Integrated Hydrological–Hydraulic Framework for Urban Flood Risk Management in Montería, Colombia: From 2D Modeling and Vulnerability Assessment to Structural, Non-Structural, and Emergency Intervention Measures

Abstract

Tropical mid-size cities on alluvial floodplains face compounded flood challenges combining pluvial accumulation from intense convective storms, regulated river overflow, and aging drainage networks. This study presents an integrated framework for Monteria, Colombia (~450,000 inhabitants; Sinu River, Caribbean lowlands), within Colombian Decree 1807/2014 and structured in four phases. (1) Hazard: A Rain-on-Grid 2D HEC-RAS 6.6 model covering 4090 ha, calibrated against four gauged events, identifies three dominant pluvial mechanisms (poor hydraulic connectivity, limited evacuation capacity, downstream channel overflow), plus 17 critical fluvial erosion points affecting ~289 properties at 100-year return period. (2) Vulnerability: Depth-damage functions from 1465 household surveys yield 36.36% of 3015 assets in high risk and 57.77% in medium risk. (3) Measures: Scenario M2 (channel widening plus dikes, land-raising, retention lagoons) removes 80 ha of flooding while displacing 28 ha at COP 845 million pre-design cost. Non-structural measures include a Sustainable Urban Drainage Master Plan, IoT-based Early Warning System, minimum construction-elevation map, and land-management instruments. A Monte Carlo residual-risk model reduces baseline risk to 19.9% under full implementation. (4) Emergency: A February 2026 cold-front event was addressed with a 4300 m perimeter dike and six pump stations deployed jointly by the Regional Environmental Authority (CVS) and Municipal Administration.

1. Introduction

Flood risk in tropical cities of the Global South is shaped by rapid urban growth on low-gradient floodplains, insufficient drainage infrastructure, and a rainfall regime dominated by high-intensity, short-duration convective events [1,2]. Monteria, capital of the Cordoba department in the Colombian Caribbean, exemplifies this compound vulnerability: the city sprawls across the right bank of the Sinu River at elevations of 15–30 m above sea level, its drainage system relies on a network of open macro-drainage channels whose hydraulic capacity has not kept pace with urban expansion, and its built environment is predominantly composed of single-story cement-block structures with minimal freeboard against shallow inundation [3,4].

Recent cross-city analyses confirm that socioeconomic vulnerability and flood exposure are spatially correlated across Latin American urban neighborhoods, with lower-income communities disproportionately located in flood-prone areas [5,6]. In the Colombian Andean context, climate-driven intensification of precipitation has been shown to increase flood damage costs non-linearly under projected warming scenarios [7], reinforcing the urgency of integrating climate adaptation into urban flood risk frameworks.

Since the Urra I hydroelectric dam commenced operation in 2000, fluvial flood risk from the Sinu has been substantially attenuated: peak discharges and stages have declined and direct river overtopping of the urban perimeter now occurs only for return periods exceeding 25 years, specifically along Street 41 on the right bank [8]. Paradoxically, this regularization has shifted public and institutional attention away from pluvial flooding, which remains the dominant hazard mechanism. Inundations in Monteria are governed by the accumulation of storm-runoff in flat urban depressions with poor connection to the channel network—a process characterized by high water depths but flow velocities generally below 1 m/s [9,10].

The Colombian regulatory framework further shapes the problem [11,12]. Decree 1807 of 2014 [11] regulates article 189 of Decree-Law 019 of 2012 and requires the incorporation of risk management into Land Use Plans (Planes de Ordenamiento Territorial, POT), establishing risk as a conditioning factor for land use and occupation with the explicit aim of avoiding the configuration of new risk conditions. Basic studies allow the delimitation and zoning of hazard areas (high, medium, low) and the identification of areas with hazard and risk conditions, in which Detailed Flood Risk Studies (Estudios Detallados de Riesgo por Inundación (Detailed Flood Risk Studies), EDRI) must be undertaken. These studies determine the risk categorization and establish mitigation measures, which may be structural (physical works) or non-structural (urban planning standards, early-warning systems), and which must be pre-dimensioned on cartography at pre-design level with the corresponding cost estimate.

Despite the maturity of two-dimensional shallow-water models for urban flood simulation—with recent benchmarking confirming that HEC-RAS 2D Rain-on-Grid approaches reproduce observed inundation patterns with acceptable accuracy when properly calibrated [13,14,15,16]—and of depth-damage functions for flood loss quantification [17,18], their integrated application in intermediate Colombian cities remains uncommon. Emerging evidence from large-scale insurance datasets demonstrates that conventional depth-damage functions often exhibit poor agreement with observed losses, with uncertainty in vulnerability translation identified as the principal bottleneck in flood risk assessment [19,20]. Furthermore, the gap between hazard assessment and pedestrian-scale risk evaluation has only recently been addressed in the Colombian context through coupled hydrological-image recognition approaches [21]. The absence of calibrated local models produces two compounding deficiencies: overconfident hazard zonification that misclassifies ponding as low-risk, and under-calibrated vulnerability curves that either over- or under-estimate economic losses. Furthermore, the link between the scientific risk assessment and the operational intervention—both long-term structural-prospective and rapid-onset emergencies are rarely formalized, leaving events such as the February 2026 cold-front episode that inundated communes 1 and 2 of Monteria without a pre-validated technical framework for containment and evacuation.

The novelty of the present contribution is threefold. First, whereas previous flood studies in intermediate Colombian cities have relied on semi-distributed or lumped hydrological models coupled with one-dimensional hydraulic routing, this work deploys a fully distributed Rain-on-Grid 2D shallow-water model (HEC-RAS 6.6) covering 4090 ha at sub-metric terrain resolution, enabling the explicit representation of bidirectional flow exchange between streets, channels, and ponding zones—a capability that is indispensable in cities with slopes as low as 0.0003 m/m but that has not been systematically applied in Caribbean Colombian urban contexts. Second, the study develops and calibrates locally derived depth-damage functions from 1465 household surveys and 156 commercial-zone interviews, filling the gap of transferable vulnerability models for cement-block typologies prevalent across the Colombian Caribbean lowlands. Third, the framework operationally links scientific risk assessment to both long-term structural-prospective planning and rapid-onset emergency response, as demonstrated by the real-time application of the calibrated model during the February 2026 cold-front event—a linkage that, to the authors’ knowledge, has not been documented for a Colombian intermediate city. This paper addresses these gaps through a four-phase integrated framework: (i) a quantitative hazard characterization using a calibrated and validated Rain-on-Grid 2D hydrodynamic model covering 4090 ha of the urban perimeter, together with a cold-front-driven Sinu River simulation; (ii) a probabilistic urban risk assessment employing locally derived depth-damage functions and field survey data from 1465 households; (iii) a portfolio of structural and non-structural intervention measures, pre-dimensioned and articulated along four management axes, with their residual risk quantified through Monte Carlo simulation; and (iv) a real-event emergency response designed and implemented during the February 2026 cold-front inundation, grounded in field-documented water-level elevations and hydraulic analysis of drainage exit options. Together, these phases constitute a replicable governance framework for flood management in Caribbean Colombian cities under the projected intensification of extreme precipitation associated with climate change [1].

2. Study Area

Monteria is situated on the bank of the Sinu River in the Caribbean lowlands of Colombia. The urban perimeter covers approximately 4090 ha at low elevations, with the river forming a natural eastern boundary and a network of 15 primary macro-drainage channels transporting pluvial runoff westward toward peripheral discharge points. The hydrological setting is characterized by a bimodal rainfall regime with a principal wet season from May to November (September being the most intense month) and a pronounced dry season from December to April. Mean annual precipitation in the urban area ranges from 1240 to 1413 mm across the 15 monitoring stations analyzed.

The geomorphology is characterized by a very low gradient (mean slopes of 0.0003–0.0017 m/m in the principal drainage sub-catchments), alluvial deposits with moderate to low infiltration capacity (Hydrologic Soil Groups B and C dominating), and the presence of urban wetlands—notably the Berlin wetland on the left bank—that interact dynamically with the channel network during extreme events. These extremely low slopes—up to two orders of magnitude below the thresholds for which kinematic-wave and 1D routing assumptions remain reliable—constitute the primary methodological justification for adopting a fully 2D Rain-on-Grid approach over lumped or 1D hydrological schemes, since at such gradients backwater effects, bidirectional flow exchange between streets and channels, and ponding dynamics cannot be reproduced by methods that assume a dominant downstream flow direction.

The left bank of the Sinu, encompassing communes 1 and 2, is particularly exposed to compound flooding. Storm-runoff from the La Caimanera sub-catchment—bounded by the Serrania de Abibe to the west, the Sinu River to the east, the Las Palomas corregimiento to the south, and La Madera (San Pelayo) to the north—accumulates in low-lying areas adjacent to the Berlin wetland, and the absence of an adequate perimeter containment barrier allows lateral intrusion of wetland water into the urban fabric during sustained wet spells. This sub-catchment is articulated by two principal streams: Caño Viejo (Caño del Bien Común), which collects runoff from the Serrania de Abibe, and Caño La Caimanera (Caño El Vidrial), together with a network of minor channels and swamp-type water bodies that function as an integrated hydrological system (Figure 1).

