IAM-FIRE: A Climate Emulator-Based Framework to Project Wildfire Impacts and Risks for Integrated Assessment Models

Area of Science:

• Climate Science and Earth System Modeling
• Environmental Science and Ecology
• Carbon Cycle and Land-Use Change

Background:

• Integrated Assessment Models (IAMs) frequently underestimate dynamic feedbacks from climate-driven disturbances like wildfires.
• This underrepresentation can lead to an overestimation of the permanence of land-based carbon sinks.
• Forest fires are projected to intensify, making their accurate representation in models increasingly critical.

Purpose of the Study:

• To introduce IAM-FIRE (Integrated Assessment Model-Fire Impacts & Risks Emulator), a novel framework for projecting wildfire burned area (BA) and carbon emissions (CE).
• To enable direct projection of wildfire impacts from IAM outputs under various socioeconomic and emissions scenarios.
• To facilitate systematic exploration of fire-climate-land feedbacks and improve assessments of mitigation permanence and climate risks.

Main Methods:

• IAM-FIRE integrates a spatial climate emulator, land-use downscaling, vegetation productivity modeling, and an empirical fire model.
• The framework generates global annual wildfire impacts at 0.5° resolution for 2020-2100.
• Model calibration was performed against GFEDv5 observations, using Global Change Analysis Model (GCAM) outputs for four Shared Socioeconomic Pathways (SSPs).

Main Results:

• IAM-FIRE accurately reproduces historical global trends in total and forest burned area, including the recent observed decline.
• Projected total burned area by 2100 varies significantly across scenarios, from 441 Mha/year (SSP1-2.6) to 794 Mha/year (SSP3-6.6).
• Projected total carbon emissions by 2100 range from 1.8 PgC/year (SSP1-2.6) to 3.6 PgC/year (SSP5-7.6), showing substantial divergence.
• Socioeconomic development primarily suppresses wildfire impacts, while climate change and CO2-driven vegetation growth amplify fire risk.

Conclusions:

• IAM-FIRE demonstrates greater sensitivity to radiative forcing and a stronger role for human fire suppression compared to CMIP6 and FireMIP models.
• The study highlights structural uncertainties in current wildfire projections.
• The developed framework provides a crucial tool for understanding future fire dynamics and their implications for climate change mitigation and risk assessment.