Overhead Transmission Line Fragility, Climate-Induced Risks and Geospatial Artificial Intelligence: A Systematic Bibliometric Review

Imagine a vast, invisible web of power lines stretching across continents, carrying the electricity that lights our homes and powers our cities. For decades, engineers have designed these lines to withstand the weather they knew: the typical storms, the usual heat, and the standard winds. But the world is changing. The climate is shifting, bringing more frequent and severe extremes—intense heatwaves, fierce storms, and raging wildfires—that these old designs were never built to handle. When these lines fail, the result is not just a flickering light; it is widespread blackouts, economic loss, and a threat to the safety of communities. To protect this critical infrastructure, scientists are turning to a powerful new tool: artificial intelligence. This technology can learn from massive amounts of data to predict where and when a line might break, but it needs to understand the complex story of the environment around it to work effectively.

A team of researchers from the University of Pretoria in South Africa set out to examine how well the scientific world is connecting these two worlds: the high-tech study of artificial intelligence and the physical reality of climate change. They conducted a systematic review, a methodical search through thousands of scientific papers published between 2004 and 2026. Their goal was to see if researchers were successfully combining computer models that predict line failures with the real-world data of weather and geography. They looked at global trends, who was doing the research, and what topics were being discussed most often. They wanted to know if the tools being built were actually ready to help power companies survive the changing climate.

The researchers found a landscape that is growing fast but remains fragmented. The number of studies on this topic has exploded in recent years, particularly since 2022, driven by a surge in interest in using machine learning to detect faults. However, the review revealed a significant gap. While there is a strong dominance of research on AI, fault diagnosis, and power-system engineering, there is limited integration of climate, environmental, and geoscience perspectives. Most studies focus on just one aspect. Some look purely at the electrical engineering side, using computers to spot a broken wire, while others look at the weather side, studying how wind or heat stresses a tower. Very few studies successfully fuse these perspectives into a single, complete picture that accounts for how climate and environment work together to cause failures.

This disconnect is not just a matter of academic interest; it has real-world consequences. The analysis showed that the research is heavily concentrated in a few specific regions, with China, the United States, and the United Kingdom leading the way. In contrast, the African continent, where the need for resilient power systems is urgent and where the climate is changing rapidly, is underrepresented in the global research map. The authors noted that while South Africa's power utility, Eskom, manages thousands of kilometers of lines that are increasingly vulnerable to extreme weather, there is very little published research that addresses the specific challenges of this region. This leaves a dangerous blind spot, as the literature remains fragmented and lacks comprehensive end-to-end geospatial artificial intelligence (Geo-AI) frameworks that adequately address the cumulative multi-hazard impacts of climate and environmental variables on overhead transmission line reliability.

The study also highlighted that current methods for predicting failures are often too simple. Many existing models look at a single event, such as one storm or one heatwave, rather than understanding the cumulative damage caused by years of changing conditions. They tend to treat the environment as a background factor rather than a central player in the story of why a line fails. The researchers found that while there are advanced tools like deep learning and digital twins—virtual models that mimic real-world systems—these are not yet being used in a way that fully integrates the messy, complex data of the natural world. The result is a collection of fragmented studies that offer pieces of the puzzle but not the whole picture.

Ultimately, the paper concludes that the field is at a critical juncture. The technology to build smarter, more resilient power grids exists, but the scientific community has not yet built the bridges needed to connect the data of climate science with the algorithms of artificial intelligence. The authors argue that for power systems to survive the future, we need a new approach that treats the environment and the technology as one interconnected system. This means developing frameworks that can track how a combination of heat, wind, and vegetation over time weakens a line, and then using that knowledge to guide maintenance before a failure occurs. Until this integration happens, the global effort to secure our power grids against a changing climate will remain incomplete, leaving many communities vulnerable to the next big storm.