3. Materials and Methods

This section describes the rainfall analysis, the hydraulic model setup and calibration, the hazard–vulnerability–risk assessment workflow, the formulation of structural and non-structural intervention measures, and the residual-risk modeling framework. The complete chain links meteorological forcing to design-level mitigation alternatives, including the cold-front-driven simulation that supported the February 2026 emergency response.

3.1. Rainfall Analysis and Design Storms

Daily precipitation records from 15 IDEAM stations within and surrounding Monteria were used. Records span 1960–2024, with all stations retaining more than 80% data availability after removal of years with more than 20% missing observations [22]. Consistency was verified through double-mass analysis (coefficient of determination greater than 0.997 for all pairs), and stationarity was assessed via the Mann–Kendall trend test and the U-test for change in the mean. Intensity–Duration–Frequency curves were synthesized following Vargas and Díaz-Granados [23] for the Caribbean hydrological region, and design storms for return periods of 2.33, 5, 10, 25, 50, and 100 years were generated for use as forcing in the hydrodynamic model.

Table 1. Summary statistics for the annual maximum 24 h precipitation at the 15 stations used in the frequency analysis. N = number of years used after removing years with >20% missing data; Mean, Std Dev, Skewness, and Kurtosis refer to the annual maximum 24 h rainfall series; CV = coefficient of variation; KS p = Kolmogorov–Smirnov p-value for the best-fit distribution; P media anual = mean annual precipitation.

Station	Record Period	N	Mean (mm)	Std Dev (mm)	CV	Skew.	Kurt.	Min (mm)	Max (mm)	KS p	Best-Fit Dist.	P Media Anual (mm)
Boca de la Ceiba	1970–2024	51	93.20	20.28	0.218	0.485	−0.115	56.0	140.0	0.13	Pearson	1413
Aero. Garzones	1976–2024	43	88.63	25.59	0.289	0.798	0.640	34.0	151.0	0.13	Pearson	1240
California	1975–2002	27	88.90	23.97	0.270	0.764	−0.453	59.4	140.0	0.12	Gumbel	1342
Coroza 1	1974–2003	24	95.65	21.73	0.227	0.073	−0.946	64.0	135.0	0.14	Normal	1399
Coroza 2	1974–2024	45	89.82	23.14	0.258	0.022	0.170	45.0	150.0	0.14	Normal	1342
Flor del Sinú	1978–2002	23	96.61	24.18	0.250	0.591	0.634	50.0	150.0	0.13	Pearson	1357
Horizonte	1975–2002	26	85.38	17.48	0.205	0.904	0.195	65.0	129.0	0.14	Pearson	1470
Lamas 3	1978–2002	22	83.55	14.05	0.168	0.394	−0.237	60.0	116.0	0.14	Normal	1371
Mocarí	1974–2002	24	83.19	19.32	0.232	1.906	7.238	47.7	153.4	0.14	Log-Normal	1268
Palma de Vino	1978–2002	21	90.29	24.46	0.271	0.658	0.619	50.0	150.0	0.14	Normal	1354
San Anterito	1973–2024	46	91.70	22.90	0.250	0.624	0.131	52.0	150.0	0.14	Log-Normal	1385
Santa Lucía	1968–2024	48	88.21	23.36	0.265	1.267	2.211	55.0	170.0	0.13	Pearson	1296
Galán	1979–2013	26	90.63	26.78	0.296	0.255	−1.096	49.0	139.0	0.14	Normal	1227
Turipaná	1960–2024	45	89.23	27.01	0.303	2.121	8.124	52.0	210.5	0.14	Log-Normal	1274
Montería	1976–2006	22	86.62	20.79	0.240	0.716	0.729	49.0	140.0	0.12	Gumbel	1403

presents the summary statistics for the 15 precipitation stations, including station name, record period, data availability (%), mean and standard deviation of annual maximum 24 h rainfall (mm), skewness, and kurtosis. Data availability across stations ranges from 69.2% (Turipaná, 1960–2024) to 96.4% (California, 1975–2002), with 12 of the 15 stations exceeding 80% availability. The stations with lower data availability (Galán, Turipaná, and Montería with 74.3%, 69.2%, and 71.0%, respectively) were retained in the analysis as the missing data is concentrated in a limited number of years and does not compromise the statistical representativity of the long-term records. Mean annual maximum 24 h rainfall across the network ranges from 83.19 mm (Mocarí) to 96.61 mm (Flor del Sinú), reflecting the spatial variability of extreme precipitation in the region. All stations exhibit positive skewness (mean skewness: 0.772), consistent with extreme-value distributions expected in precipitation frequency analysis. Kurtosis values range from −1.096 (Galán) to 8.124 (Turipaná), with elevated values at Turipaná and Mocarí driven by individual extreme events (210.5 mm and 153.4 mm respectively) that exceed the remaining series by significant margins.

Table 2. Summary statistics for the maximum annual discharge at the Montería Autónoma limnimetric station (13067020) on the Sinú River for the post-Urrá I period (2000–2024).

Station	Montería Autónoma (13067020)
River	Sinú
Analysis period	2000–2024 (post-Urrá I)
Number of years (N)	25
Mean (m3/s)	822.13
Standard deviation (m3/s)	89.05
Coefficient of variation (CV)	0.108
Median (m3/s)	820.00
Skewness	−0.282
Kurtosis (excess)	−0.167
Minimum (m3/s)	624.70 (2015)
Maximum (m3/s)	960.30 (2010)
Best-fit distribution	Log-Normal
KS distance statistic	0.099
KS p-value	0.95
μ_ln (mean of ln Q)	6.7061
σ_ln (std dev of ln Q)	0.1113

Note: Only the post-Urrá I period (2000–2024) was used, as the operation of the Urrá I hydroelectric dam (since 2000) fundamentally altered the river’s flow regime. The negative skewness (−0.282) reflects the dam’s regulation effect, which attenuates peak discharges.

presents the corresponding summary statistics for the maximum annual discharge data at the Montería Autónoma limnimetric station (13067020) for the post-Urrá I period (2000–2024), including mean (822.13 m³/s), standard deviation (89.05 m³/s), coefficient of variation (0.108), median (820.00 m³/s), skewness (−0.282), and the fitted Log-Normal distribution parameters (μ_ln = 6.7061, σ_ln = 0.1113). The analysis is restricted to the post-Urrá I hydroelectric dam period (operational since 2000) to ensure hydrologic consistency, as the dam’s regulation fundamentally altered the natural flow regime by attenuating peak discharges. The negative skewness observed in the discharge series reflects this regulatory effect, contrasting with the positive skewness characteristic of unregulated fluvial systems. The Log-Normal distribution provided the best fit to the annual maximum discharge data (Kolmogorov–Smirnov test: p = 0.95, Dist = 0.099), and these parameters were used to derive the design discharge values presented in the subsequent frequency analysis tables.

3.2. Rain-on-Grid 2D Hydrodynamic Model

The flood model was built in HEC-RAS 6.6 using a fully integrated distributed hydrological–hydrodynamic (Rain-on-Grid) approach in which rainfall is applied directly to each computational cell and the shallow-water equations govern the two-dimensional momentum-conserving routing of runoff and channel flow [24]. Recent reviews of flood inundation modeling [25] and rapid 2D flood mapping approaches [26] confirm the suitability of fully distributed hydrodynamic schemes for such settings. This methodology is particularly suited to the low-gradient, highly interconnected surface drainage of Monteria, where the distinction between overland and channel flow is blurred and empirical unit-hydrograph methods systematically underestimate travel times and flood depths [10]. The computational mesh was developed from a 20 cm-resolution photogrammetric Digital Terrain Model coupled with detailed bathymetric surveys of the Sinu River and 43 macro-drainage channels. Three mesh refinement levels were tested for mesh-independence: Mesh-A (coarse: 30 m in streets, minimum 1 cell per channel width), Mesh-B (intermediate: 20 m in streets, minimum 1.5 cells per channel width), and Mesh-C (fine: 10 m in streets, more than 2 cells per channel width). Independence was assessed by analyzing the invariance of the outlet hydrograph produced under a constant, spatially uniform rainfall applied over the principal urban sub-catchments; Mesh-C was adopted for all production simulations when the peak discharge and hydrograph shape converged with Mesh-B to within 3%.

A mesh independence analysis was already performed in the original technical study and has now been explicitly incorporated and clarified in the manuscript. Three computational meshes (M1, M2, and M3) with progressively finer spatial discretization were evaluated. The refinement included street resolution, number of computational cells across channels, and river discretization. Specifically, the river refinement increased from 50 m (>2 cells across the river width) in M1 to 15 m (>10 cells across the river width) in M3, while urban street refinement varied from 15 m to 10 m.

