Institute of Cartography and Geoinformatics Research Big Data and Machine Learning
AgrImOnIA: The impact of agriculture on air quality and the COVID-19 pandemic

AgrImOnIA: The impact of agriculture on air quality and the COVID-19 pandemic

AgrImOnIA is the project short name, being the acronym of Agricultural Impact On Italian Air. Also, Agrimonia eupatoria or Common agrimony is a plant and a flower spread in Alpine, Mediterranean and Central Europe, which is the positive symbol of this project. Lombardy has about 10 Million inhabitants and is the first Italian region for agriculture production. Bovines are about 1.5 Million and swine are about 4 Million. On average, there is a swine every 2.5 inhabitants and a bovine every 6.7 inhabitants. Considering that agriculture land is 69% of the region, there are about 245 swine and 92 bovines per rural km2. There is a large scientific consensus on the fact that livestock and fertilizers are responsible for about 95% of the ammonia emissions. After some reactions in the atmosphere, ammonia mainly turns to fine particulate matters known as PM2.5. Since PM2.5 is quite stable in the atmosphere, due to the limited air circulation in the Po Valley, the air quality is often affected by high concentrations of such pollutants. Adding the contributions of vehicle traffic and house heating, the plain of Lombardy is one of the most polluted areas in Europe. As a result, epidemiological studies found Lombardy to be the region with the highest PM2.5-related mortality rate: 164 deaths per 100 thousand inhabitants.

Using data science, AgrImOnIA aims at making a further step by estimating the local impact of ammonia to PM2.5. This additional information is crucial for the decision maker that has to prioritise the interventions. In fact, the project’s final output will be a set of scenario analysis based on spatiotemporal varying scenarios, displaying the impact of the emission changes in high resolution maps. In particular, the five Shared Socioeconomic Pathways (SSPs), used by the Intergovernmental Panel for Climate Change (IPCC) will be considered. Moreover, cases of efficient manure management, precision agriculture and a transition to organic cultivations agreed with the stakeholders will be developed.

From the technical point of view, the above results will be based will be a statistical model, capable of emulating the chemical-physical processes developing in the atmosphere, and doing scenario analysis, uncertainty included. This emulator can be considered an important intermediate output, available for further post-project uses.

AgrImOnIA will obtain these results using advanced data science methods, namely high-dimensional geostatistics and machine learning. The statistical models will be trained using a diverse blend of large datasets. Data sources will include traditional air quality monitoring networks and emission inventories. Also innovative sources such as the European Copernicus Programme (https:\\ and crowd sourced data will be used, notably: time-varying emission inventories, (reanalysis) meteorological data, satellite land use, and crowd sourced mobility data from a citizen science community.

A natural experiment with two factors will be considered to validate our approach. The first factor is the COVID-19 lockdown. Indeed, during the lockdown, one of the primary emission sources (vehicle traffic) has been vastly reduced, and agriculture emissions played a magnified effect on pollutant concentrations. This will enable us to understand if our model can separate the two effects of traffic and agriculture. The second factor is related to geography and atmospheric circulation. We will compare the behaviour of our model in Lombardy and in the German region of Lower Saxony. That is of particular interest because size, population, infrastructures, economic sectors, and agriculture production structure are similar. Vice versa, geography and climate are different. Thus the impact due to the specific climate and geography of Lombardy (compared to Lower Saxony) can be quantified.