Data for: Global net biome CO2 exchange predicted comparably well using parameter–environment relationships and plant functional types

We provide the following data to accompany our manuscript, "<strong>Global net biome CO</strong>_{<strong>2</strong>}<strong> exchange predicted comparably well using parameter–environment relationships and plant functional types</strong>", in which we compared the efficacy of multiple contrasting parameterization approaches for an intermediate-complexity carbon cycle model. These data are generated as inputs for or outputs from the model–data fusion system CARDAMOM, which is publicly available (https://github.com/CARDAMOM-framework/CARDAMOM_v2.2). Model ID #911 was used in our study. <br> Included here are: <strong>Driver files</strong> (meteorological variables and observational constraints) as .cbf files <em>('drivers/')</em>. Each 4-digit code appended to the end of each filename corresponds to a unique land surface pixel. <strong>Posterior parameter distributions for the optimally parameterized model</strong> as .cbr files <em>('parameters_opt/')</em>. Arrays are of shape [n_ensembles, n_parameters]. Each filename ends with '_{1, 2, 3, 4}', which corresponds to a different MCMC chain. Collapsing across the ensemble dimension yields parameter distributions with uncertainty. Parameter identifiers can be found in the CARDAMOM source code, linked above. <strong>Parameter estimates for the plant functional type-based (PFT-based) and environmental filtering-based (EF-based) models</strong> as .cbr files <em>('parameters_{pft, ef}/')</em>. <strong>Model output time series (fluxes and pools) for the optimal, PFT-based, and EF-based models</strong> <em>('outputs_{opt, pft, ef}/')</em>. For the optimal model, outputs are provided as .npy files (see note below). For the two alternatively parameterized models, outputs are provided as .bin files. Flux outputs are of shape [n_months, n_fluxes], while pool outputs are of shape [n_months+1, n_pools]. Here, n_months = 192. A list of flux and pool identifiers can be found in the CARDAMOM source code, linked above. * Note: Due to data volume limitations, only the median of 4 MCMC chains used in the manuscript is provided for the output of the optimal model. The full suite of ensemble members (which determine the optimal model's uncertainty) is available upon request. <br>