Makassar City, the capital of South Sulawesi Province, is the largest urban center in eastern Indonesia and ranks as the eighth-largest city in the country. Geographically, this city is situated in zone 50 S of the Universal Transverse Mercator (UTM) coordinate system, based on the World Geodetic System 1984 (WGS 84) datum, specifically located between 119°24′25.6″ E and 5°8′7.4″ S. Covering an area of 175.77 km², the city comprises 15 sub-districts and 153 wards (Statistics Indonesia, 2024). According to data from the Pompengan-Jeneberang River Basin Agency (Balai Besar Wilayah Sungai, BBWS), Makassar is passed through by two major rivers: the Jeneberang River to the south and the Tallo River to the north (Figure 1a). Makassar’s topography ranges from flat coastal plains with slopes of 0°–2° in the western and northern parts to hilly terrain with slopes of 3°–15° in the east. Lowland areas, situated at elevations between 0 and 40 m above sea level, are densely populated and dominated by residential settlements. In contrast, the eastern and upland areas are characterized by a mix of rice fields, plantations, and agricultural land, with only a small portion of the remaining forest cover (Figure 1b) (Djamaluddin et al., 2020). The city’s location, combined with rapid urbanization and limited green space, intensifies surface runoff and reduces infiltration, increasing flood risk.

Makassar’s climate is classified as tropical monsoonal, with a pronounced wet season. Based on the past decade of meteorological records, January is the wettest month with an average monthly rainfall of 369 mm, while August is the driest (Statistics Indonesia, 2024). Flooding often occurs during the rainy season, especially when heavy rainfall coincides with high tides. Several sub-districts experience recurrent inundations due to the city’s low-lying topography and inadequate drainage infrastructure. The situation is intensified by Makassar’s population density, which reached approximately 1.46 million in 2025, with an average of 8,300 people per square kilometer, which is above the threshold for a “very dense” urban area (>1,500 people/km²) (Statistics Indonesia, 2024).

Makassar’s vulnerability is part of a broader regional trend. Sulawesi, the fourth largest and third-most-populated island in Indonesia, is home to nearly 20 million people, with approximately 9 million residing in South Sulawesi Province. In the last 10 years, 2015 to 2024, floods contributed at least 30 percent of all natural disasters in the province (BNPB, 2025). The flood event of January 2019 was particularly severe, not only affecting Makassar City but also impacting other regions across South Sulawesi, including the Makassar Peninsula (Doyle et al., 2023). This extreme event highlights the urgency of understanding the hydrological dynamics and contributing factors behind flooding in Makassar as a basis for future mitigation planning.

In January 2019, Makassar experienced its most severe flooding event in over 2 decades. According to a report from the Indonesian Ministry of Public Works (PU), nine out of 15 Makassar’s sub-districts were affected. These areas were identified based on local community survey reports, reporting overflows coming from the Jeneberang and Tallo Rivers. Latos et al. (2021) described the flood because of prolonged heavy rainfall from 16:00 UTC on January 20 to 08:00 UTC on January 23, primarily affecting the southern Makassar Peninsula, which discharges into the Makassar Strait. Rainfall data recorded by the PU and the Meteorology, Climatology, and Geophysics Agency (BMKG) indicated that on 22 January 2019, the daily rainfall far exceeded the historical average daily annual maximum of 150 mm/day. Lengkese, Bawakaraeng, and Limbungan rainfall gauges recorded 329, 308, and 328 mm/day, respectively. The disaster, which affected an estimated area of 3,700 km² in the Makassar Peninsula, resulted in 53 fatalities, 47 injuries, and the displacement of over 14,000 people across South Sulawesi Province. Within Makassar City alone, there was at least 1 recorded fatality, 9,328 affected individuals, and 1,000 displaced residents (Doyle et al., 2023).

Several studies have identified contributing factors to the flood, including sedimentation in rivers and reservoirs, unregulated mining activities, unregulated land cover changes, and increased landslide risk in upstream areas (Nurdin et al., 2019). The Bili-Bili Reservoir, which receives inflow from the Jeneberang River originating from the erosion-prone Bawakaraeng mountain, had a substantial role during the event (Dhanio et al., 2008). The reservoir experienced extreme inflow, pushing its storage capacity near the operational limit. Water levels reached 101.87 m, just below the maximum allowable level of 103 m, which put the reservoir in a warning state. In response, the reservoir spillway gates needed to be fully opened to prevent overtopping. The recorded spillway discharges ranged from 1,200 m³/s to the reservoir’s maximum spillway capacity of 2,200 m³/s.

