This work proposes a framework to simplify access to STAC data under the concept of “Bridging Geospatial Technology and Humanity,” with the aim of delivering timely information and support to people, particularly during disaster situations where every second matters.

A key element of modern data management is the open-source STAC standard, which can be compared to an “intelligent knowledge repository” that systematically records changes occurring in disaster-affected areas. In Thailand, the Geo-Informatics and Space Technology Development Agency (GISTDA) has adopted this standard as a core component of its disaster management initiatives to organize spatial data related to floods, hotspots, and drought conditions. This approach enables decision-support systems to access critical information more efficiently while significantly reducing the time required for data retrieval.

To address the challenge of accessibility, we developed a human-centered AI communication approach that leverages the Gemma 3 large language model (LLM), deployed via Ollama, as the interaction and orchestration layer. The system is designed to receive natural language input from users and analyze both spatial and temporal context. It then automatically translates the request into STAC API-compliant query commands. This process removes the traditional dependency on technical experts and reduces the complexity of geospatial data discovery, allowing users to access information simply through natural-language questions and responses.

Once the system retrieves vector coordinates and satellite imagery (raster data) from GISTDA’s data repositories, the information—typically in the form of GeoJSON or metadata—is passed to a dynamic visualization process. Olama performs spatial statistical analysis to determine the event bounding box and generates JSON-based style instructions for MapLibre GL JS, which automatically configures the layer structure using vector tiles. This enables the creation of interactive maps that can instantly visualize geospatial information through a simple prompt.
The system then overlays risk-related layers, such as flooded areas or thermal hotspots, onto the map along with easily interpretable statistical summaries. As a result, complex geospatial datasets are transformed into clear visual intelligence, enabling decision-makers to quickly and accurately assess ongoing crisis situations.

This work demonstrates that when advanced geospatial technologies are combined with a deep understanding of human needs, and mediated through intelligent systems such as Olama, it becomes possible to create innovations that go beyond simply storing data. Instead, these systems can truly “listen to” and “assist” people, ultimately contributing to the safety and long-term sustainability of Thai society in the future.