Mean State

	Download Data	Period Mean (original grids) [Pg yr-1]	Model Period Mean (intersection) [Pg yr-1]	Benchmark Period Mean (intersection) [Pg yr-1]	Model Period Mean (complement) [Pg yr-1]	Benchmark Period Mean (complement) [Pg yr-1]		Bias [g m-2 d-1]	RMSE [g m-2 d-1]	Phase Shift [months]		Bias Score [1]	RMSE Score [1]	Seasonal Cycle Score [1]	Spatial Distribution Score [1]	Overall Score [1]
Benchmark	[-]	-2.94
wetlands_n04_ctsm5.4.044	[-]	-5.69	-5.32	-2.85	-0.362	-0.0236		-0.0597	0.522	2.24		0.868	0.469	0.690	0.595	0.618

Temporally integrated period mean

BENCHMARK MEAN Data not available

MODEL MEAN Data not available

BIAS

BIAS SCORE Data not available

RMSE

RMSE SCORE Data not available

BENCHMARK MAX MONTH Data not available

MODEL MAX MONTH Data not available

DIFFERENCE IN MAX MONTH Data not available

SEASONAL CYCLE SCORE Data not available

SPATIAL TAYLOR DIAGRAM Data not available

MODEL COLORS Data not available

MODEL COLORS Data not available

REGIONAL MEAN Data not available

ANNUAL CYCLE Data not available

MONTHLY ANOMALY Data not available

History:: 2026-08-12 18:09:36 UTC: downloaded CARDAMOM_satellite_constrained_terrestrial_biosphere_reanalysis.nc4 from ORNL DAAC; extracted the median ensemble member (25th/75th percentiles retained as CF bounds where the variable derives from a single source term); masked grid cells breaching fixed physical bounds [C_som any>100000 (soil C > 100 kgC m-2; exceeded only by deep peat); C_cwd any>20000 (coarse woody debris > 20 kgC m-2); rh_co2 any>20 (Rh > 20 gC m-2 d-1 is not sustained by any ecosystem); H2O_LY3 any>10000 (soil layer 3 > 10 m water column); H2O_SWE mean>3000 (mean SWE > 3 m implies a glacier, which DALEC-CWE lacks); runoff any>30 (runoff > 30 kg m-2 d-1); D_TEMP_LY1 any>323.15 (soil T > 50 C exceeds the hottest skin temperature in the forcing (41.8 C))], applied per domain (carbon bounds to carbon variables, water bounds to water variables); converted to CF-compliant netCDF for ILAMB

References:: Bilir, T. E. and Bloom, A. A. and Parazoo, N. C. and Liu, J. and Braghiere, R. K. (2026), CARDAMOM Carbon-Water-Energy Reanalysis v1100.1, 2001-2021, 10.3334/ORNLDAAC/2492

Comment:: Grid cells whose state is non-physical have been masked; see the history attribute for the exact bounds. These arise where CARDAMOM's per-pixel initial conditions are poorly constrained (producing a decaying disequilibrium transient over 2001-2003) or over ice caps, which DALEC-CWE does not represent. The bounds are evaluated on the median member only; the 25th/75th bounds are the raw ensemble interquartile range and may exceed them at retained cells. IMPORTANT: CARDAMOM assimilates GRACE/GRACE-FO water storage, mean runoff (GRUN), satellite GPP and LAI, above- and below-ground biomass, harmonised soil organic carbon, MODIS snow-covered fraction, MOPITT-CO-derived fire emissions and CMS-Flux NBE, and is forced by ERA5 meteorology, atmospheric CO2 and prescribed burned area. Comparisons against models are therefore NOT independent of the corresponding ILAMB reference datasets; see README.md for the mapping and for which variables remain informative. Additionally, nee is defined here as -NEP, taken directly from the source. This is not identical to reco - gpp as published in this collection (they differ by up to 1.7 gC m-2 d-1, median 0.006), and ILAMB derives the model-side nee as ra + rh - gpp, so a small definitional offset between reference and model is expected.