LLM anwers to earth observation questions

For testing Retrieval Augmented Generation (RAG) systems, we used GPT-4o to automatically generate a set of 70 prompts specific to the earth observation domain. In order to ensure that the generated questions are realistic and aligned with our data, we incorporate two variables within the generation prompt: (1) Question Topic: a topic from the NASA GCMD taxonomy, and (2) question intention, i.e. exploring, comparing, causality, describing, relating. An example for such a combination is topic = Climate Indicators, question_type = Descriptive. The variables are randomly chosen to create the following prompts for question generation: System PromptYou are an expert scientist and a critical thinker. Your task is to generate a realistic, concise scientific question with given criteria. User Prompt ### Question Criteria: INTENT: {intent_category} — {intent_description} TOPIC: The topics are extracted from the NASA GCMD Taxonomy. The TOPIC is described and few of its SUBJECT AREAS. <topic> {context} </topic> ### Instructions The question must strictly reflect the stated INTENT and be centered on the specified TOPIC along with its SUBJECT AREAS. Do not include any background or explanation—just the question. This question will be used to evaluate the performance of a scientific QA system under Retrieval-Augmented Generation (RAG) and zero-shot prompting conditions. ### Question: The informtation about this dataset will be updated soon once a related paper has been finished.