Next Article in Journal
Impact of Lockdown on the Surface Water Quality in Kelani River, Sri Lanka
Next Article in Special Issue
Assessment and Mapping of Riverine Flood Susceptibility (RFS) in India through Coupled Multicriteria Decision Making Models and Geospatial Techniques
Previous Article in Journal
Analysis of the Spatial Distribution Characteristics of Emerging Pollutants in China
 
 
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Cloud-Based Machine Learning for Flood Policy Recommendations in Makassar City, Indonesia

by
Andi Besse Rimba
1,*,
Andi Arumansawang
2,
I Putu Wira Utama
3,4,
Saroj Kumar Chapagain
5,
Made Nia Bunga
4,6,
Geetha Mohan
7,
Kuncoro Teguh Setiawan
8 and
Takahiro Osawa
9
1
Department of Civil Engineering, Shibaura Institute of Technology, 3-7-5 Toyosu, Tokyo 135-8548, Japan
2
Karya Alam Selaras Ltd., Citra Land, Jl. Talassa City Block 37 A Kapasa, Makassar 90245, Indonesia
3
Development Planning Agency of Bali Province, Jl. Cok Agung Tresna, Sumerta Kelod, Denpasar City 80239, Indonesia
4
Doctoral Program of Environmental Science, Udayana University, Jl. P.B. Sudirman Denpasar, Bali 80114, Indonesia
5
Institute for Integrated Management of Material Fluxes and of Resources (UNU-FLORES), United Nations University, Ammonstrasse 74, 01067 Dresden, Germany
6
Fisheries Faculty, The University of 45 Mataram, Jl. Imam Bonjol No. 45 Cakranegara, Mataram City 83239, Indonesia
7
Global Research Centre for Advanced Sustainability Science (GRASS), University of Toyama, 3190 Gofuku, Toyama City 930-8555, Japan
8
Research Center for Remote Sensing, BRIN, Jl. Raya Jakarta Bogor Km 46, Cibinong, Bogor 16911, Indonesia
9
Center for Remote and Application of Satellite Remote Sensing, Yamaguchi University, 2-16-1 Tokiwadai, Ube 755-8611, Japan
*
Author to whom correspondence should be addressed.
Water 2023, 15(21), 3783; https://doi.org/10.3390/w15213783
Submission received: 21 September 2023 / Revised: 26 October 2023 / Accepted: 27 October 2023 / Published: 29 October 2023

Abstract

:
Makassar City frequently experiences monsoonal floods, typical of a tropical city in Indonesia. However, there is no high-accuracy flood map for flood inundation. Examining the flood inundation area would help to provide a suitable flood policy. Hence, the study utilizes multiple satellite data sources on a cloud-based platform, integrating the physical factors of a flood (i.e., land use data and digital elevation model—DEM—data) with the local government’s urban land use plan and existing drainage networks. The research aims to map the inundation area, identify the most vulnerable land cover, slope, and elevation, and assess the efficiency of Makassar’s drainage system and urban land use plan. The study reveals that an uncoordinated drainage system in the Tamalanrea, Biringkanaya, and Manggala sub-districts results in severe flooding, encompassing a total area of 35.28 km2. The most affected land use type is cultivation land, constituting approximately 43.5% of the flooded area. Furthermore, 82.26% of the urban land use plan, covering 29.02 km2, is submerged. It is imperative for the local government and stakeholders to prioritize the enhancement of drainage systems and urban land use plans, particularly in low-lying and densely populated regions.