The models were forced using the same uniform rainfall event (10 mm/h during 10 h; total rainfall = 100 mm) and a Sinú River inflow of 837.7 m³/s, corresponding to a 2.33-year return period event. The analysis compared discharge, velocity, and water levels at the outlet of the Sinú River and urban drainage basins. Additionally, results from an extra hyper-refined mesh (M4), twice as refined as M3, were compared against the standard meshes. The comparison demonstrated limited variation between M3 and M4 results, indicating numerical convergence and confirming that the adopted mesh resolution provides stable hydraulic outputs while maintaining computational feasibility. This clarification has been added to the revised manuscript (Figure 2).

Surface roughness (Manning’s n) was assigned by land-cover class following Chow [27]: continuous urban fabric n = 0.018, discontinuous urban fabric n = 0.029, and roads n = 0.017. Runoff was partitioned using the SCS Curve Number method [28] under antecedent moisture condition CN(III) (wet), consistent with the sequential intense rainfall characteristic of the Caribbean wet season.

A sensitivity analysis of antecedent soil moisture conditions was incorporated to evaluate the influence of SCS Curve Number (CN) assumptions on runoff generation and flood propagation. The hydrological–hydraulic simulations adopted antecedent moisture condition CN(III) (wet conditions), consistent with the sequential intense rainfall events characteristic of the Caribbean wet season in Montería, where soils remain close to saturation for prolonged periods. CN(II) values were initially estimated from land-cover and hydrologic soil group information and subsequently transformed to CN(III) conditions for design-event simulations due to Regional Environmental Authority recommendations. The model was calibrated with 16 liquid gauging records from the Urra I power-plant team (2000–2018) at the Monteria Autonoma limnimetric station. For discharges below 600 m³/s the best-fit Manning’s value was n = 0.055; for peak discharges above 700 m³/s (the range relevant to extreme-event simulation) n = 0.06. The calibrated model was validated against observed water levels along Street 41 for two documented river-overflow events: 23 August 2007 and 17 December 2010, with simulated water levels matching observations with errors below 1%. Urban drainage performance was additionally validated against two intense pluvial events recorded at the Aeropuerto Los Garzones pluviograph: 1 August 2024 and 1 November 2024. Both events generated macro-drainage collapse and were documented through a community survey campaign using ArcGIS Survey123, which collected over 300 georeferenced depth measurements across 12 neighborhoods.

The model’s performance was evaluated using standard goodness-of-fit measures. For the Sinú River calibration dataset (14 measurement records from 2005 to 2018), with a Manning’s coefficient of n = 0.06, the Nash–Sutcliffe efficiency (NSE) was 0.916, the root mean square error (RMSE) was 0.229 m, and the coefficient of determination (R²) was 0.916. A slight systematic positive bias (mean error = +0.228 m) was identified, attributable to the progressive increase in the channel’s hydraulic conveyance capacity over time, which is conservative and acceptable for flood risk assessment.

Table 3. Observed vs. simulated water-level.

Date	Discharge (m3/s)	Obs. Level (m)	Sim. Level (m)	Delta Level (m)
29 August 2014	407.92	12.34	12.12	−0.22
21 October 2010	476.97	13.20	12.53	−0.67
15 December 2017	522.97	12.56	12.79	+0.23
10 June 2018	534.93	12.62	12.85	+0.23
5 December 2016	553.60	12.77	12.95	+0.18
26 June 2010	568.40	13.82	13.03	−0.79
19 August 2010	584.73	13.82	13.12	−0.70
28 June 2015	585.57	13.22	13.12	−0.10
18 May 2005	797.13	14.91	14.14	−0.77
11 July 2007	810.43	15.05	14.20	−0.85
27 August 2007	821.75	15.00	14.25	−0.75
26 October 2017	846.34	14.43	14.36	−0.07
13 July 2007	905.73	15.56	14.61	−0.95
28 June 2017	905.99	14.76	14.61	−0.15
Goodness-of-fit (n = 0.06)	NSE = 0.916	RMSE = 0.229 m	R2 = 0.916

Note: Delta Level = simulated minus observed. Positive values indicate overestimation; negative values indicate underestimation. NSE = Nash–Sutcliffe efficiency; RMSE = root mean square error; R² = coefficient of determination. MAE = 0.228 m; mean error = +0.228 m. Measurements sourced from the Urra I hydroelectric facility operators.

presents the observed water levels versus simulated ones, annotated with NSE, RMSE, and R². Validation against the August 2007 and December 2010 river flooding events yielded absolute water level errors ≤ 3 cm at nine monitoring points along Calle 41 (

Table 4. Revised calibration summary table with NSE, RMSE, and R².

Point	Obs. Depth (cm)	Sim. Depth (cm)	Error (cm)	Error (%)
13 August 2007—Limnimetric level = 15.85 m
0	15	18	+3.0	+20.0
1	17	20	+3.0	+17.6
2	18	20	+2.0	+11.1
3	18	19	+1.0	+5.6
17 December 2010—Limnimetric level = 15.87 m
0	20	18	−2.0	−10.0
1	23	26	+3.0	+13.0
2	28	25	−3.0	−10.7
3	30	27	−3.0	−10.0
4	35	33	−2.0	−5.7
Goodness-of-fit (9 points)	NSE = 0.846	RMSE = 2.54 cm	R2 = 0.846

Note: Observed and simulated depths are relative to local ground level at each monitoring point. Error = simulated minus observed. All absolute errors are ≤3.0 cm. NSE = 0.846; RMSE = 2.54 cm; R² = 0.846.

), corresponding to approximately 1.5–2% of the observed flood depths, with NSE = 0.846, RMSE = 2.54 cm, and R² = 0.846. The phrase “errors less than 1%” in the original text referred specifically to the percentage difference in water surface elevation at individual measurement points where Manning’s coefficient was optimized for the 2007 and 2010 events independently (e.g., 0.00% error at point 3 on 26 June 2010, with n = 0.08), prior to the final multi-year calibration.

3.3. Channel Hierarchy and Lateral-Inflow Extraction

For the mitigation analysis, channels were ordered hierarchically: first-order channels receive runoff directly from urban surfaces, while higher-order channels collect and convey lower-order contributions progressively toward the main channel. The nomenclature combines a sub-catchment prefix (e.g., CC for Canta Claro) and ascending numbering: first-order channels CC1, CC2, CC3; second-order channels CC_11, CC_12; third-order channels CC_21, CC_22; up to the outlet channel. Numbering follows a clockwise direction starting from the lower-left corner of the sub-catchment. This system is an adaptation of the Strahler [29] and Shreve’s [30] classifications to the particularities of Monteria’s urban drainage and enables a clear database identification of every reach.

The model separates effective precipitation from total precipitation and propagates the excess through the sub-catchment to the macro-drainage channels as lateral inflow. Second-order and higher channels additionally receive the accumulated discharges from upstream reaches, and the lateral contribution of each reach is computed as Q_lat = Q_downstream − Q_upstream. Zero or negative results were interpreted as absence of contribution—due to low runoff, very short reaches, or losses associated with potential overflows. This procedure distinguishes reaches that merely convey flow from those that actively contribute to it, which is essential for prioritizing structural interventions.

3.4. Hazard Classification

Flood hazard was classified into three categories following adapted Colombian technical guidance [22,31], calibrated to the specific hydrodynamic character of Monteria’s pluvial inundation (low velocity, high ponding depth). High hazard corresponds to depths ≥ 0.4 m and/or velocities ≥ 1.0 m/s; medium hazard to depths of 0.2–0.4 m and velocities of 0.5–1.0 m/s; and low hazard to depths below 0.2 m and velocities below 0.5 m/s.

The threshold h = 0.4 m for high hazard was selected based on the IDEAM damage-progression scale and validated empirically. Analysis of 12,551 randomly sampled depth pixels (0.2 m resolution; 3.35 × 10⁹ total pixels) confirms that 65.0% of pixels exceed 0.2 m and 42.5% exceed 0.4 m for the 100-year event. Velocity thresholds were correspondingly adapted, recognizing that only 20.4% of pixels reach v ≥ 0.5 m/s and only 10.7% exceed 1.0 m/s—a distribution consistent with the low-gradient, ponding-dominated hydrodynamic regime that distinguishes Monteria from high-velocity alluvial-fan or mountainous urban settings.

3.5. Vulnerability and Risk Assessment

Urban blocks within the POT 2021–2033 risk polygons [32] were identified through GIS location-selection. Critical facilities (hospitals, schools, public administration, emergency services), classified according to national risk-zoning guidelines [33], were extracted from the POT base cartography and spatially intersected with the risk polygon. Building-level exposure was characterized by 1465 household surveys (ArcGIS Survey123; 45 neighborhoods; multicriterion geographic quadrant sampling) and 156 commercial-zone interviews.

A dual-zone approach was adopted for vulnerability assessment, reflecting the distinct urban typologies within the study area. For the central commercial zone—characterized by mixed-use occupancy and higher-value contents—local depth-damage functions were derived directly from the 156 field interviews. Each respondent reported the maximum water depth experienced during the reference flood event (the 2010–2011 La Niña episode), the structural components damaged (walls, floors, doors), the contents affected (furniture, appliances), and the estimated time to functional recovery. Survey responses were converted into depth-damage pairs by assigning each record to a water-depth class (bins of 5 cm from 0 to 80 cm, and a single class above 80 cm) and estimating the damage as a percentage of replacement value based on the reported extent of structural and content losses. These empirical points were then fitted to the vulnerability function proposed by Cardona [34], following the formulation of the formulation proposed by Cardona and Ordaz [35]:

$$\mathrm{E}(\mathrm{p}|\mathrm{s})\mathrm{ }=\mathrm{ }1\mathrm{ }-\mathrm{ }\mathrm{e}\mathrm{x}\mathrm{p}\left[\mathrm{l}\mathrm{n}\left(0.5\right)·\mathrm{ }{\left({\displaystyle \frac{\mathrm{s}}{{\mathrm{s}}^0}}\right)}^{\mathsf{\epsilon }}\right]$$

(1)

where s₀ is the flood intensity (maximum depth) producing 50% expected loss and ε governs the curve slope. For the commercial zone, the fitted parameters were s₀ = 1.20 m and ε = 2.00. The fitting was performed by least-squares minimization on the binned empirical damage ratios, and the resulting curve was checked for monotonicity and plausibility (damage increasing with depth, near-zero damage below 5 cm).

For the residential neighborhoods and peripheral sectors—predominantly single-story cement-block dwellings—the construction of fully independent local curves was not feasible due to the lower density of damage observations per depth class in individual neighborhoods. Instead, the depth-damage functions developed by Cardona et al. [36] for the La Mojana Caribbean floodplain were adopted as baseline curves, on the basis that the La Mojana region shares with Montería the same dominant construction typology (unreinforced cement-block walls, concrete or tile flooring), the same hazard regime (slow-rise pluvial ponding with depths up to 1.5 m and velocities below 1 m/s), and comparable socioeconomic conditions. The La Mojana functions were then adjusted to Montería conditions using correction factors derived from the 1465 household survey damage ratios: the observed mean damage at each depth class was compared with the La Mojana prediction, and the s₀ and ε parameters were shifted to minimize the residual between survey observations and the transferred curve, yielding s₀ = 1.50 m and ε = 1.85 for one-story cement-block dwellings.

To reduce noise and avoid non-physical behavior, all fitted functions were reviewed for plausibility, ensuring that damage increased monotonically with depth and that lower depths produced limited loss. The variance of the loss was estimated using the ATC-13 functional form [37], with Vmax and D₀ parameters consistent with the cement-block structural type. Risk was computed as the convolution of hazard and vulnerability over exposed elements, producing two risk maps—one considering only exposed buildings and another that additionally incorporates road infrastructure as an exposed asset. The economic expression of risk follows Salazar [4] and Cardona [34], computing the expected annual loss as the product of the unit damage function, the probability of exceedance of each return-period event, and the number of exposed elements per depth class, consistent with the methodological recommendations for economic flood damage assessment [38] (Figure 3).

3.6. Hydraulic Assumptions for Structural Measures

For the definition of structural measures associated with macro-drainage, an integral urban-drainage analysis was carried out for a design event with a 100-year return period, based on two main premises. First, all runoff water that previously ponded in some areas is now assumed to reach the channels—that is, the micro-drainage system is assumed to convey the entire rainfall excess to the macro-drainage—so that the resulting flood patches correspond only to inundations due to macro-drainage overflow. Second, the most critical projected urban-expansion scenario is considered to evaluate its effect under current drainage conditions and under scenarios with mitigation measures implemented, specifically in the Villa Cielo sector where the expansion affects the coverage of the sub-catchment draining to the main collector channel.

Discharge hydrographs extracted from the flood model were used as input to each macro-drainage channel. Since the total volume of the raw hydrographs

$$\Sigma (\mathrm{Q}·\Delta \mathrm{t})={\mathrm{V}}_0$$

(2)

is smaller than the excess precipitation volume

$$\Sigma ({\mathrm{P}}_{\mathrm{e}}·\Delta \mathrm{t})={\mathrm{V}}_{\mathrm{e}}$$

(3)

because the rainfall-excess volume retained in ponded areas does not reach the channels—the hydrograph Q is multiplied by a correction factor

$$\mathrm{F}\mathrm{c}={\mathrm{V}}_{\mathrm{e}}/{\mathrm{V}}_0$$

(4)

so that the corrected discharge

$${\mathrm{Q}}_{\mathrm{c}}=\mathrm{F}\mathrm{c}·\mathrm{Q}$$

(5)

where V₀ is the total volume of the raw discharge hydrograph (m³), computed as the summation of the product of discharge Q (m³/s) and time step Δt (s) over all time steps; V_e is the total excess precipitation volume (m³), computed as the summation of effective precipitation P_e (m) multiplied by time step Δt (s) and the contributing area; Fc is the dimensionless volume-conservation correction factor; and Q_c is the corrected discharge (m³/s) applied to each macro-drainage channel for the structural-measure analysis generates a rainfall volume equal to the total rainfall excess. This volume-conservation correction is essential to prevent systematic underestimation of design discharges in the channel-capacity analysis.

Urban Expansion Scenarios (Villa Cielo)

For the Villa Cielo expansion area, a weighted Curve Number of CN = 75.12 was computed over the 5.27 km² polygon under current conditions, in which more than 90% of the surface is covered by forested and weedy pastures (

Table 5. Curve Number (CN) values for the land-cover units within the Villa Cielo expansion area.

Land Use	Area (km2)	% Area	CN
Forested pastures	3.59	68.04	75
Weedy pastures	1.18	22.47	72
Continuous urban fabric	0.25	4.83	89
Discontinuous urban fabric	0.24	4.53	77
Roads	0.00	0.08	92
Artificial water bodies	0.00	0.04	98
Urban green areas/bare land	0.00	0.01	80–85
Total (weighted CN = 75.12)	5.27	100.0	—

). Five prospective land-use scenarios were then defined, ranging from fully discontinuous urbanization (Scenario 1, CN = 77) to fully continuous urbanization (Scenario 5, CN = 89), including intermediate 25/75, 50/50, and 75/25 mixes. Each scenario was evaluated with the SCS method to estimate maximum discharges for several return periods (

Table 6. Maximum discharges (m³/s) for the current situation and urbanization scenarios.

TR (yr)	CN = 75	CN = 77	CN = 80	CN = 83	CN = 86	CN = 89
2.33	20.18	22.49	23.73	26.31	27.65	30.41
5	33.92	36.70	38.12	40.99	42.44	45.56
10	38.31	41.18	42.64	45.57	47.04	50.27
25	44.52	47.50	48.99	51.98	53.56	56.85
50	54.13	57.22	58.75	61.91	63.60	66.92
100	65.43	68.61	70.18	73.61	75.32	78.65

). The resulting direct-runoff hydrographs were propagated through the main collector channel to evaluate how progressive urbanization modifies the hydraulic response of the expansion sector.

3.7. Sinu River Modeling

Along the urban reach, the Sinu River presents approximately 17 points of observed fluvial erosion (points of higher shear stress) according to observations by the Regional Environmental Authority (Corporación Autónoma Regional de los Valles del Sinú y del San Jorge, CVS) [39] and to the hydraulic-modeling results of this consultancy. In these reaches, flow velocity and water force are directed toward the outer bank of the curve, generating intense pressure that undermines and erodes the riverbank material; the inner banks tend to experience lower shear stress and accumulate sediments. Hydraulic modeling shows that, for a 100-year return period, a discharge of 1021 m³/s causes the Sinu to overflow into its floodplain, generating inundations in the sector and affecting approximately 289 identified properties.

A dedicated cold-front-driven simulation was performed to reproduce the February 2026 emergency scenario on the left bank. Unlike the convective storms that dominate the wet season, cold-front forcing delivers sustained multi-day rainfall that saturates soils and progressively overwhelms the discharge capacity of the La Caimanera sub-catchment. This simulation provided the hydraulic context for the emergency response described in Section 6 and informed the design basis for the 4300 m perimeter dike and the six pump-station sectors (Figure 4).

4. Results

4.1. Flood Hazard: Mechanisms and Spatial Distribution

For the 100-year event, the calibrated 2D model identifies three pluvial inundation mechanisms: poor hydraulic connectivity (dominant; ponding depths up to 1.5 m), limited drainage evacuation capacity (most acute in the INAT sub-catchment and Villa Cielo), and downstream channel overflow. High hazard covers ~375 ha (9.2% of the urban perimeter; depth ≥ 0.4 m and/or velocity ≥ 1.0 m/s) and medium hazard ~560 ha (13.7%). The Sinú River overtops its urban banks only above the 25-year return period (100-year discharge: 837.7 m³/s), exclusively at Street 41—a consequence of Urrá I regulation since 2000 (Figure 5).

