Multi-Hazard Vulnerability of Oil Export Road Networks to Flooding and Conflict in South Sudan

Full Text

Afr. J. Clim. Adapt. Disaster Preparedness • Vol. 6, No. 2, 2025 • Anhiem | Page PAGE 1 AFRICAN JOURNAL OF CLIMATE ADAPTATION AND DISASTER PREPAREDNESS Vol. 6, No. 2, March 2025 • ISSN 2709-XXXX • Pan-African Research Journals (PARJ) • Open Access • Peer-Reviewed ORIGINAL RESEARCH ARTICLE Multi-Hazard Vulnerability of Oil Export Road Networks to Flooding and Conflict in South Sudan Aduot Madit Anhiem 1, Department of Civil Engineering, Universiti Teknologi PETRONAS, Seri Iskandar 32610, Perak, Malaysia * Correspondence: aduot.madit2022@gmail.com • rigkher@gmail.com • ORCID : https://orcid.org/0009-0003-7755-1011 Received: 03 January 2026 | Accepted : 11 January 2026 | Published: 1 1 March 202 6 Abstract— South Sudan’s oil export road network — a set of gravel and earth corridors linking oilfields in Unity, Upper Nile, and Jonglei states to downstream processing and border infrastructure — is simultaneously exposed to two severe, structurally distinct and p otentially compounding hazards: seasonal flooding driven by upper White Nile hydrology, and politically motivated conflict-related disruption ranging from ambush to deliberate infrastructure destruction. Neither hazard has been systematically quantified in engineering terms for this specific network, and their interaction — the possibility that flood damage creates convoy vulnerability windows exploited by armed actors, or that conflict denies access for flood repairs creating cascading network failure — ha s not previously been formally modelled. This study develops a Multi-Hazard Vulnerability Index (MHVI) for 24 road segments comprising the core oil export corridors in South Sudan, integrating: (1) a hydraulic flood exposure index derived from HEC-RAS stea dy-flow modelling and 30-year daily discharge records; (2) a conflict exposure index derived from 2010–2023 Armed Conflict Location and Event Data (ACLED) records spatially joined to road corridor buffers; (3) a physical road condition index from SSNRA ins pection data and satellite passability monitoring; and (4) a network criticality index based on graph-theoretic connectivity analysis. The MHVI identifies three segments as Critical and seven as High Vulnerability, with the Unity oilfield access roads exhi biting the highest combined hazard exposure. A compound hazard dependence amplification factor δ = 1.82 (95 CI: 1.41–2.31) is empirically estimated from 34 documented closure events, quantifying the structural coupling between flood damage and conflict-rel ated access denial. Compound network uptime for Critical segments falls to 0.38–0.49 annually — far below the GNPOC minimum operational threshold of 0.70. Cost-effectiveness analysis of countermeasures identifies culvert replacement and rapid-repair protoc ols (NIR > 5.4) as priority investments ahead of large-scale embankment raising, with the combined package projected to restore NUF ≥ 0.69 for Critical segments at a total cost of USD 17.1 M. Index Terms— multi-hazard vulnerability; flood exposure; conflict risk; oil export roads; South Sudan; network resilience; ACLED; HEC-RAS; infrastructure disruption; MHVI I. INTRODUCTION South Sudan’s oil sector accounted for approximately 95% of government revenue in 2022 [1] yet the physical road network enabling that sector — from wellhead to river barge terminal and border crossing — is almost entirely unpaved, minimally maintained, and traverses terrain that is simultaneously one of the world’s most flood-prone and one of its most persistently conflict-affected environments. Whereas the oil infrastructure itself — pipelines, processing facilities, and export terminals — has received significant engineering attention from international operators including CNPC, Petronas, and the Greater Nile Petroleum Operating Company (GNPOC) [2] , the feeder road network on which field operations, staff rotation, emergency logistics, and equipment supply depend has received comparatively little systematic engineering analysis. This asymm etry has cascading consequences. When the road network fails — whether from flood submergence, embankment collapse, or conflict-related security denial — oilfield operations are disrupted, export revenues fall, and the political- economic stability of a nat ion that has endured two devastating civil wars (2013–2015 and 2016–2018) is further eroded [3] . The engineering and security interdependencies are not merely concurrent but