1. Introduction

Flooding is considered to be one of the most destructive natural disasters and the most damaging in developing countries due to the low level of protection against it [1]. Indonesia is fourth-ranked in terms of the number of people exposed to high flood risk after China, India, and Bangladesh and number one in Southeast Asia, with a total high-risk population of up to 75.7 million, or 27% of the total nation’s population [2]. Jakarta, Bogor, Bandung, Semarang, and Solo, including Makassar City, are the tropical cities in Indonesia that experience flooding every year during the monsoon season. A devastating flood inundated almost all regions of Makassar City, peaking on 13 February 2023. The flood affected about 3344 houses in 19 villages—9167 people or 2695 families—according to a field survey by the Makassar Municipal Disaster Mitigation Board (BPBD), South Sulawesi [3].
Makassar City is an industrial city that supplies goods in East Indonesia and is the largest city in that region of the country. The position of Makassar City is extremely important in supporting the economic growth of Indonesia [4]. Therefore, disaster management in terms of policy recommendations is important. Several studies have been conducted on the flood in Makassar. These studies focused on potential flood risk mapping. Widodo et al. [5] found a high potential risk of flooding due to land use changes in the Makassar neighborhood region. Musliadi et al. [6] modeled the potential flood depth of the Tallo watershed in Makassar using the Log-Pearson type III method on rainfall data. Thoban and Hizbaron [7] conducted research with spatial multi-criteria evaluation (SMCE) to find the resilience level of the Makassar community to flooding. Djamaluddin et al. [8] and Indrayani et al. [9] predicted the flood vulnerability in Makassar City by considering social, physical, economic, and environmental aspects. Sudirman et al. [10] identified the watershed of high probability to flood by considering the surface runoff. Another study focused on the flood adaptation strategy conducted by the local community [11]. The verification methods for evaluating the potential flood risk in previous studies were field surveys and the flood inundation area by sub-district or village region provided by the local government. If one spot in a sub-district or village is flooded, all areas in that region are categorized as flood inundation areas because there is no study extracting the flood inundation from satellite images. Hence, the benefit of this study is that it presents a reliable inundation map that can be used to verify the potential flood risk in the future. Moreover, this study compared flood inundation to physical factors (i.e., land cover, slope and elevation), drainage distribution, and the urban land use plan of Makassar City. This helped to identify the spatial distribution of the flood map and the high-priority sub-districts in which to focus flood mitigation efforts.
In this study, the latest technology in remote sensing and geographic information systems has been proposed. Google Earth Engine (GEE) is a powerful cloud-based platform [12]. GEE can be used in detecting floods as it provides access to a large number of satellite images and geospatial datasets from various sources. Several researchers have conducted flood research utilizing the GEE database. Mehmood et al. [13] utilized the GEE cloud platform to detect floods using Landsat images. However, the limitation of optical images such as Landsat is that it is not possible to detect flood inundation during the rainy season [14]. Hence, in this study, Synthetic Aperture Radar (SAR) is utilized to detect flood inundation to gain the same advantage as previous studies. Nghia et al. [15] extracted flood information on the downstream Mekong River in Vietnam using SAR images in GEE. Tsuganskaya et al. [16] utilized SAR in the GEE platform to extract flood inundation information in the vegetated area. The advantage of GEE is that it is a cloud-based platform that provides high-performance computing capabilities, a wide range of image processing functions and algorithms that can be applied to satellite data, time-series analysis, which can compare images before and after a flood, and integration with auxiliary data [12]. Therefore, GEE has been used in this study because it has the potential to provide valuable insights and to support evidence-based decision-making for flood-related policies and interventions.
Moreover, SAR has been extensively utilized to detect flood inundation. Rimba and Miura [17] utilized ALOS PALSAR data to detect flood inundation for rapid response to flood disasters in Japan, and Tay et al. [18] employed SAR images to detect the impact of typhoons in Japan. Saleh et al. [19] utilized Sentinel-1 SAR imagery to detect flood inundation in Penang. Bernhard [20] employed 6 years of SAR data to monitor floods in Greece. SAR is intensively famous in flood-prone studies due to its capabilities. It is impossible to receive cloud-free images during rainy seasons [17,21,22,23]. SAR is not affected by cloud cover, fog, or atmospheric conditions, which can limit the usability of optical images during cloudy or rainy periods [17]. Moreover, it can penetrate through clouds and capture reliable data regardless of the weather conditions at night, ensuring continuous monitoring capability; it can also penetrate through vegetation to a certain extent and capture the underlying surface, including flooded areas [16,23]. These capabilities are beneficial when analyzing the flood extent in densely vegetated regions where optical images may be obstructed by the vegetation canopy [16,17]. Moreover, it is sensitive to the roughness of the water surface, allowing for differentiation between flooded areas and non-flooded areas based on the backscattering properties of water [17]. This sensitivity can aid in the accurate mapping of flood extents. SAR satellites often have shorter revisit times compared to optical satellites, which enables frequent monitoring and tracking of flood dynamics [24,25]. Hence, this capability is valuable for assessing flood progression, identifying temporal changes, and supporting timely decision-making. However, it is essential to note that SAR imagery has some limitations. The interpretation of SAR images may require specialized expertise due to the complex radar backscatter signals and speckle noise, and the data can also have lower spatial resolution compared to high-resolution optical images [17]. However, it is more powerful than other spectrums, especially during the monsoonal season.
The main objective of this study is to measure the extent of flood inundation during the floods in February 2023 and promote a policy recommendation. The flood map was compared to different physical factors (i.e., land cover, elevation, and slope) and flood prevention methods offered by the government (i.e., drainage systems) and the land use plans of the local government of Makassar City. Hence, the flood distribution according to different physical factors can be accessed and evaluated using the flood prevention method and future land use plan. Thus, the advantage of this study is that the local government will be able to design an appropriate policy to increase the resilience of the community against annual flooding and improve drainage systems, land use planning, early warning systems, and public awareness campaigns.

2. Materials and Methods

2.1. Study Area

Makassar, also known as Ujung Pandang, is the capital and largest city of South Sulawesi province in Indonesia; the city is known as the industrial harbor of east Indonesia. It is located on the southwestern coast of the island of Sulawesi, which is one of the largest islands in Indonesia, as shown in Figure 1. It is situated on the western side of the South Sulawesi peninsula. Makassar is primarily a coastal city, and its topography is characterized by low-lying coastal plains. The city is situated along the Makassar Strait, which separates Sulawesi from Borneo. The coastal areas are relatively flat and feature beaches. The city is traversed by several small rivers and streams that flow into the Makassar Strait [4]. These rivers have influenced the city’s development and provided freshwater resources.

2.2. Data Collection

Sentinel-1 data that were recorded on 9 February 2023 were utilized. Vertical Transmit-Vertical Receive polarization (VV) was utilized in flood detection [16,26]. VV polarization can penetrate through a thin layer of water, allowing SAR signals to reach the underlying surface [16]. This property is beneficial when trying to detect flooded areas beneath the water surface or when the floodwater is shallow [27]. The backscatter response of VV polarization from water is typically significantly lower than that of other land cover types, such as buildings, vegetation, or bare soil. This difference in backscatter values facilitates the discrimination of flooded areas from non-flooded areas. SAR data in VV polarization are not affected by weather conditions like cloud cover or precipitation, making them reliable for flood monitoring even during cloudy or rainy periods [17]. The sentinel images were made available in GEE. Moreover, a Hansen database in GEE was utilized to access the permanent water state [28]. By extracting the permanent image of water, the temporary inundation water was categorized as a flood.
The land cover was derived by averaging Landsat 8 images from 2022 in a cloud-based platform using supervised classification. The land cover classification was conducted by following the classes from the previous study [29]. The land cover was overlaid with the flood area to calculate the inundation area in specific land use types. The digital elevation data (DEM) data were derived to extract the elevation and slope. The drainage data and urban land use plan were created by the local government of Makassar City [30].