4.2. Urban Flood Risk: Exposure, Vulnerability, and Asset Classification

The 100-year inundation raster intersects 2452 urban blocks across the POT risk polygons (Commune 4: 659 blocks, 566 in high hazard) and 121 strategic facilities, of which 65 (54%) are schools. Of 3015 scored assets, 36.36% are high risk and 57.77% medium risk—only 5.91% are low risk, reflecting the city’s flat, poorly drained morphology. The 289 properties in the 100-year Sinú hydraulic corridor and 1812 dwellings within 12 m of primary channel axes account for the principal high-risk concentrations. Field surveys (1465 households) show cement-block construction dominates (70.78%); 26.5% of households reported flood-water entry, of which 65.9% recover within one month, though 30.1% rated neighbourhood drainage as inadequate. Three neighborhoods (Cantaclaro, La Gloria, Portal de La Candelaria) face heightened evacuation risk due to elevated concentrations of residents with physical or mental disabilities (Figure 6).

4.3. Ranking of Priority Hotspots

Neighborhood-level ranking (mean high-hazard area 1.83 ha, median 0.86 ha) identifies Furatena II, San Jerónimo, Los Mangos, and Portal de La Candelaria as the leading absolute hotspots, and Barrio Villa Norte as the highest proportional-exposure case (>50% of its territory in high hazard). These rankings are fed directly into the CURBA licensing platform (Section 5.2.4) to guide the spatial allocation of structural works and non-structural instruments.

5. Intervention Measures

All the analyses carried out indicate that the flood risk in the city of Monteria is characterized as mitigable in the already-consolidated urban zone and additional mitigation potential in zones that remain urbanized, provided that gradual measures are applied. The general approach for flood management in Monteria is aligned with the planning defined in the POT 2021–2032 regarding peak stormwater discharges, which requires all new developments to reduce by at least 50% the maximum discharge hydrograph (comparing runoff with and without the project) and to address the remaining 50% through recovery and expansion of the urban drainage channel system (new drainage systems, rehabilitation of existing systems, new SUDS-type structures). According to the proposed methodology, measures are organized along four axes—Sinu River, Macro-drainage, Micro-drainage, and New residential and building developments—and for each axis structural and non-structural (prospective) typologies are defined (

Table 7. Structural and non-structural typologies per management axis (adapted from Decree 1807/2014 framework). The non-structural column spans conceptually distinct sub-categories—land-use planning instruments, early-warning technology (SAT), and financial risk-transfer tools—which correspond respectively to the Governance, Enabling conditions, and Economic sub-categories of the IPCC AR6 adaptation taxonomy [1].

Axis	Structural	Non-Structural (Prospective)
Sinu River	N/A	Relocation of exposed assets; bank-stability studies for new developments; formal 12 m ronda delimitation
Macro-drainage	Channel optimization (section, dikes, land-raising, retention lagoons)	12 m buffer
Micro-drainage	Channel section optimization	12 m buffer; cleaning and maintenance; water parks; downtown drainage plan
New developments	N/A	Sustainable construction code; minimum-elevation map; land-management instruments; SAT; CURBA

).

5.1. Structural Measures: Scenarios M1 and M2

Two structural mitigation scenarios were evaluated. M1 includes channel widening in strategic reaches. M2 combines M1 with three additional components: raising the ground level in the vicinity of the confluence of the southern channels and in the northern zone near the Buenavista Shopping Center; incorporation of protection dikes in the main collector channels; and use of the north-eastern lagoons—no longer functioning as a wastewater treatment system—as a retention and peak-regulation system. These lagoons connect to the INAT channel at the Monteverde neighborhood through a system analogous to an overflow weir: water enters the lagoon during the peak and returns to the same channel further downstream, regulating the discharge that transits through the channel (Figure 7).

M1 results indicate reductions greater than 15 cm in flood elevation in the initial reaches of the modifications but increases in level in the final reaches because the widening enables faster evacuation, which increases downstream peak discharges and therefore the inundated area. Increases were equal to or less than 5 cm at the outlet of the Purgatorio channel and on the left bank. Overall, 24 hectares were removed from the inundated area, and 12 new hectares were added.

Under M2, reductions greater than 15 cm in flood elevation were observed in the initial reaches, analogous to M1, but the magnitude of the reduction is greater and the increase in level in the final reaches is concentrated around the fillings due to flood displacement. In total, 80 hectares of area no longer flood (including filled areas) and 28 hectares of new flooded areas emerge. At the local scale, at the height of the Monte Verde neighborhood at the INAT channel confluence, the flooding was eliminated. The most important assumption of this model is that the entire city drains perfectly toward one of the channel systems and what is analyzed is the hydraulic capacity of the channels to handle this discharge integrally. Ponding must therefore be resolved through the Sustainable Urban Drainage Master Plan (PMDUS) once implemented (Figure 8).

The costs associated with the PMDUS interventions required to mitigate ponding and improve local stormwater management are included in the overall economic assessment presented in

Table 8. Cost estimate of pre-dimensioned structural works (scenario M1 and M2).

Measure	Quantity	Unit Cost (COP)	Total Cost (COP)
Channel optimization	17,151.1 m	3,074,638/m	52,734,653,931
Ground-level raising	299,521 m3	95,000/m3	28,454,495,000
Protection dikes	34,806 m3	95,000/m3	3,306,541,500
TOTAL	—	—	84,495,690,431

.

These costs are estimates based on a pre-dimensioning that considers only the optimization of the hydraulic capacity of the system and does not include other costs. To refine this value, the corresponding technical, economic, social, and environmental feasibility studies must be carried out.

The COP 84,495,690,431 cost estimate for scenario M2 reflects exclusively the hydraulic-capacity optimization of the macro-drainage system (channel earthwork, fill material, and dike construction) at pre-design level and is therefore subject to significant upward sensitivity. The estimate does not incorporate property acquisition and easement costs—which in the Colombian context require lengthy land-title regularization processes, particularly in informal settlements—nor social management, environmental licensing, detailed engineering, construction supervision (AIU), or contingency provisions. Based on comparable urban drainage projects in the region, such as the Veolia downtown alternative (COP 87.7 billion including AIU and supervision for a smaller intervention area), the total executed cost could increase by 40–80% once these components are included. A formal cost–benefit analysis was not undertaken because the depth-damage functions quantify expected losses as fractional replacement values rather than absolute monetary figures, precluding direct comparison with investment costs without a property-valuation census that exceeds the scope of the present pre-design phase. Subsequent feasibility studies (required under Colombian public-investment guidelines before any procurement) must produce itemized budgets, benefit–cost ratios, and financial sensitivity analyses across discount rates and climate-change scenarios to support investment decisions.

Urban Drainage Plan for the Downtown Area

An additional structural alternative is the Urban Drainage Master Plan for the Downtown Area developed by Veolia Consulting for the Mayor’s Office of Monteria [40]. It has a direct scope of approximately 296 ha, corresponding to the downtown area from Circunvalar Avenue to Carrera 1 and from Street 21 to Street 44 with Carrera 2. The study area comprises 25 neighborhoods with 49,140 inhabitants; the project also benefits the 27,088 inhabitants of the Monteverde sector. The plan includes widening of the South Collector Channel from Av 9 with Street 22 to Street 27 with Circunvalar Ave, adaptation of the Street 21 channel, widening of the Central Collector Channel from Street 29 to Street 41, and construction of an overflow weir or gate at Street 41. The budget is COP 41,750,647,354 for Stage I and COP 45,922,677,282 for Stage II, including AIU, supervision, and Ministry of Housing review. Feasibility studies are required before execution, given the proposal to use the Sinu as a receiver of treated pluvial waters.

5.2. Non-Structural Measures

5.2.1. Sinu River Ronda Management

The actions contemplated within this measure include, first, the resettlement and relocation of exposed assets that show high vulnerability to flooding due to overflow of the Sinu River, with approximately 35 properties identified in the Municipal property database overlapping with the 100-year flood raster, medium- to long-term execution deadlines, and funding from UNGRD, the Adaptation Fund, or municipal resources. Second, technical recommendations for new urban developments: Detailed bank-stability studies of the Sinu (including geophysical tests, shear-strength tests, and slope modeling) are considered a priority, generating recommendations for slope-stabilization works to reduce the probability of a dike-failure event that may generate a breach-induced flood.

5.2.2. Sustainable Urban Drainage Master Plan (PMDUS)

According to hydraulic modeling, the identified mechanisms in urban drainage are (i) poor connection with main channels, (ii) low evacuation capacity, and (iii) channel overflow and accumulation in low-lying areas. The consultancy for the PMDUS is estimated at around COP 300 millions, and its main products are the detailed engineering of each of these problems. It articulates four integrated components.

First, urban-drainage management at the dwelling and building level: A minimum 50% reduction in the maximum discharge hydrograph in new developments is required, exceeding the national requirement of the RAS Technical Regulation (MVCT R330/2017 and its modifications) [41]. This measure targets new developments within the urban perimeter and the expansion area, with short-term implementation led by the Municipality and coordinated with the urban conservation district and the construction sector.

Second, optimization of the existing micro-drainage: the weakness of the micro-drainage network and its low connectivity with the macro-drainage is a critical challenge requiring a strategic and integral approach. Activities include identification and cartography of the existing micro-drainage network (gutters, inlets, secondary pipes) and its connection to the macro-drainage; analysis of design capacity versus actual capacity; condition evaluation; urban hydrological and hydraulic modeling; determination of design discharges for future infrastructure; and integration with the risk study.