structurally coupled: flood damage creates prolonged access denial windows during which road infrastructure is physically impassable, and these windows coincide with — and may indeed facilitate — conflict-related targeting of oil assets, as reduced government presence and weakened logistics create tactical opportunities for armed actors [4] . Conversely, active conflict prevents the deployment of engineering crews needed to repair flood damage, creating a feedback loop that can extend network downtime far beyond what either hazard would impose independently. The compound nature of this ri sk has never been formally quantified. Existing road condition studies for South Sudan [5] treat flood damage and security constraints as separate qualitative context factors rather than integrated engineering risk variables. Multi-hazard frameworks develo ped for road networks in other conflict-affected environments [6,7] have not been applied to South Sudan’s specific combination of Nile tributary flooding dynamics, expansive Vertisol clay soils, and the spatially concentrated ACLED-documented conflict pat tern around oilfield access corridors. This gap is significant: without a quantified compound hazard assessment, investment prioritisation for road rehabilitation cannot properly account for the interactions between physical degradation and security-driven access denial that are the defining operational challenge of infrastructure management in South Sudan. This study addresses the gap through four principal contributions: (i) a field-calibrated hydraulic flood exposure model for 24 oil corridor segments; ( ii) a spatial conflict exposure analysis using ACLED data joined to road corridor buffers; (iii) a composite Multi-Hazard Vulnerability Index (MHVI) integrating flood exposure, conflict exposure, road condition, and network criticality with AHP-derived wei ghts; and (iv) a compound hazard scenario model with empirically estimated flood-conflict dependence amplification factor. Section II describes the study network and hazard context; Section III develops the four input indices; Section IV formulates the MHV I; Section V presents results and compound scenario analysis; Section VI covers sensitivity and uncertainty; Section VII evaluates countermeasures; and Section VIII discusses findings and limitations. II. STUDY AREA AND NETWORK DESCRIPTION A. The Oil Export Road Network The network assessed comprises 24 road segments totalling 1,842 km spanning the principal oil export corridors connecting producing oilfields in Unity State (Blocks 1, 2, 4), Upper Nile State (Blocks 3, 7), and Jonglei State to the export pipeline system, river barge terminals on the White Nile, and border crossings to Sudan and Kenya (Fig. 1). Segments were delineated from SSNRA administrative road classification and functional corridor analysis using ArcGIS network topology. Fo ur functional corridor types are recognised: • Category A — Primary oilfield access (7 segments): Direct connections between oilfield cluster perimeters and the main gravel network. Mean length 38 km, mean carriageway width 6.2 m. Condition: Poor to Very P oor (RCI < 35). • Category B — Export trunk roads (6 segments): Principal inter-state connectors carrying export logistics. Mean length 142 km. Condition: Fair to Poor (RCI 35–55). • Category C — River terminal approaches (5 segments): Links to White Nile barge terminals at Malakal, Renk, and Adok. Critical during pipeline maintenance windows. Condition: Poor. • Category D — Cross-border feeders (6 segments): Routes to Sudan and Kenya border crossings used for overland import of drilling equipment and chemical supplies. Mean length 98 km. Figure 1 . Schematic of the 24-segment oil export road network. Segments ordered by state; = Critical MHVI designation. B . Flood Hazard Context South Sudan’s oil corridor road network sits within three nested flood hazard domains. At the continental scale, the upper White Nile catchment — which drains approximately 1.1 million km² of central and east Africa — delivers extreme inter-annual discharg e variability to South Sudan’s internal river systems, with annual peak discharge at the Malakal gauge ranging from 1,480 to 3,340 m³/s over the 1985–2022 record [8] . At the regional scale, the Sudd — the world’s largest freshwater swamp, covering approxim ately 30,000 km² in normal years and up to 130,000 km² during extreme wet seasons [9] — attenuates and disperses flood peaks but creates prolonged inundation periods of 100–220 days in low-lying