Investigating sources of measured forest-atmosphere ammonia fluxes using two-layer bi-directional modelling

Understanding and predicting the ammonia (NH3) exchange between the biosphere and the atmosphere is important due to the environmental consequences of the presence of reactive nitrogen (Nr) in the environment. The dynamics of the natural sources are, however, not well understood, especially not for forest ecosystems due to the complex nature of this soil-vegetation-atmosphere system. Furthermore, the high reactivity of NH3 makes it technically complex and expensive to measure and understand the forest-atmospheric NH3 exchange. The aim of this study is to investigate the NH3 flux partitioning between the ground layer, cuticle and stomata compartments for two temperate deciduous forest ecosystems located in Midwestern, USA (MMSF) and in Denmark (DK-Sor). This study is based on measurements and simulations of the surface energy balance, fluxes of CO2 and NH3 during two contrasted periods of the forest ecosystems, a period with full developed canopy (MMSF) and a senescent period for the DK-Sor site, with leaf fall and leaf litter build-up. Both datasets indicate emissions of NH3 from the forest to the atmosphere. The two-layer NH3 compensation point model SURFATM-NH3 was used in combination with a coupled photosynthesis-stomatal conductance model to represent seasonal variation in canopy physiological activity for simulating both net ecosystem CO2 exchange rates (R2 = 0.77 for MMSF and R2 = 0.84 for DK-Sor) and atmospheric NH3 fluxes (R2 = 0.43 for MMSF and R2 = 0.60 for DK-Sor). A scaling of the ground layer NH3 emission potential (Гg) was successfully applied using the plant area index (PAI) to represent the build-up of a litter layer in the leaf fall period. For a closed green forest canopy (MMSF), unaffected by agricultural NH3 sources, NH3 was emitted with daytime fluxes up to 50 ng NH3-N m−2 s−1 and nighttime fluxes up to 30 ng NH3-N m−2 s−1. For a senescing forest (DK-Sor), located in an agricultural region, deposition rates of 250 ng NH3-N m−2 s−1 were measured prior to leaf fall, and emission rates up to 670 ng NH3-N m−2 s−1 were measured following leaf fall. For MMSF, simulated stomatal NH3 emissions explain the daytime flux observations well, and it is hypothesized that cuticular desorption is responsible for the observed NH3 emissions at night. During leaf fall in DK-Sor, ground fluxes dominate the NH3 flux with a mean emission rate of 150 ng NH3-N m−2 s−1. This study shows that forests potentially comprise a natural source of NH3 to the atmosphere, and that it is crucial to take into account the bi-directional exchange processes related to both the stomatal, cuticular and ground layer pathways in order to realistically simulate forest–atmosphere fluxes of NH3.