Wildfires play a critical role in shaping terrestrial ecosystems and influencing the Earth's climate through intricate feedback mechanisms. However, existing Earth System Models (ESMs), including the advanced Exascale Energy Earth System Model (E3SM), struggle to accurately represent wildfire dynamics and their extensive impacts on land-atmosphere interactions. This project seeks to address these challenges by developing innovative benchmarking metrics for land-atmosphere feedback, which will be integrated into the International Land Model Benchmarking (ILAMB) framework. Moreover, the project will leverage machine learning and surrogate modeling techniques to improve E3SM's wildfire simulations. Comprehensive analyses and factorial experiments will be conducted to reduce uncertainty in the model's predictions of ecosystem post-fire resilience and the consequent atmospheric feedback. By filling these critical gaps in ESMs, the project aims to deepen our understanding of the interactions between wildfires, terrestrial ecosystems, and the global climate, thereby supporting the creation of more effective mitigation and adaptation strategies for climate change.