Integrating Urban Planning and Hydraulic Engineering: Nature-Based Solutions for Flood Mitigation in Tainan City

Abstract

Extreme rainfall events driven by climate change are increasing flood risks. Addressing flood mitigation solely from either a hydraulic engineering or urban planning perspective may overlook both feasibility and effectiveness. This study focuses on Tainan City and the Tainan Science Park in Taiwan, applying the NbS framework to assess flood mitigation strategies. From an urban planning perspective, Agricultural Development Zone Type II (Agri-DZII), parks, green spaces, and Taiwan Sugar Corporation (TSC) land were selected as flood detention sites. Hydraulic modeling was used to evaluate their effectiveness under both current and climate-change-induced rainfall conditions. Simulation results show that under current rainfall conditions, flood mitigation measures reduced inundated areas with depths exceeding 2.0 m by up to 7.8% citywide and 20.8% within the Tainan Science Park Special District Plan Area. However, under climate change scenarios, the reduction effects declined significantly, with maximum reductions of only 1.6% and 17.8%, respectively. Results indicate that, even when utilizing all available detention areas, the overall flood reduction in Tainan City remains limited. However, TSC agri-land within the Tainan Science Park overlaps with high-flood-risk zones, demonstrating significant local flood mitigation potential. This study recommends integrating hydrological analysis into urban planning to prevent high-density residential and economic zones from being designated in flood-prone areas. Additionally, policymakers should consider reserving appropriate land for flood detention to enhance climate resilience. By combining urban planning and hydraulic engineering perspectives, this study highlights the flexibility of NbS in disaster management, advocating for the integration of Natural Water Detention Measures into flood adaptation strategies to improve urban water management and climate adaptability.

1. Introduction

Flood disasters significantly threaten human safety and property, especially in densely populated urban economic hubs. A comprehensive assessment of flood risks and the enhancement of metropolitan areas’ resilience to such disasters are critical challenges [1,2,3]. From a hydrological perspective, rainfall-runoff processes in metropolitan areas are inherently complex [4] and are highly influenced by land use patterns. As urbanization progresses, the expansion of impervious surfaces significantly increases surface runoff [5,6]. Consequently, effective surface runoff management and strategic land use planning are essential flood mitigation measures for metropolitan areas [7].

According to the Sixth Assessment Report (AR6) published by the Intergovernmental Panel on Climate Change (IPCC), both the frequency and intensity of extreme rainfall events with disaster-inducing potential have shown a significant increasing trend [8,9]. Traditional engineering design standards for flood protection may fail to provide adequate protection under extreme rainfall scenarios driven by climate change. Furthermore, extensive anthropogenic land use has substantially reduced the natural infiltration capacity of the land, potentially leading to increased surface runoff that cannot be effectively drained, thereby exacerbating flood disasters [10,11,12].

The development of metropolitan areas is closely linked to urban planning processes and regulatory frameworks. However, inadequate consideration of runoff management during urban planning may result in insufficient flood detention spaces or the placement of high-density residential or key industrial zones in flood-prone areas, thereby increasing disaster risks. Echendu and Georgeou (2021) [13] examined a coastal city in Nigeria and found that urban planners had limited understanding of surface runoff and flood mitigation, resulting in inadequate urban planning and heightened flood risks under extreme rainfall events. They emphasized the need for systematic coordination between urban planning and flood management regulations. Park et al. (2021) [14] conducted a study in South Korea to assess flood vulnerability in metropolitan areas. By overlaying flood risk zones with building distributions, they identified high-risk flood areas and proposed disaster-resistant building designs to enhance urban resilience. Martínez-Graña and Gago (2018) [15] utilized a hydrological model to analyze urban flooding near river channels under different rainfall scenarios. They spatially integrated the simulation results and provided urban planners with insights to avoid situating densely populated areas in flood-prone zones during the early stages of urban development. Overall, urban planning plays a crucial role in flood prevention and protection in metropolitan areas. Comprehensive hydrological analyses can help prevent high-risk flood zones from being designated as urban development areas. Furthermore, after urban planning is implemented, integrating both planning and hydrological perspectives allows for the selection of feasible and effective flood detention locations [16,17,18]. Traditional flood mitigation strategies in metropolitan areas have predominantly relied on engineered solutions such as levees, pumping stations, stormwater drainage systems, and large-scale detention basins. However, in recent years, the International Union for Conservation of Nature (IUCN) has emphasized the importance of nature-based solutions (NbS) as a key approach to addressing the challenges posed by climate change and flood disasters. According to the IUCN, NbS refer to “actions to protect, sustainably manage, and restore natural or modified ecosystems that address societal challenges effectively and adaptively, simultaneously providing human well-being and biodiversity benefits.” Integrating NbS into flood management represents a shift from conventional engineering approaches toward solutions that prioritize both sustainability and ecological benefits. Numerous studies have explored the practical implementation of NbS in flood risk management, demonstrating its potential as an effective and adaptive strategy [19,20,21,22]. Zannat et al. (2024) [23] reviewed various NbS interventions and found that their performance varies by spatial and climatic context, underscoring the need to tailor NbS strategies to local urban planning frameworks. Thaler et al. (2025) [24] emphasized that the implementation of NbS for flood risk management is highly influenced by governance systems and institutional settings. Their study highlighted the importance of stakeholder engagement and cross-sector coordination to ensure successful NbS adoption. In 2021, the European Union (EU) Climate Adaptation Strategy emphasized the necessity of adopting integrated water management approaches in urban areas to address the risks of extreme rainfall and flooding. A key component of this strategy is the promotion of the Sponge City concept, which enhances urban water resource management through the implementation of green infrastructure and Urban Retention Ponds [25]. In 2018, the Netherlands launched the “Room for the River” program, incorporating natural flood retention areas into urban planning to mitigate the impact of river overflow on cities. This initiative involved river channel widening, levee height reduction, and the construction of artificial flood retention areas, allowing rivers greater capacity to accommodate flood discharge and reducing the risk to surrounding urban areas [26]. Similarly, Austria has implemented flood detention infrastructure and compensation mechanisms to enhance flood resilience in both agricultural and urban regions. By incorporating private land into flood management areas, Austria has strengthened its flood retention capacity, ensuring greater adaptability to flood risks [27]. In 2014, the European Union (EU) introduced policies on Natural Water Retention Measures (NWRMs) [28], aiming to restore or maintain the natural characteristics and functions of ecosystems and water bodies through a natural approach and processes. These measures seek to protect water resources and address flood challenges by utilizing agricultural areas or forests as flood retention or storage spaces, replacing conventional engineering solutions. Natural Water Retention Measures offer multiple benefits, including reducing flood risks, improving water quality, enhancing biodiversity, recharging groundwater, and creating additional recreational spaces. As a result, NWRMs are widely recognized as an application of the NbS concept [29,30,31,32,33]. Beyond Europe, the Water Resources Agency, Ministry of Economic Affairs (WRA, MOEA), Taiwan has adopted this approach by repurposing idle agricultural land as flood retention areas to mitigate flooding risks in nearby urban areas. This strategy not only reduces reliance on large-scale flood detention infrastructure but also preserves the ecological functions of agricultural land [29,32,33,34,35,36,37]. In recent years, many metropolitan areas have adopted Low Impact Development (LID) as a runoff control strategy, implementing distributed and small-scale facilities designed to enhance infiltration, filtration, storage, evaporation, and runoff delay. By integrating these measures with urban land use planning and landscape design, LID aims to improve water quality and reduce stormwater runoff volumes, making it a concrete application of the NbS concept [26,36,37]. Khodadad et al. (2023) [38] examined green infrastructure (GI) for urban flood resilience, identifying key trends, methodological approaches, and typologies from recent literature. They also noted gaps in GI application in developing regions and challenges in integrating GI with biodiversity goals.