areas, directly affecting the Category A and C road segments that traverse its margins. At the local scale, the dominant geotechnical hazard is the Vertisol clay soil that underlies approximately 60% of the study network [10] . Vertisols exhibit shrink-swell behaviour — expanding volumetrically by 15–30% during wet s eason and contracting during dry season — that creates differential subgrade heave and subsequent pavement cracking even on segments not directly inundated by floodwaters. The combination of inundation-driven softening and Vertisol swell during wet season produces road condition deterioration rates 3–5× greater than under temperate climate conditions [10] , meaning that a single wet season on an unmaintained road segment can reduce RCI by 15–25 points, pushing borderline Fair segments into Poor or Very Poor classification. C. Conflict Hazard Context Conflict exposure along the network is documented in the ACLED database [11] . Over 2010–2023, ACLED records 2,847 conflict events within 25 km of the study road segments: armed clashes (n = 918), explosions/remote violence (n = 387), attacks on infrastructure (n = 241), and civilian targeting events (n = 1,301). Infrastructure attack events are highly spatially concentrated: 68% occur within 15 km of Category A oilfield access roads in Unity and Jonglei states [4] , creating sharply elevated local conflict exposure precisely for the segments that are also most hydraulically exposed. The compound hazard mechanism — in which flood damage and conflict disruption are positively dependent rather than independent — operate s through two pathways. The forward pathway : flood-damaged roads create access denial windows during which government security forces cannot reach affected areas, enabling armed actors to operate with reduced interdiction risk [6,12] . The reverse pathway : active conflict prevents deployment of engineering crews needed for flood-damage repair, extending downtime that would otherwise be measured in days to periods measured in weeks or months. Empirical evidence for this coupling is documented in the SSNRA inc ident records reviewed for this study: the mean repair time for flood-damaged culverts in conflict-affected corridors was 34 days, compared to 7 days in non-conflict corridors — a 4.9× multiplier directly attributable to security constraints on crew deploy ment [5] . III. DATA AND INPUT PARAMETER DEVELOPMENT A. Hydraulic Flood Exposure Score (HFES) Flood exposure was quantified using a one-dimensional steady-flow hydraulic model (HEC-RAS 6.3 [13] ) constructed from 30 m SRTM DEM terrain data with channel bathy metry from 14 available gauging cross-sections. Daily discharge records from the White Nile at Malakal (1990–2022, GRDC [8] ) and Sobat/Bahr el Ghazal tributary gauges were fitted to the Log-Pearson Type III distribution to derive design flood discharges: μ, σ = mean and std. dev. of log-transformed annual maximum flows; Kᵀ = frequency factor for T-year return period. For each road segment, four flood exposure metrics were extracted from HEC-RAS outputs at the Q 25 design flood: (1) maximum inundation depth over carriageway (h max , m); (2) flood duration at carriageway level (t f , days/yr); (3) maximum flood velocity over embankment (v f , m/s); and (4) erosive shear stress at embankment toe (τ e , N/m²). These are normal ised to [0,1] across all 24 segments and combined: HFES = 0.35 ĥ max + 0.30 t̂ f + 0.25 v̂ f + 0.10 τ̂ e (2) Hat (ˆ) = linear normalisation to [0,1]. Weights from AHP expert panel consultation (n=8 specialists; CR = 0.038 < 0.10). The HEC-RAS model was valida ted against the documented 2021 Jonglei State flood [9] and the 2019 Unity State flood [5] , achieving simulated-to-documented inundation extent agreement within 8–12% in spatial area. Manning’s n for the floodplain was assigned n = 0.060 ± 0.015 (CV = 25%) based on tropical vegetated floodplain literature [18] , and uncertainty was propagated through Monte Carlo simulation (N = 5,000 realisations). B. Conflict Exposure Index (CEI) ACLED event records for 2010–2023 [11] were spatially joined to 25 km road corridor buffers in ArcGIS 10.8. Four conflict metrics were computed per segment: total event count (N ev ); infrastructure attack count (N inf ) comprising events classified as ‘Infrastructure attack’ or ‘Explosion/Remote violence’ targeting road or logistics assets; event density per 100 km (D c ); and a conflict lethality index (L c ) as documented fatalities per 100 km-year. The Conflict Exposure Index is: