Global warming and the escalating climate crisis necessitate a paradigm shift in science education, moving beyond rote conceptual mastery toward fostering climate literacy and adaptive critical thinking, essential competencies for sustainable decision-making. This study investigates the impact of a Deep Learning-integrated Problem-Based Learning (DL–PBL) model within heat transfer instruction on developing adaptive critical thinking skills among elementary school students, from which contextualized climate literacy is systematically mapped and evaluated. Utilizing a quasi-experimental research framework with a non-equivalent control group pretest–posttest design, this study engaged 50 fourth-grade students. The experimental group received the DL–PBL intervention, which anchors instructional delivery in mindful, joyful, and meaningful learning principles, while the control group followed conventional instruction. Data collection relied on a validated critical thinking instrument contextualized within thermal physics phenomena and global warming scenarios. Statistical analysis indicated that the experimental group achieved significantly higher post-test scores (M=81.6) than the control cohort (M=64.4; p<0.001). The most substantial improvements occurred within the Clarifying and Interpretation and Inference domains. The primary novelty of this study lies in its innovative reframing of classical critical thinking indicators into specialized, "adaptive" competencies, providing an empirical model where action-oriented climate literacy is directly evaluated through the student's capacity to decode thermal data and formulate concrete mitigation strategies. Ultimately, this research demonstrates that the unified DL–PBL model serves as a transformative framework for bridging abstract physics education with Education for Sustainable Development (ESD), offering a strategic pathway toward realizing Sustainable Development Goals (SDG) 4 and 13 at the primary education level.