Identifying the Limitations of the Contemporary Planetary Boundary Layer Schemes Using an Extended Self-Similarity-based Framework

Mesoscale models in conjunction with robust planetary boundary layer (PBL) schemes are needed to reliably capture the dynamical evolution of the PBL flows. Over the years, numerous PBL schemes have been proposed with diverse complexity (e.g., first-order, second-order, transilient). Several modeling studies have been conducted in parallel to identify the strengths and weaknesses of these schemes. In lieu of any rigorous benchmarks, a handful of observational or large-eddy simulation-generated datasets have been typically used for model validation. Furthermore, these conventional model validation studies usually involve statistics such as bias, root-mean-square error, mean absolute error, correlation coefficient, threat score, etc. However, due to a variety of unavoidable errors (e.g., errors in initial conditions, boundary conditions), the simulated fields are always displaced in space and/or time from their observed counterparts and the standard verification metrics often lead to wrong conclusions. Thus, we believe that the modeling community needs rigorous benchmarks which are capable of capturing the essential multiscale structure of observed and modeled atmospheric fields and which can also provide constructive feedback to the model developers.

In this presentation, we will introduce a novel scaling-based benchmarking approach based on the extended self-similarity (ESS) framework. The proposed approach has roots in the classical turbulence literature and it is expected to be quasi-universal in nature. We will document the performance of several contemporary PBL schemes in capturing ESS by utilizing long-term observed and simulated datasets from Cabauw (the Netherlands), Høvsøre (Denmark), and FINO1 (the North Sea).