Representing the Heterogeneity of Land-Atmosphere Interactions in Earth System Modeling

Abstract

Many scientific and operational challenges exist in evaluating and modeling the coupling of a complex, heterogeneous surface with the overlying turbulent atmosphere. Traditional models for the interaction between the atmospheric boundary layer (ABL) and the underlying surface rely on widely applied assumptions of homogeneity and stationarity, despite the fact that these are regularly violated in natural systems. A number of important processes are modeled poorly in models due to the dismissal of the significant influence of surface and atmospheric heterogeneity. In this work, we look at how both atmospheric surface layer turbulence (ASL) and ABL turbulence can be better parameterized in earth system models (ESMs) and numerical weather prediction schemes (NWPs). In the ASL, where the dominant Monin-Obukhov Similarity Theory is applied, we examine MOST based models of potential temperature variance in the first chapter and show how heterogeneity correlates with poorer accuracy of MOST based models using the National Ecological Observation Network (NEON) of eddy-flux towers. The variances of other atmospheric variables, including the velocity components, moisture and carbon dioxide, are then examined. This analysis shows opportunities for model improvement using higher order turbulence statistics, specific the anisotropy of turbulence, to address the shortcomings of traditional MOST relations through a generalized form proposed by Stiperski and Calaf. In the ABL, I examine the issues of circulations driven by spatial gradients of surface fluxes. In large scale models these circulations, analogous to sea breezes, are sub-grid and therefore need to be parameterized, a task yet to be accomplished by the modeling community. In this chapter, I develop a two-column model for sub-grid thermally driven circulations and compare them to large-eddy simulations (LES) of heterogeneous and homogeneous surfaces. The work qualitatively reproduces the impacts observed in LES runs, and provides a route for implementation in multi-plume schemes in ESMs. Continental scale impacts of discarding surface heterogeneity are also explored. 3km WRF simulations of the Continental United States over three summers, one set of simulations where fluxes are averaged to some scale to mimic the averaging that occurs with tiling approaches in ESMs and another set with no flux averaging, show that reduced representation of surface heterogeneity can change long term precipitation statistics and patterns of mesoscale flow in models, with some regions experiencing a doubling of precipitation due to flux homogenization at a 60 kilometer scale. The sum of this work shows that accurate representation of the heterogeneous coupling between the land and atmosphere is necessary for development of realistic climate and weather simulations, and contributes to development of new model parameterizations.