Global transportation infrastructure exposure to the change of precipitation in a warmer world

The study addresses how climate warming alters the frequency (return periods) of extreme precipitation and, in turn, challenges the adequacy of transport infrastructure drainage systems that were designed based on historical climate stationarity. Reliable transport networks underpin trade and economic development, yet are increasingly threatened by natural hazards and projected intensification of extremes with warming. Meeting Sustainable Development Goal 9 requires substantial investment in resilient infrastructure, particularly in low- and middle-income countries. However, limited work has assessed how design standards must evolve under future precipitation regimes. This research quantifies global changes in precipitation return periods and evaluates where and by how much road and railway drainage design standards need to adjust to maintain intended reliability, ultimately proposing practical safety factors for climate change adaptation.

Prior work has estimated multi-hazard risks and direct damages to transport assets, and examined indirect economic impacts from network disruptions. Climate science indicates extreme precipitation generally intensifies with warming (approximately 6–8% per °C on average) but with strong regional variability, a pattern expected to continue. Studies have highlighted insufficiencies of existing hydrologic design standards under increasing rainfall extremes and emphasized the need to maintain infrastructure reliability amid climate uncertainty. Global assessments of infrastructure exposure and flood risk exist, as do analyses of road network completeness and growth, yet there remains a gap linking projected extreme precipitation changes to specific updates in infrastructure design standards, especially drainage return periods. This study advances that gap by integrating climate projections, infrastructure inventories, and design assumptions to quantify required adaptations.

The analysis evaluates future changes in extreme precipitation frequency and intensity and the exposure of global road and railway assets under mid- (2030–2059) and late-21st century (2070–2099) horizons for RCP4.5 and RCP8.5. Climate data come from NASA NEX-GDDP (CMIP5 downscaled) at 0.25° (~25 km) resolution. For each grid, annual maximum daily precipitation (RX1D/RXID) series were fitted with a generalized extreme value (GEV) distribution. Parameters were estimated via L-moments (lmoments3 package) and goodness-of-fit was tested using the Kolmogorov–Smirnov test (5% significance). Present-day (1971–2000) precipitation intensities were derived for standard return periods (2–50 years). The same intensities were mapped onto future GEV distributions to obtain future return periods, characterizing shifts in exceedance probabilities. Infrastructure data (roads and railways) were sourced from OpenStreetMap via Geofabrik (updated May 6, 2021). Road classes included motorway, trunk, primary, secondary, and tertiary; railways included full-sized passenger or freight standard-gauge lines. Assets were rasterized to the climate grid, summing lengths within each cell. Base design return periods for drainage were assigned by World Bank income group and by asset type (railways; motorway/trunk/primary/secondary; tertiary), with higher and lower design standard variants (Supplementary Table 1). Exposure was assessed where the design return period shortened by more than 25% relative to 1971–2000. Absolute exposure (AE) is the total exposed length within a grid; relative exposure (RE) is AE divided by the total asset length in the grid. Country-level rankings aggregate grid results. A safety factor (SF) for climate adaptation was defined as the ratio of future to contemporary precipitation intensity at the contemporary design return period, computed per asset type and grid and then averaged across asset types. Spatial patterns were summarized using multi-model medians to reduce sensitivity to outliers and ensure robustness. Sensitivity analysis assessed SF under higher design standards. Key outputs include global maps of future return period shifts, absolute and relative exposure, and safety factors by scenario and time slice.