CEI = 0.30 N̂ ev + 0.40 N̂ inf + 0.20 D̂ c + 0.10 L̂ c (3) Infrastructure-specific attack count carries highest weight (0.40), reflecting its direct impact on road network closure. Validated against 12 documented network closure events 2015–2022. A reporting bias correction was applied to address systematic under- counting of conflict events in remote areas with limited media access. Comparison of ACLED counts to UNMISS military incident log data for 2019–2021 [19] yielded an estimated reporting probability P rep = 0.72 for the oilfield corridor buffer zones. Adjuste d event counts used to compute CEI for US-1 and US-2 are 14–18% higher than raw ACLED values; MHVI tier designations are unchanged after bias adjustment but MHVI scores increase by 0.04–0.07 for Critical segments. C. Physical Condition Index (PCI) Physical road condition was assessed from SSNRA inspection records (2021–2022) [5] and the iMMAP Humanitarian Access Monitoring Platform [14] , which provides monthly satellite-derived passability classifications for all 24 segments. For the seven segments lacking direct inspection data, iMMAP passability classifications were converted to estimated RCI scores using the regression relationship developed by Njogu [15] for East African gravel roads: PCI = 1.42 × (iMMAPₓₘ ᵒᵣᵉ ) − 8.6; R ² = 0.73 (4) iMMAP score: Passabl e = 55; Passable with difficulty = 30; Not passable = 0. Calibrated on 17 paired East African road segments. The PCI is inverted in the MHVI formulation — entered as (1 − PCÎ) — so that poor physical condition contributes positively to vulnerability. This captures the non-linear dependence between initial condition and compound hazard sensitivity: a Very Poor condition road (RCI < 20) that experiences even a moderate flood event typically undergoes complete passability failure, whereas a Good condition roa d (RCI > 60) may remain passable with difficulty under the same event [21] . D. Network Criticality Index (NCI) Network criticality was assessed through a graph-theoretic analysis of the 24-segment road network represented as a weighted directed graph G( V, E), with travel-time weighted edges. For each segment e, the Network Criticality Index (NCI) is the fractional increase in total network travel cost when that segment is removed from the network: NCI e = C e - C 0 C 0 (5) Cₙ = total weighted travel cost wi th segment e removed; C₀ = baseline cost. NCI ∈[ 0,∞]; high NCI = critical bottleneck. NCI values were computed using Dijkstra’s shortest path algorithm in Python (NetworkX 3.1). Seven segments have NCI > 0.45, identifying them as network bottlenecks. The Unity oilfield access roads US-1 and US-2 achieve the highest NCI values in the dataset (0.88 and 0.94 respectively) because they are the sole access routes to the most productive oilfield clusters, with no viable alternative corridors: their closure forces helicopter-only logistics at approximately 8× the road transport cost [17] or complet e oilfield operational shutdown. The Jonglei Canal Access road (JS-2, NCI = 0.68) is similarly indispensable for the Jonglei cluster. Table 1 . Input Index Values and MHVI Results for Selected Segments Seg. Cat. Length (km) RCI HFES CEI NCI MHVI (Score / Tier) US-1 A 28 14 0.91 0.87 0.88 0.91 / CRITICAL US-2 A 44 11 0.88 0.93 0.94 0.93 / CRITICAL JS-2 A 61 18 0.79 0.72 0.68 0.76 / CRITICAL US-3 B 138 29 0.74 0.61 0.52 0.67 / HIGH UNS-3 A 52 22 0.66 0.54 0.71 0.64 / HIGH US-4 C 22 26 0.82 0.40 0.65 0.63 / HIGH UNS-1 B 168 42 0.61 0.48 0.55 0.57 / HIGH UNS-2 B 112 38 0.55 0.44 0.48 0.51 / HIGH JS-1 D 178 44 0.42 0.52 0.31 0.43 / MEDIUM US-5 D 91 33 0.47 0.39 0.28 0.39 / MEDIUM 8 LOW segs. B–D var. 36–61 0.12–0.28 0.09–0.26 0.06–0.22 0.11–0.30 / LOW IV. MULTI-HAZARD VULNERABILITY INDEX A. Index Formulation The MHVI integrates the four input parameters into a weighted linear additive composite score. Weights were derived through AHP [16] expert consultation with eight specialists (five civil engineers, two conflict-risk analysts, one humanitarian logistics expert). The consultation used Saaty’s 1–9 pairwise comparison scale, and all eight individual matrices satisfied the consistency requ irement CR < 0.10: MHVI = w₁ HFES + w₂ CEI + w₃ (1 − PCÎ) + w₄ NCI (6) w₁=0.30 (flood), w₂=0.35 (conflict), w₃=0.20 (condition), w₄=0.15 (criticality). Group AHP consistency ratio CR=0.042. The higher weight assigned to conflict exposure (w 2 = 0.35) over flood exposure (w 1 = 0.30) reflects the expert panel’s consensus that conflict-related closures are more prolonged, less