Search

The Kwinana Coastal Fumigation Study took place in early 1995 at Kwinana near Perth in Western Australia. The study involved surface and elevated meteorological and plume fumigation measurements in sea-breeze flows near the coast, and has yielded a comprehensive data set that is suitable for assessing meteorological and plume dispersion models. In this paper, we simulate the meteorology and turbulence on four case study days, and compare model results with the detailed surface and aircraft measurements taken during the study. These days had surface synoptic winds ranging from southerly to northeasterly, with either stable or near-neutral temperature profiles over the sea. The model used was based on that developed by Hurley ( Boundary-Layer Meteorol. 83, 43–73, 1997), but extended here to allow domain nesting, optional non-hydrostatic simulations, and a vegetative canopy at the surface. The model was forced by standard weather service synoptic data, and the simulations have captured the essential features of the strong sea-breeze circulation observed on these days. The boundary-layer structure over the sea was predicted to be near-neutral or stable in agreement with the observations on the particular day. The wind speed and direction in the sea-breeze flow were generally predicted well, although the predicted maximum inflow speed over the land was a little too high. The potential temperature was generally over-predicted, but temperature gradients agreed well. Predicted turbulence levels in the bottom-half of the thermal internal boundary layer compared well to the observations, but under-estimated the observations in the in the upper half of this layer. Near-surface measurements of meteorological variables were predicted well over the entire diurnal cycle, although the predicted sea-breeze onset was generally too early. A quantitative model evaluation for the near-surface sites showed the model performance to be better than that from other studies, with Index of Agreement (IOA) values of 0.8 (wind speed) and 0.96 (temperature), compared with values of 0.5–0.6 (wind speed) and 0.33 (temperature) obtained from other studies. The availability of new higher resolution synoptic analyses should obviate the lack of spatial and temporal resolution in synoptic inputs. The incorporation of these higher resolution synoptic inputs and new parameterisation schemes should improve future model performance. [ABSTRACT FROM AUTHOR]