However, while Low-Impact Development (LID) facilities are effective in mitigating flooding under low-return-period rainfall events, their effectiveness is limited under extreme rainfall scenarios driven by climate change [9,27,28]. Therefore, it is essential to evaluate LID performance under various rainfall conditions before implementation. For urban planners, a lack of communication with hydraulic engineers may lead to systematic design flaws in urban planning. This could result in high-value industrial zones being located in flood-prone areas or excessive development of flood detention spaces, compromising runoff management in metropolitan areas. Similarly, for hydraulic engineers, relying solely on hydrodynamic modeling to designate flood detention areas without a clear understanding of urban planning regulations may lead to impractical flood detention site selections that cannot be effectively implemented.

Given these insights, this study focuses on Tainan City, located in southwestern Taiwan, and integrates both urban planning and hydraulic engineering perspectives to evaluate the feasibility and flood mitigation effectiveness of implementing Natural Water Detention Measures as a climate adaptation strategy for flood reduction within the study area.

2. Materials and Methods

2.1. Study Area

This study focuses on Tainan City, a metropolitan area in southwestern Taiwan. Covering approximately 2191 km², the city’s topography slopes from east to west, with its coastal regions characterized by alluvial plains and lagoons. Tainan City spans six major river basins, listed from north to south as Bajhang River, Jishui River, Jiangjun River, Zengwen River, Yanshuei River, and Erren River. The city receives an annual rainfall of approximately 1500 to 2500 mm, primarily influenced by the Mei-yu season (May–June) and typhoon events (July–September). Due to the concentration of rainfall within short durations, intense precipitation often leads to flooding.

As of February 2025, Tainan City had a total population of approximately 1.86 million, according to statistics from the Bureau of Civil Affairs, Tainan City Government [39]. Both the central and local governments are actively promoting the development of the technology and manufacturing sectors. In 2025, the Executive Yuan introduced the Southern Taiwan Silicon Valley Program, aiming to establish a comprehensive artificial intelligence (AI) industry chain in southwestern Taiwan by integrating the region’s science parks. Among these, Tainan Science Park Special District Plan Area is located at the junction of Shinshih District, Shanhua District, and Anding District in Tainan City, covering an area of approximately 3287.40 hectares.

Situated within this special district, the Tainan Science Park serves as a key hub for semiconductor and precision manufacturing industries, including wafer fabrication, optoelectronics, and biotechnology. According to statistics from the Department of Economic Development, Executive Yuan, the park’s economic output reached New Taiwan Dollars (NTD) 1.5855 trillion in 2023, the highest among all science parks in Taiwan. The park’s primary industry is integrated circuits, and Taiwan Semiconductor Manufacturing Company, Ltd. (TSMC) has a major facility in this area, underscoring its significance.

However, due to the flat terrain of Tainan Science Park Special District Plan Area and the limited flood discharge capacity of the surrounding drainage system, the area experienced flooding during Typhoon Morakot in 2009. Nevertheless, the potential increase in rainfall intensity due to future climate change poses a significant flood risk that cannot be ignored [33]. Currently, multiple flood detention facilities have been implemented within Tainan Science Park, which help mitigate flooding during typhoons and heavy rainfall events. However, whether the existing flood control infrastructure and drainage system are sufficient to address the adverse impacts of future climate change remains a key concern of this study. The study area, as shown in Figure 1, encompasses the entire Tainan City region, including Tainan Science Park Special District Plan Area.

2.2. Methodology

Urban surface runoff is influenced by the spatiotemporal distribution of rainfall and surface water flow dynamics. Therefore, the simulation process requires an analysis of the topography, geomorphology, and land cover within the study area. To achieve this, a non-structured computational grid is implemented, along with an appropriate flow equation to simulate surface water movement. Based on the quasi-two-dimensional flow concept, this study employs the Physiographic Drainage-Inundation Model (PHD model) to simulate flooding and drainage processes.

The PHD model incorporates digital elevation data, hydrological networks, and physiographic environment characteristics within the computational domain, ensuring a relatively uniform and appropriately generated non-structured grid for accurate simulations. The key advantages of using the PHD model for flood simulation are as follows:

• Efficient runoff computation with reduced computational time.
• Capability to output flow and hourly water level hydrographs for specific regions.
• Applicability in real-time flood forecasting and early warning systems.
• Enhanced adaptability to actual land use patterns, improving the model’s reliability.

The following sections provide an overview of the PHD model construction and the principles governing the design of the non-structured computational grid.

2.2.1. PHD Model Construction

The PHD model calculates surface water flow based on land use, topography, rainfall, and tidal conditions, enabling the simulation of surface inundation scenarios. In inundation simulations, a grid-based approach is employed to compute water flow, where adjacent grid cells are connected using flow continuity equations derived from quasi-two-dimensional flow theory, along with appropriate discharge formulas. This allows for the analysis of water levels within each grid cell and flow exchange between adjacent cells.

