Testing of local turbulence closures in mesoscale model to be applied in the SAGA decission support system

The mesoscale meteorological model ARPS (Advanced Regional Prediction System, Xue et al, 2001) is being used at the As Pontes coal-fired power plant (over the Galicia region, a coastal and complex terrain domain located in the northwestern of the Iberian Peninsula) as part of an air quality decision support system (namely, SAGA). This model is nested to the NCEP GFS operational weather forecast at different resolutions in order to provide inputs of winds, temperature and stability profiles, mixings depths and other turbulence parameters that are required for different Lagrangian dispersion models (both puff and particle models) to simulate the plume transport. One of the main critical points in order to obtain accurate results from this system is to obtain a good representation of the vertical structure of the planetary boundary layer, as the main pollutants plume is dispersed on it.

In this work, an evaluation of results obtained using two types of local turbulence closures, over the As Pontes power plant environment, is presented. The first one is the local closure scheme developed by Smagorinsky (1963) and Lilly (1962), based on a first-order closure method. The second one is a 1.5-order closure scheme that predicts turbulent kinetic energy (1.5-TKE), tested with three different parameterizations: Moeng & Wyngaard (1988), Deardorff (1980), and Sun & Chang (1986).

Evaluations were made along three different dates, selected according to the typical synoptic situations over Galicia: High pressure with ENE winds from light to moderate (March, 19th and July 7th 2003) and low pressure with strong SW winds (November 29th 2003).

The performance of these schemes was analyzed by comparison of vertical profiles of temperature, wind speed and wind direction, as the main meteorological parameters used by the Lagrangian models. These results were validated against measurements from the rawinsonde launched by the Spanish National Meteorological Institute at A Coruña (around the power plant domain), twice a day (at 12Z and 24Z).

With the different formulations of the 1.5-TKE scheme, temperature and wind direction results showed similar values, whereas for wind speed the Sun & Chang (1986) parameterization provided better results, since the other two formulations provided wind speed with too strong changes, in comparison with the experimental data. Temperature and wind direction results from the first-order closure method were very similar to the results from the different 1.5-TKE parameterizations, but wind speed results from the first-order closure scheme were more accurate, especially during nightime and from 250 meters height on, despite some worsening of surface data due to the overestimation of rawinsonde data values.

In all cases, vertical profiles of temperature and wind direction were better represented than wind speed profiles, since the simulated wind speed values were quite overestimated until around 500 meters in height and underestimated above 750 meters. These results shows the weakness of the local closure schemes included in ARPS model in order to represent the vertical structure of the planetary boundary layer, as its application to the estimation of local air pollution dispersion requires additional developments.

Acknowledgements

This work was financially supported by Endesa company. The computational time provided by CESGA is acknowledged.