Ensemble predictions across multiple models — invert_d2H

Runs invert_d2H against each model in models and combines the per-draw reconstructions per site, preserving the per-site dimension. Useful for downstream uncertainty estimates that should span structural model uncertainty rather than condition on one calibration variant.

Usage

invert_d2H_ensemble(
  ...,
  models = c("full_sp", "full_interact_sp", "elevation_c4_interact_sp"),
  ensemble_method = c("equal", "all")
)

Arguments

...: Arguments passed to invert_d2H (e.g., d2H_wax, d2H_wax_sd, longitude, latitude, optional covariates).
models: Character vector of v10 model names to include in the ensemble. Defaults to three structurally distinct variants: full_sp (all covariates + spatial GP), full_interact_sp (full + elevation x C4 interaction + spatial GP), and elevation_c4_interact_sp (elevation x C4 interaction with spatial GP, no PFT).
ensemble_method: "equal" (default) pools per-draw reconstructions per site across models with equal weighting and returns a per-site posterior. "all" returns the per-model results without pooling.

Value

If ensemble_method = "equal", a list with: posterior_draws (an n_draws x n_sites matrix of pooled per-site, per-draw reconstructions), ensemble_summary (a data frame with one row per site: mean, median, sd, ci_90_lower/ci_90_upper, ci_95_lower/ci_95_upper, n_models_used), model_results (the per-model output from invert_d2H() with return_full = TRUE), models_used (the models actually pooled), and ensemble_method. If ensemble_method = "all", only model_results and ensemble_method are returned.