An Integrated Approach to Assessing the Impacts of Urbanization on Urban Flood Hazards in Hanoi, Vietnam

Abstract

Urban flooding is a major challenge to sustainable development in rapidly urbanizing cities. This study applies an integrated approach that combines Sentinel-1 SAR data, geomorphological analysis, and the DPSIR (Drivers–Pressures–State–Impacts–Responses) framework to assess the relationship between urbanization and flooding in Hanoi during the 2010–2024 period (with Sentinel-1 time-series data for 2015–2024). A time series of Sentinel-1 images (2015–2024) was processed on Google Earth Engine to detect inundation and construct a flood frequency map, which was validated against 148 field survey points (overall accuracy = 87%, Kappa = 0.79). The results show that approximately 80% of newly urbanized areas are situated on geomorphologically sensitive units, including inside- and outside-dike floodplains, fluvio-marine plains, paleochannels, and karst terrains, characterized by low elevation and high flood susceptibility. Meanwhile, about 73% of the total inundated area occurs within newly developed urban zones, primarily in western and southwestern Hanoi, where rapid expansion on flood-prone terrain has intensified hazards. The DPSIR analysis highlights rapid population growth, land use change, and inadequate drainage infrastructure as the main pressures driving both the frequency and extent of flooding. To our knowledge, this is the first study integrating geomorphology, Sentinel-1, and DPSIR for Hanoi, thereby providing robust evidence to support sustainable urban planning and climate-resilient development.

1. Introduction

Urban flooding has long been recognized as one of the most severe environmental and social hazards, causing major loss of life and property, disrupting economic activities, and degrading community well-being [1,2]. According to statistics from the United Nations and various international organizations, floods and inundation accounted for approximately 44% of all natural disasters from 1970 to 2019, causing over 31% of total global economic losses [3]. Between 2010 and 2019, 1298 major floods were recorded worldwide, defined by the Dartmouth Flood Observatory as events with documented human or economic impacts [4]. In tropical regions, the frequency of major floods quadrupled between 1985 and 2015, while global flood risk increased by 20–24% from 2000 to 2018 [5,6,7]. These trends indicate a rapid increase in flood risk driven by the combined impacts of climate change and urbanization, underscoring the urgent need for sustainable disaster risk management [8].

The reasons for urban flooding are not only linked to extreme rainfall but also complex anthropogenic factors, such as unplanned growth [9]; land use and surface cover change [10]; degradation of the urban ecosystem, which reduces infiltration and stormwater retention [11]; outdated or inconsistent drainage networks [10,12,13,14]; and ineffective management of green infrastructure [15]. In the context of unpredictable climate change and rapid urbanization, population growth and the expansion of urban areas have become key drivers exacerbating flood hazards [13,16]. Therefore, urban flooding should be understood as the cumulative outcome of interactions among climate, geomorphology, and uncontrolled human-induced factors [17,18,19].

Southeast Asia, characterized by low-lying terrain and high population density, is particularly vulnerable to flooding. Severe floods in recent decades in Hanoi (2008), Bangkok (2011), Beijing (2012), and Ho Chi Minh City (2018) illustrate the escalating flood risk across rapidly urbanizing metropolises [1,20,21,22,23].

In Vietnam, Hanoi is a typical example of urban flooding. Its low-lying terrain and gentle slope hinder natural drainage, making many areas prone to inundation. Older streets are less affected by flooding, while many new urban areas frequently experience severe flooding, highlighting the imbalance between urban spatial expansion and the capacity of drainage infrastructure [24,25,26,27]. Since 2012, repeated heavy rainfall events have caused severe floods in rapidly developing districts, resulting in substantial socioeconomic damage and directly impacting people’s lives [27]. Recent studies have proposed various solutions to control floods, ranging from technical measures to risk-based prevention planning [27,28].

Numerous methods have been developed to analyze and predict flooding. Hydrological-hydraulic models, such as HEC-RAS, LISFLOOD-FP, MIKE NAM, MIKE 11, and Flo-2D, are widely used in flow simulation and flood mapping [29,30,31,32]. Raster models [33], GIS-based flood modeling [34], and uncertainty analysis in flooding mapping [35] have further improved simulation accuracy and provided more reliable information for management. However, these models require detailed input data, which poses challenges for application in data-scarce urban contexts. Another approach is to combine GIS with multi-criteria decision analysis (MCDA, AHP) to classify flood risk zones based on elevation, rainfall, land use, and infrastructure factors [36,37,38,39,40]. Despite its efficiency, it is still affected by subjectivity in determining weights [41,42,43,44].

In recent decades, remote sensing technology has played an increasingly important role in monitoring and assessing inundation [45,46,47]. In particular, Sentinel-1 SAR imagery offers data acquisition under all weather conditions and through cloud penetration [48,49]. Recent studies have integrated Sentinel-1 data with deep learning models (CNN–LSTM) and MODIS data to create high-resolution maps, supporting urban planning and risk management [50,51]. Beyond satellite-based analyses, alternative data streams have also been explored for flood detection and impact assessment. For instance, Pastor-Escuredo et al. (2014) integrated mobile-phone activity (call detail records, CDRs) with rainfall and Landsat imagery to infer flood-affected areas in Tabasco, Mexico [52]. Their event-based, human behavior approach provided valuable insights into near-real-time population responses during flood events, complementing traditional physical models.

In Vietnam, the use of Sentinel-1 for urban flooding studies remains limited, mainly focusing on large river basins or pilot-scale areas [53,54,55,56]. Applications in large metropolitan areas remain challenging due to the complexity of built environments, which cause radar layover and shadowing effects [57], the limited availability of ground-truth flood data [58], and rapid land-cover changes resulting from urban expansion [59]. Nevertheless, such studies offer significant opportunities for advancing flood-monitoring methodologies in densely populated cities. The present study builds on these foundations by employing a multi-temporal Sentinel-1 SAR series (2015–2024) combined with geomorphological analysis to characterize long-term flood susceptibility across Hanoi.

