Improving Spatial and Temporal Variation of Ammonia Emissions for the Netherlands Using Livestock Housing Information and a Sentinel-2-Derived Crop Map

Ammonia emissions to the atmosphere have a range of negative impacts on environmental quality, human health, and biodiversity. Despite the considerable efforts in quantifying spatially explicit ammonia emissions, there are significant uncertainties in ammonia emission estimates at regional scales. We aimed to improve the modeling of atmospheric ammonia emission variability in space and time across the Netherlands by updating an agricultural ammonia emission model using a newly derived high-resolution crop map and animal housing locations in the Netherlands. We applied a random forest classification to the multi-temporal features of the spectral metrics and vegetation indices derived from Sentinel-2 to generate a high-resolution crop map of 12 agricultural land cover classes. The crop statistics per region were used to calculate manure and mineral fertilizer and ammonia emission distribution based on nitrogen demand of different crop types using the INTEGRATOR model. Next, the crop map was used to spatially allocate the ammonia emissions to a high-resolution grid across the Netherlands. In addition, point sources emissions from animal housing and manure storage systems were introduced as point sources using data from the GIAB (Geographic Information Agricultural Business) system. The temporal emission variability was updated using a recently developed TIMELINES module.The derived crop map of the Netherlands has an average accuracy score of 0.73. It was compared with Dutch national statistics, and the results showed that the relative differences of the crop map statistics lie within 5% except for wheat. Using the new crop and housing information in CTM LOTOS-EUROS, the modeled monthly ammonia surface concentrations compared better with observations of the temporal variation of ammonia concentrations than those derived with the original model, indicating that the spatial distribution of ammonia emissions was improved. The model captured the magnitude and temporal variability presented in the in situ measurements if either field application or animal housing dominates local emission. The comparison of modeled and measured annual averaged surface concentrations indicated that the spatial distribution of ammonia emission was also improved. All error measures significantly dropped, and the performance of the updated model was more stable and robust. The improvement is more evident at the stations where housing dominates emissions than at other stations. This study illustrates that apart from including satellite-derived crop distributions, information on the locations of housing systems also plays an essential role in improving the spatial and temporal distribution of ammonia emissions by modeling and that it can be worthwhile to extrapolate the method to other regions in Europe and elsewhere