© 2026 Dendro Analytics · R. Alex Thompson, PhD
How to read this scenario
NPP change is expressed relative to expected NPP at each forecast date under reference weather, including the fitted trend. Project values average cell percentage responses using polygon–grid intersection areas. They are not estimates of project-total NPP, biomass loss or carbon-credit losses.
2026 uses forecast weather for October–December; 2027 uses January–June forecasts and normal weather thereafter. The 2028–2031 trajectory reflects delayed responses to earlier climate differences, with normal weather assumed in both scenarios. These are conditional model estimates; archived weather forecasts have not consistently outperformed a normal-weather baseline.
The shaded bands are 16th–84th percentile ranges (68%). Combined bands use joint model-refit and sampled-weather draws; weather-only bands use 32 CFSv2 members with the fitted model fixed. Neither is a calibrated predictive interval or Bayesian credible interval. Ensemble agreement is not a calibrated probability.
Coverage measures overlap with modeled forest grid cells, not actual forest cover within a project. Coarse cells do not capture project-specific age, species or management. Neighboring projects may share a signal. The 80% coverage screen is not a forecast-skill criterion. Boundaries and metadata are from the historical Karnik et al. database; current registry status has not been verified.
Basemap: Natural Earth, bundled with this site; no API key required. Projects without an NPP estimate are excluded from this explorer and its download.
Sources: Karnik et al. · Original project database · GLASS/MuSyQ NPP historical model · ERA5 · CFSv2.