(1)

Continuity Equation of Flow

Flow exchange between adjacent grid cells without localized obstructions is modeled as overland flow, with discharge across the shared interface calculated using the Manning equation or other relevant hydraulic formulas. Interfaces bounded by physical features, such as roads, levees, field ridges, aquaculture embankments, natural riverbanks, or overflow weirs of reservoirs, are treated as broad-crested weirs, and corresponding discharge is computed using the weir flow equation. For boundaries containing pumping stations or culverts, flow is determined based on specified pumping capacity and structural dimensions. After establishing the computational grid, the flow continuity equation governing water exchange between any given grid cell $i$ and its adjacent grid cells can be expressed as follows:

$$A{s}_i{\displaystyle \frac{d{h}_i}{dt}}=P{e}_i+\sum _k{Q}_{i,k}\left(h_i,h_k\right)$$

(1)

where ${As}_i$ denotes the surface area of grid cell $i$ at time $t$; ${Pe}_i$ denotes the excess rainfall volume per unit time for grid cell $i$, which is calculated as the product of the excess rainfall intensity and the grid cell area; $Q_{i,k}$ is the flow rate from grid cell $k$ into grid cell $i$, where a positive value indicates inflow from grid cell $k$ to $i$, and a negative value indicates outflow from $i$ to $k$; $h_i$ is the water level of grid cell $i$ at time $t$; and $h_k$ is the water level of grid cell $k$ at time $t$. Infiltration within each computational grid cell is estimated using the Soil Conservation Service (SCS) curve number (CN) method [40]. Curve numbers are assigned based on land use types and hydrologic soil group classifications to reflect surface permeability conditions. The estimated infiltration volume is then subtracted from total rainfall to calculate excess rainfall (${Pe}_i$), which serves as the driving input for surface runoff generation in the PHD model. The SCS method is expressed as follows:

$$P_e={\displaystyle \frac{{\left(P-0.2S\right)}^2}{P+0.8S}}$$

(2)

$$S=2.54\left({\displaystyle \frac{1000}{CN}}-10\right)$$

(3)

where $S$ is the potential maximum soil retention and the unit is cm, $P_e$ is the excess rainfall, P is the total precipitation, and S is the maximum potential retention of the catchment. The Curve Number (CN) reflects factors such as land use type, antecedent moisture conditions, and soil group. In this study, appropriate CN values were primarily determined based on land use classifications.

The PHD model provides only a coarse approximation of the flow circulation in the urban zone at a large scale. Therefore, for site-specific decision making or detailed engineering design, a more refined two-dimensional hydrodynamic model should be used. The PHD model, based on the continuity equation, offers the advantage of simplifying computational procedures to enhance efficiency while reasonably capturing inundation patterns. It is well-suited for simulating large-scale flooding scenarios and evaluating flood mitigation strategies across extensive catchment areas. However, its limitations in spatial detail should be recognized when used for precise applications.

(2)

Numerical Method

The PHD model is developed based on quasi-two-dimensional flow equations using an explicit finite difference method. The flow continuity in Equation (1) is discretized using this approach, as follows:

$$\Delta h_i=[P{e}_i+\sum Q_{i,k}(h_i,h_k)]\Delta t/A{s}_i$$

(4)

where ${\Delta h}_i$ denotes the water level increment over the time step $\Delta t$, which, in the absence of bed erosion or deposition, is equivalent to the water depth increment. $Q_{i,k}$ is determined using an appropriate flow discharge formula depending on boundary conditions and flow regime. The water level (or depth) at time $\left(t+\Delta t\right)$ is obtained by adding the increment ${\Delta h}_i$ to the water level (or depth) at time $t$.

The PHD model aims to approximate real-world conditions while simplifying certain complex parameters under practical constraints. During computation, a key factor to consider is the surface overland flow roughness coefficient, which is closely related to land use characteristics. Therefore, this model is particularly suitable for rainfall-runoff simulations in areas where land use data is available.

2.2.2. Principles for Constructing the Non-Structured Computational Grid

Before conducting flood simulations using the PHD model, it is essential to establish a relatively uniform and appropriate computational grid based on the topography, geomorphology, and land cover distribution within the study area. The grid resolution should be determined according to the research objectives and spatial scale.

During runoff calculations, the model assumes homogeneous physiographic and hydrological conditions within each grid cell. If a grid cell contains diverse topographic or hydrological characteristics, it should be further subdivided. Additionally, areas with similar elevation and land use types should be grouped into the same grid cell whenever possible. For features such as levees, river channels, farmlands, and aquaculture ponds, their boundaries must be explicitly delineated in the grid design.

The principles for designing the unstructured computational grid are as follows:

(1)

Grid cells should share the same hydrometeorological conditions.

(2)

Roads, levees, or natural terrain features should be used as grid boundaries.

(3)

The area difference between adjacent grid cells should be minimized.

Following these principles, this study establishes a computational grid for the study area, which spans multiple river basins, as shown in Figure 2. The total computational area is 2446.62 km², with a total of 40,147 grid cells.

2.2.3. Validation of PHD Model

To evaluate whether the PHD model can accurately simulate the rainfall-runoff process, two major extreme rainfall events in the study area were selected as test cases: the extreme rainfall event on 23 August 2018 and that on 5 June 2021.

The simulation results were compared with observed water level records from water level stations. The model’s accuracy was validated using the Nash–Sutcliffe efficiency coefficient (NSE), calculated as shown in Equation (5).

$$NSE=1-{\displaystyle \frac{\sum _{t=1}^T{(S_o^t-S_s^t)}^2}{\sum _{t=1}^T{(S_o^t-\overline{S_o})}^2}}$$

(5)

where $S_o^t$ denotes observed water level at time $t,S_s^t$ denotes simulated water level at time $t,\mathrm{and} \overline{S_o}$ denotes mean observed water level.

The NSE values of the PHD model simulation results are summarized in

Table 1. NSE values for flood simulation results of the 0823 and 0605 storm events.

Water Level Station	NSE
	0823 Storm Event	0605 Storm Event
Peishihchou Bridge
Yufong Bridge
HuaYi Bridge

, and the location of the water level station is shown in Figure 3. The results show that:

For the extreme rainfall events on 23 August 2018, and 5 June 2021, the NSE values from the simulation results range from 0.70 to 0.93 and from 0.81 to 0.95, respectively. These results indicate that the PHD model can reasonably reproduce the observed rainfall-runoff processes, demonstrating its applicability for inundation simulations in the study area.

