Uncertainties in estimating the mixing depth-comparing three mixing-depth models with profiler measurements

Abstract

Determining the temporal evolution of the mixing depth is a complex problem in air pollution modeling. In this paper, mixing depths derived from three algorithms, RAMMET-X, MIXEMUP, and CALMET, are compared with estimates derived from a 915-MHZ radar profiler and Radio Acoustic Sounding System (RASS) located at Schenectady, NY, for nine clear summer days. Besides wind and temperature soundings, the profiler system also provided estimates of the mixing depth based on refractive index structure parameter ( C n 2) measurements. For the nine test days studied, mixing depths ranged from 1.6 km in midafternoon to 150–250 m at night, based on C n 2. Mixing depths obtained from CALMET and MIXEMUP were in good agreement with C n 2 estimates throughout the day, but especially in the afternoon when they agreed to within 100 m. In contrast, estimates from RAMMET-X displayed considerably less diurnal variation, with afternoon mixing depths 300–400 m lower than profiler estimates, and nighttime values similarly too high. A separate “analytical method” based on an analysis of the profiler system's wind and temperature profiles, is offered as an alternate way for estimating the mixing depth. Mixing depths derived from the analytical method agreed well with C n 2 estimates at the times of the maximum and minimum, but displayed a much faster growth rate during the morning and a slightly slower decay rate in the evening. It is well known that ozone concentrations predicted by photochemical models are very sensitive to the mixing depth. The impact of the uncertainties inherent in the estimation of the mixing-depth profile on the ozone concentrations is examined through Ozone Isopleth Research Model (OZIPR) simulations. The results reveal that the variability in the OZIPR-predicted ozone concentration due to the uncertainties in the specification of the evolution of the mixing height is comparable or greater than that of different chemical mechanisms in the OZIPR model.