The NOAA Twin Otter and its role in BRACE: A comparison of aircraft and surface trace gas measurements

A DeHavilland DHC-6 Twin Otter, operated by the National Oceanic and Atmospheric Administration, was deployed in Tampa, FL to measure aerosols and primary and secondary trace gases in support of the Bay Regional Atmospheric Chemistry Experiment (BRACE). The Twin Otter repeatedly overflew the surface chemistry monitoring super site near Sydney, FL to assess the comparability of surface and airborne datasets and the spatial representativeness of the surface measurements. Prior to comparing the chemical datasets, we evaluated the comparability of the standards used to calibrate surface and airborne detectors, as well as the uniformity of wind fields aloft and at the surface. Under easterly flow, when the dearth of significant upwind emission sources promoted chemical homogeneity at Sydney, trace gas concentrations at the surface and aloft were generally well correlated; R² ranged from 0.4396 for H₂O₂ to 0.9738 for O₃, and was typically better than 0.70 for NO, NO₂, NO_Y, HNO₃, HCHO, and SO₂. Mean ratios of aircraft-to-surface concentrations during 10 overflights of Sydney were as follows: 1.002±0.265 (NO), 0.948±0.183 (NO₂), 1.010±0.214 (NO_Y), 0.941±0.263 (HCHO), and 0.952±0.046 (O₃). Poorer agreement and larger variability in measured ratios were noted for SO₂ (1.764±0.559), HNO₃ (1.291±0.391), and H₂O₂ (1.200±0.657). Under easterly flow, surface measurements at Sydney were representative of conditions over horizontal scales as large as 50 km and agreed well with airborne values throughout the depth of the turbulently mixed boundary layer at mid-day. Westerly flow advected the Tampa urban plume over the site; under these conditions, as well as during transitional periods associated with the development of the land–sea breeze, surface conditions were representative of smaller spatial scales. Finally, we estimate possible errors in future measurement-model comparisons likely to arise from fine scale (or subgrid;<2 km) variability of trace gas concentrations. Large subgrid variations in concentration fields were observed downwind of large emission point sources, and persisted across multiple model grid cells (distances>4 km) in coherent plumes. Variability at the edges of the well-mixed urban plume, and at the interface of the land–sea breeze circulation, was significantly smaller. This suggests that even a failure of modeled wind fields to resolve the sea breeze return can induce moderate, but not overwhelming, errors in simulated concentration fields and dependent chemical processes.