Prompt-Driven Map Generation: A Generative GeoAI Framework for Automated Spatial Analytics and Dynamic Cartographic Visualization

The steep learning curves and high workflow latency inherent to traditional Geographic Information Systems (GIS) present significant accessibility barriers for decision-makers in urban planning and environmental monitoring. This research introduces a novel Generative GeoAI framework designed to democratize multi-dimensional spatial analytics through a conversational interface. The proposed architecture utilizes a Large Language Model (Google Gemini 1.5 Flash), operating strictly as a semantic logical compiler rather than a standard text generator. This model translates unstructured natural language queries into a rigorous JSON schema, which subsequently functions as an execution token for an asynchronous Python backend. This backend concurrently harvests planetary-scale, multispectral raster data via Google Earth Engine (e.g., Sentinel-2, MODIS) and high-fidelity vector topology via OpenStreetMap. A primary contribution of this framework is the automated mathematical fusion of these disparate datasets, alongside the programmatic enforcement of international cartographic standards, including dynamic scale bars, coordinate graticules, and native 300 DPI high-fidelity exports. To validate the system’s operational efficacy, a case study was executed for the Ahmedabad region, India. The pipeline successfully generated complex raster–vector fusions—overlaying critical administrative infrastructure onto multispectral imagery—and automated a Normalized Difference Vegetation Index (NDVI) mapping workflow. Crucially, the framework condensed a multi-hour manual data-harvesting process into a 15-second execution sequence. By bridging the gap between natural language and satellite telemetry, this conversational spatial intelligence ecosystem provides a highly scalable, real-time foundation for rapid disaster response, precision agriculture, and continuous planetary monitoring.