2.3. Site Selection for Flood Mitigation Measures and Estimation of Flood Detention Capacity

Building on the successful implementation of Natural Water Retention Measures (NWRMs) in the EU and the proven effectiveness of On-Site Flood Detention and green infrastructure initiatives led by water engineering and urban planning authorities in Taiwan [34]. Applying Nature-based Flood Detention Measures in Tainan City and Tainan Science Park Special District Plan Area could serve as a promising NbS strategy. Proper site selection for these measures would help mitigate flood risks exacerbated by climate change, enhancing the resilience of both the city and the industrial zone, ultimately supporting sustainable urban and industrial development.

Green infrastructure serves as a key strategy for integrating runoff allocation plans into spatial planning. For example, parks and green spaces can function as essential green infrastructure measures for flood mitigation within urban planning areas, while agricultural zones can be designated as natural flood detention areas to enhance flood mitigation. This study, based on a spatial planning perspective, examines the effectiveness of agricultural land and parks/green spaces in flood detention within different land use zoning categories. These findings provide a basis for government authorities to develop adaptive measures in response to the flood risks posed by climate change. The following sections detail the selection principles and results of the flood mitigation measure site selection process.

2.3.1. Identification of Suitable Land for Flood Mitigation Measures

According to the Spatial Planning Act of Taiwan, national land is demarcated into four major functional zones: Environmental Conservation Zones, Marine Resource Zones, Agricultural Development Zones, and Urban-Rural Development Zones. Based on Chapter 4, Article 20, Paragraph 1, Item 3, Sub-item 2 of the Spatial Planning Act of Taiwan, Agricultural Development Zone Type two (Agri-DZII) is defined as “areas with decent environments for agricultural production that can contribute to food security and support agricultural diversity.” From an urban planning perspective, publicly owned and undeveloped Agri-DZII land offers greater regulatory flexibility under the Spatial Planning Act of Taiwan, making it more adaptable for flood detention operations. In addition, parks and green spaces are often designated as LID (Low-Impact Development) sites, where green infrastructure is implemented to enhance flood mitigation. Therefore, this study identifies publicly owned and undeveloped Agri-DZII land, parks, and green spaces as potential flood detention areas.

From a basin perspective in Tainan City, an analysis of the undeveloped Agri-DZII land, parks, and green spaces within the study area reveals that the Yanshuei River basin contains the largest land area, totaling approximately 100.97 hectares. This basin also encompasses Tainan’s major urban development focus areas, resulting in a higher concentration of parks and green spaces. The Zengwen River basin has the second-largest area of parks and green spaces, totaling approximately 83.34 hectares, following the Yanshuei River basin. The Jishui River basin ranks third, with 26.04 hectares, while the Erren River basin is fourth, with 22.27 hectares. The Bajhang River basin ranks fifth, containing 20.85 hectares, and the Jiangjun River basin has the least available land, with just 11.28 hectares. With the exception of the Yanshuei River basin, Agri-DZII land constitutes the largest portion of undeveloped land in all other basins.

In addition to considerations under the Spatial Planning Act of Taiwan, land owned by state-owned enterprises under the Ministry of Economic Affairs, Executive Yuan offers greater flexibility for implementing flood mitigation measures. Therefore, this study also includes an analysis of such lands. The Ministry of Economic Affairs oversees ten state-owned enterprises, and based on the location of the study area and the characteristics of these enterprises, this study selects land owned by the Taiwan Sugar Corporation (TSC) for flood detention operations.

Given its role in sugarcane cultivation and processing, TSC holds vast agricultural land and undeveloped areas, making it a viable option for flood mitigation site selection. The Water Resources Agency, Ministry of Economic Affairs [37] previously designated Youcai Village in Yunlin County, Taiwan, as a demonstration site for natural flood detention, utilizing TSC-owned agricultural land (TSC agri-land) to enhance flood mitigation in metropolitan areas. Given its effectiveness, this study identifies and assesses TSC agri-land in Tainan City as potential sites for flood mitigation measures.

The spatial distribution of land owned by the Taiwan Sugar Corporation (TSC) in Tainan City varies substantially across its river basins. The Yanshuei River basin contains the largest TSC-owned agricultural land area, covering approximately 383.78 hectares, followed by the Erren River basin, with 274.75 hectares. The Jishui River basin ranks third, with 72.69 hectares, while the Jiangjun River basin ranks fourth, with 60.98 hectares. The Zengwen River basin ranks fifth, with only 25.04 hectares. In contrast, the Bajhang River basin does not contain any undeveloped TSC agri-land in Tainan City that could be considered for flood mitigation measures.

Based on the land use zoning framework, the compilation of Agri-DZII land, green spaces, parks, and TSC agri-land provides an overview of the spatial distribution of potential flood detention areas within the study area, as illustrated in Figure 3. The availability of flood detention areas across different river basins is summarized in

Table 2. Area distribution of AGRI-DZII, green space, park, and TSC agri-land across river basins in the study area.

River Basin		AGRI-DZII, Green Space, and Park		TSC Agri-Land
Name	Area (ha)	Area (ha)	Propotion (%)	Area (ha)	Propotion (%)
Erren River Basin	35,000	22.27	0.06	274.75	0.79
Bajhang River Basin	47,474	20.85	0.04	0.00	0.00
Jishui River Basin	37,900	26.04	0.07	72.69	0.19
Jiangjun River Basin	15,840	11.28	0.07	60.98	0.38
Zengwun River Basin	117,664	83.34	0.07	25.04	0.02
Yanshuei River Basin	34,317	100.97	0.29	383.78	1.12
Total	288,195	264.74	0.61	817.24	2.50

, which indicates that TSC agri-land constitutes a significant portion of the available areas. Tainan Science Park Special District Plan Area, located within the Yanshuei River basin, has potential flood detention areas mapped in Figure 4. Compared to other urban planning districts, the TSC agri-land in this area is relatively more extensive, making it a viable option for flood mitigation implementation.

