APPLICATION OF REVERSE MODELLING TO EVALUATION OF AMMONIA EMISSION FLOW IN AGRICULTURAL SECTOR

Abstract

Reverse modelling consists in the use of a local dispersion model to calibrate the value of emissions produced by specific sources or to determine the value of emission factors that can be also used in other contexts. It is a procedure widely applied in scientific literature, in particular for the identification of emission sources of greenhouse gases and for the creation of top-down emission inventories. The main steps to carry out reverse modelling require the definition of some information, such as meteorological data, geographical data related to emissive sources and their hypothetical emission flows, that are given in input to an air quality dispersion model. The model calculates pollutant concentration in air due to the emissive sources and then a regression between measured concentrations and those estimated by the model is applied in order to optimize emission flows’ values. The aim of this study is the implementation of reverse modelling to estimate ammonia (NH₃) emission flows from housing, storage and slurry application, the typical sources related to agricultural sector. In this work, two farms located in Lombardia Region are analysed; they deal with livestock (cattle and pigs) and cultivation of surrounding fields by applying different technologies such as band spreading on soil surface and injection. Since reverse modelling requires an air quality dispersion model, Aermod, a gaussian steady-state plume model that incorporates air dispersion based on planetary boundary layer turbulence structure, is used: in this model pollutant concentration is simulated as a continuous plume. Ammonia emission flows are then calculated through linear regression between the concentrations of NH₃ simulated by Aermod, and field measurements of same pollutant carried out by ARPA Lombardia (Regional Environmental Protection Agency) in the farms analysed during slurry application. The results obtained from the implementation of reverse modelling show the consistency of the procedure: ammonia emission flows are comparable with values reported by scientific literature and with results obtained by the implementation of other methodologies. First, reverse modelling confirms the difference in terms of ammonia emission between different technique of spreading since emission flows related to band spreading on soil surface are higher than the ones estimated during injection: for example, by considering slurry application carried out in May 2018 in the cattle farm, Aermod estimates an ammonia concentration of 232 µg m^-3 and of 40,2 µg m^-3 respectively for band spreading and injection. These results comply with data reported by UN-ECE (2014) in Guidance document on preventing and abating ammonia emissions from agricultural sources. Emission flows estimated by linear regression are also comparable with the ones calculated from emission factors defined in the emission inventory of Lombardia region since values of the same order of size are provided by both the methodologies for the main agricultural sources. BAT-tool model (CRPA, 2019), a software developed by LIFE-PRPAIR (Progetto PREPAIR – LIFE15 IPE IT013, https://www.lifeprepair.eu/) project, whose aim is the assessment of NH₃ emissions from intensive pigs and poultry livestock, also calculate ammonia losses comparable to ones got by reverse modelling. Nevertheless, reverse modelling concentrations are more similar to measured data because this procedure is representative of farms analysed and of the period of slurry application, whereas emission inventory and BAT-Tool model estimate the annual mean concentrations. The results of this study show the consistency of reverse modelling procedure, whose main critical issues is overfitting; moreover, if the number of sources is much higher than the number of observations, then air quality dispersion model cannot easily allocate emissions to specific sources.