Toward combining observations and model experiments to improve soil moisture-precipitation feedbacks in Earth system models
Authors
Ian N. Williams — Lawrence Berkeley National Laboratory
Shaoyue Qiu — Lawrence Berkeley National Laboratory
Jungmin Lee — Lawrence Livermore National Laboratory
Christina Patricola — Lawrence Berkeley National Laboratory
Jovan Tadic — Lawrence Berkeley National Laboratory
Yunyan Zhang — Lawrence Livermore National Laboratory
Category
Boundary layer structure, including land-atmosphere interactions and turbulence
Description
Soil moisture-precipitation feedbacks are a major source of uncertainty in water cycle projections. Understanding these feedbacks at a process level has been difficult due to the multiple scales involved and confounding influences of external factors. We will report progress toward these challenges in two areas. To address the dependence of feedbacks on atmospheric state, we classified ARM US Southern Great Plains (SGP) observations into days with positive-, negative-, or no-feedback, based on hindcast experiments with perturbed soil moisture. The classification gives a clearer picture of the convective initiation process and its connection to surface state. The observed difference in surface energy partitioning and boundary layer depth between positive- and negative-feedback days suggests the importance of surface-boundary-layer interactions in initiating deep convection over drier surfaces and less stable atmospheres. To explore these mechanisms, convection-permitting hindcast experiments will be analyzed within each feedback regime. To address the land-surface segment of the feedback, we developed and tested a land-model configuration for convection-permitting simulations (3-km CLM4 with subgrid tiling) in the SGP domain and surrounding region. Multifilter radiometers were used to develop model forcing datasets that capture the seasonal distinction between crops and grasses, which enables consistent comparisons to ground-based flux measurements across the ARM extended facilities (years 2000-2016). The results show a coherent spatial pattern across the SGP domain, with winter crops and summer grasses exhibiting out-of-phase seasonality that contributes to spatial heterogeneity in surface forcing for deep convection. We will discuss how these more accurate characterizations of surface and atmospheric state, combined with cloud-permitting hindcasts, can be used to improve model parameterizations for water cycle projections.
