Constraining uncertainty of modelled N2O emissions using isotopic composition

Process-based models provide an avenue to assess greenhouse gas emissions on scales where measurements alone are impractical. However, outcomes obtained from models are also subject to sources of error, which include model parameter uncertainty, model input uncertainty and model bias. For process-based models, the high-dimensional parameter spaces lead to large uncertainty contributions arising from model parameterization. Here, we show how time series measurements of 15N intramolecular N2O isotopic composition, i.e., site preference (SP), can be used to constrain parameter uncertainty in the process-based biogeochemical model LandscapeDNDC in connection with the Stable Isotope MOdel for Nutrient cyclEs (SIMONE). We develop a multivariable calibration framework that incorporates isotope tracing simulations from SIMONE into the calibration of LandscapeDNDC parameters, based on measurements of both N2O and SP simultaneously. We perform site-scale calibrations using the SP and N2O flux measurements from Swiss grassland at Chamau, and use a Sampling Importance Resampling scheme to estimate model parameter uncertainties, both with and without using SP as a calibration variable. Our results show that including SP into a calibration-uncertainty estimation framework for N2O emissions significantly reduces model parameter uncertainty.