DISPERSION EXPERIMENTS IN CENTRAL LONDON: The 2007 dapple project

摘要

A successful full-scale tracer dispersion campaign was undertaken in central London during the summer of 2007, continuing the series that began in 2004. In this article, we demonstrated the synthesis that DAPPLE can bring by the incorporation of flowfeatures—such as channeling and corner vortices—into the tracer data analysis and reviewed results from the analysis of data from this campaign. This clearly justifies the experimental design, which aimed to allow successful sampling of much of the spatial variability in near-field urban dispersion under a range of background wind speeds and directions. Two sets of long-term reference measurements were taken to generate a climatology of the winds at the site—in particular, the relationship between mean rooftop (18 m) and tower-top (190 m) winds. An important conclusion was that for analysis of tracer transport, the local flow in the street network was better represented by the mean wind at rooftop than tower top. However, the associated wind tunnelwork was most readily characterized in terms of the "free stream" wind speed, so it was therefore important to understand the relationships between the two reference winds. A result of the continuing analysis was finding a ratio of 4.1 between the 15-minmean wind speeds at rooftop and tower top in near-neutral conditions. This ratio is useful in assigning the same reference flow in nondimensionalized data in both wind tunnel and full-scale fieldwork. Here, the rooftop wind speed has been used as the reference with which to nondimensionalize the dosage data for comparison with models and other tracer release experiments. Each field experiment is simply a single realization from the ensemble of possible outcomes, given the boundary conditions. At best,two or three experiments might be undertaken with very similar boundary conditions. In the wind tunnel, however, 100 realizations have been found necessary to determine reliably the statistical properties of dispersion behavior (Robins 2008). Hence, thewind tunnel is the key tool that can provide a full analysis of variability, but this too has its limitations. For example, at lower wind speeds, traffic-produced turbulence (TPT2; Di Sabatino et al. 2003; Kastner-Klein et al. 2003) probably becomes important but was not included in the wind tunnel modeling. Further work is in progress to investigate such issues using the advanced large-eddy simulation model, FLUIDITY (Pain 2000; Wang et al. 2005), which includes the simulation of vehicles moving through the study area and the additional turbulence and pollutant dispersion that they induce. The key point is that all of these research techniques need to be used in a complementary way to reach an adequate understanding of the flow and dispersion processes. The decay of the maximum downstream contamination with distance in these full-scale experiments was consistent with the inverse square relationship. Channeling flow was observed at site Y on tracer day 1, which explained the extremely high dosages that extended northward along Gloucester Place (indeed, these were the data that formed the upper limit of dosage as a function of distance from the source in Fig. 8). One of the key questions addressed by DAPPLE was whether there are simple dispersion guidelines that can be reliably applied in emergency situations. From the work summarized in this article, we can conclude that regardless of the mean above-roof wind direction i) the most contaminated areas were within one block of the source in all directions and ii) above-background dosages are possible up to around 6-8H in any direction (where H is mean building height). Beyond this distance, the greatest dosages were in the downwind sector following the inverse square decay relationship and particularly at sites where there was an uninterrupted path from source to receptor along a street that was parallel to the mean above-roof flow. Simple dispersion guidelines are essential and need careful develo