Land-Atmosphere Coupling Uncertainty due to Soil Moisture and Atmospheric Parameterization Schemes
To examine coupling strength over Australia we used a GLACE-like methodology with the Weather and Research Forecasting (WRF) model. The computational efficiency of WRF allowed multiple iterations of the experimental design to establish the sensitivity of the simulated coupling strength for this region. Within WRF we implemented two planetary boundary layer (PBL) and two cumulus (Cu) schemes used previously over Australia. Our aim was to replicate the spread of coupling strengths obtained in GLACE within one model, and to consider the impact of PBL and Cu combinations on convective triggering from a perturbed land surface state. We therefore also implemented two soil moisture states representative of El Nino (dry) and La Nina (wet) conditions.
We show that obtaining a robust estimate for the land-atmosphere coupling strength is challenging. We could replicate the multiple-model range of GLACE coupling strengths in one model by perturbing the model physics. We could also strongly vary coupling strength by changing the soil moisture state according to the El Nino/La Nina phase. These two interact such that the sensitivity of coupling strength to model physics is dependent upon the soil moisture state. Drier soils show greater coupling sensitivity to model physics and wetter soils show less coupling sensitivity. Overall, we could replicate the GLACE results either by changing the model physics in the atmosphere, or by changing the soil moisture reflecting inter annual variability. Our results therefore suggest that coupling strength, as calculated in GLACE, is very sensitive to the year chosen to evaluate the metric, and on physical parameterizations in the modeling system. To resolve this into the future will require an exploration of land-atmosphere feedbacks using perturbed physics ensembles for enough years to sample inter-annual soil moisture variability.