Land‐atmosphere coupling and climate prediction over the U.S. Southern Great Plains

Abstract

Biases in land‐atmosphere coupling in climate models can contribute to climate prediction biases, but land models are rarely evaluated in the context of this coupling. We tested land‐atmosphere coupling and explored effects of land surface parameterizations on climate prediction in a single‐column version of the National Center for Atmospheric Research Community Earth System Model (CESM1.2.2) and an off‐line Community Land Model (CLM4.5). The correlation between leaf area index (LAI) and surface evaporative fraction (ratio of latent to total turbulent heat flux) was substantially underpredicted compared to observations in the U.S. Southern Great Plains, while the correlation between soil moisture and evaporative fraction was overpredicted by CLM4.5. To estimate the impacts of these errors on climate prediction, we modified CLM4.5 by prescribing observed LAI, increasing soil resistance to evaporation, increasing minimum stomatal conductance, and increasing leaf reflectance. The modifications improved the predicted soil moisture‐evaporative fraction (EF) and LAI‐EF correlations in off‐line CLM4.5 and reduced the root‐mean‐square error in summer 2 m air temperature and precipitation in the coupled model. The modifications had the largest effect on prediction during a drought in summer 2006, when a warm bias in daytime 2 m air temperature was reduced from +6°C to a smaller cold bias of −1.3°C, and a corresponding dry bias in precipitation was reduced from −111 mm to −23 mm. The role of vegetation in droughts and heat waves is underpredicted in CESM1.2.2, and improvements in land surface models can improve prediction of climate extremes.

Using ARM Observations to Evaluate Climate Model Simulations of Land‐Atmosphere Coupling on the U.S. Southern Great Plains

“…Williams et al () also found a similar disproportion in the SM‐EF versus LAI‐EF coupling strengths occurring in a single‐column version of the National Center for Atmospheric Research Community Earth System Model (CESM1.2.2) atmosphere, when centered on the SGP‐CF site and coupled to the CLM4.5 land model (Oleson et al, ). In attempting to correct these coupling biases, Williams et al () modified selected properties of the CLM4.5 model: they prescribed model LAI according to the observational estimates of Williams and Torn (), while also increasing bare‐soil resistance to evaporation, the minimum moisture conductance of vegetation stomata, and leaf reflectance. These modifications improved the single‐column model predictions for the warm seasons at the CF site, especially during the dry 2006 summer, when large negative biases in precipitation and positive biases in surface temperature were greatly reduced.…”

Evaluating Soil Moisture Feedback on Convective Triggering: Roles of Convective and Land‐Model Parameterizations

“…The NCAR Community Earth System Model (CESM1.2.2; Hurrell et al, ) was run here as a single‐column model (SCM). SCMs are used to evaluate climate model parameterizations (Randall et al, ; I. N. Williams et al, ; Xie et al, ). An SCM represents a specific grid‐cell of a full climate model, and would reproduce the same grid‐cell climate if the time‐varying advective tendencies and vertical velocities were prescribed to match those of the parent climate model.…”

“…The second set of experiments (CLM‐veg) implemented the land‐model parameter modifications in I. N. Williams et al () to explore the effects of weakening the coupling between soil moisture and ET (i.e., weakening the correlation between soil moisture and ET). Specifically, the modified parameters reduced the stomatal resistance to transpiration and increased the soil resistance to evaporation (by factors of 2–6 depending on the parameter), with the effect of increasing the fraction of transpiration in ET.…”

“…The parameterization of deep convection is a major source of uncertainty in climate model projections of future changes in the water cycle over land (Klein et al, ; Qian et al, ; Qiao & Liang, ; Van Weverberg et al, ; Wilcox & Donner, ; Yang et al, ). A related source of uncertainty is the parameterization of evapotranspiration (ET) and its relationship to soil moisture and vegetation state (Cheruy et al, ; Y. Lin et al, ; I. N. Williams et al, ). In particular, poor representation of the feedback between soil moisture and precipitation can contribute to the persistence and amplification of climate prediction biases (Bonan & Levis, ; Boussetta et al, ).…”

“…However, such approaches can be confounded by covarying factors such as low‐level jets and related large‐scale meteorological forcing (Wang et al, ), and vegetation processes. For example, correlations based only on soil moisture can underestimate the role of the land‐surface in cloud and precipitation processes, since vegetation state (e.g., leaf area index, LAI) can also influence surface heat fluxes (Guillod et al, ; I. N. Williams et al, ; I. N. Williams & Torn, ). New approaches that integrate observations and complex models are ultimately needed to address causality, and to inform development of model parameterizations.…”