DEVELOPMENT AND EVALUATION OF EMULATORS FOR THE ASSESSMENT OF THE GLOBAL-SCALE ECONOMIC IMPACT OF CLIMATE CHANGE

Abstract

 Process-based simulations of global-scale economic impact of climate change are computationally costly. This limits the number of scenarios that can be explored in the assessment. To overcome this difficulty, in this study, we developed and tested the emulators which mimic the behavior of the simulators with much lower computational cost. The developed emulators could better reproduce the output of the simulators by considering information other than temperature rise and by representing interactions among input variables. The performances of the emulators were better particularly when the effects of climate on the economic outcome are large. The results also suggested that spatially and temporally disaggregated information on the climate condition may further improve the accuracy of emulations.