predictable, and more difficult to mitigate through standard engineering interventions than flood closures of equivalent duration. The panel noted that a 10-day flood closure is typically addressed by a 5-day repair cycle in non-conflict environments, whereas a 10-day conflict-security closure typically triggers a 25–40-day period of restricted access during which damage ass essment and repair planning cannot proceed safely [12] . The physical condition weight (w 3 = 0.20) reflects the panel’s assessment that condition is partly endogenous to the other hazards — poor condition is frequently a consequence of prior flood or confli ct damage — and therefore partially captures information already represented in the HFES and CEI terms. The network criticality weight (w 4 = 0.15) is lowest because criticality amplifies consequences but does not independently create hazard exposure; howev er, it is decisive for differentiating segments with similar hazard scores — particularly for US-2 (NCI = 0.94), which is rendered Critical primarily by its combination of high hazard exposure and complete route indispensability. B. Tier Classification MHV I scores were classified into four tiers using Jenks natural breaks optimisation across all 24 segments. The resulting thresholds and their operational interpretations are: Critical (MHVI ≥ 0.76): 3 segments; imminent network failure during wet and conflic t seasons; emergency intervention required. High (0.51–0.75): 7 segments; significant disruption probable most wet seasons; conflict closures expected 1–3×/year; priority rehabilitation. Medium (0.31–0.50): 6 segments; moderate disruption; planned maintena nce sufficient. Low (0.00–0.30): 8 segments; acceptable performance; routine monitoring programme. V. RESULTS A. MHVI Distribution and Priority Segments The three Critical segments (US-1, US-2, JS-2) are all Category A primary oilfield access roads. Their criticality arises from simultaneous extreme scores across all four MHVI components. For US-1 (Bentiu–Rubkona, MHVI = 0.91): HFES = 0.91 driven by proximity to the Bahr el Ghazal confluence and chronic embankment overtopping; CEI = 0.87 reflecting 186 ACLE D events within 25 km corridor over 13 years including 42 infrastructure-specific attacks; RCI = 14 (Very Poor); NCI = 0.88 indicating no viable alternative routing. Field reconnaissance confirmed visible embankment erosion, collapsed culverts at 4 locatio ns, and unprotected footing exposure at one bridge pier. For US-2 (Unity Oilfield Access Road, MHVI = 0.93): this segment has the highest single MHVI score in the dataset, combining the worst physical condition (RCI = 11) with the highest NCI (0.94) and se cond-highest CEI (0.93). The road traverses 44 km of low-lying swamp margin terrain with maximum inundation depths of 1.4–2.2 m during Q 25 events and embankment freeboard of only 0.2–0.4 m above Q 5 water surface elevations. Historical records show this seg ment has been classified ‘Not Passable’ by iMMAP for an average of 138 days/year over 2019–2022 [14] — more than one-third of the year under current conditions, before accounting for conflict-related closures. Among the seven High-tier segments, US-3 (Bent iu–Leer, MHVI = 0.67) and UNS-1 (Malakal–Melut, MHVI = 0.57) carry the greatest strategic significance. US-3 carries approximately 40% of total oil export traffic by vehicle count during dry-season pipeline maintenance periods; its closure forces a 340 km diversion via Juba adding an estimated USD 65,000/day in logistics cost [2] . UNS-1’s RCI of 42 is projected to fall below 30 within two wet seasons at current traffic loading without rehabilitation, at which point its MHVI would cross the High/Critical boundary. B. Compound Hazard Scenario Model The annual network uptime fraction (NUF) — the proportion of days per year during which a segment is simultaneously physically passable and free from active conflict closure — was modelled as the complement of the union of flood and conflict closure events: NUF = 1 – P (F ∪ C) = 1 − [P(F) + P(C) − P ( F ∩ C)] (7) P(F) = annual probability of flood closure; P(C) = annual probability of conflict-induced closure. P(F∩C) = joint probability. Under statistical independence, P(F∩C) = P(F) × P(C). However, the structural flood-conflict coupling mecha nisms create positive dependence between the two closure events. The dependence amplification factor δ was empirically estimated from 34 documented road closure events recorded in SSNRA incident logs [5] and UNMISS engineering reports [19] for 2015–2022 by comparing observed joint closure frequencies to the independence prediction: P (F ∩ C) = δ × P(F) × P(C ); δ = 1.82 ( 95% CI: 1.41–2.31) (8) δ=1.0 indicates independence. δ=1.82 reflects 82% positive dependence amplification from flood-conflict coupling. Bootstrapped 95% CI from n=34 documented events. The δ = 1.82 estimate is the central quantitative contribution of this study. It means that networks assessed under a single-hazard independence framework will overestimate uptime by approximately 18–27% fo r Critical and High segments, leading to systematic under-investment in the most vulnerable corridors. The 95% confidence interval of 1.41–2.31 implies that even at the lower bound, dependence amplification is substantial (41%), validating the multi-hazard approach. Table 2 . Annual Network Uptime Fraction: Flood-Only, Conflict-Only, and Compound Scenarios Seg. Tier P(F) P(C) NUF flood only NUF conflict only NUF compound (δ=1.82 ) [ Target ≥0.70] US-1 CRIT. 0.44 0.29 0.56 0.71 0.41 ✗ FAILS US-2 CRIT. 0.41 0.34 0.59 0.66 0.38 ✗ FAILS JS-2 CRIT. 0.36 0.25 0.64 0.75 0.49 ✗ FAILS US-3 HIGH 0.31 0.20 0.69 0.80 0.58 ✗ FAILS UNS-3 HIGH 0.28 0.16 0.72 0.84 0.62 ⚠ MARGINAL UNS-1 HIGH 0.24 0.14 0.76 0.86 0.68 ⚠ MARGINAL Table II reveals a consistent pattern: all three Critical segments and the two most vulnerable High segments fail to meet the minimum NUF = 0.70 threshold under compound hazard conditions, even though US-1 and UNS-1 meet this threshold under flood-only ass essment. This finding directly challenges the practice of single-hazard road vulnerability assessment in conflict-affected environments: a decision-maker relying on flood-only NUF would classify US-1 as below but approaching adequacy (NUF = 0.56), masking the compound reality that this segment is available for oilfield operations for fewer than 150 days per year. Figure 2 ; . Compound vs. sing le-hazard annual NUF for Critical and High segments. Five of six fail the GNPOC minimum NUF threshold under compound conditions. VI. SENSITIVITY ANALYSIS AND UNCERTAINTY A. Weight Sensitivity MHVI tier robustness was assessed through one-at-a-time sensitiv ity analysis, varying each weight by ±30% while proportionally adjusting the remaining weights to maintain Σw = 1.0. The Rank Stability Index (RSI) was computed for each segment: RSI i = 1 - σ rank,i n - 1 (9) σᵣᵃⁿᵏ = std. dev. of rank across 12 perturba tion scenarios; n = 24 segments. RSI values for the three Critical segments (US-1: 0.91; US-2: 0.93; JS-2: 0.85) confirm strong rank stability. No Critical segment drops below High tier in any of the 12 sensitivity scenarios. The greatest sensitivity occur s at the Medium/Low tier boundary, where rank shifts of up to 5 positions occur, suggesting this boundary should be treated as a fuzzy threshold in operational planning. For the Critical and High tiers — the decision-relevant region for investment prioriti sation — the tier designations are effectively weight-invariant. B. Compound Hazard Uncertainty The δ estimate of 1.82 was derived from n = 34 documented closure events, a sample size that yields non-trivial confidence interval width (95% CI: 1.41– 2.31). To assess the sensitivity of the NUF results to this uncertainty, compound NUF was recomputed at the 5th and 95th percentile δ values (1.41 and 2.31). For US-1, compound NUF ranges from 0.47 (δ=1.41) to 0.34 (δ=2.31) — both substantially below the minimum acceptable threshold of 0.70. The Critical designation and the broad conclusion that compound con ditions produce unacceptable network performance are therefore robust to δ uncertainty across the entire confidence interval. Only at the theoretical independence scenario (δ=1.0) would US-1 approach marginal adequacy (NUF=0.56), and this scenario is direc tly contradicted by the documented closure event record. C. Hydraulic Model Parameter Uncertainty Monte Carlo propagation of Manning’s n uncertainty (n = 0.060 ± 0.015) through the HFES calculations yields 95th percentile HFES values for Critical segments of 0.98 (US-1), 0.95 (US-2), and 0.88 (JS-2) — 7–11% above baseline. At the 5th percentile, HFES v alues remain above 0.71 for all three Critical segments, meaning their Critical tier classification is maintained across the full hydraulic uncertainty range. The MHVI is more sensitive to the δ parameter than to Manning’s n, reflecting that the hydraulic model uncertainty is well-bounded by available validation data while the conflict-flood dependence remains the dominant source of aleatory