2.3. Image Processing

Image processing was applied to Sentinel-1 SAR GRD image data. The dataset was pre-processed using Sentinel-1 Toolbox for thermal noise removal, radiometric calibration, and terrain correction using SRTM 30. The final terrain-corrected values were converted into decibels via log scaling (10xlog10(x)). Moreover, the filtering was applied to remove the speckle noise, enhancing the algorithm for detecting flood inundation. The thresholding techniques to differentiate flooded areas from non-flooded areas were applied from the backscattering properties of VV polarization [16]. A histogram of the temporal water property identifies the thresholding value [31,32].

2.4. Data Integration and Validation

In the realm of flood analysis, the integration of geospatial data plays a pivotal role in improving our understanding of flood-prone areas and their potential impact. DEMs provide critical information about the terrain’s topography, aiding in the identification of low-lying regions susceptible to flooding. Land cover maps offer insights into the types of land surfaces, such as forests, urban areas, or water bodies, enabling us to assess how various land types influence flood propagation.
Additionally, drainage information is indispensable for comprehending the natural flow of water, allowing for better predictions of flood pathways and flood-prone areas. Urban land use plans are instrumental in understanding how human activities and infrastructure development may exacerbate or mitigate flooding risks in urban areas. Integrating these datasets into flood analysis greatly enhances our ability to model and predict flood events accurately.
In order to validate the flood distribution predictions generated by machine learning algorithms, a comprehensive approach was taken. This involved interviewing local citizens in flood-prone areas to gather firsthand accounts and utilizing digital platforms to access news reports and social media data related to recent floods. By comparing these real-world observations with the machine learning predictions, the overall accuracy of the flood models was assessed [33]. The total amount of classified data matched the reference data divided by the total number of samples.
The assessment revealed promising results in its overall accuracy. The classified data were provided from machine learning, and the reference data were from field surveys, digital news, and social media surveys. The land cover accuracy achieved an impressive 90%, although it should be noted that this accuracy might be affected by land use changes that occurred in the year between the data source and the survey. Furthermore, the flood inundation accuracy was 91%; this discrepancy can be attributed to the satellite data’s limited availability, which were collected only two days prior to the flood peak. The accuracy assessment can be seen in Table 1 and Table 2.

3. Results and Discussion

The study conducted an extensive analysis of the flood-prone region in Makassar during the February 2023 monsoonal flood. Advanced machine learning tools were harnessed, providing invaluable insights into the extent of the flooding. These tools played a pivotal role in generating a comprehensive flood inundation map, which serves as a critical resource for disaster management and mitigation efforts.
Furthermore, the study delved into the characterization of the land cover within the region. The detailed land cover map was created by employing cloud-based analysis and machine learning techniques. Moreover, the flood map was integrated with the elevation and slope map. This study helps to discern the different types of terrain and land usage patterns, aiding urban planners and environmentalists in their decision-making processes.
The most significant achievement of this study was the seamless integration of machine learning data with the local government’s information. This synergy yielded a holistic overview, encompassing the severe flood distribution, drainage system assessment, land cover analysis, and urban land use planning. The resulting amalgamation of data, as illustrated in Figure 2, is an invaluable resource for policy makers, urban planners, and disaster response teams, enabling them to make informed decisions and take proactive measures in mitigating future flood-related crises in Makassar.

3.1. Flood Inundation Area and Drainage

Figure 2a provides a comprehensive visualization of the flood inundation area, shedding light on the critical role played by drainage systems in mitigating these inundations. The map illustrates that drainage systems are instrumental in sub-districts such as Panakukang, Ujung Panjang, Mamajang, Mariso, and Rappocini, effectively preventing flood inundation. However, in other regions, a stark contrast emerges as they remain devoid of any drainage infrastructure, rendering them highly susceptible to flooding.
Figure 3, on the other hand, zooms in on the situation in Tamalanrea, where the impact of flooding is most pronounced. This sub-district experiences the most significant inundation, closely followed by the Biringkanaya, Tamalete, and Manggala sub-districts. A notable point of interest is that these areas lack interconnection through drainage systems, except for the Tamalate sub-district. Surprisingly, even though Tamalate is equipped with a drainage system, it continues to grapple with flooding issues, primarily attributed to a slope degree ranging from 0 to 0.1, as clearly depicted in Figure 2c.
It becomes increasingly evident that drainage systems assume a pivotal role in alleviating the consequences of flood inundations, as visibly demonstrated in Figure 2a and Figure 3. Notably, the Panakukang sub-district stands out as a prime example of effective flood management. This region boasts a well-connected drainage network extending over 233 m, leading to a substantially reduced inundation area. Even in areas with low-gradient slopes, as highlighted in Figure 2c, Panakukang remains relatively resilient in the face of flooding due to its robust drainage infrastructure.
The intricate interplay between drainage systems and flood inundation areas is vividly portrayed in Figure 3. While some areas benefit from well-designed drainage systems, others suffer the consequences of their absence, emphasizing the crucial need for comprehensive flood management strategies to safeguard vulnerable regions.

