Abstract:

Dealing with uncertainty in climate-economic modeling is a challenging task and a major point of disagreement in many climate policy studies. Uncertainty exists in both the natural climate system and the socioeconomic system，as well as in their interactions，such as the socio-economic impacts of climate change and human adaptation. In addition，the limited nature of human knowledge（uncertainty）regarding the complex climate economic systems also contributes to uncertainty in modeling. The study of uncertainty in climate-economy complex systems has gained increasing attention since the development of the first fully meaningful climate-economy complex systems model by the Nobel Prize winner in economic sciences， William D Nordhaus，in the early 1990s. With the development of uncertainty decision theory and methodology，climate science， and computational technology， there have been many advances in the study of uncertainty in climate-economy complex systems. At the same time，there are still many challenges in the understanding of the mechanism of the system，model coupling，parameter calibration，solution algorithms and other aspects that need to be overcome. Climate-economy complex systems models generally consist of the climate system module，climate impact module，socio-economic system module，and mitigation and adaptation（climate policy）module，which are coupled into a closed-loop system. The key uncertainties are categorized and classified in these four key modules in the climate economy complex system， the research progress in handling uncertainties is summaried in system modeling，and the future research directions are explored， with a view to better support climate economy modeling and decision-making. In terms of methodologies to address uncertainty，the stochastic dynamic programming approach is gradually replacing the classical sensitivity analysis and Monte Carlo simulation methods for dealing with risk-based uncertainty in climate-economy models；and the decision analysis approach is emerging as a new trend for dealing with deep uncertainties （ambiguities and misspecification） in climate-economy models. In terms of specific modeling of uncertainty： （1） the simple “carbon-climate response” is often incorporated into cost-benefit assessments under uncertainty. （2） stochastic process modeling is commonly used to model uncertainty in economic growth and technological progress. （3） catastrophic risks such as climate tipping points，and the non-economic losses from climate change are gradually incorporated into models. （4） criteria for evaluating utility or social welfare are more focused on inter- and intra-generational equity，as well as the uncertainty averse preferences. （5） endogenous technological progress in emission reduction is more reflected in the model. （6） negative-emission technologies and solar geo-engineering may become hot spots in the modeling of climate-economy complex systems. Finally，responding to climate change is a cross-cutting scientific issue that requires the full cooperation of multidisciplinary experts in earth sciences，economic and management sciences，computational sciences，and other disciplines. The intrinsic characteristics of uncertainty in the climate economic system and the subjective conditions that are not yet fully recognized scientifically make the modeling of the climate-economy under uncertainty face many challenges. Model-solving remains one of the bottlenecks in the development of climate economic modeling under uncertainty. In addition to balancing model complexity，data availability and computational feasibility，future modeling of climate economic uncertainty needs to incorporate the heterogeneity and gaming mechanisms of different levels of decision makers and groups；the analytical framework for climate decision making should not be limited to cost-benefit analysis；and climate economic modeling should also incorporate the political-economic factors as well as the complex impacts of the different players.

Key words: climate change, climate-economy complex system, uncertainty, decision analysis