Numerical experiments with assimilation of the mean and unresolved meteorological conditions into large-eddy simulation model

TL;DR

This paper explores data assimilation and unresolved surface structure incorporation in large-eddy simulation models to improve micro-scale meteorological predictions relevant to urban and environmental applications.

Contribution

It demonstrates the applicability of nudging techniques for turbulence data assimilation and proposes neural network filtering for surface stress estimation in LES models.

Findings

Nudging effectively recovers turbulence fields from mean profiles.

Artificial neural networks can filter unresolved surface features.

Methods improve LES utility for micro-scale meteorology.

Abstract

Micrometeorology, city comfort, land use management and air quality monitoring increasingly become important environmental issues. To serve the needs, meteorology needs to achieve a serious advance in representation and forecast on micro-scales (meters to 100 km) called meteorological terra incognita. There is a suitable numerical tool, namely, the large-eddy simulation modelling (LES) to support the development. However, at present, the LES is of limited utility for applications. The study addresses two problems. First, the data assimilation problem on micro-scales is investigated as a possibility to recover the turbulent fields consistent with the mean meteorological profiles. Second, the methods to incorporate of the unresolved surface structures are investigated in a priopi numerical experiments. The numerical experiments demonstrated that the simplest nudging or Newtonian relaxation technique for the data assimilation is applicable on the turbulence scales. It is also shown that the filtering property of the three layers artificial neural network (ANN) can be used for formulation of the surface stress from the unresolved surface features.