• Global land change in extreme precipitation frequency: Approximately 91.7–94.6% of land areas are projected to experience decreasing return periods (more frequent extremes). Under RCP4.5, 58.7% (mid-century, ~−1.8 °C) and 73.8% (late-century, ~−2.5 °C) of land see >25% return period decreases; under RCP8.5, 71.5% (mid, ~−2 °C) and 86.6% (late, ~−4 °C). Hotspots include Greenland, eastern and western North America, northern South America, Central Africa, eastern Siberian Plateau, Central India, Southwest China, and Southeast Asia. • Asset-level design return period shortening: 88.4–94.6% of global transport assets will have shorter design return periods than in 1971–2000. Average decreases are 24.6% (mid-century) and 34.2% (late-century) with standard deviations 13.3% and 14.9% (means across RCP4.5 and RCP8.5). • Absolute exposure (assets with >25% design return period decrease): RCP4.5—6.8 million km (28.8%) mid-century; 11.0 million km (46.6%) late-century. RCP8.5—10.3 million km (43.6%) mid-century; 16.5 million km (69.9%) late-century. High absolute exposure clusters in eastern North America, northern Western and Central Europe, and East Asia; rankings are influenced by both infrastructure density and climate change (e.g., Sweden’s rank rises due to precipitation change). • Relative exposure (share of assets within a grid exposed): Under RCP4.5, grids with >80% exposure ratio constitute 22.5% (mid) and 40.5% (late) of exposed grids; under RCP8.5, 36.6% (mid) and 69.8% (late). Countries with small networks can have high relative exposure; e.g., Panama sees 99% of grids >80% exposure late-century under RCP8.5 while ranking 107th in absolute exposure. • Country specifics: United States leads absolute exposure; under RCP4.5, 1.14 million km (32.7%) mid and 2.09 million km (60.1%) late are exposed. China has 1.27 million km (42.9%) mid and 1.94 million km (65.5%) late exposed. Largest increases from mid to late (RCP4.5) in absolute exposure occur in Romania (+509%), Ireland (+216%), and India (+206%). In relative exposure (RCP4.5), Finland reaches 54.9% (mid) and 92.7% (late). • Safety factor (SF) for design: Under RCP4.5, 90.6% (mid) and 92.1% (late) of land has SF between 1 and 1.5; only 0.10% (mid) and 0.15% (late) exceed 1.5 (notably Tibet, Central India, Andes). Under RCP8.5, 91.2% (mid) and 90.7% (late) lie in 1–1.5; 0.19% (mid) and 1.7% (late) exceed 1.5 (notably India, Southwest China/Indo-China, East Africa, Andes, and >50°N). A safety factor of 1.2 suffices for quick design approximations in 87.6% (mid) and 82.0% (late) under RCP4.5, but only 83.6% (mid) and 45.5% (late) under RCP8.5. Sensitivity to higher design standards remains modest (RCP4.5: 87.7% and 90.9% land within SF 1–1.5; 0.05% and 0.10% >1.5).

The findings directly address how changing extreme precipitation probabilities undermine historically derived drainage design standards for roads and railways. A widespread shortening of return periods increases exceedance probabilities, potentially elevating disruption risks, maintenance needs, and lifecycle costs. By quantifying both absolute and relative exposure, the study highlights not only infrastructure-rich regions (e.g., the United States, Europe, East Asia) but also countries with smaller networks facing disproportionately high exposure. The proposed safety factor framework provides a pragmatic adaptation tool to approximate climate change impacts in design, enabling asset managers to maintain intended reliability levels without immediate, detailed local studies. Results underscore the benefits of pathways consistent with RCP4.5 in reducing exposure compared to RCP8.5 and identify regions where higher safety factors are warranted, informing prioritization for adaptation investments and planning. Overall, integrating climate-informed safety factors into design can help bridge gaps between climate projections and engineering practice, supporting resilient infrastructure development aligned with SDG 9.

This study synthesizes downscaled multi-model climate projections with global transport infrastructure inventories to quantify future exposure to intensified extreme precipitation and provides an actionable safety factor approach for updating drainage design. Key contributions include global maps of return period changes, quantification of absolute and relative exposure across countries, and recommended safety factors (notably ~1.2 under RCP4.5 for most regions) to maintain design reliability. The analysis identifies exposure hotspots and demonstrates that following a stabilization pathway (RCP4.5) substantially reduces exposure relative to higher-emissions futures (RCP8.5). Future research should advance cost–benefit assessments of adaptation options, integrate indirect economic impacts from network disruptions, refine local design standards and maintenance considerations, and apply high-resolution local hydrologic–hydraulic models and data to support site-specific design and prioritization.

Key limitations include: (1) reliance on assumed drainage design return periods by World Bank income group and asset type due to lack of global, detailed design data; (2) use of a uniform baseline (1971–2000) and stationarity assumption for contemporary design; (3) omission of infrastructure maintenance quality, deterioration rates, and subnational variability in standards; (4) exposure defined by a fixed 25% shortening threshold, which may not capture all risk nuances; (5) dependence on NEX-GDDP downscaled CMIP5 projections and multi-model medians, with inherent climate model and downscaling uncertainties; and (6) no explicit assessment of the feasibility, costs, or efficacy of specific adaptation measures. Given the global scale, results are intended to identify hotspots and guide further local analyses with detailed data.