Scaling Soil Greenhouse Gas Emissions to the National Level

Agricultural soils are a major source of greenhouse gas (GHG) emissions globally. In Ireland, 81% of the land is devoted to agriculture, which means that there is potential for significant mitigation of agricultural GHG emissions through land-use change. However, current tools for assessing GHG emission savings through land-use change – Intergovernmental Panel on Climate Change Tier 1 and Tier 2 – are limited; Tier 3 approaches, in which process-based models are used to estimate GHG emissions for given land-use and climatic scenarios, are more flexible. This project was concerned with improving the national inventory of GHG emissions from Irish soils through the use of Tier 2 and 3 methodologies, effectively upscaling regional data on N2 O and CO2 (ecosystem respiration, Recos) fluxes to the national level through a combined process-based model and geographic information system (GIS) approach. A 5 × 5 km GIS map framework for Ireland has been successfully developed that will allow calculation of nationwide annual emissions of N2 O and CO2 (Recos) from grasslands and arable soils. This has been linked to climate, land use and soil type using the Irish Soil Information System (SIS). Preliminary runs for background emissions of N2 O and CO2 have been carried out using soil and climate parameters as the main drivers and using DailyDayCent, DeNitrification DeComposition (DNDC 9.5) and ECOSSE (Estimation of Carbon in Organic Soils – Sequestration and Emissions) as simulation models. The total area of grassland and cropland calculated from the GIS maps was 3 and 0.3 million hectares, respectively, which is in close agreement with the total area of cropland and pasture land recorded in the 2016 census (3.9 million hectares). Upscaling DNDC 9.5 outputs using the GIS map gave combined Recos and N2 O background emissions of between 0.45 and 0.5 Mt CO2 e for grassland and between 0.074 and 0.08 Mt CO2 e for arable land. These are in broad agreement with inventory values, considering that the effects of fertiliser additions and management were not considered. Using an extensive validation dataset, we evaluated the capability of the four process-based models representative of the current ecosystem of biogeochemical models – DailyDayCent, DNDC 9.4 and 9.5, and ECOSSE – for simulating soil N2 O emissions and Recos at a range of arable and grassland agricultural sites. The majority of model simulations underestimated cumulative N2O emissions and many performed poorly when used to simulate sites with no fertiliser additions. Simulation performance was highly site specific but, in general, a combination of the DailyDayCent and ECOSSE models performed well at arable sites, whereas both versions of the DNDC model were useful for simulating grassland N2 O emissions. The model performance of DNDC 9.4 and 9.5 was significantly better for Recos than for N2 O, producing low root mean square error values, correlation coefficients approaching 90% and final simulated cumulative fluxes that followed measured values closely. In this case, we suggest a combination of DNDC 9.4 and 9.5 as suitable for providing the most reliable estimates of Recos flux. We performed Monte Carlo simulations and determined that, in general, the variation in precipitation and temperature made the greatest contribution to uncertainty in the model outputs. Although we have successfully produced a workable GIS map framework for calculation of nationwide fluxes of N2 O and CO2 from grassland and arable systems, the fit of modelled to measured N2 O emissions is so poor as to suggest that a simpler empirical approach should be used to upscale N2 O fluxes to the national level. This study is the first of its kind in Ireland and represents a useful objective baseline from which to establish priorities for model parameterisation and improvements for application in Irish agricultural systems