3.2. Flood Inundation Area and Land Cover

Figure 4 provides a compelling snapshot of the land cover, revealing that cultivation occupies the largest portion of the flooded land cover. To be precise, it spans an extensive 15.24 km2, constituting a staggering 43.2% of the total flooded area. Following closely behind it are water bodies, vegetation, and built-up areas, which account for 25.8%, 17% and 11.5%, respectively. This distribution, as illustrated in Figure 2b, is not uniform but concentrated predominantly in the Biringkanaya, Manggala, and Tamalanrea sub-districts.
The inundation in this study area presents a twofold dilemma. On the one hand, it leads to significant economic losses due to failed harvests, impacting local livelihoods. On the other hand, the inundation of built-up areas poses a grave threat to the population. An alarming statistic emerges from the field surveys: an average of over five people inhabit a single household in the affected regions. Consequently, the flood’s consequences in built-up areas are deeply intertwined with the well-being of the local population.
Furthermore, the extended duration of inundation in built-up areas can be attributed to impermeable surfaces and exacerbated by an inadequate drainage network. The resulting water entrapment hampers swift water flow, prolonging the period of flooding. This exacerbates the hardships faced by the affected residents and further emphasizes the importance of addressing this issue promptly.
In contrast, the impact of flooding on water bodies and vegetation, while not to be underestimated, may not directly translate into economic losses as significant as those incurred by cultivation and built-up areas. Nonetheless, a comprehensive strategy is imperative to mitigate these challenges across the spectrum of land cover types. It is evident that a multifaceted approach is needed to safeguard both the livelihoods and well-being of the local population.

3.3. Flood Inundation Area, Slope and Elevation

The extent of the flood inundation in a given area is significantly influenced by the terrain’s slope and elevation, as illustrated in Figure 5. Notably, this study examined slopes ranging from 0° to 0.3°. A substantial increase in the inundation area becomes evident when compared to steeper gradients exceeding 0.3°. This phenomenon can be attributed to the fact that the 0.2° to 0.3° slope range boasts the highest inundation levels in this study. This is primarily due to the extensive land cultivation practices occurring within this gradient range.
Focusing specifically on Makassar City, it is essential to highlight the vulnerability of the Tamalanrea and Manggala sub-districts to floods, as revealed in Figure 2c. These areas are situated on remarkably flat terrain, featuring slopes within the 0° to 0.3° range. Compounding the issue, the absence of a proper drainage network in these sub-districts exacerbates the flood risk.
Moreover, Figure 6 illustrates that the flooded regions are primarily concentrated in areas with elevations ranging from 0 to 8 m. These findings emphasize the critical role of slope and elevation in understanding and mitigating flood vulnerabilities in the region.

3.4. Flood Inundation Area and Urban Land Use Plan

In a forward-looking plan spanning from 2015 to 2035, Makassar’s city government has embarked on a comprehensive development strategy that envisions the transformation of the city into two distinct types of areas: cultural and protected. Protected areas encompass a diverse range of landscapes, including city forests, green zones, and the boundaries of lakes, beaches, and rivers. The cultural areas, in particular, are designated for a myriad of purposes, such as ports, industries, businesses, recreational activities, educational institutions, residential zones, military installations, healthcare facilities, and various other functions essential to supporting human life.
A significant facet of this development is revealed in Figure 7, illustrating Makassar’s predominantly urban land use plan with cultural areas taking the lead. However, a notable concern arises from the limited coverage of drainage systems in several sub-districts. This issue became acutely evident during the February 2023 floods, which inundated a staggering 29.02 km2 or 82.26% of the cultural areas.
In light of these challenges, it is imperative that this study advocates for heightened attention to be directed toward the development of drainage systems, particularly in the cultural areas of the Tamalanrea, Manggala, and Biringkanaya sub-districts. Recognizing that the cultural zones serve as the epicenter of socio-economic development, placing a sharper focus on their resilience against floods will undoubtedly yield substantial benefits in terms of flood management and urban sustainability. Ultimately, by fortifying the cultural areas, Makassar can aspire to achieve a more resilient and prosperous future.

3.5. Current Status of Flood Adaptation

The current flood adaptation strategy constructed by the local government is a drainage system, as shown in Figure 3. The sub-districts with the top three longest drainage systems were Panakukang, Tamalate, and Bontoala, with drainage lengths of 233 km, 87 km, and 25 km, respectively. The sub-districts with no drainage system coverage are Wajo, Ujung Tanah, Tamalanrea, and Biringkanaya. The Rappocini, Mariso, Mamajang, Manggala, Ujung Pandang, Makassar, and Tallo sub-districts were covered by drainage systems less than 6 km. Thus, over 14 sub-districts, only 3 sub-districts have sufficient drainage systems. The impact of this poorly integrated drainage system is severe flooding for areas with systems less than 6 km long or no drainage system. These conditions pushed the community to prepare a self-adaptation strategy by seeking refuge in a flood-free area in the event of a severe flood. However, approximately 65% of the affected community stays in their houses and prepares rubber tires as boats for emergency transportation if there is flooding greater than one meter [11].