Geomorphology also plays a vital role in flooding analysis. Numerous Vietnamese studies have demonstrated that terrain characteristics strongly influence flood behavior across different regions. Research in the Thu Bon Delta showed that low-lying plains and alluvial surfaces act as major flood-accumulation zones due to poor drainage conditions [17]. Studies in Hanoi revealed that urban expansion over floodplains and paleochannels has increased local inundation frequency and vulnerability to surface water flooding [19,25]. In the Central Coast and Hue Delta, geomorphological analyses confirmed that coastal lowlands and gentle slopes exhibit high flood retention capacity and limited drainage efficiency, directly controlling flood sensitivity [60,61]. In addition, investigations on the ancient Day–Nhue riverbeds demonstrated that inherited geomorphological structures continue to affect present-day urban flooding patterns in Hanoi [62]. Collectively, these studies confirm that geomorphology governs flood distribution, drainage performance, and re-flooding potential in Vietnam. However, most remain descriptive and lack integration with modern remote sensing approaches such as Sentinel-1.

Urban flooding is also associated with institutions and planning [63,64]. Planning that does not integrate hydrological factors, climate change, natural drainage mechanisms, or account for increased impervious surfaces and unreasonable ground elevation can overload drainage systems and cause localized flooding. Modern urban planning requires integrating nature-based solutions such as flood corridors, detention basins, and ecological buffers from the master planning stage. From a system analysis perspective, the DPSIR (Drivers–Pressures–State–Impacts–Responses) framework, proposed by the European Environment Agency (EEA) in 1999, provides a systematic approach to analyze cause–and–effect relationships of environmental issues [65,66]. Many international studies have applied the DPSIR framework to analyze floods and inundation, constructing aggregate indices such as the Integrated Flood Risk Index (IFRI) or the Coastal Risk Index (CRI) [66,67]. In Vietnam, the DPSIR framework has been applied in only a few cases, primarily for environmental or hydrological assessments, rather than for urban flood analysis. In the Vu Gia–Thu Bon Basin, Pham et al. (2015) combined hydrological–hydraulic modeling (MIKE NAM and MIKE-11) with GIS-based mapping and the conceptual DPSIR framework to analyze flood hazards and potential impacts [68]. Ha and Thao (2016) applied DPSIR to evaluate surface-water quality and pollution pressures in Cu Khe Commune, Hanoi [69]. Although both studies provided valuable methodological insights, they did not integrate DPSIR with quantitative geomorphological or remote sensing data, limiting their applicability for spatially explicit flood-risk assessment.

This literature overview shows that each approach has its strengths and limitations: Hydraulic models can reproduce detailed flood dynamics but require extensive hydrological and infrastructure data, which are often unavailable in rapidly urbanizing cities [31,33,35]; GIS–AHP provides flexible spatial analysis but relies heavily on expert judgment, leading to subjectivity and limited reproducibility [38,39,70]; Sentinel-1 performs well for flood detection and status monitoring, yet its outputs require integration with spatial and geomorphological data for proper contextual interpretation [51,54]; the DPSIR framework offers a systemic cause–effect perspective but has not yet been deeply integrated with quantitative spatial data such as geomorphology and remote sensing [66,67,69]. These methodological gaps collectively underscore the necessity for an integrated approach that combines geomorphological analysis and remote–sensing data within the DPSIR framework to assess urban flooding risk comprehensively.

In this context, this study adopts an integrated geomorphology–Sentinel-1–DPSIR approach to assess urban flooding risks in Hanoi. Study objectives include: (i) geomorphological analysis to identify flood-prone areas; (ii) using Sentinel-1 to detect and monitor actual flood areas; and (iii) integrating these findings into the DPSIR framework to analyze the system of drivers–pressures–state–impacts–responses. The study is expected to provide a scientific basis for urban planning and propose solutions for sustainable flood risk management and climate change adaptation.

In parallel, Hanoi’s rapid digital transformation underscores the need to integrate flood risk information into smart management systems. Combining geospatial analysis and inundation mapping with digital decision-support tools can strengthen data sharing, transparency, and coordination among sectors. This approach ensures that scientific assessments are not isolated from practice but become part of a broader framework for adaptive and resilient urban management.

2. Study Area

Hanoi city is geographically located between 20°53′ and 21°23′ North latitude and 105°44′ to 106°02′ East longitude, situated in the center of the Red River Delta. It is the capital and one of the major political, administrative, economic, and cultural centers of Vietnam. Hanoi has a high level of urbanization and population density, covering an area of 3359.84 km² with a population of over 8.5 million [71,72]. As one of the two main economic growth poles of Vietnam, the city experiences increasing pressure on its infrastructure and urban environment [72,73] (Figure 1).

Hanoi’s topography is relatively low and flat, characterized by a dense river network and a tropical monsoon climate. Rainfall is concentrated mainly during the wet season (May–October) [74], accounting for approximately 80% of the annual total, which makes the city highly prone to flooding, especially in the low-lying central districts during heavy rain events [72,73] (Figure 2).

Hanoi’s terrain is also profoundly shaped by its river system. Rivers such as the Day, Nhue, To Lich, and Ca Lo once had strong natural flows but, through geological and hydrological evolution, have left behind traces in the form of paleochannels, residual lakes, and interlaced ridges of varying elevation. Large water bodies such as West Lake, Yen So Lagoon, and a chain of smaller lakes along riverbanks are remnants of these processes, forming the city’s distinctive “mound–depression” landscape pattern [75].

