This paper reviews the opportunities and constraints of applying daytime satellite imagery and Geospatial

Artificial Intelligence(GeoAI) in economic geography. Reliable and timely socio-economic statistics are often

unavailable in many regions due to institutional, financial, and political constraints, thereby limiting spatial

economic analysis and evidence-based policymaking. Recent advances in computer vision and the increasing

availability of high-resolution daytime satellite imagery offer a promising alternative by enabling the extraction

of spatial indicators related to economic activity at fine spatial and temporal scales. The paper organizes the

existing GeoAI-based approaches into three strands: (1) object-based image analysis, (2) socio-economic indicator

estimation, and (3) multimodal approaches with large language models(LLMs). It further highlights key

limitations of these approaches, particularly their limited ability to capture economic decline. I conclude that

GeoAI-based remote sensing constitutes a transformative complementary tool for advancing methods in economic

geography, especially when combined with nighttime imagery and other sensor data.



KeyWords : GeoAI, remote sensing, computer vision, machine learning, indicator estimation