3.6. Policy Recommendations

Makassar, a city on the rise, is fast becoming the largest industrial center in the eastern part of Indonesia. This remarkable growth has positioned Makassar as a potential economic leader in Asia. However, this rapid development has also led to intensive land use changes as a consequence of urbanization. The implications of such changes are far-reaching, particularly in terms of increased vulnerability to flood-related damage. To navigate these challenges effectively, the government must take proactive steps, primarily via meticulous urban land use planning and the establishment of a well-connected drainage network.
The city’s transformation into an industrial powerhouse has attracted a surge in population and economic activities. As more people flock to Makassar in search of opportunities, urbanization has surged, exerting immense pressure on land resources. This has translated into an unprecedented transformation characterized by the proliferation of residential, commercial, and industrial areas in the city’s landscape.
With this rapid urban expansion comes a heightened risk of flooding, especially during the rainy season. Inadequate drainage infrastructure coupled with haphazard urban development exacerbates the city’s susceptibility to flood-related damage. Without effective planning and a well-connected drainage network, Makassar faces not only the immediate threat of flooding but also long-term consequences such as infrastructure damage, economic losses, and public health hazards.
To safeguard the city’s future and foster sustainable growth, the government must prioritize urban land use planning. This entails zoning regulations, building codes, and infrastructure development that can withstand the challenges posed by urbanization. Moreover, a well-connected drainage network is essential for managing excess water during heavy rainfall and mitigating flood risks.
Furthermore, the local government should prioritize drainage system development because the local government has the crucial role of implementing drainage systems to mitigate flood impacts, prioritizing the development and maintenance of drainage systems in flood-prone areas. The focus should be on areas like the Tamalanrea, Manggala, and Biringkanaya sub-districts, which were identified as highly vulnerable due to the absence of drainage systems in them. Moreover, it is necessary to expand the drainage coverage by increasing the coverage of drainage systems in urban areas, particularly in locations with a high population density and significant economic activity. This includes the cultivation areas that were heavily flooded, causing economic losses. Expanding the drainage coverage will help reduce the impact of floods on agriculture and businesses. Flat-slope areas and low elevation areas are recognized as influencing flood vulnerability. Hence, the development of drainage systems in these areas with slopes ranging from 0° to 0.3° and elevations between 0 and 8 m should be prioritized. These flatter areas are more susceptible to flooding and require special attention to enhance flood resilience.
The local government should integrate flood risk reduction into urban land use planning. The new urban planning should ensure that cultural areas are designed to include effective drainage systems. This will not only protect cultural sites but also contribute to the overall flood resilience. Moreover, they should implement land use policies that discourage cultivation in flood-prone areas, especially in locations with inadequate drainage. By implementing these policy recommendations, Makassar City may significantly enhance its resilience to monsoonal floods and reduce the economic, social, and environmental impacts associated with such events.
Globally, numerous large cities successfully manage floods with well-connected drainage. Singapore, a small island nation with a tropical climate, experiences frequent heavy rainfall. To manage this challenge, the city-state has implemented a comprehensive drainage system known as the Marina Barrage. This system includes a dam that creates a freshwater reservoir and regulates water flow into the sea, thus preventing flooding while ensuring a stable water supply for the city [34]; likewise, Hong Kong has installed drainage systems for the prevention of stormwater-related disasters [34,35]. Another success story comes from Amsterdam, a city built partly below sea level, which is known for its sophisticated canal system. The city uses a combination of canals, locks, and sluices to manage water levels. Additionally, Amsterdam has developed advanced stormwater management techniques, such as green roofs and permeable pavements, to absorb rainwater and reduce the burden on its drainage infrastructure [36,37,38]. Additionally, Shanghai, a rapidly growing metropolis, has faced severe flooding in the past. To address the issue, the city has expanded its network of drainage canals and built large reservoirs. Suzhou Creek, once heavily polluted, has been transformed into a clean waterway that also serves as a flood control channel, protecting the city during heavy rains [39]. The most successful example comes from Tokyo, which is one of the most densely populated cities globally and prone to heavy rainfall and typhoons. The city has invested heavily in an intricate drainage system to combat floods. This system includes a network of underground tunnels, reservoirs, and canals designed to divert and manage excess rainwater. One notable example is the G-Cans Project, a massive underground reservoir that can hold up to 10,000 tons of water. This system has significantly reduced the occurrence of flooding in Tokyo [40]. These cities’ experiences illustrate how well-connected drainage systems and innovative stormwater management strategies are essential for large cities to protect against floods and ensure the safety and resilience of their communities. These infrastructure investments not only prevent property damage but also contribute to sustainable water resource management.

3.7. Study Limitation and Future Study

This study had a specific focus: assessing the physical flood inundation during the February 2023 flood season. A comprehensive survey was undertaken to achieve this goal, aiming to quantify the accuracy of land cover classification and determine the flood inundation areas. However, it is important to note that this analysis was solely dedicated to the physical aspects and did not take into account the socio-economic factors affecting the populations residing in flood-prone regions. Consequently, the study did not incorporate an assessment of the economic losses incurred during these flooding events. Moreover, this flood inundation study evaluated a single severe flood event. Identifying historical severe flood events would give more detail about flood-prone areas. Nevertheless, it is worth highlighting that this omission is deliberate, with the intention of addressing this critical aspect in future research endeavors. This holistic approach will offer a more comprehensive understanding of the overall impact of flooding, encompassing both the physical and economic dimensions.

4. Conclusions

Makassar’s ascent as an industrial giant in Asia brings immense promise, but it also carries the burden of intensified land use changes due to urbanization. To ensure a resilient future, the government must lead the way with strategic urban planning and investment in drainage infrastructure. Only then can Makassar continue to prosper while safeguarding its residents from the increasing threats of flooding and damage.
The integration of geospatial data and the validation process have substantially improved flood analysis and prediction. While there are challenges, such as data source gaps and timing constraints, the accuracy achieved is a testament to the potential of combining machine learning and geospatial information for more effective flood risk assessment and management.
This study focused on assessing the flood map in Makassar during the February 2023 monsoonal flood using machine learning tools. The flood inundation map revealed that several sub-districts, including Panakukang, Ujung Pandang, Mamajang, Mariso, and Rappocini, had drainage systems, while information for others was not available. Tamalanrea experienced the most significant flooding because it had no drainage system and more land used for cultivation, with inundation exacerbated by its low degree of slope. Drainage systems played a crucial role in mitigating flood impacts, with the Panakukang sub-district demonstrating effective drainage connectivity and less flooding despite its location in a flat area.
The study found that cultivation areas were the most heavily flooded land cover, causing economic losses due to failed harvests. Floods in built-up areas had a severe impact on the population, given the high population density, and prolonged inundation of impermeable surfaces. In this study, slope and elevation were identified as factors influencing inundation, with flatter areas more prone to flooding.

