“…Published feedback is sparse on the quality of GIMMS NDVI 3g.v1 (1981–2015), which uses a new calibration approach and has a conservative measurement of uncertainty of about ±0.005 compared with SeaWiFS images, as reported by the authors (Pinzon & Tucker, 2014). NDVI images have been used in cluster analyses, most often at small scales and with predetermined numbers of clusters (i.e., k‐mean or related approaches; Damian et al, 2020; Romani et al, 2011) but sometimes at large scales (e.g., Mills et al, 2011; White, Hoffman, Hargrove, & Nemani, 2005) or hierarchically (Boone et al, 2000). Summaries of productivity or metrics representing phenology, or their changes through time, could be included in clustering, but I followed the most parsimonious approach, using only the raw GIMMS greenness profiles (Bunker, Tullis, Cothren, Casana, & Aly, 2016).…”