Between 2010 and 2024, Hanoi experienced rapid urbanization, with an urbanization rate of 49.1%, 5.4 percentage points higher than the national average in 2024. Urban expansion has occurred primarily toward the southern and southwestern peripheries, accompanied by increases in both building density and height. However, this expansion has not been synchronized with infrastructure development, posing significant challenges for sustainable urban management and planning (Figure 3).

Hanoi has a comprehensive system of rivers, lakes, and dikes designed for flood control; however, recent inundations are primarily caused by localized heavy rainfall that exceeds the drainage capacity of these systems. The existing drainage infrastructure is insufficient to meet the growing demands under rapid urbanization and population increase. In recent years, intense rainfall events have repeatedly caused severe flooding in many inner-city districts, resulting in considerable damage to assets and public infrastructure.

The scope of this study encompasses 12 inner-city districts and four suburban districts (Dan Phuong, Hoai Duc, Gia Lam, and Thanh Tri) (Figure 1), representing areas that have undergone intensive urbanization and extensive land use conversion from agricultural to non-agricultural purposes. These areas have also recorded numerous severe flooding locations between 2010 and 2024. Furthermore, they face critical planning challenges, particularly regarding drainage systems, land use management, and urban development control under conditions of climate change and rapid population growth.

3. Materials and Methods

3.1. Data Sources

The study integrates multi-source datasets, including remote sensing data, topographic and geomorphological maps, statistical data, and field surveys to assess the current status and flooding risk in Hanoi (

Table 1. Data sources for the study.

No.	Types of Data	Formats	Duration	Source/References	Purposes of Use
1	Sentinel-1 SAR (C-band, IW)	Raster (10 m)	2015–2024 (July–October)	ESA/Copernicus Hub	Flood mapping, constructing maps of flood frequency
2	ALOS AVNIR-2	Raster (30 m)	2010, 2020	JAXA	Analysis of urban fluctuation
3	VLULC Vietnam Land Use Land Cover	Vector/Raster	2010–2020	Ministry of Natural Resources and Environment, VLULC update	Map of land use, urban expansion
4	DEM (SRTM v3)	Raster (30 m)	2011	USGS	Topographic analysis, geomorphological indices
5	Geomorphological map of Hanoi	Vector (1:320,000)	Original 2015; Reclassified 2024	Dao Dinh Bac (1984) [76]; Authors (reclassification and GIS update 2024)	Classification of 9 geomorphological units
6	Historical flood points	Vector (GPS points)	2012–2022	Reports of authority, press, and field surveys	Flood map validation
7	Population (Statistical yearbook)	Data tables	2013–2021	Hanoi Statistics Office	Assessment of “Drivers” in DPSIR (non-GIS)
8	Drainage systems	Documents	2018–2022	Hanoi Sewerage and Drainage Company Limited	Evaluation of “Pressures” in DPSIR (non-GIS)

Notes: Sentinel-1 data were used to construct a flood frequency map; VLULC and ALOS datasets supported urban expansion analysis; DEM and geomorphological maps were used to evaluate terrain-related flood risks, while population and drainage data contributed to the DPSIR framework analysis.

). Data were selected according to three criteria: (i) reliability of the sources; (ii) spatial and temporal resolution suitable for Geographic Information System (GIS) analysis; and (iii) direct relevance to each component of the Drivers–Pressures–State–Impacts–Responses (DPSIR) framework.

Remote sensing data were obtained mainly from Sentinel-1 Synthetic Aperture Radar (SAR) imagery provided by the European Space Agency (ESA) and from Advanced Land Observing Satellite (ALOS) AVNIR-2 optical images distributed by the Japan Aerospace Exploration Agency (JAXA). The Vietnam Land Use Land Cover (VLULC) dataset was sourced from the Ministry of Natural Resources and Environment. Digital Elevation Model (DEM) data were derived from the Shuttle Radar Topography Mission (SRTM) version 3, distributed by the United States Geological Survey (USGS). All spatial data were standardized to the World Geodetic System 1984/Universal Transverse Mercator (WGS84/UTM) Zone 48N coordinate system.

3.2. Constructing Flood Maps from Sentinel-1 SAR

Sentinel-1 (A and B) is a spaceborne Synthetic Aperture Radar (SAR) mission operated by the European Space Agency (ESA) under the Copernicus Programme. It operates in the C-band (5.405 GHz). It provides all-weather, day-and-night imaging capability with a 6–12-day revisit cycle, ensuring consistent Earth observation independent of cloud cover or illumination conditions. Sentinel-1 SAR data have been widely utilized for various environmental and geospatial applications, including flood mapping, surface-water monitoring, land deformation detection, and other Earth observation analyses [77].

In this study, Sentinel-1 SAR images (VV and VH polarizations) acquired during the rainy season (May–October, 2015–2024) were processed on Google Earth Engine (GEE) following these steps:

• Preprocessing: Orbit and terrain correction, radiometric calibration, and speckle filtering using the Refined Lee filter (5 × 5 kernel).
• Flood classification: Calculation of the SAR-based NDWI index, application of Otsu’s automatic thresholding method to classify water and non-water pixels, and generation of binary flood maps for each image [78].
• Flood frequency mapping: The image series was compiled to count the number of times each pixel was classified as inundated during the period from 2015 to 2024. The resulting pixel-occurrence values were then categorized into five flood-frequency levels—very low, low, medium, high, and very high—using the equal-interval classification method based on the statistical distribution of inundation counts [57,79].
• Validation: Using 148 historical flood points (2012–2024) collected from official reports, media sources, and GPS surveys for spatial–temporal matching. The validation focused on flood occurrence and distribution rather than flood depth or duration, due to the limited resolution of reference data.

