Spatial Drivers of Atmospheric NH3 in North-West Europe Intern

Laboratory: Centre of Education, Research and Innovation on Energy and Environment (CERI EE)

Supervisors: Anna FONT

Type of contract and duration: Research internship – 6 months

Workplace: Bourseul Campus, Douai, France

Context:

IMT Nord Europe is a public higher-education and research institution belonging to the Institut Mines-Télécom (IMT) group, under the supervision of the French Ministry of Industry. Its missions include engineering education, scientific research, and innovation in support of sustainable territorial development.
For more details, please visit: www.imt-nord-europe.fr/en

This internship will be carried out within the CERI EE at IMT Nord Europe, in the research axis “Observations, Sources and Processes in the Atmosphere” (OSPA). The host team develops advanced experimental and data-analysis approaches to characterize atmospheric aerosols, their sources, and their impacts on air quality and health.

The work will fit within the CDP AREA project (Aérosols au cœur du système Terre-Atmosphère), funded by the University of Lille. The AREA project aims to improve the measurement of complex interactions between aerosols and gaseous precursors with the environment across different spatial and temporal scales. Its ultimate goal is to deepen our understanding of aerosol behaviour in the atmosphere and improve atmospheric and climate models.

Ammonia (NH3) is the most abundant alkaline gas in the atmosphere playing a crucial role in the neutralization of atmospheric acids leading to particle formation (Seinfeld and Pandis, 2006). Further, wet and dry deposition of NH3 to soils, water bodies and ecosystems leads to acidification, eutrophication and biodiversity loss (Asman et al., 1998). Most NH3 emissions in Europe are from agricultural and farming activities but also waste handling, road transportation and residential heating (EEA, 2018). The increase in temperature associated with global warming is expected to increase NH3 emissions worldwide in the next century (Jiang et al., 2026) with new potential emissions such as green fuel. Therefore, it is crucial to identify and quantify the drivers of NH3 atmospheric concentrations (emissions, meteorology and other environmental factors) to define efficient NH3 reduction policies.

Ground-level NH3 monitoring is sparse. The longest and widest NH3 monitoring in Europe are in the Netherlands (van Zanten et al., 2017) and in the United Kingdom (Tang et al., 2018) with programs starting in the mid-1990s. However, the number of monitoring programs have increased and currently exist in Belgium, France and Spain (Liu et al., 2024). Despite increasing ground-level monitoring, the influence of non-agricultural emissions on atmospheric NH3 is still unknown. To overcome the spatial limitations of ground-level NH3 monitoring, satellite observations could be used to fill the gaps. Currently, NH3 from IASI (Van Damme et al., 2015) and CrIS (Shephard et al., 2020) are available from 2008 and 2012, respectively, and proven to observed consistent temporal and spatial patterns (Guo et al., 2021; Wang et al., 2023). Machine-learning methods applied to both ground-level and satellite data have proved to inform about the factors driving the spatial NH3 variability (Wang et al., 2021, 2026).

This M2 internship aims to develop a machine-learning analysis framework to understand the driving mechanisms controlling atmospheric NH3 concentrations in north-west Europe using a combination of in-situ ground-level and satellite data. The analysis is expected to help us understand which drivers explain the spatial distribution of NH3 concentrations, including meteorological conditions, environmental factors and anthropogenic emissions.

Missions:

The internship will follow these steps:

• Literature review: Investigate existing statistical modelling and machine-learning methods (e.g., random forests, gradient boosting, etc.) for analysing the drivers of NH₃ spatial distribution;

• Exploratory data analysis: Characterize available datasets (e.g., NH₃ concentration measurements from ground-level and satellite data, meteorological data, land use, agricultural emissions) and study correlations between variables;

• Methodological development: Process and implement the modelling framework; select relevant features; implement and optimize machine-learning models to predict or explain NH₃ distribution;

• Interpretation and visualization: map results (e.g., heatmaps of NH₃ concentration, feature importance by area); policy implications, etc.