<p>Understanding the complex combustion dynamics within scramjet engines is critical for advancing high-speed propulsion technologies. However, the large-scale, high-dimensional nature of temporal flow field data from numerical simulations presents significant challenges for visual interpretation, feature differentiation, and cross-case comparison. In this paper, we present TemporalFlowViz, a parameter-aware visual analytics workflow and system designed to support expert-driven clustering, visualization, and interpretation of temporal flow fields derived from scramjet combustion simulations. Our approach leverages hundreds of simulated combustion cases with varying initial conditions, each yielding temporal sequences of flow field images. We employ pretrained Vision Transformers to extract high-dimensional embeddings from flow field frames, apply dimensionality reduction and density-based clustering to uncover latent combustion modes, and construct temporal trajectories in the embedding space to track the evolution of each simulation over time. To bridge the gap between latent representations and domain knowledge, specialists annotate representative cluster centroids with descriptive labels. These annotations serve as contextual prompts for a vision-language model, which generates natural language summaries for individual frames or complete simulation cases. Furthermore, the system supports parameter-based case filtering, similarity-based case retrieval, and coordinated multi-view exploration to facilitate in-depth analysis. We demonstrate the effectiveness of TemporalFlowViz through two expert-informed case studies and domain expert feedback, showing that TemporalFlowViz facilitates hypothesis generation, supports interpretable pattern discovery, and accelerates knowledge discovery in large-scale scramjet combustion analysis.</p> Graphical abstract <p></p>