Examining Parameterizations of Potential Temperature Variance Across Varied Landscapes for Use in Earth System Models.

Earth system models (ESMs) and mesoscale models have come to employ increasingly complex parameterization schemes for the atmospheric boundary layer, requiring surface boundary conditions for numerous higher order turbulence statistics. Of particular interest is the potential temperature variance (PTV), which is used not only as a boundary condition itself but also to close boundary conditions of other statistics. The existing schemes in ESMs largely rely on the assumptions of Monin‐Obukhov similarity theory (MOST) and are not necessarily applicable over complex and heterogeneous surfaces where large scale circulations and roughness sub‐layer effects may cause deviations from MOST. The National Ecological Observation Network is used here to evaluate existing parameterizations for the surface boundary of PTV, note key deficiencies, and explore possible remedies. The results indicate that existing schemes are acceptable over a variety of surface conditions provided the analysis of a priori filters out low frequency variability not associated with turbulent time scales. There was, however, significant inter‐site variability in observed similarity constants and a significant bias when compared to the textbook values of these parameters. Existing models displayed the poorest performance over heterogeneous sites and rough landscapes. Attempts to use canopy structure and surface roughness characteristics to improve the results confirmed a relation between these variables and PTV but failed to significantly improve the predictive power of the models. The results did not find strong evidence indicating that large scale circulations caused substantial deviations from textbook models, although additional analysis is required to assess their full impacts. Plain Language Summary: Modern models of the lower atmosphere, which are used to analyze climate change and weather, resolve increasingly complex characteristics of the turbulence in the atmosphere. An estimate for the value of many of these characteristics at the land surface is required to set boundary conditions for these models. An important boundary condition is the variance of very small temperature fluctuations that occur in the atmosphere due to turbulence. Currently, model estimates for these values assume the surface is flat and its characteristics do not change in space, which does not represent many of the conditions we wish to model over the earth. In addition, existing studies tend to only analyze data from a small number of locations. We analyzed data from a network of 39 sites and found that the current estimates work fairly well across a large variety of conditions, but that there is a bias in the constants often used and there are notable differences over forests, complex surfaces, and heterogeneous terrain. There is a clear relationship between surface characteristics such as tree canopy height and the performance of the model, however, it was not clear enough to improve our ability to predict the surface boundary condition. Key Points: Models of potential temperature variance in the surface layer based on similarity theory were evaluated using data from 39 varied sitesExisting schemes perform well across most surfaces, although the data shows a significant bias in the values of the similarity constantsCanopy structure and surface heterogeneity drive a large portion of inter‐site variability in model performance [ABSTRACT FROM AUTHOR]