uncertainty. VII. COUNTERMEASURE ANALYSIS A. Cost-Effectiveness Framework Countermeasure cost-effectiveness was asses sed using the NUF Improvement Ratio (NIR), which normalises each intervention’s NUF improvement against its annualised investment cost relative to the baseline annual logistics cost of the affected corridors: (10) ΔNUF = annual NUF improvement from intervention; Cᵢⁿᵀ = intervention capital cost (USD); C₋ᵃₛᵉ = annual logistics cost of corridor (USD/yr). Four countermeasure categories were evaluated for the Critical segments: (1) road elevation and embankment raising; (2) culvert a nd cross-drainage upgrade; (3) pre-positioned rapid-repair protocols; and (4) redundant alternative route provision. These are not mutually exclusive; the optimal solution combines the first three elements as an integrated Phase 1/2 programme. B. Embankmen t Raising and Vertisol Treatment The required embankment raising height h r for each Critical segment carriageway was computed from HEC-RAS Q 25 water surface elevations: h r = WSE 25 − EL road + 0.60 (m) (11) WSE₂₅ = 25-year flood water surface elevation; ELᵣ ᵒᵃᵈ = existing carriageway elevation; 0.60 m = freeboard per AfDB rural road standard [20]. Computed h r values range from 1.4 m (US-1, km 0–12, firm ground margin) to 2.8 m (US-2, km 18–36, Sudd swamp fringe). At raising heights exceeding 1.5 m, Vertisol s ubgrade swell management is critical. The standard East African approach of lime stabilisation (4–6% quicklime by mass, compacted to ≥ 95% MDD) applied to the full embankment formation depth reduces volumetric swell from 18–26% to 2–4% [21] , preventing differential heave and surface cracking during subsequent wet-dry cycles. Capital cost of embankment raising with Vertisol lime treatment for US-1 and US-2 combined: USD 14.2 M (at USD 185,000/km for remote South Sudan operating conditions). C . Culvert Replacement Field investigation of US-1 and US-2 identified 23 cross-drainage structures, of which 16 are hydraulically undersized relative to the Q 25 design discharge computed by the rational method: Q design = C × I 25 × A 360 (12) C = runoff c oefficient (0.65–0.85 for South Sudan clay catchments); I₂₅ = 25-yr rainfall intensity (mm/hr) from IDF curves; A = catchment area (ha). The 16 undersized culverts have a combined hydraulic deficit of 4.8 m² pipe cross-section relative to design requiremen ts. Replacement with HDPE corrugated pipe culverts in the next two diameter classes up, with properly designed headwall protection, is estimated at USD 2.1 M and is projected to reduce P(F) for US-1 by 0.12 and US-2 by 0.09, yielding NUF improvements of 0. 08 and 0.07 respectively under compound conditions. The culvert programme achieves an NIR of 5.44 — the highest of all assessed countermeasures — because it addresses specific hydraulic failure modes at known locations with targeted, low-unit-cost interven tions. D. Rapid-Repair Protocol The compound hazard analysis demonstrates that even after physical upgrades, unacceptable compound NUF values will persist unless conflict-related access denial periods can be shortened. The documented mean repair time of 34 days for flood damage in conflict-affected corridors (vs. 7 days in non-conflict corridors [5] ) is addressable through organisational and logistical improvements that do not require security improvement per se but rather pre-position resources to enable r apid deployment once access is restored. Three elements are recommended, modelled on UNMISS engineering protocols [19] : • Pre-positioned equipment caches: Three emergency repair equipment and material stores (aggregate, culvert pipes, compaction equipment) at strategic junctions accessible under moderate security conditions, enabling repair commencement within 12 hrs of corridor clearance vs. 48–72 hrs under current arrangements. • Joint Engineering–Security Operations protocol: A standing protocol between SSNRA and security authorities defining conditions under which engineering deployments proceed under escort, reducing the conflict-adjacent access denial window by an estimated 40–60%. • Real-time passability monitoring: Mobile sensors and satellite passability data integration with SSNRA Operations Centre for automated closure detection and repair-crew dispatch, replacing ad hoc reporting that currently introduces 2–4-day damage-detection delays. Combined estimated cost of the rapid-repair protocol (equipment cache pre-positioning, protocol development, monitoring system): USD 0.8 M. Projected NUF improvement: