Measuring the environmental impact of computational processes is a crucial step towards developing more sustainable digital technologies. While artificial intelligence (AI) is playing a transformative role in various fields, its energy-intensive nature gives way to concerns about carbon emissions. In this paper, we present a hybrid approach that combines a weighted average framework and a dynamic model matching methodology to improve the accuracy and adaptability of carbon footprint estimates. The weighted average method integrates multiple carbon tracking tools and assigns weights based on reliability, accuracy and applicability to provide a unified emissions estimate. Meanwhile, the dynamic method categorizes AI models based on their computational characteristics and assigns them to the most appropriate emissions estimation framework. Future research will focus on refining weighting strategies, integrating real-time energy consumption data, and embedding sustainability considerations into AI development workflows. Our proposed methodology contributes to a more standardized and comprehensive assessment of the environmental impact of AI, encouraging responsible AI innovation.