Modelling economic and biophysical drivers of agricultural land-use change. Calibration and evaluation of the Nexus Land-Use model over 1961–2006

Other literature type, Article English OPEN
Souty, F. ; Dorin, B. ; Brunelle, T. ; Dumas, P. ; Ciais, P. (2013)
  • Journal: (issn: 1991-9603, eissn: 1991-9603)
  • Related identifiers: doi: 10.5194/gmdd-6-6975-2013
  • Subject: | | Modèle mathématique | Prix | E11 - Economie et politique foncières | | Rendement des cultures | Démographie | F01 - Culture des plantes | Agriculture | Conditions météorologiques | Terre cultivée | | | | Pâturages | Engrais | | | | | Modèle économétrique | Modèle de simulation | U10 - Méthodes mathématiques et statistiques | Utilisation des terres | A01 - Agriculture - Considérations générales | | | Évaluation | | Élevage | Environnement socioéconomique | | L01 - Elevage - Considérations générales | |

The central role of land-use change in the Earth System and its implications for food security, biodiversity and climate has spurred the development of global models that combine economical and agro-ecological drivers and constraints. With such a development of integrated approaches, evaluating the performance of global models of land-use against observed historical changes recorded by agricultural data becomes increasingly challenging. The Nexus Land-Use model is an example of land-use model integrating both biophysical and economical processes and constraints. This paper is an attempt to evaluate its ability to simulate historical agricultural land-use changes over 12 large but economically coherent regions of the world since 1961. The evaluation focuses on the intensification vs. extensification response of crop and livestock production in response to changes of socio-economic drivers over time, such as fertiliser price, population and diet. We examine how well the Nexus model can reproduce annual observation-based estimates of cropland vs. pasture areas from 1961 to 2006. Food trade, consumption of fertilisers and food price are also evaluated against historical data. Over the 12 regions considered, the total relative error on simulated cropland area is 2% yr<sup>&minus;1</sup> over 1980–2006. During the period 1961–2006, the error is larger (4% yr<sup>&minus;1</sup>) due to an overestimation of the cropland area in China and Former Soviet Union over 1961–1980. Food prices tend to be underestimated while the performances of the trade module vary widely among regions (net imports are underestimated in Western countries at the expense of Brazil and Asia). Finally, a sensitivity analysis over a sample of input datasets provides some insights on the robustness of this evaluation.
Share - Bookmark