ACINN Graduate Seminar - WS 2023/24

2023-10-18 at 12:00 (on-line and on-site)

Mesoscale model evaluation techniques over complex terrain

Gaspard Simonet

ACINN, University of Innsbruck, Austria

 

Both operational and research mesoscale model simulations, with a horizontal grid spacing ranging from a few hundred meters to a few kilometers, are often evaluated against surface station observations by typically selecting the grid point with the smallest distance from the station's location (i.e., the nearest grid point). However, in complex terrain, strong land-cover heterogeneity occurs together with orographic features at multiple different scales (e.g., large mountain  idges and smaller tributary valleys), which impact atmospheric processes. With the intention of "comparing apples with apples" in mind, we propose in this work that distance should not be the only parameter considered in the grid point selection for model evaluation. The idea is that, by selecting a grid point that is representative of the measurement site, model errors can be attributed to the correct sources, with the goal of ensuring that the model produces accurate results for the right reasons.

We will provide a brief overview of an experiment conducted with the Weather Research and Forecasting (WRF) model. First, a new physically consistent grid point selection method is developed, taking into account orographic parameters (i.e., slope angle and aspect), the most relevant landcover parameters (i.e., albedo and roughness length), and a combination of orographic and land-cover parameters, which typically vary simultaneously in complex terrain. Once the grid point selection is made, elevation-dependent variables are corrected for height differences between the model and the real terrain. Similarly, to allow for a model evaluation with observations at heights other than the typical 2- and 10-m AGL, model variables are extrapolated from the first model level to the respective sensor height using Monin-Obukhov Similarity Theory consistent with the WRF 2-m and 10-m diagnostic output variables. The physically consistent grid point selection is compared with the traditional nearest grid point selection for different horizontal model resolutions by evaluating the model output at the respective grid points against observations from a number of automatic weather stations and from a small network of surface-energy balance stations (i-Box eddy-covariance stations) in the Inn Valley, Austria.

Another challenge when evaluating modeled sensible and latent heat fluxes with measurements is that many observational studies have revealed an under-closure of the surface-energy budget, resulting in a large residual of the surface-energy budget. This under-closure is generally assumed to be the result of advection by quasi-stationary circulation systems and large eddies, which arise from surface heterogeneities and are not captured by traditional eddy-covariance systems, as well as within-canopy storage. Numerical models do not account for advection or storage terms, and the surface-energy budget is closed by definition. Therefore, a direct comparison of modeled and observed turbulent fluxes should not be expected to show perfect agreement when the observed surface-energy budget remains unclosed. Particularly during daytime, when fluxes are generally larger than during nighttime, this discrepancy may result in significant absolute differences between the model and the observations. To address this challenge, we will also introduce a novel metric that facilitates a comparison between the closed energy balance components in the model and the non-closed components observed in reality

 

 

The content of this webpage is the author’s opinion and the author’s intellectual property. ACINN has the author’s consent to make this information available on the webpage for the announcement of the seminar presentation and in the seminar archive.