Author Contributions

A.B.R.: conceptualization, methodology, machine learning scripting, investigating, writing—original draft; A.A.: methodology, investigating, writing related to drainage and reviewing; I.P.W.U.: writing related to urban land use and reviewing; S.K.C.: writing related to policy recommendation and reviewing; M.N.B.: writing related to urban land use and reviewing; G.M.: writing related to policy recommendation and reviewing; K.T.S.: writing related to machine learning recommendation and reviewing; T.O.: writing related to machine learning and reviewing. All authors have read and agreed to the published version of the manuscript.

Funding

This research received no external funding.

Institutional Review Board Statement

Not applicable.

Informed Consent Statement

Not applicable.

Data Availability Statement

We utilized the dataset from Google Earth Engine, Makassar City government information, and field surveys.

Acknowledgments

Thank you very much to those who have provided valuable comments in their reviews, helping us enhance this study.

Conflicts of Interest

The authors declare no conflict of interest.

References

  1. Tanoue, M.; Taguchi, R.; Alifu, H.; Hirabayashi, Y. Residual Flood Damage under Intensive Adaptation. Nat. Clim. Change 2021, 11, 823–826. [Google Scholar] [CrossRef]
  2. Rentschler, J.; Salhab, M. People in Harm’s Way: Flood Exposure and Poverty in 189 Countries; Policy Research Working Paper; No. 9447; The World Bank: Washington, DC, USA, 2020; p. Policy Working Paper 9447. [Google Scholar]
  3. BNPB. Banjir Melanda Kota Makassar Sebanyak 1.869 Jiwa Mengungsi (Floods Hit Makassar City, As Many As 1,869 People Displaced). Available online: https://bnpb.go.id/berita/banjir-melanda-kota-makasar-sebanyak-1869-jiwa-mengungsi (accessed on 7 July 2023).
  4. BPS-Statistics of Makassar Municipality. Kota Makassar Dalam Angka 2023 (Makassar Municipality in Figure 2023); 73710.2302; BPS-Statistics of Makassar Municipality: Makassar City, Indonesia, 2023; p. 426. [Google Scholar]
  5. Widodo, T.N.; Zubair, H.; Padjung, R. Land Use Change Study and the Increased Risk of Floods Disaster in Jeneberang Watershed at Gowa Regency, South Sulawesi, Indonesia. IOP Conf. Ser. Earth Environ. Sci. 2021, 824, 012045. [Google Scholar] [CrossRef]
  6. Musliadi; Chaerul, M.; Gusty, S. Study on Flood Simulation of Tallo Watershed, Makassar City, South Sulawesi Province. J. Phys. Conf. Ser. 2021, 1899, 012063. [Google Scholar] [CrossRef]
  7. Thoban, M.I.; Hizbaron, D.R. Urban Resilience to Floods in Parts of Makassar, Indonesia. E3S Web Conf. 2020, 200, 01007. [Google Scholar] [CrossRef]
  8. Djamaluddin, I.; Indrayani, P.; Caronge, M.A. A GIS Analysis Approach for Flood Vulnerability and Risk Assessment Index Models at Sub-District Scale. IOP Conf. Ser. Earth Environ. Sci. 2020, 419, 012019. [Google Scholar] [CrossRef]
  9. Indrayani, P.; Djamaluddin, I.; Cai, Y. Employing a Local Framework and GIS to Evaluate the Flood Risk Index Maps of Makassar City, Indonesia. Arab. J. Geosci. 2023, 16, 568. [Google Scholar] [CrossRef]
  10. Sudirman; Trisutomo, S.; Barkey, R.A.; Mukti Ali, M. Watershed Identification and its Effect toward Flood (case study: Makassar CITY). Int. J. Adv. Res. 2018, 6, 513–519. [Google Scholar] [CrossRef]
  11. Halim, H.; Arifin, A.; Nonci, N.; Zainuddin, R.; Anriani, H.B.; Kamaruddin, S.A. Flood Disaster and Risk Anticipation Strategy. IOP Conf. Ser. Earth Environ. Sci. 2019, 235, 012032. [Google Scholar] [CrossRef]
  12. Google Earth Engine Google Earth Engine. Available online: https://earthengine.google.com (accessed on 7 July 2022).
  13. Mehmood, H.; Conway, C.; Perera, D. Mapping of Flood Areas Using Landsat with Google Earth Engine Cloud Platform. Atmosphere 2021, 12, 866. [Google Scholar] [CrossRef]
  14. Soria-Ruiz, J.; Fernandez-Ordoñez, Y.M.; Ambrosio-Ambrosio, J.P.; Escalona-Maurice, M.J.; Medina-García, G.; Sotelo-Ruiz, E.D.; Ramirez-Guzman, M.E. Flooded Extent and Depth Analysis Using Optical and SAR Remote Sensing with Machine Learning Algorithms. Atmosphere 2022, 13, 1852. [Google Scholar] [CrossRef]
