Large scale spatial modelling of soil organic carbon dynamics
Under the Kyoto Protocol, participating nations are required to reduce National CO₂emissions according to their 'reduction commitment' or 'quantified emissions limitation', over the first commitment period, 2008-2012. One way in which nations could achieve this would be by increasing soil carbon storage through different management practices. Most former estimates of regional scale C sequestration potential have made use of either linear regressions based on long-term experimental data, whilst some have used dynamic soil organic matter (SOM) models linked to spatial databases. Few studies have compared these two methods, and none have compared regressions with two different SOM models. This thesis presents a case study investigation of the potential of different land management practices to sequester carbon in soil in arable land, and preliminary estimates of other potential C savings. Two dynamic SOM models were chosen for this study, RothC (a soil process model) and CENTURY (a general ecosystem model). RothC and CENTURY are the two most widely used and validated SOM models world-wide. Methods were developed to enhance use and comparability of the models in a predictive mode. These methods included a) estimation of the IOM pool for RothC, b) estimation of C inputs to soil, c) investigation of pool size distributions in CENTURY, and d) creation of a program to allow use of C inputs derived from CENTURY with the RothC model. This thesis has also investigated the importance of errors in C inputs to soil for predictive SOM modelling, and performed sensitivity analyses to investigate how errors in setting the size refractory SOM pools might affect predictions of SOC. RothC and CENTURY were compared at the site scale using datasets from seven European long-term experiments, in order to a) verify their ability to predict SOC changes under changes in land use and management relevant to studies of C sequestration potential, b) evaluate model performance under European climatic conditions, and c) compare the performance of the two models. Finally, a Geographic Information System (GIS) containing soil, land use and climate layers, was assembled for a case study region in Central Hungary. GIS interfaces were developed for the RothC and CENTURY models, thus linking them to spatial datasets at the regional level. This allowed a comparison of estimates of the C sequestration potential of different land management practices obtained using the two models and using regression-based estimates. Although estimates obtained by the different approaches were of the same order of magnitude, differences were observed. Encouragingly, some of the land management scenarios studied here showed sufficient C mitigation potential to meet Hungarian CO₂reduction commitments.