INTRODUCTIONA number of studies have shown evidence of apparent emission of small particles (e.g. Gallagher et al., 1993, 1997). Particle flux measurements are still difficult to interpret and prone to a variety of errors. For example, humidity gradients can lead to aerosol growth with height and humidity fluctuations can effect the measurements. The phenomenon of apparent emission of small particles, however, can sometimes be related to periods of strong NH 3 emission (Gallagher et al., 1993) and has therefore been thought not to be an artefact. Instead it has been attributed to the condensation of ammonia (NH 3 ) and nitric acid (HNO 3 ) to the surface of existing particles (gas-to-particle conversion, gtpc), under conditions where the vapour product (NH 3 HNO 3 ) exceeds its theoretical value. Since the species involved undergo different surface exchange processes, this reaction may result in non-linear concentration profiles, flux divergence, apparent aerosol emission and apparent deposition of HNO 3 faster than permitted by turbulence. Approaches to explicitly model the profiles of the species involved usually consider only the bulk concentration of NH 4 NO 3 aerosol. This approach is restricted by the limited knowledge of the reaction kinetics of the process and the lack of measurements to assess the performance of the models.Wexler and Seinfeld (1990) have shown the chemical time-scale to be highly dependent on the composition and size spectrum of the particles and provide mechanistic expressions for the chemical time-scale and the modified equilibrium products. Both composition and size spectrum change with height if gtpc modifies the aerosol and the deposition rate of the aerosol is size dependent. Hence more accurate modified gradients of NH 4 NO 3 NH 3 and HNO 3 should be obtained by explicitly modelling not just the bulk concentration of the aerosol, but also its distribution. Here a first implementation of the approach is outlined and first results are presented which indicate that the size dependent gtpc modelling can indeed qualitatively predict some of the observations made in the fieldMODEL DESCRIPTIONThe model utilizes the first order relaxation description of the chemical source/sink term (Q N H 3 ) according to Brost et al. (1988), which leads to a divergence of the flux (F N H 3 ) with height (z):where the chemical time-scale is calculated as a function of aerosol size and composition distribution (Wexler and Seinfeld, 1990) and the theoretical equilibrium ammonia concentration ([NH 3 ] e q ) is calculated from the measured concentrations of NH 3 , HNO 3 and NH 4 NO 3 and the dissociation constant after Wexler and Seinfeld (1991), which also depends on the aerosol composition. Similar equations to (1) describe the production / destruction of HNO 3 and N aerosol size classes. The model simultaneously solves a set of N+2 coupled non-linear 2 n d order differential equations, two for the mass concentrations of NH 3 and HNO 3 and a further N for the particle densities of the different size classes. Boundary conditions for the aerosol are given by measured concentrations and size spectrum at a reference height and by the parametrization of the particle deposition rate using the model by Slinn (1982), whereas for NH 3 and HNO 3 boundary conditions are obtained by fitting the modelled profiles to gradient measurements.PRELIMINARY RESULTSEddy correlation measurements at Leende Heide, NL, reported by Gallagher et al. (1993), have shown simultaneous emission of particles with a particle radius R p < 0.15 μm and deposition of larger particles. In first model runs this behaviour could qualitatively be reproduced, although the sign change in the particle flux was only found at certain heights (e.g. 0.34 m in Figure 1b).