In addition to the Bili-Bili outflow, the Jenelata River contributed a peak discharge of approximately 1,200 m³/s during the event, compounding downstream flood conditions in Makassar (Kompas.id, 2019). According to BNPB reports, inundation depths ranged from 1.5 to 4 m in affected areas, causing extensive damage to infrastructure, homes, and agricultural lands. A total of 32 houses were completely swept away, 25 were severely damaged, 14 were moderately damaged, and 55 were buried by landslides. In addition, 2,694 houses and 114.33 km² of farmland were inundated. Numerous public facilities, roads, and bridges suffered significant structural damage, further worsening the impact on affected communities.

In this study, these datasets were used to support hydrological and hydraulic modeling. Additionally, a historical flood inundation map was utilized to validate the simulation results. Hydrological data, particularly rainfall data, were also essential for driving the flood discharge scenarios. The types of data, their descriptions, specifications, and sources used in this study are summarized in Table 1. Topographic data were derived from the FABDEM, a global Digital Elevation Model (DEM) product with forest and building features removed. FABDEM has a spatial resolution of 1 arc-second (∼30 m) (Hawker et al., 2022; Uhe et al., 2022). To improve the accuracy of urban topography, the FABDEM dataset was complemented with cross-section data obtained from the PU and building footprint data from OpenStreetMap (OSM-B) (Haklay and Weber, 2008; OpenStreetMap, 2025). The integration of FABDEM and OSM-B layers allowed for a more realistic representation of Makassar’s densely built urban environment, which is crucial for simulating flood behavior in city settings.

Hydrological data in this study consisted primarily of rainfall data. Precipitation information was obtained from the GSMaP product, developed by the Japan Aerospace Exploration Agency (JAXA). This data offers consistent and high-resolution spatiotemporal coverage (Kubota et al., 2020), which is especially valuable in data-limited settings such as Makassar. Specifically, version 7 of GSMaP was used, which provides high-resolution satellite-based rainfall estimates. To capture the temporal dynamics of the January 2019 extreme flood event, hourly rainfall data for the period of 20–26 January 2019 were extracted using a custom script (Rohmat, 2024). This period was selected to represent conditions before, during, and after the peak flood event. The rainfall data were then used in HEC-HMS to simulate runoff and generate river discharge inputs for the flood modeling scenarios. The historical flood inundation map was obtained from BBWS Pompengan–Jeneberang. This map, which represents observed flood extents during the January 2019 event, was used as a benchmark to validate the model outputs. It is important to note that the information used to develop this map was gathered from eyewitness reports, based on observations of flood inundations in their respective areas.

This study employed an integrated hydrologic–hydrodynamic modeling approach to simulate the January 2019 flood event. The hydrologic simulation was carried out using HEC-HMS version 4.1.1 (USACE Hydrologic Engineering Center, 2024a). HEC-HMS is a widely used physically based model developed by the U.S. Army Corps of Engineers (USACE), it processes rainfall, evapotranspiration, and watershed characteristics such as area, slope, and land use to simulate runoff generation, infiltration (using the SCS Curve Number method), and hydrograph transformation (using the SCS Unit Hydrograph method). Flow routing is also incorporated to simulate water movement through channels and reservoirs, accounting for interflow and baseflow contributions (Table 2) (Abdelkebir et al., 2024; Belina et al., 2024; Guduru and Mohammed, 2024; Rohmat et al., 2025). HEC-HMS was used to model the hydrologic response of the Tallo, Jenelata, and Jeneberang Rivers. Special attention to the Bili-Bili Reservoir, which regulates outflows into the Jeneberang River. FABDEM terrain data were used to extract watershed parameters such as flow paths and time of concentration. GSMaP rainfall intensity, distribution, and duration were critical for simulating runoff generation. Technical information on the Bili-Bili Reservoir (Table 3) was also integrated to simulate its operation during the event. The resulting discharge hydrographs were then used as upstream inputs for hydraulic simulation using HEC-RAS.

The hydraulic modeling was conducted using HEC-RAS version 6.4.1, which supports two-dimensional unsteady flow simulation (USACE Hydrologic Engineering Center, 2024b). HEC-RAS is widely applied in flood studies for its capabilities in modeling water surface profiles, sediment transport, and hydrodynamic processes (Pinos and Timbe, 2019; Garcia et al., 2020; Rohmat et al., 2022; Peker et al., 2024; Pratiwi et al., 2024; Zainal and Talib, 2024). In this study, the terrain model was developed by integrating FABDEM elevation data with OSM-B data. Since OSM-B lacks elevation attributes, average building heights were estimated using zonal statistics derived from FABDEM. A uniform height of 4 m was added to represent typical single-story structures. The combined dataset was resampled to a 1-m resolution to improve spatial accuracy in urban areas.

