Evaluation of leaf phenology of different vegetation types from local to hemispheric scale in CLM

Accurate simulation of plant phenology is important in Earth system models as phenology modulates land-atmosphere coupling and the carbon cycle. Evaluations based on grid-cell average leaf area index (LAI) can be misleading because multiple plant functional types (PFT) may be present in one model grid cell and PFTs with different phenology schemes have different LAI seasonal cycles. Here we examined PFT-specific LAI amplitudes and seasonal cycles in the Community Land Model versions 5.0 and 4.5 (CLM5.0 and CLM4.5) and their relationship with the onset of growing season triggers in the Northern Hemisphere. LAI seasonal cycle and spring onset in CLM show the best agreement with MODIS for temperature-dominated deciduous PFTs. Although the agreement in LAI amplitude between CLM5.0 and MODIS is better than CLM4.5, the agreement in seasonal cycles is worse in CLM5.0. CLM5.0 also simulates higher soil moisture and shows lower influences of soil moisture on LAI amplitudes and seasonal cycles. While productivity depends on the environmental factors to which the plant is exposed during any given growing season, differences in phenology sensitivity to its environment necessitate a decoupling between the seasonality of LAI and GPP, which in turn could lead to biases in the carbon cycle as well as surface energy balance and hence land-atmosphere interactions. Because the discrepancy not only depends on parameterizing phenology but phenology-environment relationship, future improvements to other model components (e.g., soil moisture) could better align the seasonal cycle of LAI and GPP.