Third, water parks (“giant sponges”): These parks are conceived to temporarily retain and store large volumes of rainwater through capture of the rainfall peak, temporary retention, controlled pumped evacuation, and residual-volume management through evaporation. The standardized parametric design has a maximum depth of 80 cm (including storage volume and freeboard); for the design storm, an event with a 25-year return period is selected—95.0 mm in 1 h 23 min. Specific measures prevent vector proliferation, and OPEX is estimated at USD 0.55–2.20 per m²/year.

Fourth, channel cleaning, maintenance, and citizen awareness: Deficiencies in citizen behavior (waste disposal into channels) are a recurrent factor in flooding. A comprehensive plan comprises continuous preventive maintenance (removal of garbage, sediment, and obstructing materials), correct sediment-disposal protocols, active collaboration with waste-collection companies as a strategic channel for household outreach, and permanent condition evaluation through regular inspections and technology (drones, sediment-level sensors). The reference unit cost reported for Monteria is COP 312,968.36 per linear meter.

In addition to gray infrastructure, the PMDUS framework explicitly integrates Nature-Based Solutions and blue-green infrastructure as complementary stormwater management measures. The study highlights the potential of distributed retention and controlled evacuation through water parks, which are described as “giant sponges,” and presents them as part of a sustainable drainage approach with a maximum storage depth of 80 cm and controlled discharge. It also recommends reinforcing drainage governance at the building and neighborhood scales through measures such as improved microdrainage, drainage maintenance, and the integration of results into planning and licensing tools. At the same time, the document stresses that the viability of such measures depends on local conditions and that the urban flood strategy should combine structural works with non-structural actions, maintenance, and periodic updating (Figure 9 and Figure 10).

5.2.3. Early Warning System (SAT) Integrated with Urban Mobility

An IoT-based SAT connected to the urban mobility system is proposed, based on four pillars. Monitoring integrates a network of IoT sensors for real-time river/channel stages and precipitation, three automatic meteorological stations, and a network of trained local observers providing contextual validation. Analysis processes data through IoT platforms and advanced hydrological-forecasting models to predict urban ponding (magnitude and duration), complemented by infiltration-capacity analysis to estimate excess water that the terrain cannot absorb. Dissemination combines SMS alerts, email notifications, sirens, megaphones, flags, and emergency-light systems to reach the maximum population, especially vulnerable communities with limited technological access. Preparation activates clear alarms, evacuation of risk zones, and continuous communication with the Monteria Risk Management Group, together with detailed evacuation plans and inter-institutional coordination.

Integration with the mobility system includes automatic identification of alternative routes based on real-time flood information, implemented through Intelligent Transportation Systems (ITS) that automatically divert traffic or adjust signal timing; variable-message signs (VMS), mobile applications, and on-road water-level sensors activating flashing warning beacons; and a unified command center integrating SAT and traffic-management data for real-time decision-making with transit authorities. The initial investment is estimated at COP 300 millions.

The operational feasibility of the proposed SAT rests on three pillars. First, the sensor network comprises commercially available IoT devices (ultrasonic water-level sensors with LoRaWAN or 4G telemetry, tipping-bucket rain gauges) with unit costs of USD 500–1500 per station, installed at 3–4 critical nodes identified from the hydraulic model (channel junctions, ponding hotspots, river-bank reference points). Second, annual operation and maintenance (O&M) costs are estimated at 10–15% of the initial investment (approximately COP 30–45 million/year), covering sensor calibration, battery replacement, data hosting, and software licensing. Third, institutional sustainability is ensured through integration with the Montería Municipal Risk Management Group (CMGRD), which already operates a basic hydrometeorological monitoring network. The SAT design follows the four-pillar structure (monitoring, analysis, dissemination, preparation) recommended by UNGRD [42], and its long-term sustainability depends on the establishment of the proposed Urban Drainage Management Unit with dedicated budgetary autonomy (Section 7.1).

5.2.4. CURBA Integration and Minimum Construction-Elevation Map

CURBA is the Monteria municipal web-based information system, designed to administer land-use-concept requests; it uses QR codes to prevent plagiarism of land-use certificates and has a CURBA Rural version that functions as a GIS for predictive mapping. Integrating the detailed risk-study outputs (mitigable and non-mitigable high-risk zones, minimum construction-elevation criteria, structural determinants) directly into CURBA enables informed decision-making for urbanization, optimization of the licensing process (eliminating additional procedures currently required of developers and reducing costs and timelines), reduction in high-risk zones on the city’s risk cartography, and strengthening of proactive urban planning.

The minimum construction-elevation map was produced from advanced hydraulic modeling for a 100-year return-period event and complemented with a 60 cm freeboard safety factor following FEMA risk-management standards [43]. This additional height incorporates a protection margin for model uncertainties, climatic variations, and unforeseen drainage obstructions. The map is a technical-guidance tool and not a rigid normative instrument: it does not substitute detailed studies at the property scale, which may refine the suggested elevations through site-specific modeling. Wetland filling is prohibited; Nature-Based Solutions (connectivity corridors, green infrastructure, passive-recreation spaces) are preferred instead (Figure 11).

5.2.5. Land-Management Instruments

Four instruments are proposed to guide development away from high-risk areas and to finance mitigation. Transfer of Development Rights (TDR, existing in Colombia since 1997) allows the transfer of construction rights from flood-prone or hydrologically vital areas (wetlands, floodplains, zones along the Sinu) to higher-elevation, lower-risk zones, incentivizing owners not to develop in risk areas. Value Capture allows the city to capture a portion of the increase in land value generated by public urban actions such as green-infrastructure investments. Betterment Levies allow the city to recover part of the public investment in drainage improvements or flood-mitigation projects through contributions from benefited property owners, fostering co-responsibility for maintenance. Impact/Development Fees are contributions that developers make to the city (in land or money) for the impact of their projects; new developments near the river or drainage system can be required to incorporate flood-mitigation measures within their own projects, and monetary contributions can be allocated to a specific flood-risk-management fund.

5.3. Residual Risk Analysis (Monte Carlo Model)

A theoretical residual-risk model was formulated to be validated as measures are implemented and evaluated. The model aims to analyze the reduction in residual risk under different implementation scenarios, identify the most sensitive and effective measures, and provide a theoretical basis for decision-making. Base risk is set at 1.0 (100%), representing the current state. Measure efficiencies and impact weights are: improved drainage (70% efficiency, 30% impact), early-warning systems (80%, 25%), institutional capacity (75%, 20%), community preparedness (65%, 15%), and quality of hydrological–hydraulic models (85%, 10%). Six implementation scenarios were considered, from 0% (risk 100%) to 100% (expected risk 15%). The model uses Monte Carlo simulation with 10,000 iterations per scenario, a Beta distribution for efficiencies adjusted to the implementation level, and a normal distribution with variable standard deviation for uncertainty. The residual risk is calculated as

$$\mathrm{R}\mathrm{e}\mathrm{s}\mathrm{i}\mathrm{d}\mathrm{u}\mathrm{a}\mathrm{l}\mathrm{ }\mathrm{R}\mathrm{i}\mathrm{s}\mathrm{k}=\mathrm{B}\mathrm{a}\mathrm{s}\mathrm{e}\mathrm{ }\mathrm{R}\mathrm{i}\mathrm{s}\mathrm{k}\times (1-\Sigma \mathrm{ }\left({\mathrm{E}\mathrm{f}\mathrm{f}\mathrm{i}\mathrm{c}\mathrm{i}\mathrm{e}\mathrm{n}\mathrm{c}\mathrm{y}}_{\mathrm{i}}\times {\mathrm{I}\mathrm{m}\mathrm{p}\mathrm{a}\mathrm{c}\mathrm{t}}_{\mathrm{i}}\right))\mathrm{ }$$

(6)

The residual-risk analysis was developed as a theoretical Monte Carlo model to estimate flood-risk reduction under progressive implementation of mitigation measures. The model starts from a base risk ${\mathrm{R}}_0=100\%$ and estimates residual risk as

$${\mathrm{R}}_{\mathrm{r}}={\mathrm{R}}_0(1-\sum \mathrm{ }{\mathrm{E}}_{\mathrm{i}}{\mathrm{I}}_{\mathrm{i}})$$

(7)

where ${\mathrm{R}}_{\mathrm{r}}$ is the residual risk, ${\mathrm{R}}_0$ is the base risk, ${\mathrm{E}}_{\mathrm{i}}$ is the efficiency of measure i, and ${\mathrm{I}}_{\mathrm{i}}$ is its impact weight. The model includes five measures with predefined efficiencies and weights: improved drainage, early warning systems, institutional response capacity, community preparedness, and hydrological/hydraulic model quality. To propagate uncertainty, the simulation was run 10.000 times per scenario.

The model assumes independent effects among measures, linear aggregation of their contributions, gradual implementation over time, stable efficiencies, and adequate maintenance. Measure efficiencies were sampled probabilistically, and uncertainty varied by implementation scenario. This design captures both the expected effect of the measures and the variability associated with partial implementation.