  15. Nghia, B.P.Q.; Pal, I.; Chollacoop, N.; Mukhopadhyay, A. Applying Google Earth Engine for Flood Mapping and Monitoring in the Downstream Provinces of Mekong River. Prog. Disaster Sci. 2022, 14, 100235. [Google Scholar] [CrossRef]
  16. Tsyganskaya, V.; Martinis, S.; Marzahn, P. Flood Monitoring in Vegetated Areas Using Multitemporal Sentinel-1 Data: Impact of Time Series Features. Water 2019, 11, 1938. [Google Scholar] [CrossRef]
  17. Rimba, A.B.; Miura, F. Evaluating the Extraction Approaches of Flood Extended Area by Using ALOS-2/PALSAR-2 Images as a Rapid Response to Flood Disaster. J. Geosci. Environ. Prot. 2017, 5, 40–61. [Google Scholar] [CrossRef]
  18. Tay, C.W.J.; Yun, S.-H.; Chin, S.T.; Bhardwaj, A.; Jung, J.; Hill, E.M. Rapid Flood and Damage Mapping Using Synthetic Aperture Radar in Response to Typhoon Hagibis, Japan. Sci. Data 2020, 7, 100. [Google Scholar] [CrossRef] [PubMed]
  19. Saleh, A.; Yuzir, A.; Abustan, I. Flood Mapping Using Sentinel-1 SAR Imagery: Case Study of the November 2017 Flood in Penang. IOP Conf. Ser. Earth Environ. Sci. 2020, 479, 012013. [Google Scholar] [CrossRef]
  20. Bauer-Marschallinger, B.; Cao, S.; Tupas, M.E.; Roth, F.; Navacchi, C.; Melzer, T.; Freeman, V.; Wagner, W. Satellite-Based Flood Mapping through Bayesian Inference from a Sentinel-1 SAR Datacube. Remote Sens. 2022, 14, 3673. [Google Scholar] [CrossRef]
  21. Tarpanelli, A.; Mondini, A.C.; Camici, S. Effectiveness of Sentinel-1 and Sentinel-2 for Flood Detection Assessment in Europe. Nat. Hazards Earth Syst. Sci. 2022, 22, 2473–2489. [Google Scholar] [CrossRef]
  22. Foroughnia, F.; Alfieri, S.M.; Menenti, M.; Lindenbergh, R. Evaluation of SAR and Optical Data for Flood Delineation Using Supervised and Unsupervised Classification. Remote Sens. 2022, 14, 3718. [Google Scholar] [CrossRef]
  23. Brisco, B.; Shelat, Y.; Murnaghan, K.; Montgomery, J.; Fuss, C.; Olthof, I.; Hopkinson, C.; Deschamps, A.; Poncos, V. Evaluation of C-Band SAR for Identification of Flooded Vegetation in Emergency Response Products. Can. J. Remote Sens. 2019, 45, 73–87. [Google Scholar] [CrossRef]
  24. Pulvirenti, L.; Pierdicca, N.; Chini, M.; Guerriero, L. An Algorithm for Operational Flood Mapping from Synthetic Aperture Radar (SAR) Data Using Fuzzy Logic. Nat. Hazards Earth Syst. Sci. 2011, 11, 529–540. [Google Scholar] [CrossRef]
  25. García-Pintado, J.; Neal, J.C.; Mason, D.C.; Dance, S.L.; Bates, P.D. Scheduling Satellite-Based SAR Acquisition for Sequential Assimilation of Water Level Observations into Flood Modelling. J. Hydrol. 2013, 495, 252–266. [Google Scholar] [CrossRef]
  26. Kim, J.; Kim, H.; Kim, D.; Song, J.; Li, C. Deep Learning-Based Flood Area Extraction for Fully Automated and Persistent Flood Monitoring Using Cloud Computing. Remote Sens. 2022, 14, 6373. [Google Scholar] [CrossRef]
  27. Manjusree, P.; Prasanna Kumar, L.; Bhatt, C.M.; Rao, G.S.; Bhanumurthy, V. Optimization of Threshold Ranges for Rapid Flood Inundation Mapping by Evaluating Backscatter Profiles of High Incidence Angle SAR Images. Int. J. Disaster Risk Sci. 2012, 3, 113–122. [Google Scholar] [CrossRef]
  28. Hansen, M.C.; Potapov, P.V.; Moore, R.; Hancher, M.; Turubanova, S.A.; Tyukavina, A.; Thau, D.; Stehman, S.V.; Goetz, S.J.; Loveland, T.R.; et al. High-Resolution Global Maps of 21st-Century Forest Cover Change. Science 2013, 342, 850–853. [Google Scholar] [CrossRef] [PubMed]
  29. Rahman, M.; Ningsheng, C.; Mahmud, G.I.; Islam, M.M.; Pourghasemi, H.R.; Ahmad, H.; Habumugisha, J.M.; Washakh, R.M.A.; Alam, M.; Liu, E.; et al. Flooding and Its Relationship with Land Cover Change, Population Growth, and Road Density. Geosci. Front. 2021, 12, 101224. [Google Scholar] [CrossRef]
  30. BPPD Kota Makassar. RTRW Kota Makassar (Urban Land Use Plan of Makassar City) 2015-2034; BPBD Kota Makassar: Makassar City, Indonesia, 2015; p. 182. [Google Scholar]
  31. Long, S.; Fatoyinbo, T.E.; Policelli, F. Flood Extent Mapping for Namibia Using Change Detection and Thresholding with SAR. Environ. Res. Lett. 2014, 9, 35002–35009. [Google Scholar] [CrossRef]
  32. Voormansik, K.; Praks, J.; Antropov, O.; Jagomagi, J.; Zalite, K. Flood Mapping With TerraSAR-X in Forested Regions in Estonia. IEEE J. Sel. Top. Appl. Earth Obs. Remote Sens. 2014, 7, 562–577. [Google Scholar] [CrossRef]
