Remote sensing of spring phenology in northeastern forests: A comparison of methods, field metrics and sources of uncertainty
- Download PDF (664286)
- This publication is available only online.
Remote Sensing of Environment. 148: 97-107.
Current remote sensing studies of phenology have been limited to coarse spatial or temporal resolution and often lack a direct link to field measurements. To address this gap, we compared remote sensing methodologies using Landsat Thematic Mapper (TM) imagery to extensive field measurements in a mixed northern hardwood forest. Five vegetation indices, five mathematical fits to model a continuous temporal response, and a suite of threshold estimates for "start of spring/season" (SOS) assessments were compared to field measurements of bud burst stage and hemispherical photo derived canopy structural metrics (transparency, leaf area index, greenness). Results indicated that a four-parameter logistic model based on at least five spring coverages of the Enhanced Vegetation Index (EVI) and a SOS threshold of 0.3 was most closely related to field metrics and most accurate in predicting the date of full leaf out. Plot level SOS was predicted with a mean absolute error of 11 days for all species and elevation combinations, but improved to 9 days for hardwood dominated plots and 7 days for sugar maple dominated plots. Mean absolute error was improved to 8 days when forest type (mixed, conifer hardwood) was used to refine predictions. The consistency of prediction errors across forest types indicates that while overall accuracy across pixels may be low, inter-annual comparisons of changes in phenology on a pixel basis may provide accurate assessments of changes in phenology over time. This was confirmed by application to seven years of independent phenology data predicted with 12 days of mean absolute error. However, image availability will be a limiting factor in areas of frequent cloud cover.
KeywordsLandsat; Green leaf phenology; Vegetation indices; Field to sensor scaling; EVI; NDVI
White, Katharine; Pontius, Jennifer; Schaberg, Paul. 2014. Remote sensing of spring phenology in northeastern forests: A comparison of methods, field metrics and sources of uncertainty. Remote Sensing of Environment. 148: 97-107. https://doi.org/10.1016/j.rse.2014.03.017.