Climate change is already increasing the exposure of global road infrastructure to extreme rainfall and landslides, and this trend will continue in the future. Global road risk is projected to rise under all climate scenarios, with the strongest increase reaching about 30.6% by 2100 under high emissions. Even under a sustainable pathway, risk still grows by about 23.9%, showing that adaptation is necessary regardless of mitigation efforts.

At present, about 8.3% of global road areas are classified as high or very high risk, concentrated in key hotspots such as Western Europe, South Asia, East Asia and Southeast Asia. These regions combine dense infrastructure, high traffic volumes, and intense rainfall patterns, which amplify vulnerability. In some cases, such as East Asia, road risk levels are up to nine times higher than the global average.

Climate adaptation plays a decisive role in shaping future outcomes. Countries with strong adaptive capacity can limit risk growth even when exposure is high. For example, regions that invest in drainage upgrades and resilient design standards can partially offset increasing rainfall extremes. In contrast, areas with limited adaptive capacity - such as rapidly urbanizing regions or small island states - experience faster risk escalation. In high-risk hotspots, road risk can increase by more than 70% in the far future, highlighting the consequences of insufficient adaptation.

Economic exposure further reinforces the importance of adaptation. More than 40% of global road assets, valued at over $ 19 billion, are located in high-income countries. Because of this concentration, even small physical damage (for example, 5% degradation) can generate large economic losses. Overall, global damages to transport infrastructure already range between $ 3,1 and $ 22 billion annually, and are expected to rise with climate change.

Adaptation is therefore not optional but essential. Without targeted strategies, especially in high-risk regions, road networks will face increasing failures, reduced reliability, and higher maintenance costs.

Climate change increases extreme precipitation intensity by about 6-8% for every 1°C of warming, directly raising landslide probability through soil instability and higher pore water pressure. As a result, landslide frequency is projected to increase by around 10% under high-emission scenarios.

Globally, landslide susceptibility is uneven but persistent. About 57.5% of land areas have very low susceptibility, while 16% fall into high or very high categories, mainly in tropical and mountainous regions such as the Himalayas, Andes, and Southeast Asia. These patterns remain stable over time, meaning that climate change primarily intensifies existing risks rather than creating new ones.

Road infrastructure is particularly vulnerable because of its spatial distribution. Networks extend over large areas and often cross hazard-prone zones, increasing exposure. At the same time, infrastructure aging and growing traffic loads accelerate deterioration and raise maintenance costs. To meet future needs, low and middle income countries alone must invest between 0.5% and 3.3% of GDP annually (up to $ 1 trillion) in new infrastructure, plus 1-2% of GDP for maintenance.

Climate adaptation strategies are therefore critical. Evidence shows that proactive adaptation such as resilient design, improved drainage systems, and better planning can stabilize or reduce risk growth. In contrast, limited adaptation capacity leads to rapidly increasing vulnerability, especially in regions with fast urban expansion or constrained financial resources.

This study developed a global framework to assess road risk under climate change by combining landslide susceptibility, infrastructure exposure, and adaptive capacity. The analysis uses climate projections from 13 CMIP6 models across four scenarios (SSP1-2.6 to SSP5-8.5) and three time periods: 1981-2020, 2021-2060, and 2061-2100.

Landslide susceptibility was estimated using seven machine learning models integrated into an ensemble approach, achieving high accuracy (AUC = 0.9892). The model incorporates key drivers such as extreme precipitation, soil type, slope, vegetation, and land use. A dataset of 9,875 rainfall-induced landslides was used to calibrate and validate results.

Infrastructure exposure was quantified using global road data from OpenStreetMap, including both road length and economic value. Roads were classified into four categories, and their value was estimated using global construction cost benchmarks. 

Road risk was calculated as the interaction between hazard and vulnerability (R = S × V), where vulnerability includes exposure and coping capacity. Coping capacity was approximated using economic density, capturing the ability of regions to adapt through investments and resilience measures.

This framework allows the identification of regions where climate adaptation can have the greatest impact, highlighting that risk reduction depends not only on hazard intensity but also on infrastructure design, investment levels, and adaptive capacity.