Established a Damage Prediction Technology to Visualize Road Disruption Risk During Heavy Rainfall
Quick and simple nationwide road assessment without on-site surveys

News Highlights:

Figure 1 Overview

Background

In recent years, heavy rainfall disasters have intensified, and there have been increasing cases of tilted utility poles and severed communication cables caused by landslides and embankment collapses during such events. To enable proactive responses to these disasters, NTT has developed a technology that predicts utility pole damage using publicly available data and past inspection records. This technology makes it possible to simulate and predict utility pole damage across Japan without fieldwork, using only desktop analysis.³

Meanwhile, heavy rainfall disasters not only damage communication infrastructure but can also lead to landslides and embankment failures on roads, causing entire communities to become isolated. While it is important to assess such risks in advance and take preventive measures, creating hazard maps requires on-site surveys and trained engineers, making it difficult to forecast damage across all roads before disasters occur.

Technology Highlights

Building on this background, NTT has developed a foundational technology for predicting road damage by adapting its previously established utility pole damage prediction model. As shown in Figure 2, virtual utility poles are placed along the roads to be analyzed, and the damage prediction model is applied using only the road's location data. This makes it possible to quantitatively assess road damage risk on a grid-by-grid basis (10-meter resolution).

Since many utility pole failures are caused by road collapses or falling trees due to landslides, this approach leverages the similarity in damage factors between poles and roads to enable accurate predictions.

Figure 2 Overview of Road Damage Prediction

Overview of the Experiment

Using a prediction model trained on inspection data that recorded whether utility poles were damaged during a heavy rainfall event in July 2020, along with surrounding environmental data such as elevation, we predicted part of the road damage caused by the heavy rainfall in the Noto Peninsula in September 2024, and quantified the associated risk.

Approximately 122,000 virtual utility poles were placed along roads and surrounding areas within the target region. The model's predictions of damage to each pole were then compared with slope failure, debris flow, and sediment distribution data⁴. When using hazard maps for interpretation, the accuracy was 77%, correctly identifying approximately 94,000 poles. In contrast, our technology achieved a higher accuracy of 80%, correctly identifying approximately 98,000 poles, confirming the superior performance of this prediction model. Additionally, we confirmed that this technology was able to successfully predict damage in certain areas within the region that currently lack hazard map coverage but were nonetheless affected by the disaster.

Future Developments

We aim to build partnerships with private companies and local governments to conduct trial implementations of this technology for use in real evacuation planning and pre-disaster planning. At the same time, we will continue research and development to expand damage prediction not only for heavy rainfall but also for earthquakes, as well as to apply the technology to other types of infrastructure. By leveraging the data that NTT has accumulated over time, we strive to contribute to disaster prevention, damage mitigation, and rapid recovery of social infrastructure, ultimately working toward the realization of a more disaster-resilient society.

[Glossary]

¹Hazard map
Refers to areas at risk of landslides (landslide warning zones and special landslide warning zones) as published on the hazard map portal site.
URL: https://disaportal.gsi.go.jp/

²Tsukuba Forum 2025
URL: https://www.rd.ntt/e/as/tforum/

³Press release "Building an AI to predict infrastructure damage from disasters"
URL: https://group.ntt/en/newsrelease/2024/04/25/240425a.html

⁴Slope failure, debris flow, and sediment distribution data
Based on aerial photographs taken by the Geospatial Information Authority of Japan on September 23 and 24, this dataset identifies areas of slope failure, debris flow extents, and sediment deposition believed to have resulted from the heavy rainfall starting on September 20, 2024.
URL: https://www.gsi.go.jp/BOUSAI/R6_noto_heavyrain.html

About NTT

NTT contributes to a sustainable society through the power of innovation. We are a leading global technology company providing services to consumers and businesses as a mobile operator, infrastructure, networks, applications, and consulting provider. Our offerings include digital business consulting, managed application services, workplace and cloud solutions, data center and edge computing, all supported by our deep global industry expertise. We are over $92B in revenue and 330,000 employees, with $3.6B in annual R&D investments. Our operations span across 80+ countries and regions, allowing us to serve clients in over 190 of them. We serve over 75% of Fortune Global 100 companies, thousands of other enterprise and government clients and millions of consumers.