Results show a non-linear reduction in residual risk as implementation increases, with diminishing returns at higher levels of deployment. The strongest gains occur between 20% and 60% implementation, and the minimum residual risk reached in the optimal scenario is 19.9% (Figure 12). Sensitivity analysis indicates that improved drainage is the most influential measure, followed by early warning systems and institutional response capacity.

Risk reduction does not follow a linear progression: a diminishing-returns phenomenon is observed, where initial efforts are notably more effective than later ones. The first 20 percentage points of implementation achieve a risk reduction of 14.1%, while the following 20 points achieve a smaller reduction of 12.4%. The inherent uncertainty is maximum in intermediate scenarios (40–60% implementation) and minimum in the extreme 0% and 100% states. The staged implementation confirms incremental benefits: basic measures (20%) reduce risk to 85.8%, moderate (40%) to 73.9%, advanced (60%) to 58.2%, complete (80%) to 40.1%, and optimal (100%) to 19.9% (Figure 12).

Sensitivity analysis identifies improved drainage as the dominant measure, with a correlation of 0.632 with risk reduction—the highest base efficiency among structural measures (70%) and the highest weight in the model (30%). Early warning systems follow with a correlation of 0.521 and the highest base efficiency of the entire model (80%), making them the second priority because of their favorable cost–benefit ratio. Institutional response capacity shows medium dominance (correlation 0.424). Economic implications are clear: a relatively low initial investment produces a significant risk reduction, with the 20–60% implementation range being the most cost-effective and 60% implementation marking the inflection point where reduction becomes substantial (Figure 13).

6. Emergency Response: February 2026 Cold-Front Event

6.1. Event Characterization and Hydrological Context

In February 2026, an atypical cold front entered the Colombian Caribbean [44]—an event type uncommon during the dry season (December–March) but consistent with anomalous atmospheric circulation documented for early 2026. Unlike the convective storms that dominate the wet season, the cold front delivered sustained multi-day rainfall that saturated soils and progressively overwhelmed the discharge capacity of the La Caimanera sub-catchment on the left bank. The affected area—communes 1 and 2—covers the left-bank urban and peri-urban strip bounded by the Berlin wetland to the west and the Sinu River to the east. Lateral wetland intrusion, surface-runoff accumulation, and inadequate perimeter containment produced prolonged inundation with depths of 50–70 cm in critical zones (notably Barrio El Dorado and Vallejo) and submergence of road surfaces in La Ribera, República de Panamá, Villa Nazaret, and adjacent sectors (Figure 14).

6.2. Field Documentation of Water-Surface Elevations

Field teams conducted systematic topographic-leveling campaigns across 12 neighborhoods within the inundated zone, documenting the maximum water-surface elevation (cota máxima de lámina de agua) at selected reference points and at the junctions of the primary urban channels. These observations served two complementary purposes: validation of the hydraulic model for a real cold-front forcing scenario, and design basis for the gravity-outlet and pumping alternatives analyzed for emergency water evacuation.

An important key finding was located at the Vallejo–Berlin interface, the maximum water-surface elevation coincided with the highest historical inundation level recorded in adjacent dwellings, confirming that the Berlin wetland level controls the lateral boundary condition for the left-bank urban drainage. The Channel Vallejo–Dorado carries water at near-bankfull capacity, with the earthen berm separating Channel Dorado from the adjoining borrow pit limiting the effective hydraulic cross-section and impeding rapid drainage of the Vallejo and El Poblado sub-sectors. Channel INAT showed free-surface elevations consistent with blockage-induced backwater at its downstream discharge point into the Sinu. The documented elevations were used directly to dimension the crest elevation (minimum 1.7 m above street level in the most critical section at Vallejo–Berlin) and the discharge capacity of the proposed containment dike.

6.3. Three-Front Mitigation Strategy

Based on the hydraulic analysis and field documentation, a multi-component emergency mitigation strategy was designed and implemented jointly by Consorcio Riesgos Monteria, the Regional Environmental Authority (CVS), and the Municipal Administration of Monteria. The strategy comprises three integrated fronts of operation.

6.3.1. Front 1—Perimeter Containment Dike (4300 m, Commune 1)

A perimeter dike was constructed and/or reinforced along a 4300 m alignment encircling Commune 1, from Barrio Los Colores (northern limit) to Barrio El Dorado (southern limit). Construction used heavy machinery wherever soil conditions and maneuvering space were allowed, through two simultaneous fronts progressing toward the center of the perimeter (approximately behind the Vallejo urbanization). The first front, in the Los Colores sector, began from the Vía a Las Palomas, bordering the neighborhood to the channel at its end and then along the channel dikes. The second front, in the El Dorado sector, began from the Arboletes road. The dike crest was designed to the maximum recorded water-surface elevation plus a freeboard of at least 0.5 m, with a minimum height of 1.7 m at the Vallejo–Berlin interface.

Two construction methods were evaluated (Table 4): geotextile bags with compacted impervious fill (sandbags filled with calcareous or clayey material, encased in non-woven geotextile to form a large containment bag), and geocontainers or hydrotubes (large-format geotextile bags filled with impervious material). Anti-return gates were installed at channel–street intersections within the dike alignment to prevent backflow from the channels into the protected urban area during high Sinu stages; their locations were selected at the junctions of the channels to guarantee water evacuation and prevent the generation of a piston effect. In parallel, Channel Vallejo–Dorado hydraulic optimization was identified as imperative: reshaping and raising the banks of the earthen berm to the documented maximum elevation is required to provide an adequate hydraulic gradient and prevent backwater toward the urbanized area (

Table 9. Construction alternatives for the 4300 m perimeter dike: advantages and disadvantages.

Alternative	Advantages	Disadvantages
Geotextile bags with impervious bag-soil fill	Lower cost	Reduced maneuverability
Geocontainers/hydrotubes	Greater installation speed; superior hydraulic containment	Higher unit cost

).

6.3.2. Front 2—High-Capacity Pump Stations

In parallel with dike construction, pumping equipment was deployed in six operational sectors within communes 1 and 2. Major stations were sized for at least 1000 L/s (1 m³/s), complemented by 3-inch motor pumps for localized ponding zones. The objective was to control evacuation of water trapped between the urbanized zone and the perimeter dike through pumping to the external side of the dike, without allowing water to re-enter dried sectors. Sectors that become fully or partially dry retain oversaturated soils from which water may re-emerge—since hydrostatic pressure from the column outside the dike drives re-infiltration—and pumping equipment must therefore remain on standby for residual evacuation.

Table 10. Operational sectors of Communes 1 and 2 for pumping-equipment deployment during the February 2026 event.

Sector	Neighborhoods	Observation
1	La Ribera; Panamá neighborhood	Evacuating through Channel Centenario, redirected to Pitolandia property.
2	La Navarra; El Portal I–III; Los Colores; Los Ébanos; Villa Nazaret; La Vid	Much of the territory is already dry.
3	La Palma; Rancho Grande; Mi Ranchito; Nuevo Horizonte; Sector Campano	Mostly dry; lower-capacity pumps are required to remove stagnant water.
4	Caracolí; El Níspero I–II; Vallejo	Critical zone adjacent to Berlin wetland. Strategy: perimeter dike + 6–8 × 3 motor pumps; tractor pump at Channel Vallejo intersection (highest cota).
5	El Dorado; El Poblado	Standing water 50–70 cm average in El Poblado; intervention on Channel Pitolandia dike and high-capacity tractor pump recommended.
6	Commune 2	Water has already evacuated to Channel Centenario.

summarizes the operational sectors and field observations.

6.3.3. Front 3—Channel Vallejo–Dorado Hydraulic Optimization

The earthen berm dividing Channel Dorado from the adjacent borrow pit was identified as the critical hydraulic bottleneck. Field measurements of maximum water-surface elevations in adjacent dwellings defined the required design section: the berm must be reconformed and raised to at least the documented maximum cota to provide an adequate hydraulic gradient and prevent backwater toward the urbanized area. This structural intervention is recommended as a priority permanent measure following the emergency phase.

7. Discussion

The results are discussed around three interpretive themes: (1) the pluvial-dominance finding and its implications for investment priorities; (2) system-scale evaluation of structural measures under current and projected climate conditions; and (3) the limitations of the modeling framework and directions for future work.

7.1. Pluvial Dominance and Its Investment Implications

The four-phase workflow demonstrates the operational value of maintaining a calibrated urban hydrodynamic model as a standing decision-support tool rather than a one-time planning artifact. The November 2024 validation event closely approached the 100-year design storm and its simulated inundation extent matched field observations, providing the technical credibility that underpinned both the emergency decisions during the February 2026 cold-front episode and the design basis for the long-term structural portfolio.

The dominance of pluvial ponding over fluvial overflow—with the latter restricted to return periods exceeding 25 years since Urrá I regulation commenced—carries a direct investment implication: expenditure focused exclusively on fluvial protection addresses only the residual tail of the risk distribution. The dominant return on investment lies in improving hydraulic connectivity between residential ponding zones and the macro-drainage channel network, a conclusion that redirects the conventional emphasis on riverbank works toward dispersed micro- and macro-drainage interventions across the urban fabric.