Accuracy metrics were computed from the confusion matrix following Congalton (1991) [80]. The Overall Accuracy (OA) and Kappa Coefficient (κ) were calculated using the following equations:

$$\mathrm{O}\mathrm{A}={\displaystyle \frac{{\sum }_{i=1}^r{n}_{ii}}{\mathrm{N}}}$$

$$\kappa ={\displaystyle \frac{\mathrm{N}{\sum }_{i=1}^r{n}_{ii}-{\sum }_{i=1}^r(n_{i+}{n}_{+i})}{{\mathrm{N}}^2-{\sum }_{i=1}^r(n_{i+}{n}_{+i})}}$$

where $n_{ii}$ represents correctly classified samples in class i, N is the total number of samples, $n_{i+}$ and $n_{+i}$ are the total numbers in the reference and classified datasets for class i, respectively.

The overall accuracy reached 87% (Kappa = 0.79), consistent with international studies [81,82]. The GIS–remote sensing framework developed in this study not only supports quantitative flood risk assessment but also provides a standardized spatial database that can be incorporated into digital flood management systems. This interoperability enables continuous data updating, integration with monitoring networks, and long-term application in e-government platforms for flood control and urban planning.

While the present method estimates inundation based on multi-temporal Sentinel-1 SAR observations, other studies have utilized non-physical or behavioral data streams—such as mobile-phone activity combined with satellite imagery and rainfall records—to infer flood-affected zones [52]. These complementary approaches capture real-time human responses to flooding, whereas the SAR-based analysis adopted here provides consistent, spatially explicit monitoring of long-term inundation dynamics across multiple years, offering a more stable basis for urban flood assessment.

3.3. Analysis of Urbanization and Land Use

ALOS AVNIR-2 images (2010 and 2020), combined with VLULC datasets (2010–2020), were used to detect changes in land cover and identify newly urbanized areas in comparison to pre-existing urban zones.

Old urban areas were defined as those urbanized before 2010, while new urban areas refer to regions developed after 2010. Urbanization layers were overlaid with the geomorphological map to determine which geomorphological units were most affected by new urban expansion, particularly those in low-lying terrains (inner and outer-dike floodplains and fluvio-marine plains).

Previous studies have consistently shown that urban expansion in low-elevation zones increases flood susceptibility [83].

3.4. Flood Risk Assessment by Geomorphology

The geomorphological map of Hanoi (scale 1:320,000) delineates nine geomorphological units, including inner and outer-dike floodplains, paleochannels, karst areas, fluvio-marine plains, denudation surfaces, fluvial surfaces, mixed surfaces, and river terraces.

The flood frequency map (Section 3.2) and urbanization layer (Section 3.3) were overlaid with this geomorphological map to identify terrain units most vulnerable to inundation.

The flood risk map was constructed using the Weighted Overlay method in GIS, which assigns weights to each geomorphological unit based on its level of vulnerability, and then overlays them with flood frequency and urbanization data. This method has been widely applied in numerous studies assessing disaster and flood risk [39,84,85].

A Weighted Overlay Analysis was performed in GIS, assigning each geomorphological unit a weight from 0 (very low) to 5 (very high) based on its susceptibility to flooding. Weighting values were derived from a synthesis of international studies [86] and adjusted to reflect Hanoi’s specific topographic features, including elevation, slope, and natural drainage capacity.

Based on the weighting table, combined with the flood frequency and urbanization data layers, a geomorphology-based flood risk map was developed.

3.5. DPSIR Framework

The Drivers–Pressures–State–Impacts–Responses (DPSIR) framework was applied to systematically examine the cause–and–effect relationships among factors influencing flooding in Hanoi.

• Drivers: Population growth, rapid urbanization, and climate change.
• Pressures: Increasing impervious surfaces, infilling of lakes and ponds, and reduced drainage capacity.
• State: Current flood frequency and urbanization maps, combined with geomorphological conditions.
• Impacts: Results from field and sociological surveys assessing effects on livelihoods, infrastructure, and the economy.
• Responses: Policies, planning solutions, and infrastructure projects that have been implemented or are under development.

The DPSIR framework has been widely and successfully applied in environmental and risk management research, both in Vietnam and internationally [87]. Its use in this study enables a comprehensive analysis of the root causes and evolution of flooding, also supporting the formulation of adaptive and sustainable planning strategies.

3.6. Field Surveys and Sociological Investigations

Field surveys were conducted from 22 to 30 March 2025 in four districts with high flood frequency (Thanh Xuan, Hoang Mai, Nam Tu Liem, and Hoai Duc)—the surveys documented flooded sites, drainage infrastructure, photographic evidence, and GPS coordinates.

For the sociological component, 80 structured questionnaires were administered directly to households in flood-affected areas to collect perceptions of flood causes, impacts, and proposed planning solutions. Respondents were recruited through local community groups using purposive sampling, ensuring inclusion of both long-term residents and newly urbanized households. Participation was voluntary and anonymous, and all 80 forms were validly completed (100% response rate).

The questionnaire comprised 20 items organized into four thematic parts: (1) respondent information; (2) factors influencing flooding, including flood frequency, duration, and drainage performance; (3) impacts of flooding (e.g., on daily life, health, and property); and (4) expectations and proposed measures for future planning and flood management.

Most questions were close-ended using a five-point Likert scale (1 = no impact/strongly disagree, 5 = very strong impact/strongly agree), while the final section contained open-ended items for qualitative feedback. The collected data were analyzed descriptively to complement the Sentinel-1 flood mapping (State) and to assess Impacts and Responses within the DPSIR framework. The full questionnaire and summary of residents’ responses are presented in Supplementary Material File S1.

