Temperature and heat are difficult concepts for children to grasp due to their abstractness. An image-to-writing approach, guided by the visualisation practices of scientists, was designed to engage elementary students with constructing images to represent their ideas about phenomena and translating these images into text using scientific terminologies. Taking a quasi-experimental approach, the experimental group students received inquiry-based instruction based on the image-to-writing approach, while the control group students received a mix of direct instruction and inquiry activities without explicit focus on multimodal representations. An instrument consisting of four free response questions was developed and administered to 129 primary 4 students (aged 9-10) before (pre-test) and after (post-test) instruction to determine their conceptual understanding and representational competences. ANCOVA showed that students in the experimental group perform significantly better than those in the control group in their conceptual understanding. Further analysis revealed that a larger percentage of students in the experimental group demonstrated higher levels of conceptual understanding after instruction, compared to the control group for more complex phenomena, even though both groups showed similar levels of representational competences. The findings suggest that an image-to-writing approach can help students develop deeper conceptual understanding as well as use representations to demonstrate their conceptual understanding. The use of images could have helped students in their thinking and learning of complex phenomena, which allowed them to better convey their understanding of the concepts.