2.3.2. Estimation Approach for Flood Detention Capacity

To estimate the flood detention capacity of the selected flood detention sites identified in the previous section, this study applies an approach based on the Urban Planning Act, defining building coverage ratios and statutory vacant space ratios for different land use zoning categories. Additionally, the Runoff Allocation Technical Manual is referenced to apply the generalized formula for calculating potential runoff allocation capacity, expressed as [37]:

Q (m³) = Usable Area (m²) × Permissible Detention Depth (m)

(6)

A certain proportion of the statutory vacant space ratio is deducted as a baseline criterion, while the allowable retention depth is determined for parkland, green spaces, and agricultural zones. The specific estimation methods for potential runoff allocation capacity in these land use categories are summarized in

Table 3. Methods for estimating runoff allocation potential in park, green space, and agricultural land.

Land Use Districts	Urban Planning Act		Runoff Diversion Volume Assessment
	Building Coverage Ratio (%)	Statutory Vacant Space Ratio (%)	Detention Area Ratio (%)	Permissible Detention Depth ($\mathbf{m}$)	Land Detention Capacity Calculation (${\mathbf{m}}^3$)
Park (<5 ha)	15	85	65	0.2	$Q=A\times 0.65\times 0.2$ $Q=0.13A$
Park (>5 ha)	12	88	68	0.5	$Q=A\times 0.68\times 0.5$ $Q=0.34A$
Green space	0	100	25	0.3	$Q=A\times 0.25\times 0.3$ $Q=0.075A$
Agricultural area	-	-	100	0.5	$Q=A\times 0.5=0.5A$

.

2.4. Rainfall Scenarios

Within the computational domain of Tainan City, a total of 19 rainfall stations are utilized, including 10 stations operated by the Central Weather Administration (CWA) and 9 stations managed by the Water Resources Agency (WRA). Based on hydrological frequency analysis for each station, this study constructs 24 h rainfall events corresponding to a 10-year return period and estimates the current rainfall scenario accordingly.

For climate change scenario analysis, this study utilizes downscaled daily statistical rainfall data produced by the National Science and Technology Center for Disaster Reduction (NCDR) based on the Intergovernmental Panel on Climate Change (IPCC) Sixth Assessment Report (AR6). The dataset is derived from Atmospheric General Circulation Models (AGCMs) under the Coupled Model Intercomparison Project Phase 6 (CMIP6) framework. The data is spatially resolved using the WGS84 coordinate system at a 0.05° × 0.05° (approximately 5 km) grid, with a historical baseline period from 1960 to 2014 and future projections from 2015 to 2100. The dataset includes simulation results for five Shared Socioeconomic Pathways (SSP) scenarios, which define different greenhouse gas concentration trajectories: SSP1-1.9, SSP1-2.6, SSP2-4.5, SSP3-7.0, and SSP5-8.5. This study references the “Analysis and Application of Hydrological Scenarios” report published in 2023 by the Water Resources Planning Branch, Water Resources Agency, Ministry of Economic Affairs (WRA) [39]. The report employs AR6 downscaled daily statistical rainfall data and primarily adopts Shared Socioeconomic Pathways (SSP) scenarios along with Global Warming Levels (GWLs) to establish climate projections. The SSP scenarios represent potential global socioeconomic development pathways, with warming levels expressed in terms of radiative forcing at the end of the century. Specifically, five SSP scenarios are considered: SSP1-1.9, SSP1-2.6, SSP2-4.5, SSP3-7.0, and SSP5-8.5. Additionally, the GWL approach simplifies complex climate scenarios into different global mean temperature increase thresholds, specifically 1.5 °C, 2.0 °C, 3.0 °C, and 4.0 °C. Each scenario includes 20-year datasets of daily rainfall and temperature projections. For this study, the 2.0 °C GWL scenario is selected as the climate change scenario for analysis.

The rainfall stations within the study area and the downscaled grid cells from AR6 are shown in Figure 5. In this study, the 19 rainfall stations were matched to their corresponding AR6 statistical downscaled daily rainfall grid cells, followed by an analysis of daily rainfall increment percentages under different climate change models. To mitigate potential biases from extreme values, the highest and lowest rainfall increment percentages were first excluded. The average of the remaining increment percentages was then calculated and used as the rainfall increment percentage for each station under climate change scenarios. Using the derived rainfall increment percentages, the projected climate change-adjusted rainfall for each station was estimated based on its 10-year return period rainfall. The results are presented in

Table 4. Accumulated 24 h rainfall with a 10-year return period at rainfall stations under current and climate change scenarios.

Competent Authority	Rainfall Station	Increment (%)	Scenarios
			Current (Case0-P, mm)	Climate Change (Case0-C, mm)
Water Resources Agency	Guanziling (2)	37.4	658.74	904.86
	Liuxi	41.5	557.38	788.78
	Dongyuan	38.5	554.81	768.22
	Beiliao	30.7	679.32	887.77
	Nanhua (2)	44.0	562.89	810.75
	Qiding	29.0	435.59	562.10
	Hutoupi	30.4	447.43	583.27
	Wangye Temple	44.4	501.29	723.99
Central Weather Administration	Guanshan	36.3	796.32	1085.17
	Tainan	27.1	381.05	484.48
	Yongkang	22.4	382.01	467.42
	Zengwen	36.4	722.20	985.33
	Shanhua	22.9	403.16	495.51
	Yujing	40.6	551.71	775.48
	Chiali	35.3	365.38	494.37
	Huanhu	37.5	527.71	725.82
	Dadongshan	30.7	547.57	715.60
	Guanshan	36.3	574.40	782.74
	Dongyuan	41.5	541.65	766.52

, showing an increase in rainfall at all stations, with the largest increase reaching 44.0%, highlighting the significant impact of climate change on metropolitan areas in Taiwan.

3. Flood Simulation Results and Findings

3.1. Simulation Results and Impact of Climate Change on Inundation Depth and Extent

The maximum potential inundation depth simulation results for Case0-P are shown in Figure 6, while those for Case0-C are presented in Figure 7. A comparison of the inundated areas corresponding to different inundation depths in Tainan City is provided in

Table 5. Inundation area of Tainan City under current (Case0-P) and climate change (Case0-C) scenarios.

Inundation Depth (m)	Inundation Area (ha)		Change in Area (ha)	Change (%)
	Case0-P	Case0-C
>0.1	78,703.6	95,314.1	16,610.5	21
>0.3	47,705.9	59,046.9	11,341.0	24
>0.5	35,190.7	44,514.8	9324.1	26
>1.0	17,735.2	24,803.9	7068.6	40
>2.0	3496.6	6148.4	2651.8	76

, whereas

Table 6. Inundation area of Tainan Science Park Special District Plan Area under current (Case0-P) and climate change (Case0-C) scenarios.

