Evaluation of GCAM Agricultural Land Use Modeling: A Hindcasting Experiment Comparing Model Behavior to History

Monday, May 12, 2014 - 07:00
Add to Calendar

The Global Change Assessment Model (GCAM) is an integrated assessment model linking global energy, agriculture, land use, emissions, and climate systems. GCAM currently divides the world into 151 land subregions determined by geopolitical boundaries and agro-ecological zones within those boundaries. Within each of the land subregions, GCAM performs an economic allocation of land uses among crop production, forests, pasture, and several non-commercial land cover categories based on relative profit rates, both market-based and implied, with a market equilibrium approach. Non-linear profit functions are calibrated to reproduce a historical base period. Future model period behavior is not constrained but instead evolves from history when changes in key economic drivers such as technologies, policies, climate, and other conditions change the relative profits of alternative land uses. As part of a larger effort towards model evaluation, we present here a hindcasting experiment where we compare GCAM model land use allocations to history. For this study, we will construct scenarios to evaluate GCAM global and regional land use modeling against the last two decades of historical data. Specifically, we will perform our standard calibration process with historical data for the year 1990, and then run GCAM forward annually through the year 2010 without updating the calibration for more recent data. We will compare the time trajectory of modeled global and select regional results for land use and crop production to historical data. We will identify reasons for any major differences from history and explore what differences are due to short-term effects and what may be more relevant for improving decisions in long-term integrated assessment modeling.