The sensitivity of simulated competition between different plant functional types to subgrid‐scale representation of vegetation in a land surface model

Abstract

The Canadian Land Surface Scheme coupled to the Canadian Terrestrial Ecosystem Model is used to simulate competition between the model's seven non-crop plant functional types (PFTs) for available space. Our objective is to assess if the model is successfully able to reproduce the observed mix of PFTs and their fractional coverages and to what extent the simulated competition is affected by the manner in which the subgrid-scale variability of vegetation is represented. The model can be run either in a composite (single tile) configuration, where structural vegetation attributes of PFTs are aggregated for use in grid-averaged energy and water balance calculations, or a mosaic (multiple tiles) configuration, where separate energy and water balance calculations are performed for each PFT. The model realistically simulates the fractional coverages of trees, grasses, and bare ground, as well as that of individual tree and grass PFTs and their succession patterns. Our results show that the model is not overly sensitive to the manner in which subgrid-scale variability of vegetation is represented. Of the seven sites chosen across the globe to illustrate the difference between the two configurations, the simulated fractional coverage of PFTs are generally very similar (root-mean-square difference, RMSD, < 5%) between the composite and mosaic configurations at locations characterized by low heterogeneity (e.g., Amazonia, Vancouver Island, and the Tibetan Plateau), whereas at locations characterized by high heterogeneity (e.g., India, South Sudan and California), the two configurations yield somewhat different results (RMSD > 5%).

Numerical Evaluation of JULES Surface Tiling Scheme with High-Resolution Atmospheric Forcing and Land Cover Data

“…3e and 3f). These numerical results support those of previous studies that reported significant improvements in model performance by incorporating the surface tiling, or mosaic, approach (Essery et al 2003;Giorgi et al 2003;Shrestha et al 2016). The seasonality of the errors in GPP estimation might be caused by differences in either the initial soil moisture or the phenological response.…”

“…The results indicate that the current JULES LSM surface tiling scheme does not fully reflect the effects of SSGH, partly because it fails to consider the effects of subsurface heterogeneity. This may result in much greater errors under conditions of both limited soil water and the coexistence of different PFTs with distinct characteristics (tree-grass or tree-shrub mixtures) (Shrestha et al 2016). The JULES community is attempting to introduce consideration of subsurface heterogeneity by incorporating the soil tiling scheme; We suggest that, in addition to the soil tiling scheme, more precise representation of interconnections with subsurface processes should be introduced in LSMs.…”

“…The wide degree of land surface heterogeneity and the dependences of fluxes on surface characteristics result in spatially nonlinear and heterogeneous fluxes. Thus, a critical issue in land surface modeling is the reliable simulation of the effects of land surface heterogeneity, including its interconnection with biogeochemical exchanges between land surfaces and the atmosphere (Bonan et al 1993;de Vrese and Hagemann 2016;Essery et al 2003;Giorgi et al 2003;Shrestha et al 2016;Tian and Xie 2008).…”

An assessment of geographical distribution of different plant functional types over North America simulated using the CLASS–CTEM modelling framework

“…For example, models with coarse grid resolutions cannot explicitly resolve climatic niches, which in turn potentially contributes to biases in simulated vegetation distribution Shrestha et al, 2016).…”

“…Natural vegetation is typically characterized in dynamic global vegetation models (DGVMs) based on a limited number of plant functional types (PFTs; Sitch et al, 2003) because it is impossible to represent thousands of species in a model. Species characterized by similar attributes, mainly based on their form and interactions with the environment (Box, 1996), are grouped together as a single PFT.…”