Testing a coupled biophysical/dynamic vegetation model (SSiB-4/TRIFFID) in different climate zones using satellite-derived and ground-measured data

Studies have shown that seasonal vegetation phenological variation influences regional climate patterns through its interaction with land surface water, carbon, and energy exchange. Therefore, vegetation-memory should not be ignored in climate study. A coupled biophysical model and dynamic vegetation model is used to further explore the effect of vegetation memory. . SSiB is a biophysical model of land-atmosphere interactions intended for global and regional studies. In version SSiB-4, parameterizations of plant photosynthesis and stomatal conductance, developed by Collatz et al., for C3 and C4 plants are included using quasi-analytical solutions. Meanwhile, a new scaling methodology for the canopy scale, which includes sun-lid and leaf-shading effects, has also been developed. The TRIFFID (Top-down Representation of Interactive Foliage and Flora Including Dynamics) is a top-down dynamic vegetation model, in which the relevant land-surface characteristics of vegetation cover and structure are modeled directly. The land-surface temperature and soil moisture, as well as net primary production (NPP) are calculated within SSiB-4, and then passed to TRIFFID. TRIFFID allocates the average NPP over the coupling period into the growth of the existing vegetation. This process is strongly influenced by soil moisture and soil temperature. After each call to TRIFFID, the land-surface parameters required by SSiB-4 (e.g., albedo, roughness length) are updated, based on the new vegetation state, so that changes in the biophysical properties of the land surface feedback to the atmosphere.

We have used 7 observational data sets from different latitudes and landform conditions to test the off-line TRIFFID/SSiB-4, including Sahel savanna and shrubs, Amazon tropical rainforest, and boreal forest. Some of them have more than 10 years of data. Satellite-derived and ground measured LAIs are used for comparison. In general, the model properly simulates the vegetation annual cycle. It has been found that the seasonality in the model simulation is mainly controlled by surface temperature in high latitude and by soil moisture in tropical region, respectively. In mid-latitude, both factors play a role. In addition to vegetation property, observed surface latent heat, sensible heat fluxes and carbon flux, whenever available, are also used for evaluation. The testing results show that in offline experiments, in which atmospheric conditions are fixed, the coupled SSiB-4/TRIFFID model and the SSiB-4 model with satellite derived vegetation properties produce similar results. In general, simulated results in forest sites are better. The preliminary results from the coupled GCM simulations are also discussed.