Inundation Depth (m)	Inundation Area (ha)		Change in Area (ha)	Change (%)
	Case0-P	Case0-C
>0.1	2021.1	2320.4	299.3	13
>0.3	1297.6	1463.2	165.6	11
>0.5	960.1	1115.0	154.9	14
>1.0	530.1	665.0	134.9	20
>2.0	43.0	104.6	61.6	59

summarizes the same analysis for Tainan Science Park Special District Plan Area. Additionally, Figure 8 illustrates the differences in inundation depth and extent between Case0-P and Case0-C.

The simulation results indicate that climate change will substantially increase rainfall intensity, leading to a notable rise in both inundation depth and affected area across Tainan City and Tainan Science Park Special District Plan Area. In Tainan City, areas with inundation depths exceeding 0.3 m increase from 47,705.9 hectares (Case0-P) to 59,046.9 hectares (Case0-C), representing a 24% increase in affected area (Table 5). Similarly, in Tainan Science Park Special District Plan Area, the inundated area for depths exceeding 0.3 m increases from 1297.6 hectares to 1463.2 hectares, reflecting an 11% increase (Table 6). These findings suggest that shallow to moderate flooding will become more widespread under climate change conditions, exacerbating urban drainage challenges and increasing flood risks in both urban and industrial areas. For severe flooding (inundation depths exceeding 2.0 m), the affected area in Tainan City increases from 3496.6 hectares to 6148.4 hectares, representing a 76% increase (Table 5). In Tainan Science Park Special District Plan Area, the inundated area for depths exceeding 2.0 m increases from 43.0 hectares to 104.6 hectares, marking a 59% increase (Table 6). Although the relative percentage increase in Tainan Science Park Special District Plan Area is slightly lower than in Tainan City as a whole, the increase remains substantial, highlighting the vulnerability of the region to future extreme rainfall events.

3.2. Flood Mitigation Implementation Results in Tainan City

The flood mitigation simulation results for Case0-P (current rainfall scenario) are presented in

Table 7. Flood mitigation results in Tainan City under Case0-P (current rainfall scenario).

Inundation Depth (m)	Case0-P	Case 1		Case 2		Case 3
	Inundation Area (ha)	Inundation Area (ha)	Change (%)	Inundation Area (ha)	Change (%)	Inundation Area(ha)	Change (%)
>0.3	47,705.9	46,682.3	2.1	46,849.1	1.8	46,779.3	1.9
>0.5	35,190.7	34,241.6	2.7	34,385.1	2.3	34,282.2	2.6
>1.0	17,735.2	16,848.0	5.0	17,054.0	3.8	16,867.4	4.9
>2.0	3496.6	3224.0	7.8	3253.0	7.0	3224.8	7.8

. The results indicate that Case 1 provides the most effective flood mitigation, followed by Case 3, while Case 2 shows the least reduction in inundation extent.

(1)

Impacts of Inundation Depths Exceeding 0.3 m

In Tainan City, under Case0-P, areas with inundation depths exceeding 0.3 m decrease from 47,705.9 hectares to 46,682.3 hectares in Case 1, representing a 2.1% decrease. The reductions in Case 2 and Case 3 are 1.8% and 1.9%, respectively. This suggests that flood detention measures provide limited mitigation for moderate flooding but still contribute to overall flood reduction. However, under Case0-C (climate change rainfall scenario), the impact of flood mitigation measures is substantially smaller. The inundated area for depths exceeding 0.3 m decreases by only 0.3% in Case 1 and Case 3, while Case 2 shows almost no decrease (0.0%). These findings indicate that climate change exacerbates overall flood severity and weakens the effectiveness of existing flood mitigation measures, particularly for moderate flooding depths.

(2)

Severe Flooding with Inundation Depths Exceeding 2.0 m

For severe flooding (inundation depths exceeding 2.0 m), flood mitigation effects become more pronounced under Case0-P. The inundated area decreases from 3496.6 hectares to 3224.0 hectares in Case 1, representing a 7.8% decrease. Similar decreases are observed in Case 3 (7.8%), while Case 2 achieves a 7.0% decrease. These results suggest that larger flood detention spaces are more effective in mitigating extreme flooding events. In contrast, under Case0-C, the reduction in severely inundated areas is substantially lower. The flooded area for depths exceeding 2.0 m decreases by only 1.6% in Case 1, 1.4% in Case 3, and 0.0% in Case 2. This indicates that flood mitigation measures have limited effectiveness in addressing extreme flooding under intensified rainfall conditions, emphasizing the challenges posed by increasing precipitation levels due to climate change.

The simulation results indicate that under current rainfall conditions (Case0-P), flood mitigation measures remain relatively effective, particularly for areas experiencing severe inundation (>2.0 m). Among the three scenarios, Case 1 (utilizing Agri-DZII land, green spaces, parks, and TSC agri-land) demonstrates the most significant decrease in inundation extent, followed by Case 3 (utilizing only TSC agri-land). Case 2 (utilizing only Agri-DZII land, green spaces, and parks) shows the least effectiveness. The greatest flood mitigation benefits occur in areas where inundation depths exceed 2.0 m, suggesting that designated flood detention spaces contribute more effectively to extreme flood mitigation. Under climate change conditions (Case0-C), the effectiveness of flood mitigation measures is substantially reduced. Moderate flooding (depths >0.3 m) shows a minimal decrease, with Case 2 providing almost no improvement. Even for severe flooding (depths >2.0 m), the decrease percentages are substantially lower compared to Case0-P, reinforcing the challenges of managing extreme rainfall events in the future.

3.3. Flood Mitigation Implementation Results in Tainan Science Park Special District Plan Area

The flood mitigation simulation results for Case0-P (current rainfall scenario) are presented in

Table 8. Flood mitigation results in Tainan Science Park Special District Plan Area under Case0-P (current rainfall scenario).

