ifs-landbench
Offline single-point ecLand runs scored against flux-tower observations — latent and sensible heat, net ecosystem exchange, methane flux, and soil moisture and temperature. Dashboards are served from the ifs-landbench repository.
Curated Best-Sites Subset
The towers recommended for automated benchmarking, chosen for spatial and biome diversity: 36 available here, 664 site-years.
All 775 Sites
The full FLUXNET-Shuttle pool across AmeriFlux, ICOS and TERN: 775 towers, 1507 site-years. Median r of 0.79 (Qle) and 0.82 (Qh).
ifs-lakebench
Offline single-point ecLand (FLake lake scheme) runs scored against satellite lake-surface-water-temperature observations — overpass and daily-mean matching. Dashboards are served from the ifs-lakebench repository.
Lakes vs. ESA CCI Lakes LSWT
23 lakes worldwide spun up over 2017–2022, model AvgSurfT vs. ESA CCI Lakes surface temperature, at satellite overpass and daily-mean instants.
ifs-riverbench
River discharge from the IFS river routing component, scored against gauge observations. Dashboards are served from the ifs-riverbench repository.
Discharge vs. Gauges
River discharge model experiments vs. gauge observations, by station, with a river-network diagnostic view.
ifs-floodbench
Flood extent from CaMa-Flood coupled to the IFS, scored against satellite flood mapping (VIIRS, GFM, MODIS) over three event catalogues. Dashboards are served from the ifs-floodbench repository.
KuroSiwo Flood Events
Model vs. VIIRS / GFM / MODIS observations for 43 events from the KuroSiwo benchmark catalogue.
Post-2016 Named Events
Model vs. observations for 17 hand-picked notable global floods since 2016.
DE 2026 Events
Model vs. observations for 10 recent, currently-monitored flood events.