On the sensitivity of flow statistics to parameter uncertainty in flow over plant canopies

Abstract:

Accurately modeling the flow within and above a realistic vegetation canopy is a task fraught with difficulty, owing to the multi-scale and multi-physics nature of the problem. The past decades have seen significant efforts devoted to advancing the current understanding and modeling capabilities of flow in plant canopies, which, for the majority of cases, have been based on deterministic models. Despite recent advancements, accurately representing the air flow and related processes over and within plant canopies remains a challenge. The bane is the specification of the parameters, which are often impossible to be accurately measured, and whose uncertainty inevitably impacts the accuracy of model predictions. To address this problem, a general Bayesian framework is introduced that allows to quantify the uncertainty of model parameters in one-dimensional steady Reynolds-averaged Navier-Stokes simulations of flow over horizontally-homogeneous dense and sparse plant canopies. Multiple sources of uncertainty are considered, ranging from the collection and post-process of leaf area density measurements to the inherent heterogeneity of the canopy structure across scales. The same uncertainty is then propagated, via a collocation method and a Monte-Carlo approach, into the vertical profiles of the mean streamwise velocity, the turbulent kinetic energy and the Reynolds stress. To the authors' knowledge, the present study represents the first attempt to formalize a framework for uncertainty quantification in this context.