Assessing the Impact of Vegetation Data on HadCM3L Phanerozoic Climate Simulations

Climate-model simulations are important tools for testing hypotheses about the drivers of shifts in climate and ecosystem distributions throughout the Phanerozoic. Initial simulations of Phanerozoic climates have been carried out using the HadCM3L climate model, with 109 time slices across the 540 million years. Each time slice represents a distinct stage, with CO₂ concentrations prescribed to align the modeled global mean surface temperatures (GMST) with estimates of past GMST.

However, these simulations utilized modern plant functional types (PFT) and globally homogeneous surface properties across all Phanerozoic timescales. In reality, vegetation has evolved through time. So, use of modern PFT may introduce significant errors in climatically relevant variables (e.g., albedo). Consequently, the estimated values of modeled temperatures through time may be inaccurate.

The aim of this project is to implement more realistic representations of vegetation in the simulations, by utilizing PFTs that are appropriate for each time slice. For example, the early Ordovician would be characterized by low-lying, sparse vegetation dominated by bryophyte-like plants, which likely exhibited simple anatomy and physiology, were restricted to moist lowland habitats, and lacked deep anchoring structures.

As a first step towards this aim, we have set up a series of simulations that are simple continuations of the existing simulations, run for 110 years, but including more vegetation-specific outputs. Our analysis included visualizations of the Phanerozoic vegetation fraction, which pointed out clear inaccuracies, such as the unrealistic representation of vegetation during the early Phanerozoic. These findings emphasize the limitations of the original model’s assumptions about vegetation. Furthermore, we demonstrated that vegetation significantly influences surface temperature and found strong relationships between climate variables (such as precipitation and surface air temperature) and vegetation distribution. Our results underscore the need to make realistic adjustments to vegetation parameters in HadCM3L simulations.