3.7. Study Process

The overall study process comprised four main stages (Figure 4):

• Data collection and preprocessing: Compilation and preparation of Sentinel-1, ALOS, VLULC, DEM, geomorphological, flood point, population, and drainage datasets.
• Geomorphological and urbanization analysis: Generation of old and new urban maps and analysis by geomorphological unit.
• Flood mapping and frequency analysis: SAR image processing, flood frequency calculation, and validation using historical flood points.
• Risk assessment and adaptive planning proposals: Integration of Weighted Overlay results with the DPSIR framework and social survey data to identify high-risk zones and propose adaptation-oriented solutions.

Figure 4. Research flow.

Figure 4. Research flow.

4. Results and Discussion

4.1. Flood Frequency in Hanoi City (2015–2024)

The flood frequency map derived from Sentinel-1 image series (2015–2024) reveals clear spatial differentiation in both extent and frequency of inundation across Hanoi. Each flood event corresponds to a distinct Sentinel-1 acquisition date during which water pixels exceeded the Otsu-based threshold [78]. Based on the number of flood occurrences, the study area was classified into five levels: very low (0–1 event), low (2–3 events), medium (4 events), high (5–6 events), and very high (>six events) (

Table 2. Statistics of the area of flood frequency levels in Hanoi (2015–2024).

Flood Frequency Level	Area (ha)	Rate (%)
Very low	225,626.2	67
Low	47,631.2	14
Medium	24,061.7	7
High	30,347.2	9
Very high	7435.1	2

). This classification represents the relative recurrence of inundation, where low-frequency areas correspond to occasional, short-term events, and high-frequency zones denote persistent or annually recurring floods.

Approximately 67% of the area experienced a very low flood frequency, whereas 18% fell into the medium to very high categories. Flood-prone areas are mainly concentrated in the western and southern districts (Thach That, Chuong My, My Duc, Phu Xuyen, Thuong Tin) and Soc Son, Dong Anh, which have low-lying terrain and rely on dikes and natural drainage capacity. In contrast, inner-city districts primarily experienced localized flooding events associated with extreme rainfall that exceeded the capacity of the existing drainage system (Figure 5).

The flood-frequency map was validated using 148 historical flood points (2012–2024) collected from official reports, media sources, and GPS field surveys. Each point was georeferenced and matched with pixel values from the Sentinel-1 flood-frequency map using the “Extract Values to Points” function in ArcGIS Pro (version 3.0.2). The comparison yielded 129 correctly classified (flooded) points and 19 misclassified points, resulting in an overall accuracy of 87% (κ = 0.79).

Major flood events in Bac Son (Soc Son) and Xuan Mai (Chuong My) were clearly reflected, demonstrating the consistency between the remote sensing results and actual conditions. However, the validation data mainly focused on large-scale flood events, with limited information on minor and short-duration inundation. In addition, the recorded locations were often reported only at the commune or village level, which limited the precision of spatial matching. Future studies should incorporate higher-resolution field surveys, community-based (crowdsourced) flood records, or IoT-based hydrological sensors to enhance spatial accuracy and validation robustness [48,88].

4.2. Urbanization Trends on Geomorphological Units in Hanoi

The urbanization map, constructed by overlaying data from 2010 and 2020, shows that Hanoi’s urban area increased rapidly, reaching 56,543.7 ha in 2020. Of this total, 66% corresponds to old urban areas (existing since 2010) and 34% to new urban areas (developed between 2010 and 2020), mainly concentrated in the West and Southwest (Thach That, Thanh Tri, Thuong Tin, My Duc) (

Table 3. Urbanization area and rate in Hanoi.

Urban Types	Area (ha)	Rate (%)
Old urban area	37,122.1	66
New urban area	19,421.6	34

).

When integrated with the geomorphological map, the results show that urbanization has primarily occurred on two low-lying geomorphological units: inner-dike floodplains (39.5%) and outer-dike floodplains (23.0%) (Figure 6). These areas are prone to flooding due to low elevation, limited drainage capacity, and high dependence on river–lake systems.

This reflects an imbalance between urban spatial expansion and flood adaptability. Previous studies have also indicated that many newly urbanized areas of Hanoi overlap with flood-prone geomorphological units, increasing their vulnerability to heavy rainfall events [40,89]. The findings reinforce the observation that the trend of urban development on low-lying alluvial grounds is one of the key factors contributing to increased flood risk (Figure 7 and Figure 8).

Specifically, the overlay analysis indicates that approximately 80% of newly urbanized areas are distributed across geomorphologically sensitive units, including inside- and outside-dike floodplains, fluvio-marine plains, paleochannels, and karst, confirming a strong spatial association between urban expansion and flood-prone terrains (Figure 8).

4.3. Statistics of Inundated Areas by Geomorphological Units

The results indicate that the three geomorphological units with the largest inundated areas are inner-dike floodplains (36.4%), outer-dike floodplains (16.6%), and fluvio-marine plains (15.6%), mainly concentrated in low-lying zones (

Table 4. Statistics of inundated areas by geomorphological units in Hanoi.

	Inundated Area	Percentage
Denudation–erosion origin	11,336.1	9.5%
Fluvial-origin surfaces	4595.2	3.8%
River terrace	7554.0	6.3%
Inner-dike floodplains	43,638.6	36.4%
Outer-dike floodplains	19,950.7	16.7%
Paleochannels	4534.4	3.8%
Mixed surfaces	2145.1	1.8%
Karst	7250.1	6.1%
Fluvio-marine plain	18,727.5	15.6%

). These areas are characterized by low elevation, limited drainage capacity, and direct exposure to heavy rainfall and river flooding, particularly in the southwestern subsided zones (Chuong My, My Duc, Phu Xuyen, Thuong Tin), which function as the “flood outlets” of Hanoi.