Inundation Depth (m)	Case0-P	Case 1		Case 2		Case 3
	Inundation Area (ha)	Inundation Area (ha)	Change (%)	Inundation Area (ha)	Change (%)	Inundation Area(ha)	Change (%)
>0.1	2021.1	1984.2	1.8	2006.8	0.7	2005.1	0.8
>0.3	1297.6	1259.3	2.9	1306.5	−0.7	1283.6	1.1
>0.5	960.1	955.5	0.5	961.8	−0.2	955.5	0.5
>1.0	530.1	502.4	5.2	523.7	1.2	509.9	3.8
>2.0	43.0	34.0	20.8	43.0	0.0	34.0	20.8

Note(s): Positive values in the ‘Change (%)’ column represents a decrease in inundation area, while negative values (−) indicate an increase in inundation area.

, while results for Case0-C (climate change rainfall scenario) are shown in

Table 9. Flood mitigation results in Tainan Science Park Special District Plan Area under Case0-C (climate change rainfall scenario).

Inundation Depth (m)	Case0-C	Case 1		Case 2		Case 3
	Inundation Area (ha)	Inundation Area (ha)	Change (%)	Inundation Area (ha)	Change (%)	Inundation Area(ha)	Change (%)
>0.1	2320.4	2303.4	0.7	2320.4	0.0	2316.9	0.2
>0.3	1463.2	1448.1	1.0	1463.2	0.0	1461.1	0.1
>0.5	1115.0	1099.7	1.4	1115.0	0.0	1109.1	0.5
>1.0	665.0	634.9	4.5	657.7	1.1	636.8	4.2
>2.0	104.6	86.0	17.8	104.6	0.0	100.3	4.2

.

(1)

Impacts of Inundation Depths Exceeding 0.3 m

Under Case0-P, areas with inundation depths exceeding 0.3 m slightly decrease in Case 1, with inundated areas reducing from 1297.6 hectares to 1259.3 hectares, representing a 2.9% decrease. However, the decrease is much less significant in Case 3 (1.1%), and in Case 2, the inundated area slightly increases by 0.7%, indicating that in some cases, the flood mitigation measures may have limited or negligible effectiveness in addressing moderate flooding. Under Case0-C, the impact of flood mitigation measures is even more limited. The inundated area for depths exceeding 0.3 m in Case 1 decreases by only 1.0%, while Case 3 sees only a 0.1% decrease, and Case 2 shows no decrease at all (0.0%). This suggests that under climate change conditions, existing flood mitigation measures provide minimal relief for moderate flooding, making urban areas more vulnerable to increased rainfall.

(2)

Severe Flooding with Inundation Depths Exceeding 2.0 m

For severe flooding (inundation depths exceeding 2.0 m), flood mitigation measures exhibit more noticeable effects under Case0-P. The inundated area decreases from 43.0 hectares to 34.0 hectares in both Case 1 and Case 3, representing a 20.8% reduction. However, Case 2 provides no decrease (0.0%), indicating that TSC agri-land plays a crucial role in mitigating extreme flooding. Under Case0-C, the flood mitigation effects on severe flooding are substantially reduced. The inundated area for depths exceeding 2.0 m in Case 1 decreases by 17.8%, while Case 3 shows a 4.2% decrease, and Case 2 remains at 0.0%, meaning that flood mitigation in Case 2 is entirely ineffective in addressing severe inundation under climate change conditions.

The simulation results indicate that under current rainfall conditions (Case0-P), flood mitigation measures in Tainan Science Park Special District Plan Area are somewhat effective, particularly for severe inundation (>2.0 m). Case 1 (utilizing Agri-DZII land, green spaces, parks, and TSC agri-land) provides the most significant reduction in inundation extent, followed by Case 3 (utilizing only TSC agri-land). However, Case 2 (utilizing only Agri-DZII land, green spaces, and parks) shows little to no impact, especially for extreme flooding events. Under climate change conditions (Case0-C), the effectiveness of flood mitigation measures is substantially reduced. Moderate flooding (depths > 0.3 m) shows only minimal reduction, and Case 2 remains ineffective across all depth categories. Even for severe flooding (depths > 2.0 m), the decrease percentages are substantially lower compared to Case0-P, reinforcing the challenges of managing extreme rainfall events in the future.

A comparison between Case0-P and Case0-C across all mitigation scenarios in both Tainan City and Tainan Science Park Special District Plan Area reveals that none of the measures are sufficient to reduce the inundation area to the levels observed under current rainfall conditions. The differences in inundation extent between the two scenarios remain substantial, highlighting the urgent need for adaptive and enhanced flood mitigation strategies to address the increasing risks posed by climate-change-driven extreme precipitation in both urban and industrial areas.

4. Discussion

4.1. Importance of Integrated Urban Planning and Hydraulic Engineering in Flood Risk Management

Flood disasters threaten human safety and property. Urban planners and hydraulic engineers, due to differing expertise, may overlook surface runoff behavior, leading to the placement of high-density areas or industrial parks in flood-prone zones. Misidentification of flood detention sites further hinders mitigation efforts. Climate-change-driven extreme rainfall exacerbates flood risks, making it crucial to assess flood conditions under both current and future scenarios while evaluating the effectiveness and feasibility of various flood detention strategies.

4.2. Study Scope and Rainfall Scenario Selection

For the current scenario, rainfall data from 19 stations were used to generate a 10-year return period, 24 h rainfall event. The AR6-projected +2.0 °C Global Warming Level (GWL) scenario was selected for the climate change scenario, analyzing rainfall increases at each station. Results show significant rainfall increases, with a maximum rise of 44%, used to estimate future rainfall based on station-specific trends.

4.3. NbS-Based Flood Mitigation Strategies and Land Use Considerations

This study applies an urban planning perspective and the Nature-based Solutions (NbS) framework to identify flood detention areas for Low-Impact Development (LID). Agri-DZII, parks, and green spaces, as public lands, offer flexibility for localized flood detention or green infrastructure. Additionally, Tainan City’s TSC agri-land, managed by the state-owned Taiwan Sugar Corporation (TSC), allows for adaptable land use. Given its feasibility, this study prioritizes Agri-DZII and TSC agri-land as key flood detention sites. Previous studies have highlighted the effectiveness of Nature-based Solutions (NbS) in mitigating urban flood risks. For instance, Zannat et al. (2024) [23] reviewed a range of NbS interventions and emphasized their adaptability under various climatic and spatial conditions, particularly at localized scales [20]. Similarly, Rosmadi et al. (2024) demonstrated the potential of NbS strategies in enhancing urban flood resilience in Southeast Asia [22]. Compared to these studies, the present research extends the spatial scale of analysis to a metropolitan context—Tainan City, with an area of 2446 km²—and offers empirical evidence showing that TSC agri-land located in flood-prone zones provides substantial flood detention capacity. Additionally, while Khodadad et al. (2023) [38] focused on green infrastructure typologies and their ecological benefits [30], this study incorporates land use zoning classifications (e.g., Agri-DZII) and applies the runoff detention estimation parameters outlined in the Runoff Allocation Technical Manual [34]. By quantifying flood detention potential for Agri-DZII land, green spaces, and TSC agri-land, the results bridge the gap between spatial planning, regulatory interpretation, and hydrological modeling. This integrated approach contributes a context-specific perspective to the existing literature and provides planners and engineers with operationalizable strategies for flood mitigation under climate change.

