Statistical comparison of modelled and SODAR measured turbulence data in a coastal area

Abstract

The standard deviations of wind components (σu, σv, σw) are used by Particle Dispersion (PD) models to calculate the random motion causing the dispersion of pollutants. The goodness of concentration results is known to be dependent on the accuracy of the evaluations of wind component fluctuations. A study has been conducted to investigate the ability of the boundary layer formulations used in the PD model SPRAY in estimating the actual variance of wind components. A comparison of the σw and u* data calculated by SPRAY, with those measured by means of a sonic anemometer and a SODAR system, both located in a coastal area, has been carried out. Modelled and observed σwdata have been studied under different atmospheric turbulence and wind conditions. The effect of the land use of the model cell and its representativeness with respect to the land/sea wind direction, were also investigated. A seasonal statistical comparison was conducted by means of least-square and frequency of occurrence methods, both applied at different heights. Results exhibit a general underestimation of σwdata predicted by SPRAY with respect to the measured one. The level of agreement is found to be dependent on the atmospheric stability and on land/sea wind regimes. A strong dependence on friction velocity estimations is found as well. A way to improve the model estimations is also presented.