Conversely, elevated geomorphological units such as river terraces, karst, and denudation–erosion surfaces, exhibit a lower flooding risk due to their greater topographic height and more effective natural drainage capacity (Figure 9). Previous studies have also indicated that floodplains and alluvial grounds face a 2–3 times higher flood risk compared to elevated geomorphological units [81], confirming the decisive role of geomorphology in the spatial distribution of inundation [90].

4.4. Characteristics of Inundation in Urban Areas (New and Old Urban Areas)

Analysis of the overlay between flood data and the urban spatial layer revealed a clear distinction between old and new urban areas. The total inundated area was 4779.2 ha, of which 73% occurred in new urban areas and 27% in old urban areas (

Table 5. Statistics of the inundated area in Hanoi’s urban areas.

Urban Types	Inundated Area (ha)	Rate (%)
New urban area	3500.12	73
Old urban area	1279.0	27

).

Although new urban areas cover only 34% of the city’s total urban land (Table 2), they account for 73% of the inundated surface (Table 4). This pattern indicates that newly urbanized zones—especially those in the western and southwestern sectors—are more exposed to flooding due to their low-lying geomorphological settings and underdeveloped drainage infrastructure.

New urban areas are more prone to inundation due to rapid development on soft, compressible alluvial and fluvio-marine soils with low bearing capacity and poor drainage, particularly in the western and southwestern zones (Figure 10). Old urban areas experience less flooding thanks to their higher terrain and earlier-constructed drainage systems. However, localized inundation still occurs in degraded streets such as Nguyen Trai, Giai Phong, and Truong Chinh.

In fact, newly developed urban zones in the western part of Hanoi frequently experience flooding during heavy rainfall events, a finding consistent with previous analyses [27,89]. This finding suggests that inundation is not only a consequence of rapid urbanization but also closely linked to geomorphological characteristics, underscoring the need to integrate topographic and geomorphological factors into sustainable urban planning.

4.5. Assessment of Flood Risk Using a Geomorphological Approach

The weighting scale consists of five levels (0, 2, 3, 4, and 5), corresponding to very low, low, medium, high, and very high risk. This classification employs a multi-criteria evaluation (MCE) framework, commonly used in GIS-based flood-susceptibility mapping, in which geomorphological units are assigned relative weights reflecting terrain elevation, slope, drainage efficiency, and sedimentary structure [84,85].

This scale was determined from: (i) previous geomorphological research in Hanoi [19,25,62]; (ii) comparative studies applying weighting methods for flood-risk zoning [84,85]; and (iii) analytical results of flood-frequency mapping for 2015–2024.

Particularly, areas of denudation–erosion origin and karst terrain at higher elevations were assigned negligible to moderate risk values (0–3). For the first-order river terrace, flood risk depends on the combined influence of alluvial deposition and human impacts (0–3). The remaining geomorphological units (floodplains, paleochannels, fluvio-marine plains, and fluvial-origin surfaces) were assigned the full five-level scale (0–5), as they are located in low-lying terrain with poor drainage and are highly susceptible to river floods (

Table 6. Weighting system for flood risk assessment by geomorphological units.

Risk Level	Denudation Origin	Fluvial-origin Surfaces	River Terrace	Inner-Dike Floodplains	Outer-Dike Floodplains	Paleochannels	Mixed Surfaces	Karst	Fluvio-Marine Plain
Very low	0	0	0	0	0	0	0	0	0
Low	0	2	0	2	2	2	2	0	2
Medium	0	3	0	3	3	3	3	0	3
High	2	4	2	4	4	4	4	2	4
Very high	3	5	3	5	5	5	5	3	5

Notes: Applied consistently to both old and new urban areas.

).

A qualitative sensitivity check was also performed by modifying selected weights (±1 level) to test the robustness of classification outcomes. The overall distribution of high-risk zones on the resulting flood-risk map (Figure 11) remained stable, confirming that the weighting system is theoretically sound and relatively insensitive to minor variations in the data. This supports the reliability of the geomorphology-based flood-risk classification summarized in Table 6.

The results show that areas with medium to very high risk are mainly concentrated in suburban zones, particularly in Thuong Tin, My Duc, and Phu Xuyen—natural “flood outlet” areas of Hanoi. Several other districts, such as Gia Lam and Me Linh, also exhibit varying levels of risk, which should be taken into consideration in urban planning.

The map shows a clear five-level differentiation between inner-city and outer-city areas, providing a scientific basis for identifying high-risk zones and serving as a foundation for DPSIR-based analysis to propose solutions for urban flood management.

4.6. DPSIR Framework in Assessing Urban Flooding and Spatial Planning in Hanoi

4.6.1. Drivers

Urban flooding is strongly influenced by socioeconomic drivers such as population growth, rapid urbanization, inadequate integration of hydrological considerations into urban planning, and the impacts of climate change. Between 2010 and 2023, Hanoi’s population increased by more than 2 million people, reaching a density of approximately 2556 people per km², which has placed considerable pressure on the drainage infrastructure [71,91].

The impervious surface area expanded by about 35%, reducing water permeability and the city’s natural regulation capacity. New urban areas have primarily developed toward the West and South, along major transportation corridors, resulting in drainage demands that exceed existing capacity.

Spatial planning remains administrative mainly, with limited integration of climate change and hydrological conditions, and is sometimes influenced by short-term interests, leading to fragmented urban development and infrastructure overload. Drainage policies still depend primarily on centralized systems, thereby reducing the city’s adaptive capacity and resilience [92].