This modified terrain enabled HEC-RAS to simulate more realistic flood propagation by accounting for the presence of buildings, ensuring that flow is routed around structures rather than through them. Such detail is particularly important in densely built environments like Makassar, where urban features significantly influence water movement and inundation patterns. The resulting flood model geometry included 2D flow areas, break lines, terrain elevations, and unsteady flow boundary conditions derived from HEC-HMS outputs. Figure 3 illustrates the difference between the original FABDEM and the modified version incorporating OSM-B. Figure 3b highlights elevated features corresponding to buildings, capturing Makassar’s urban complexity and allowing for improved simulation of flood extents and depths.

The hydraulic simulation in this study was configured using HEC-RAS 2D using the Shallow Water Equations using the Explicit Leapfrog Method (SWE-ELM), which is optimized for computational speed without sacrificing solution stability or accuracy. Time steps were dynamically adjusted based on the Courant number to ensure numerical stability and convergence throughout the simulation domain. Flow resistance was represented using variable Manning’s roughness coefficients assigned according to land cover classifications, which were extracted from spatial land cover datasets. Infiltration losses were modeled using the SCS Curve Number method, consistent with the hydrologic modeling approach in HEC-HMS. The computational domain was discretized using a curvilinear mesh, which allows for better alignment with natural and urban terrain features, improving model realism in areas with complex topography. Accurate flood modeling requires comprehensive data on flood inundation parameters. The selected hydraulic model setup parameters used in HEC-RAS 2D are summarized in Table 4.

Flood modeling in study area was conducted in two main stages: (1) simulation of the observed January 2019 flood, and (2) simulation of various discharge scenarios. The first stage simulated a 7-day period corresponding to the January 2019 flood, and the resulting inundation extents were validated using the PU flood inundation map. Validation aimed to ensure that the model represented observed flood conditions. Once the model was validated, it was used to simulate hypothetical discharge scenarios based on Q20. As outlined in Table 5, Q0 represents a condition in which no water flows in the river (dry conditions), while Q20 refers to flow conditions equivalent to a 20-year discharge event. The selection of Q20 follows PU guidelines, which suggest that rivers passing through provincial capitals should be designed for at least Q20 discharges. The rivers modeled in these scenarios include the Tallo River, the Jenelata River, and the Bili-Bili Reservoir outflow through the Jeneberang River. The purpose of these scenarios was to assess the relative contribution of each river system to urban flooding and to evaluate the effectiveness of the Bili-Bili Reservoir in mitigating flood impacts.

Hydrologic simulations were conducted to convert rainfall data into river discharge, providing essential input for flood modeling. The primary outputs of this stage are the discharge hydrographs for the Tallo and Jenelata Rivers, as shown in Figure 4. These hydrographs illustrate the temporal variations in flow during the flood event and highlight the differing responses of each river system to extreme rainfall. The discharge results from HEC-HMS were subsequently used as input for the hydrodynamic simulations in HEC-RAS, enabling a detailed spatial analysis of flood propagation during the January 2019 event. To ensure the reliability of the model, the simulation results were compared with existing reports. Notably, the peak discharge simulated for the Jenelata River, which was approximately 1,000 m³/s, was consistent with the report from BBWS Pompengan–Jeneberang (Kompas.id, 2019), indicating that the HEC-HMS model was suitable for further hydraulic simulation. Following the validation of the hydrologic model, additional simulations were performed using a 20-year return period discharge (Q20) under multiple scenarios to evaluate the potential impact of future flood events of greater magnitude.

The hydraulic simulation was validated by comparing the modeled flood inundation extents with the official flood inundation map for the January 2019 event, provided by the PU. The historical map displays flooded areas using a red hatch (Figure 5), which serve as reference points based on eyewitness reports collected during the event. Locations without red squares were considered unaffected by flooding during that time. The simulation results from HEC-RAS are shown on the same map using shades of blue, where darker shades indicate greater flood depth (Figure 5). The validation showed a strong agreement between the simulation and reported flood extents in three sub-districts: Manggala, Tamalanrea, and Biringkanaya (labeled 1, 2, and 3), suggesting that the model accurately represented riverine flood mechanisms in these areas. However, discrepancies were observed in six other sub-districts, i.e., Panakkukang, Tallo, Rappocini, Tamalate, Mamajang, and Makassar (labeled 4–9), where the simulation did not fully match the reported inundation extents.

