Mechanistic modelling of cropland and grassland ecosystems: focus on the water cycle and on cattle grazing

Abstract

Climate change stirs up more and more citizens’ interest and concern, and the role of greenhouse gases (GHG) in climate change has now been widely discussed. The agricultural sector is pointed as one of the main causes of climate change, mostly for its emissions of biogenic GHG. However, the impact of ecosystems environment and management practices on these emissions is yet not fully understood, and the response of agroecosystems to a changing environment is still questioned. This context highlights the necessity of studying and understanding ecosystem dynamics in order to design climate change mitigation strategies. To do so, mechanistic models reproducing the carbon, nitrogen and water cycles of ecosystems can be developed. This thesis aims at modifying the ASPECTS model, developed by Rasse et al. (2001) to simulate the evolution of forest stands, into a cropland and grassland model. The main dissimilarities between forests and croplands or grasslands were identified, and the required modifications were implemented in a new version of ASPECTS, called the Terrestrial Agroecosystems Dynamics Analysis (TADA) model. This new model was then calibrated against data acquired in two cropland and grassland sites, both equipped with eddy covariance (EC) systems and meteorological stations. In this work, the attention is paid to the water cycle. Soil evaporation and canopy transpiration were calibrated against evapotranspiration fluxes (ET) measurements, and the dynamics of water infiltration and percolation within the soil profile was compared to measures of soil water content (SWC). Soil evaporation was calibrated during bare soil conditions and resulted in a calibration RMSE of 1.37 and 2.57 mm day-1 and a validation RMSE of 1.82 and 2.75 mm day-1 for, respectively, the cropland and the grassland sites. These results could not be transferred to soils covered with vegetation, making plant transpiration impossible to calibrate. Canopy aerodynamic resistance was identified as a possible cause of this problem and a new methodology is proposed to calibrate these two processes with a wide and diverse dataset in terms of environmental conditions. In addition to this calibration, the new grazing module was tested by comparing measured and modelled grass height. The discrepancies are mainly due to the partitioning of assimilated carbon between the shoot and root compartments and to uncertainties in the estimation of cattle intake capacity. Paths of improvement are provided for a future calibration of this grazing module, considering both available data and potentially measurable variables.