4.6.2. Pressures

From the above drivers, three main groups of pressures are pushing Hanoi’s urban system to its limits of exploitation:

(1) Expansion of impervious surfaces and reduction of natural regulation capacity: Between 2014 and 2023, many districts (Nam Tu Liem, Hoang Mai, Ha Dong, Hoai Duc) reported increases of 11.3–39.8% in special-use land, while agricultural land decreased by 3.6–24.1% [71,93]. The number of lakes and ponds declined from 122 (2010) to 112 (2015), with a surface water loss of approximately 72,540 m² [94], gradually eroding the city’s natural “buffer zones.”

(2) Lack of synchronization between planning and infrastructure: New urban areas have often been constructed on low-lying ground with limited drainage outlets. In Linh Dam (Hoang Mai), the master plan was adjusted to nearly double the residential land, reducing green spaces and detention basins; construction density exceeded 50%, rapidly overloading the drainage network [92]. Surveys indicate that over 50% of residents rate the drainage system as degraded and frequently clogged. Hanoi still employs a combined drainage system for stormwater and wastewater, with a treatment capacity of only about 30% (1 million m³/day) across six plants (Figure 12).

(3) Extreme climate and heavy rainfall: Although total annual rainfall during 1961–2020 decreased by 8.7%, the frequency of high-intensity rainfall events has significantly increased. In this study, extreme rainfall refers to precipitation exceeding the 95th percentile of the local long-term rainfall distribution ([95]). In contrast, heavy rainfall is defined as daily totals of 100 mm or more, following the operational standard of the Vietnam Meteorological and Hydrological Administration. For instance, Typhoon No. 3 (Yagi) on 7 September 2024 caused 200–500 mm of rainfall over two days, far exceeding the design capacity of the drainage system (70–100 mm in 2 h). The combination of heavy rainfall, rapid urbanization, and inadequate infrastructure has turned many districts into frequent hotspots for flooding.

4.6.3. State

Between 2012 and 2022, Hanoi experienced over 50 heavy rainfall events, resulting in flooding at more than 200 locations across 16 districts [27]. The number of flooded points increased from 21 (2012) to 144 (2022), reflecting an upward trend in both spatial extent and frequency [27,92].

Mapping results indicate that flooded points frequently form clusters along major urban corridors including Ring Road 3—Thang Long Bridge (North), Thang Long Avenue (West), National Highway 6 (Southwest), and Duong Bridge axis (Northeast). Flood frequency was distributed as follows: 18% very high (9–24 events), 15.5% high (6–8 events), 34.5% medium (3–5 events), and 32% low (1–2 events) [27]. Thanh Xuan, Hoan Kiem, Hoang Mai, Cau Giay, and Ha Dong were identified as flooding “hotspots” due to high construction density, aging infrastructure, and loss of permeable surfaces (Figure 13).

These results not only confirm the widespread occurrence of urban flooding but also indicate a systematic increase exceeding the design capacity of existing drainage infrastructure, highlighting limitations in management and planning [92].

4.6.4. Impacts

Flooding in Hanoi has a multidimensional impact on society, the economy, and infrastructure. Socially, 74% of surveyed residents reported that their daily life and work were directly affected. Indirect economic damage due to traffic congestion and livelihood disruptions is estimated at tens of billions of VND per hour of flooding, particularly for small traders and businesses (52%).

Repeated flooding also accelerates the deterioration of transport infrastructure, sewers, and underground facilities. Road surfaces are degraded, basements and maintenance holes are regularly clogged with sediment, waste, and grease, while maintenance mechanisms remain passive and lack systematic monitoring. Consequently, maintenance costs have risen sharply, creating a burden on the municipal budget and reducing the efficiency of public investment.

4.6.5. Responses

At the household level, residents have adopted temporary adaptive measures such as raising floor levels, installing pumps, elevating furniture, and sharing flood warnings on social media. However, these solutions only mitigate short-term damages without addressing the root causes.

At the city level, Hanoi has implemented major projects, including the Yen Xa, Phu Do, and An Lac wastewater treatment plants; rehabilitation of the To Lich and Kim Nguu rivers; and gradual separation of stormwater and wastewater systems under Decision 725/QD-TTg, alongside promoting integrated planning following Decision 1569/QD-TTg (2024).

Nevertheless, many projects remain delayed, facing capital shortages and a lack of cross-sector coordination. The absence of a real-time monitoring system has further reduced operational efficiency. To improve long-term resilience, an integrated management strategy is necessary, combining technical measures (such as green infrastructure, detention basins, and permeable materials) with governance solutions (including basin-based planning and smart monitoring).

Analysis using the DPSIR framework (Figure 14) systematizes the cause–and–effect relationships between urbanization and flooding in Hanoi. The drivers of population growth, spatial expansion, and socioeconomic development have created intense pressures on land and infrastructure. The pressures—such as increased impervious surfaces, infilling of lakes and ponds, and climate change—have overloaded the drainage system, resulting in expanded and recurrent seasonal flooding.

The consequences of this phenomenon are reflected in numerous negative impacts, including economic losses, traffic disruptions, a decreased quality of life, and accelerated infrastructure degradation. Although responses have been implemented through the 2030 master plan, detention basins, and wastewater treatment projects, their effectiveness remains limited due to a lack of intersectoral coordination, slow implementation, and insufficient integration of nature-based solutions.