The counterintuitive result whereby isolated channel enlargement (M1) increases downstream water levels by up to 5 cm illustrates the peak-flow synchronization mechanism: widening one reach accelerates hydraulic transit and causes tributary peaks to arrive at the collector simultaneously with peaks from other sub-catchments that previously lagged in time. In flat, low-gradient systems where channel storage buffers timing differences, this synchronization can offset or reverse local conveyance gains. This finding—consistent with documented experience in low-lying coastal cities but rarely systematically reported in Caribbean Colombian contexts—underscores that structural works must be evaluated at the system scale and that the retention and peak-regulation components of M2 (dikes, land-raising, lagoon activation) are necessary precisely to counteract synchronization. The equity dimension reinforces this logic: neighborhoods combining high hydraulic exposure, poor infrastructure quality, and limited socioeconomic recovery capacity should receive preferential attention—a principle embedded in the risk-classification framework of Decree 1077/2015 [31].

7.2. System-Scale Evaluation of Structural Measures and Climate-Change Resilience

A critical question for long-term planning is whether the proposed structural and non-structural measures retain their effectiveness under the precipitation intensification projected for the Colombian Caribbean. Three representative Shared Socioeconomic Pathways are considered.

Under SSP 1–2.6 (low-emission, sustainability-oriented pathway), CMIP6 multi-model ensembles project moderate increases in extreme short-duration rainfall intensity of approximately 5–15% by mid-century for the Caribbean region [1]. Under this scenario, the 100-year design storm total would increase from the current ~95 mm to approximately 100–109 mm. The M2 structural portfolio, which was dimensioned for 100-year conditions and already incorporates a 60 cm freeboard on the minimum construction-elevation map, provides an adequate safety margin. The retention lagoons activated under M2 operate below their storage capacity, and the PMDUS micro-drainage improvements would further absorb the incremental runoff volume. Under this pathway, the framework requires only routine updating of IDF curves as new gauge data accumulate, without fundamental redesign.

Under SSP 2–4.5 (intermediate pathway), projected increases in extreme precipitation intensity reach 15–40% by late century [1]. The 100-year design storm would rise to approximately 109–133 mm, expanding high-hazard areas particularly in the INAT and Villa Cielo sub-catchments, where the extremely low gradients (0.0003 m/m) amplify the sensitivity of ponding depth to incremental rainfall volume. The M2 channel widening and dike components remain functional but with reduced freeboard margins; the retention lagoons approach their storage limits during sustained events. Under this scenario, the non-structural portfolio becomes critical: the water parks (“giant sponges”) with their 80 cm storage depth provide distributed peak attenuation that partially compensates for the increased rainfall loading, and the SAT system enables anticipatory evacuation that reduces exposure even when hazard intensity exceeds structural design capacity. The minimum construction-elevation map would require recalculation with updated IDF curves, increasing the recommended elevations by 10–25 cm in the most sensitive zones.

Under SSP 5–8.5 (high-emission pathway), late-21st-century precipitation intensification of up to 120% for the Caribbean region [1] would increase 100-year design-storm totals to approximately 155–165 mm. At this level, the M2 structural measures would be overtopped in several sub-catchments, and the 60-cm freeboard would be consumed in the most exposed neighborhoods. This scenario would require a second generation of structural interventions—deeper retention basins, additional pump capacity, and potentially controlled flooding of designated sacrifice zones—together with substantial revision of land-use restrictions in the POT.

The Urrá I dam provides an important but bounded buffering role. By attenuating peak Sinú discharges (the post-regulation 100-year flow of 1021 m³/s is substantially below pre-regulation estimates), the dam effectively removes fluvial overflow as a design-controlling mechanism for return periods below 25 years. However, this buffering is contingent on the dam’s operational rules remaining unchanged and on the structural integrity of its spillway and outlet works over the coming decades. Furthermore, the dam provides no attenuation of pluvial flooding, which is the dominant urban hazard. Climate-driven increases in sustained wet-season rainfall could simultaneously elevate both Sinú base levels (reducing the hydraulic gradient available for gravity-based urban drainage) and pluvial volumes, producing a compound scenario not explicitly represented in the current single-mechanism design storms. Future model updates should incorporate joint-probability analysis of coincident fluvial and pluvial forcing under each SSP pathway.

The framework’s iterative architecture—whereby hazard rasters, depth-damage curves, and risk classifications can be recalculated without rebuilding the model—is specifically designed for climate-adaptive updating. The February 2026 cold-front event adds a documented non-convective forcing scenario to the calibration database, strengthening the model’s capacity to represent the sustained multi-day rainfall patterns that climate projections indicate will become more frequent under all SSP pathways.

7.3. Limitations and Future Work

Several limitations should be explicitly acknowledged. First, the structural-measure analysis assumes that the micro-drainage conveys runoff perfectly to the macro-drainage—an idealization that justifies the separate treatment of ponding through the PMDUS but that must be progressively replaced as PMDUS components are implemented and monitored. Second, the Monte Carlo residual-risk model is a conceptual screening tool, not a validated predictive instrument. The five measure efficiencies and impact weights are expert-elicited values without empirical calibration, as post-intervention performance data do not yet exist for Montería. The aggregation formula (Equation (7)) assumes independent, linearly additive contributions, ignoring potential synergies (e.g., improved drainage enhancing early-warning lead times) and antagonisms (e.g., channel widening accelerating peak-flow synchronization that offsets downstream dike performance). The linear scaling of efficiencies with implementation level presupposes proportionality between investment and risk reduction, whereas threshold effects, institutional bottlenecks, and maintenance decay introduce strong non-linearities in practice. Consequently, the reported minimum residual risk of 19.9% should be interpreted as an indicative order of magnitude rather than a precise prediction, and validation requires post-implementation monitoring over multiple flood seasons to calibrate efficiencies, quantify inter-measure interactions, and replace assumed distributions with empirically derived ones. Third, the depth-damage functions derived in this study are calibrated for the cement-block construction typology that predominates across the Colombian Caribbean lowlands (s₀ = 1.50 m, ε = 1.85 for residential; s₀ = 1.20 m, ε = 2.00 for commercial), making them in principle transferable to intermediate cities with comparable building stock (e.g., Sincelejo, Cereté, Lorica). Nonetheless, direct application without local adjustment is not recommended, as differences in floor elevation, wall finishing, and content density can shift both parameters. Cities with significant proportions of timber or reinforced-concrete construction require independent curve fitting. Fourth, the fluvial-erosion analysis of the 17 critical Sinu points is observational rather than geotechnical; dedicated slope-stability and dike-breach studies are required to quantify the associated probability of occurrence. Fifth, cost estimates are pre-design figures based on hydraulic-capacity optimization alone and must be refined through feasibility studies that include property acquisition, social management, and environmental permits. Future work will address these limitations through the PMDUS detailed engineering, post-event validation of the February 2026 emergency works, and the integration of climate-downscaling products under IPCC SSP scenarios into the hazard rasters.

8. Conclusions

This study developed and applied an integrated flood-risk framework for Montería combining a Rain-on-Grid 2D HEC-RAS 6.6 model (4090 ha, 20-cm DTM resolution), locally calibrated depth-damage functions from 1465 household surveys, a structural and non-structural intervention portfolio, and a documented real-event emergency response. Four key findings emerge.

First, pluvial ponding—driven by poor hydraulic connectivity, limited evacuation capacity, and downstream channel overflow—governs urban flood damage, while Sinú River overflow is restricted to return periods exceeding 25 years under Urrá I regulation. This establishes that investment in dispersed micro- and macro-drainage connectivity yields higher risk reduction per unit cost than fluvial-focused protection.

Second, of 3015 scored urban assets, 94.13% fall in high or medium risk, reflecting Montería’s fundamentally flat and poorly drained morphology. The depth-damage functions calibrated for the predominant cement-block typology (s₀ = 1.50 m, ε = 1.85) constitute a transferable vulnerability model for similar Caribbean Colombian cities.

Third, the combined structural scenario M2 (channel widening, dikes, land-raising, retention lagoons) removes 80 ha from flooding while displacing 28 ha, but isolated channel enlargement (M1) produces counterproductive downstream aggravation through peak-flow synchronization—a finding that mandates system-scale evaluation of all structural interventions. The Monte Carlo residual-risk model indicates that full implementation of structural and non-structural measures reduces baseline risk to 19.9%, with the 20–60% implementation band as the most cost-effective range.

Fourth, the February 2026 cold-front emergency validates the operational model: field-documented water-surface elevations directly informed the design of the 4300 m perimeter dike and six pump-station sectors, demonstrating that a calibrated hydrodynamic model coupled with institutional coordination can deliver technically sound real-time response.

Five priority recommendations are addressed to competent authorities: (i) permanent updating of models incorporating real-time monitored data; (ii) prioritization of PMDUS execution at critical connectivity points; (iii) establishment of an Urban Drainage Management Unit with budgetary autonomy; (iv) urgent SAT implementation with IoT sensors integrated with the mobility system; and (v) development of an innovative financing program combining land-management instruments with multilateral and national resources.