The differences in the result point to the possibility that flooding in these locations may not have been caused primarily by river overflow, or there are discrepancies between the report and the actual flood, which comparably reported in the similar case study region (Rohmat et al., 2022). The spatial distribution of the discrepancies suggests that the observed flood events were likely driven by urban drainage system limitations rather than fluvial processes. Localized flooding in these areas could have resulted from high-intensity rainfall exceeding the capacity of the drainage network, leading to surface water accumulation independent of river discharge levels. This interpretation is supported by the non-correlation between observed flood locations and river-based inundation patterns. To confirm this hypothesis, a separate analysis incorporating urban drainage infrastructure would be required to assess system performance during extreme rainfall conditions.

Further discrepancies were identified in rural or non-urban zones, particularly agricultural areas such as rice fields, as shown by the yellow squares in Figure 6. While the model indicated inundation in these areas, the official flood map did not report any flooding. This inconsistency is likely due to underreporting, as rural flooding often lacks direct observation or formal documentation. Additionally, rice fields are routinely inundated during the planting season (October–March) (Susilowati, 2017), making it difficult to distinguish between agricultural water management and actual flood events. This highlights a common limitation in flood reporting systems: rural and agricultural areas are frequently overlooked, as official attention is typically focused on residential and urban zones. Consequently, the absence of recorded data from these areas can lead to discrepancies during model validation. This underscores the importance of improving ground-based flood observations and validation protocols, especially in rural landscapes, to ensure that modeling results accurately reflect real-world conditions.

Following model validation, further simulations were conducted using modified discharge inputs to identify the primary contributors to flooding in Makassar City and assess the role of the Bili-Bili Reservoir in mitigating flood impacts. Eight scenarios were tested using a 20-year return period discharge (Q20), as detailed in Table 5. The simulation results are presented in Figure 7, showing a range of inundation depths from 0 to 1.5 m, with “max” areas indicating depths exceeding 1.5 m. Among the eight scenarios, four (Scenarios 2, 5, 6, and 8) resulted in substantial inundation. These scenarios all involved Q20 discharge from the Tallo River, strongly indicating that the Tallo River is the dominant contributor to flooding in Makassar City.

A comparative analysis of Scenarios 3 and 5 (no release from the Bili-Bili Reservoir) versus Scenarios 7 and 8 (with reservoir releases) highlights the reservoir’s influence on downstream flooding. While Scenarios 3 and 5 show relatively limited inundation, Scenarios 7 and 8 simulate wider flood extents, particularly in coastal and nearshore areas. These findings highlight that while the Bili-Bili Reservoir is operational, its current management may be suboptimal for effective flood control. Factors such as the timing and volume of releases or coordination with downstream channel capacity may be limiting its effectiveness.

Further insights can be drawn from scenarios where the Jenelata River was excluded (Scenarios 2, 4, and 6). Scenario 2, which excludes both the Jenelata River and Bili-Bili outflow, shows no inundation along the Jeneberang River, indicating the downstream area is primarily affected by reservoir and tributary flows. In contrast, Scenarios 4 and 6 include controlled releases from Bili-Bili and show inundation along the Jeneberang. With the addition of flow from the Jenelata River in Scenarios 7 and 8, flooding extends toward the coastal zones, confirming the cumulative effect of multiple sources on downstream flood severity.

The patterns observed in Scenarios 7 and 8 emphasize that although the Bili-Bili Reservoir plays a key regulatory role, it alone is insufficient under extreme conditions. This calls for reevaluation of reservoir operations and consideration of complementary measures such as upstream retention basins or diversion channels. Furthermore, the Jenelata River should be included in future flood mitigation planning due to its significant contribution during high rainfall events. Assessing the Jeneberang River’s capacity to accommodate combined flows is also critical, particularly for protecting low-lying coastal areas vulnerable to sedimentation, subsidence, and tidal influences (Isma et al., 2024; Nugroho et al., 2024).

Scenario 8 produced the most severe flooding, resulting from the combined effects of Tallo River discharge, Jenelata River flow, and Bili-Bili reservoir releases. This confirms the Tallo River as the most influential flood source in the city. Therefore, mitigation efforts should prioritize both the Tallo and Jenelata Rivers. Additionally, further research into urban drainage capacity is necessary to identify other flood mechanisms, particularly pluvial flooding caused by insufficient stormwater infrastructure (Liu et al., 2022; Zhang et al., 2024).