LoginSign Up
← Back to all sparks
R

rsofun

INFRA · APIS
Velocity0.0

The P-Model and BiomeE Modelling Framework

github.com/geco-bern/rsofun ↗

An ecosystem model starts tracking carbon isotopes and land-use change.

ecosystem-modellingcarbon-isotopesland-use-changefortrancalibration
◆Current state
rsofun wraps the P-model and BiomeE vegetation models in R with Fortran cores, covering photosynthesis, water balance and forest demography, plus Bayesian calibration. The 5.1.0 release is the first in the window to widen what the models simulate rather than reorganise them. Before it, the history is renaming, cost-function rewrites and output-format consistency work.
◆Where it's heading
The direction is from a calibration harness toward a model that can answer different questions: isotope fractionation now comes out of the P-model, BiomeE handles land use and land-use change, and forcing can be recycled when a simulation outruns its data. Version stamps are unreliable here, with a v5.0 tag carrying only a build fix and predating v4.4, so the arc reads better through content than through numbering.
◆Prediction
The isotope work is explicitly unfinished, with a constant atmospheric signature standing in for daily d13c forcing, so the next likely step is accepting that as model input.

◆Recent moves

  1. 10mo ago

    Carbon isotope tracking and LULUC support across both models

    ⚡ SPARK

    The release that widens the model rather than the interface: carbon isotope tracking arrives in the P-model with BiomeE prepared for it, BiomeE gains land use and land-use change, and a new one-step leaf-level entry point appears. The Vcmax and Jmax temperature dependencies change, so existing calibrations will not reproduce.

    View source ↗
  2. 1y ago

    LM3-PPA renamed BiomeE; cost function and stress functions rewritten

    A consolidation release: LM3-PPA becomes BiomeE, the cost function and the temperature and soil moisture stress functions are rewritten, water balance variables join the P-model output, and the output format is made consistent across models. Names and shapes change, but the science surface stays where it was.

    View source ↗
  3. 1y ago

    Parallel make fix on the v5.0 tag

    A v5.0 tag carrying only a parallel make fix, stamped before the later v4.4 release. Version numbering on this feed does not track the order of the work.

    View source ↗
  4. 2y ago

    cnmodel 0.1 research snapshot tag

    A snapshot tag marking simulations for a draft nitrogen-cycle study rather than a package release. The body is truncated and carries no user-facing changes.

    View source ↗
  5. 4y ago

    Fortran crash guards and consistent P-model variable names

    Collects the 4.1 to 4.3 work in one entry: Fortran aborts are caught so they no longer take the R session down, canopy transfer is solved for LM3-PPA with the P-model, variable names are made consistent across the two models, and BayesianTools optimization is stabilised. The release notes warn that dependent code needs checking.

    View source ↗
  6. 4y ago

    Public release following a code refactor

    A one-line entry marking the public release after a refactor. It gives no detail on what changed, and sets the baseline the later 4.x work builds on.

    View source ↗