The results highlight the urgent need for an integrated flood management strategy that combines green–gray infrastructure, basin-based planning, and real-time monitoring technologies to enhance resilience and move toward a sustainable and climate-adaptive Hanoi [96]. The integration of geomorphological and Sentinel-1 analyses provides reliable spatial patterns that reflect the interactions among topography, land use, and observed inundation [48,88]. These results could be linked with real-time monitoring platforms and digital spatial management systems to support proactive flood response and infrastructure planning. Such integration also enhances coordination between scientific assessment and administrative decision-making, contributing to more adaptive and transparent urban governance [40].

4.7. Planning Orientations for Flood-Adaptive Urban Development in Hanoi

Flooding in Hanoi is a consequence of rapid urbanization on low-lying terrain, the expansion of impervious surfaces, the infilling of lakes and ponds, fragmented planning, and drainage systems that fail to meet increasing demands [19,25,26,27,92]. Flooding is concentrated in newly developed urban areas and low-lying geomorphological zones, which account for over 70% of total inundated areas [40]. This highlights the urgent need for sustainable urban planning and enhanced resilience based on risk management and the integration of spatial data.

(1) Basin-based planning and risk zoning: Flood frequency maps combined with geomorphological and urbanization analyses should be used to identify high-risk areas (such as Thuong Tin, My Duc, Phu Xuyen, and Ha Dong). These zones should be protected from high-density development, preserved as natural drainage corridors, and be incorporated into climate-adaptive spatial planning. Experience from the Sponge City program (China) and GIS–AHP studies in Hanoi demonstrates that pre-planning risk zoning helps reduce damage and optimize investment costs [40,97].

(2) Development of green–gray infrastructure and nature-based solutions: The results presented in Section 4.2, Section 4.3, Section 4.4 and Section 4.5 indicate that the loss of water bodies and permeable land is a major driver of increased flood risk. Hanoi should prioritize the restoration of lakes, ponds, flood drainage corridors, and natural water retention zones, in combination with green infrastructure (rain gardens, green roofs, permeable pavements) and gray infrastructure (detention basins, interceptor sewers). The green–blue infrastructure approach, along with Low-Impact Development (LID) and sunken green spaces, can reduce surface runoff by up to 75% while simultaneously improving landscape quality and water environments [96,98].

(3) Upgrading and separation of the drainage system: Currently, more than 70% of inner-city areas still rely on a combined drainage network for stormwater and wastewater. It is essential to separate the two systems, expand detention basin capacity, and integrate decentralized regulating tanks with smart drainage management technologies to reduce overload risks during extreme rainfall events [96].

(4) Governance mechanisms and financial policies: New drainage design standards should be established based on extreme rainfall scenarios (≥100 mm/h). Flood risk assessment must be required before urban project approval, and an anti-flood fund should be developed under public–private partnership (PPP) models. Land use planning, drainage, and infrastructure development must be synchronized to ensure coherence and long-term effectiveness [92,99].

(5) Monitoring and implementation roadmap: A real-time rainfall and flood monitoring system integrating GIS and remote sensing should be established to evaluate the effectiveness of interventions. Pilot projects should be implemented in heavily flooded districts (Hoang Mai, Ha Dong, Thanh Xuan), followed by large-scale deployment over a 5–10-year roadmap aligned with Hanoi’s sustainable development strategy.

Based on the analytical findings, several orientations are proposed to promote flood-adaptive and intelligent urban development in Hanoi. Beyond conventional physical and institutional measures, urban management should also adopt digital technologies and data-driven tools to enhance its effectiveness. Integrating GIS- and remote sensing-based flood databases with IoT monitoring and e-government platforms can improve early warning capabilities, optimize resource allocation, and facilitate public participation. This integrated approach connects scientific analysis with governance practice, strengthening the city’s adaptive capacity to climate-induced flooding.

These orientations will not only help Hanoi reduce flood risks but also contribute to achieving SDG 11—building safe, adaptive, and sustainable cities.

5. Conclusions

Urban flooding has become a signification challenge for sustainable development in Hanoi, as its frequency, extent, and severity continue to increase. This study employed an integrated approach combining geomorphological analysis, Sentinel-1 remote sensing data, and the DPSIR framework to assess the relationship between urbanization and flood risk.

The results indicate that more than 70% of inundated areas are situated in low-lying geomorphological units, including floodplains, paleochannels, and fluvio-marine terrains. Notably, approximately 80% of newly urbanized areas have been developed in sensitive zones, of which 36.5% are situated on inner-dike floodplains. The flood map validated using 148 field points achieved an overall accuracy of 87% (Kappa = 0.79), confirming the effectiveness of radar data in spatio-temporal monitoring of urban flooding.

The DPSIR analysis identified three major groups of pressures: the reduction of natural water retention spaces, unsynchronized infrastructure planning and management, and the increase in extreme rainfall events caused by climate change. Among these, uncontrolled urbanization and weak enforcement of planning regulations were identified as root causes, leading to recurrent and increasingly severe flood incidents.

This study presents scientific evidence of the combined impacts of urbanization on flooding. It proposes adaptive planning strategies for Hanoi, including the preservation of water retention spaces, the development of green–gray infrastructure, and the upgrading of the drainage system. However, the scope of this study was limited to Hanoi from 2015 to 2024, and it did not consider extreme climate scenarios. Future research should expand the study area, employ high-resolution remote sensing data, integrate urban hydrological models, and apply artificial intelligence techniques to enhance reliability and support the development of a solid scientific foundation for sustainable urban planning in Vietnam. The findings of this study also demonstrate the potential of integrating geomorphological and remote sensing data into digital management frameworks. Future work should focus on developing interoperable geospatial infrastructures and applying artificial intelligence for dynamic flood prediction and urban monitoring. Embedding these analytical frameworks into smart-city and e-government systems will enhance resilience, transparency, and the overall efficiency of urban governance in Vietnam.