4.4. Citywide Flood Mitigation Effectiveness

Citywide simulations show that even with all available flood detention spaces, flood reduction remains limited. Table 8 and

Table 10. Flood mitigation results in Tainan City under Case0-C (climate change rainfall scenario).

Inundation Depth (m)	Case0-C	Case 1		Case 2		Case 3
	Inundation Area (ha)	Inundation Area (ha)	Change (%)	Inundation Area (ha)	Change (%)	Inundation Area(ha)	Change (%)
>0.3	59,046.9	58,842.7	0.3	59,031.9	0.0	59,980.9	0.3
>0.5	44,514.8	44,397.7	0.3	44,511.5	0.0	44,405.5	0.2
>1.0	24,803.9	24,551.4	1.0	24,767.0	0.1	24,648.1	0.6
>2.0	6148.4	24,551.4	1.6	6148.0	0.0	6059.6	1.4

reveal that under climate change rainfall, the inundation area expands substantially, reducing mitigation effectiveness. Rainfall at all stations increases over 20%, causing greater surface runoff. Thus, green infrastructure and localized flood detention alone are insufficient for effective flood mitigation.

4.5. Flood Mitigation Effectiveness in Tainan Science Park Special District Plan Area

For Tainan Science Park Special District Plan Area, Table 9 and Table 10 show greater flood mitigation effectiveness than citywide results. Under Case 1, areas with flood depths exceeding 2.0 m are reduced by 20.8% under current rainfall and 17.8% under climate change.

From a conservation of mass perspective, flood detention capacity directly impacts mitigation effectiveness. However, detention sites should prioritize flood-prone areas for maximum impact. Since Agri-DZII, parks, green spaces, and TSC agri-land make up a small portion of Tainan City and are not all in high-flood-risk zones, Table 7 and Table 8 suggest that solely using these areas has limited impact.

Figure 4 indicates that the Science Park’s southwestern and eastern sections contain more suitable land, particularly TSC agri-land. Figure 6 and Figure 7 further show greater flood depths in the southwestern portion, making it an ideal flood detention site to improve mitigation efficiency.

4.6. Implications for Future Flood Mitigation Strategies

The findings of this study highlight the importance of communication and collaboration between urban planning and hydraulic engineering disciplines. Analysis shows that the southwestern section of Tainan Science Park Special District Plan Area experiences greater flood depths, requiring flood detention measures to enhance mitigation efficiency, especially under climate-change-induced rainfall.

Although LID measures have limitations in extreme rainfall, expanding LID in high-flood-risk areas and integrating diverse flood detention strategies could improve disaster resilience. Utilizing agricultural land and green spaces for localized flood detention and green infrastructure, instead of relying solely on traditional levees and detention basins, offers a sustainable approach to reducing flood extent and depth while preserving natural landforms.

In terms of practical implementation, the proposed flood detention strategy primarily utilizes publicly owned lands, including Agri-DZII zones, green spaces, and parks. These land types are not privately held and thus offer greater flexibility for policy-driven adaptation measures. Government authorities can designate and repurpose these spaces under the objective of flood reduction and climate adaptation without the need for complex land acquisition procedures. This institutional accessibility reinforces the feasibility of implementing NbS-aligned flood mitigation strategies, as developed in this study.

In addition, TSC agri-land is owned by a state-owned enterprise, the Taiwan Sugar Corporation, and is therefore also considered public land under administrative oversight. This status allows for a more adaptable use of the land for flood detention purposes. Taiwan has previous experience in utilizing agricultural land for temporary flood retention, which has proven effective in mitigating flood risks in urban and peri-urban areas. These precedents support the practical applicability of the proposed spatial planning approach and reinforce the role of land governance in enabling NbS in real-world urban flood management.

5. Conclusions

Given the varying geomorphic and hydrological characteristics across regions, the feasibility and effectiveness of climate change flood adaptation strategies under the NbS framework must be carefully evaluated. Feasibility relies on urban planning expertise in land use, while effectiveness depends on hydrological and hydraulic analysis in water engineering. However, interdisciplinary collaboration often faces knowledge gaps or barriers, leading to inefficiencies or conflicts.

Hydraulic simulations show that flood extent and depth will substantially increase across Tainan City under climate change, with Tainan Science Park Special District Plan Area also at risk. The simulation results further confirm that implementing flood detention measures—particularly those utilizing Agri-DZII, green spaces, parks, and TSC agri-land—can effectively reduce flood impacts. Under current rainfall conditions, areas with inundation depths exceeding 2.0 m were reduced by 7.8% citywide and up to 20.8% within the Tainan Science Park Special District Plan Area. However, under climate change-induced rainfall scenarios, the mitigation effects decreased, with corresponding reductions of only 1.6% and 17.8%, respectively. These results underscore the importance of expanding flood detention capacity and spatially targeting high-risk zones to maintain adaptation effectiveness under intensifying climate conditions. As TSC agri-land is located in flood-prone areas, implementing LID measures offers both high flood mitigation efficiency and feasibility from an urban planning perspective. This study highlights the necessity of integrating urban planning and water engineering.

Although LID alone cannot eliminate flooding under extreme climate conditions, targeting high-risk zones and incorporating diverse flood detention strategies could enhance regional disaster resilience. By utilizing agricultural land and green spaces for flood detention and green infrastructure, rather than traditional engineering solutions, this study presents a sustainable flood mitigation approach, providing valuable insights for policymakers.

Declaration of generative AI and AI-assisted technologies in the writing process: During the preparation of this work, the author(s) used ChatGPT-4-turbo in order to improve language and readability. After using this tool, the author(s) reviewed and edited the content as needed and take(s) full responsibility for the content of the publication.