitive to latitude. This study demonstrates that the satellite-derived green wave index (GWI)can accurately predict the timing of goose migration, irrespective of latitude and therefore is suggested as a reliable green wave index for predicting the timing of avian herbivores spring migration.
Original languageUndefined
Pages (from-to)322-331
Number of pages10
JournalEcological indicators
Publication statusPublished - Nov 2015


  • EWI-26120
  • IR-96991
  • METIS-310866
  • 2023 OA procedure

Cite this