  33. Lillesand, T.M.; Kiefer, R.W.; Chipman, J.W. Remote Sensing and Image Interpretation, 7th ed.; John Wiley & Sons, Inc: Hoboken, NJ, USA, 2015; ISBN 978-1-118-34328-9. [Google Scholar]
  34. Chan, F.K.S.; Chuah, C.J.; Ziegler, A.D.; Dąbrowski, M.; Varis, O. Towards Resilient Flood Risk Management for Asian Coastal Cities: Lessons Learned from Hong Kong and Singapore. J. Clean. Prod. 2018, 187, 576–589. [Google Scholar] [CrossRef]
  35. Chui, S.K.; Leung, J.K.Y.; Chu, C.K. The Development of a Comprehensive Flood Prevention Strategy for Hong Kong. Int. J. River Basin Manag. 2006, 4, 5–15. [Google Scholar] [CrossRef]
  36. Dierickx, W. Manual of Surface Drainage Engineering. Vol. II, Stream Flow Engineering and Flood Protection. Agric. Water Manag. 1985, 10, 184–185. [Google Scholar] [CrossRef]
  37. Van Der Nat, A.; Vellinga, P.; Leemans, R.; Van Slobbe, E. Ranking Coastal Flood Protection Designs from Engineered to Nature-Based. Ecol. Eng. 2016, 87, 80–90. [Google Scholar] [CrossRef]
  38. Nillesen, A.L. A New Nature-Based Approach for Floodproofing the Metropolitan Region Amsterdam. In Coastal Flood Risk Reduction; Elsevier: Amsterdam, The Netherlands, 2022; pp. 209–225. ISBN 978-0-323-85251-7. [Google Scholar]
  39. Sun, X.; Li, R.; Shan, X.; Xu, H.; Wang, J. Assessment of Climate Change Impacts and Urban Flood Management Schemes in Central Shanghai. Int. J. Disaster Risk Reduct. 2021, 65, 102563. [Google Scholar] [CrossRef]
  40. Nakamura, H.; Oosawa, M. Effects of the Underground Discharge Channel/Reservoir for Small Urban Rivers in the Tokyo Area. IOP Conf. Ser. Earth Environ. Sci. 2021, 703, 012029. [Google Scholar] [CrossRef]
Figure 1. Study area.
Figure 1. Study area.
Water 15 03783 g001
Figure 2. Comparison between flood inundation map and physical factors of flood. (a) Drainage, (b) land use/land cover, (LULC), (c) slope, (d) elevation.
Figure 2. Comparison between flood inundation map and physical factors of flood. (a) Drainage, (b) land use/land cover, (LULC), (c) slope, (d) elevation.
Water 15 03783 g002
Figure 3. Flood inundation (km2) and drainage length (km) per sub-district.
Figure 3. Flood inundation (km2) and drainage length (km) per sub-district.
Water 15 03783 g003
Figure 4. Total area of flooded land cover.
Figure 4. Total area of flooded land cover.
Water 15 03783 g004
Figure 5. Flood inundation (km2) based on slope (°).
Figure 5. Flood inundation (km2) based on slope (°).
Water 15 03783 g005
Figure 6. Flood inundation (km2) and elevation (m).
Figure 6. Flood inundation (km2) and elevation (m).
Water 15 03783 g006
Figure 7. Flood inundation map and urban land use plan.
Figure 7. Flood inundation map and urban land use plan.
Water 15 03783 g007
Table 1. Accuracy assessment of land use land cover.
Table 1. Accuracy assessment of land use land cover.
Land Use Land CoverReference Data
Built-UpCultivationWater BodyVegetationBare SoilTotal
Classified dataBuilt-up67000067
Cultivation54202251
Water body33480054
Vegetation22035039
Bare soil51015259
Total8248483854270
Overall accuracy(67 + 42 + 48 + 35 + 52)/2700.90
Table 2. Accuracy assessment of flood inundation.
Table 2. Accuracy assessment of flood inundation.
Object ClassificationReference Data
FloodNo FloodTotal
Classified dataFlood1214125
No flood146175
Total13565200
Overall accuracy(121 + 61)/2000.91
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Rimba, A.B.; Arumansawang, A.; Utama, I.P.W.; Chapagain, S.K.; Bunga, M.N.; Mohan, G.; Setiawan, K.T.; Osawa, T. Cloud-Based Machine Learning for Flood Policy Recommendations in Makassar City, Indonesia. Water 2023, 15, 3783. https://doi.org/10.3390/w15213783

AMA Style

Rimba AB, Arumansawang A, Utama IPW, Chapagain SK, Bunga MN, Mohan G, Setiawan KT, Osawa T. Cloud-Based Machine Learning for Flood Policy Recommendations in Makassar City, Indonesia. Water. 2023; 15(21):3783. https://doi.org/10.3390/w15213783

Chicago/Turabian Style

Rimba, Andi Besse, Andi Arumansawang, I Putu Wira Utama, Saroj Kumar Chapagain, Made Nia Bunga, Geetha Mohan, Kuncoro Teguh Setiawan, and Takahiro Osawa. 2023. "Cloud-Based Machine Learning for Flood Policy Recommendations in Makassar City, Indonesia" Water 15, no. 21: 3783. https://doi.org/10.3390/w15213783

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop