Numerical simulation and redesign of Kamar Kajang Levee against Mount Semeru lahar flood based on HEC-RAS 2D non-newtonian flow model

Mount Semeru is one of the most active volcanoes in Indonesia where the pyroclastic, debris, and mudflows from its activities have severely damaged its surrounding areas. On December 4, 2021, Mount Semeru erupted, followed by intense rainfall, and damaged the Kamar Kajang Levee after a lahar flood event. The collapsed Levee was located along the Leprak Lumajang River, one of the natural drainage systems for Mount Semeru. The dyke failure had led to severe economic loss. Thus, reconstruction of the training dyke is highly important. In response, this research aims to propose an improvement to the Kamar Kajang Levee. This study was done through field observation, numerical simulation, and levee design calculation. The Non-Newtonian algorithm in HEC-RAS 6.2 was employed to model the flood events. The 2D model has been verified using a Sentinel-2 image and has shown a 90.7% similarity of inundated area after flooded by at least 214 m3/s of flood discharge, equivalent to a 5-year design discharge. The engineering design analysis shows that an additional 3.5 m height is needed to anticipate future floods and mitigate events effectively.


Introduction
Indonesia is located in the Pacific Ring of Fire, which stretches from the northern islands of Sumatra and Java to Nusa Tenggara and Sulawesi.This geographical positioning makes Indonesia home to numerous volcanoes and makes it vulnerable to volcanic eruptions.Disasters caused by volcanic activity occur directly during the eruption and secondarily, such as lahars and ashfall, which can persist for several days [1].Volcanic eruption leads to the deposition of residual eruption materials, and when rainfall occurs, these deposits on steep slopes could be carried away, resulting in lahars or debris floods [2], [3].
Lahars or debris floods present a formidable and destructive nature that poses a significant threat to human existence.Consequently, proactive measures must be taken to anticipate potential disasters from these events.The development of forecasting systems and early warning systems serves as an effort to anticipate the possibility of debris flood disasters if they occur at any time [4].
One of the active volcanoes in Indonesia that frequently exhibits volcanic activity is Mount Semeru [5], [6].This mountain erupted on December 4, 2021, at 15:20 local time [7].Unfortunately, it was followed by an intense rainfall.Thus, the combined forces of the volcanic eruption from Mount Semeru and the intense precipitation overwhelmed the levee's capacity to hold the lahar flow in the Leprak River, leading to the collapse of the Kamar Kajang Levee.The collapse of the Kamar Kajang Levee not only resulted in immediate damage but also emphasized the urgent need for repairs and reinforcement to enhance its resilience against future lahar flood occurrences.This incident serves as a stark reminder of the critical importance of continuous monitoring, maintenance, and improvement of levee systems to safeguard communities and mitigate the devastating impacts of natural disasters.
A scientific report described that sediment deposition on Mount Semeru does not experience a significant decline in the first year following an eruption.This phenomenon could be attributed to the continuous release of sediments through daily pyroclastic flows that envelop the mountain's summit cone.Subsequently, during the rainy season, these sediments are carried through the three main natural drainage systems of Mount Semeru: Glidik River, Rejali River, and Mujur River [8].
One of the natural drainage systems in the Rejali watershed is Leprak River.It importantly serves to carry lahars resulting from heavy rainfall on Mount Semeru.To anticipate and mitigate the destructive consequences of lahars in the Leprak River, levees have been constructed at vulnerable points to prevent debris flood overflow, one of which is located in Kamar Kajang, Candipuro District, Lumajang Regency (Figure 1).The Kamar Kajang lahars levee plays a critical role in protecting the residents of Candipuro.Built in 1990, it safeguards the residents' lands, primarily used for plantations and rice fields, from lahar flows.As the levee collapsed during the event on December 4, 2021, reconstruction of this levee is needed.
This study aimed to analyze the reconstruction of the Lahar Levee in Kamar Kajang.It assessed the current condition and capacity of the levee and identified potential improvements that can be implemented to mitigate the risks associated with volcanic lahars.The research began with modeling lahar flow in the Leprak River, Lumajang, to predict the potentially inundated areas during lahar floods.Subsequently, engineering improvements are implemented on the Kamar Kajang Levee to anticipate potential disasters in the future.These modeling and engineering efforts would be mitigation measures against lahar floods by predicting the characteristics of the floods, enabling the identification and controlling of high-risk or hazardous areas along the Leprak River.

Materials and Methods
The research was conducted through a combination of numerical simulation using HEC-RAS 2D and field observations in the impacted Kamar Kajang area, which suffered a devastating lahar flood following the collapse of the levee.This integrated methodology facilitated a comprehensive examination and analysis of the hydrological dynamics and hydraulic behaviour in response to the lahar event.

Field Observation
Field observations were conducted on August 25, 2022, in the Leprak River to gather valuable information and evidence related to the previous lahar flood event on December 4, 2021.Although the site visit was carried out more than half a year after the eruption, the damage and impact of the flooding were still visibly present.
During the field observation, temporary repair works were observed to address the collapsed levee and restore the riverbed to its normal state.However, the absence of a more permanent structure was notable.As observed in Figure 2 and Figure 3, the levee remained buried under extensive sediment, spanning across acres of the river's vicinity.These temporary repair works represented the initial efforts to stabilize the river and mitigate future flood risks.Further measures and long-term solutions were required to ensure the resilience and proper functionality of the levee system.
Based on the field observations, it became evident that the Leprak River typically remained dry in the absence of rainfall events around the peak of Mount Semeru.However, even with moderate rainfalls occurring near the mountain, the water level in the river rose rapidly.This observation underlined the significant influence of rainfall events on Mount Semeru's peak in triggering a sudden and substantial water elevation in the Leprak River.Combining the field observations and the analysis of aerial photos and videos captured in the immediate aftermath, a more comprehensive understanding of the impact and the recovery efforts were captured.These various sources of data contributed to a deeper understanding of the flood event, the condition of the levee, and the need for both temporary and long-term measures to mitigate future flood risks.

Hydrological Analysis
A. Catchment Area.The boundary of the catchment area was obtained through data processing of the Digital Elevation Model (DEM) of Lumajang Regency, which was published by Badan Informasi Geospatial (BIG).The control point for the catchment area, located at the Geladak Perak Bridge upstream of the levee, is shown in Figure 5.Using ArcGIS 10.8.2, the DEM data was processed, resulting in the delineation of the Leprak Watershed with a main river length of 12.06 km and a total area of 25.8 km2.The land cover was estimated by referring to the Indonesian Land Cover Map published by the Geospatial Information Agency or through satellite image delineation.In this study, land cover was determined by delineating SENTINEL-2 imagery taken on July 24, 2021, as depicted in Figure 6.The lahar flood examined in this study occurred on December 4, 2021, which was the rainy season.During this period, the soil becomes wet, reducing its ability to absorb rainfall.Therefore, the Curve Number value needed to be adjusted using the Soil Conservation Service Curve Number (SCS-CN) method [10].For wet conditions, the Curve Number could be corrected using the CN(III) formula as in equation (1).
where The Nakayasu Synthetic Unit Hydrograph (SUH) was employed to calculate the flow discharge.This method was favored for its simplicity and requirement of minimal input parameters, while still providing accurate estimates of flood discharge comparable to other techniques [11].Furthermore, the Nakayasu SUH method has been successfully utilized in previous researches [12], [13] to model different types of floods, including regular floods, flash floods, and debris floods.To determine the Nakayasu Synthetic Unit Hydrograph, several parameters were required, as presented in Table 2. Before calculating the discharge, the rainfall depth should be determined.In this study, the rainfall depth was estimated using the available rainfall data from the Kobokan and Kamar A rainfall stations, located within the Rejali Watershed.The data from these stations were obtained from the River Basin Management Agency (Balai Besar Wilayah Sungai) Brantas.
Once the rainfall depth for the area is determined, the Flood Discharge was then calculated based on the Synthetic Unit Hydrograph and Effective Rainfall of the Leprak Watershed.The rainfall distribution time was estimated to be 6 hours, with the hourly rainfall distribution for each time interval presented in Table 3.
Based on the distribution of hourly rainfall and the Nakayasu Synthetic Unit Hydrograph (SUH), the flood discharge for each return period in the Leprak River is displayed in Figure 7.The lahar concentration was calculated using Takahashi's debris concentration equation (2) as follows [14] : where:   is sediment concentration,  is water density (kg/m 3 ),  is channel slope (ᵒ),  is sediment density (kg/m 3 ) and  is friction slope (ᵒ).The river parameters used to calculate the Cd value are presented in Table 4 The previously obtained flood discharge (Q) was then used to calculate the debris flow discharge (Qd) using the equation ( 3) [14] : ) where α represents the sediment content coefficient which was calculated using equation ( 4).
where C* is the volumetric sediment concentration in the deposited debris flow.With a value of C* set at 0.65, the sediment content coefficient (α) is determined to be 1.857.The values of Qd obtained for each return period can be seen in Figure 8.

Model Data and Parameters A. Domain Setup
The model utilized the Non-Newtonian fluid algorithm in the HEC-RAS 6.2 software.This approach allows for the accurate modelling of lahar or debris flow areas [15].The flow domain was modeled based on 1-meter LiDAR data recorded after the levee collapse, capturing sediment deposits from the aftermath of the December 4, 2021 lahar flood event.Additionally, 10-meter DEMNAS data was used to estimate the pre-collapse condition of the levee.The model was constructed using a base cell size of 40 meters, which was further refined to a 10-meter resolution in the river area and the impacted Kamar Kajang area using refinement regions.Breaklines with a cell size of 5 meters were also included along the Levee to simulate overtopping events accurately.

B. Non-newtonian characteristics of the lahars
This model does not possess data regarding the concentration or rheology of the lahar flow during the actual disaster event.Obtaining such data during the occurrence of a lahar flood would be extremely challenging and hazardous.Instead, the characteristic parameters of the lahar utilized in this study are based on research conducted by [16], which measured the rheological conditions in the Lengkong River, part of the Rejali Watershed.Based on the study, it can be determined that the appropriate non-Newtonian flow model is the Generalized-Herschel Bulkley model with the parameters: K = 2.76, n = 0.5, and τy = 0.0239.

C. Calibration Data
The simulation result was then compared to the deposited sediment extent captured in the SENTINEL-2A satellite imagery on June 29, 2022.The reason this satellite image is that the sediment coverage is visible and free from cloud cover.Although there is a considerable time gap between the flood event and the satellite image acquisition date, the satellite data still indicates the boundaries of the flood-affected area that resemble the lahar flood event in the Kamar Kajang area.To obtain the boundaries of the remaining sediment area, the SENTINEL-2A imagery was processed using interactive supervised image classification in ArcGIS 10.8.2.The classification is conducted by aligning the researcher's visual perception to the imagery processing software.Considering the classification process algorithm's sensitivity, a good ability to select training samples of the area is required so the software is able to effectively interpret the classification [17].

Model Verification and Simulation
During the flood event on December 4, 2021, the lahar initially filled the river channel and overflowed due to the abundance of sediment from the eruption that occurred on the same date.When the lahar overflowed, a collapse occurred due to overtopping at one point of the Kamar Kajang Levee.After the levee collapsed, the lahar flow spread to the back of the levee and swept over the surrounding paddy The DEMNAS data used to model the lahar flow before the levee collapse had a relatively low resolution, resulting in an inadequate representation of the levee condition.Therefore, modifications to the DEM were necessary to accurately model the levee before the collapse.The DEMNAS data are also prone to error as the river bed captured in DEMNAS data may not represent the river bed elevation at the time of the disaster, so the model that was developed is a rough estimation.Based on experiments, it was found that overflow occurred at the 5-year return period discharge, as shown in Figure 10.Subsequently, the 5-year return period discharge was used to simulate the post-collapse condition of the Levee.
The creation of the model for the post-collapse condition of the Levee is utilizing LiDAR data.This choice was made because the captured sediments in the LiDAR data represented the accumulated sedimentation before and after the collapse of the levee.Consequently, the modelling efforts primarily focused on the impacted area located at the Kamar Kajang side of the levee.The simulation results for both pre-collapse and post-collapse levee conditions were then combined and compared with the sediment area recorded on June 29, 2022.Based on Figure 12, the results sufficiently depicted the impacted area of the lahar flood.The estimated total area of flow during the event was 358.49hectares, with 107.34 hectares in the Kamar Kajang area.In contrast, the simulation results indicated a flow area of 325.29 hectares, with 121.59 hectares in the Kamar Kajang area.Overall, the simulation successfully captured 90.7% of the recorded sediment area in the satellite image.This demonstrates that the parameters used in the simulation model are applicable for planning purposes.The simulation result can also be expanded in future research by modelling the sediment accumulation with the respective flow model.

Proposed Levee Reconstruction Design
After conducting the simulation, we developed a comprehensive design for the reconstruction of the Levee.We aim to propose an effective reconstruction design that is tailored to address the specific challenges and requirements of the current situation.The design considers various factors such as the optimal location, dimensions, and other relevant considerations.By conducting a simulation of the proposed design, we can ensure that the design will be capable of mitigating future flood events and safeguarding the surrounding areas.

A. Location
The proposed Levee is located at coordinates 113.0253°E, -8.1816° S to coordinates 113.0341°E, -8.1906° S. The total length of the Levee is 1488 meters.This location corresponds to the current levee system.The reasons to choose this location are motivated by the intention to utilize and improve the existing levee system, which is deemed more cost-effective compared to constructing a new system in a different location.

B. Dimension
The dimensions of the Levee are calculated according to the guidelines provided by the Ministry of Public Works and Housing's Construction and Building Standards Pd T-16-2004-A.These guidelines consider the appropriate safety factors required for water infrastructure projects.The 100-year return period discharge was used to design the Levee.Based on the calculation, it is determined that an additional 3.52 m, including a 0.8 m freeboard, needs to be added to the existing height of 10 m, resulting in a total height of 13.52 m.To support this height addition, an extension of the base length is also required, resulting in a total length of 50.32.

Figure 1 .
Figure 1.Map of the study area (yellow line on the right panel shows the center line of Kamar Kajang Levee)

Figure 2 .Figure 3 .
Figure 2. Satellite image (SENTINEL-2A) showing the location of the collapsed Kamar Kajang Levee, acquired on June 29, 2022 (a yellow line represents the levee's centerline) Figure 3.The collapsed levee is still buried in sediment, taken on August 25, 2022 (a yellow line represents levee's centerline)

Figure 5 .
Figure 5. Watershed control point and the Catchment Area

Figure 8 .
Figure 8. Debris flood hydrograph of 2 to 100 years return period

Figure 9 .
Figure 9. Sediment deposit captured on June 29, 2022 (The sediment extent is depicted by a black border) areas.In this study, two models were developed for verification purposes: one representing the condition of the Levee before the collapse, and another simulating the post-collapse scenario.

Figure 12 .
Figure 12.Comparison of the simulation results (yellow hatch) to satellite-captured sediment extent (black line)

Figure 13 .
Figure 13.Typical cross-section of the proposed levee reconstruction designC.Simulation of the proposed designNext, the lahar flow simulation is conducted using the proposed levee design that has been calculated.In this simulation, the geometry is based on the existing LiDAR DEM data at the levee location.The DEM data needs to be modified in the HEC-RAS software to incorporate the planned levee elevation, which is an additional 3.52 m.Elevation control points are established every 50 m to maintain the desired height of the levee according to the design.The simulation results indicate that the levee can withstand

Table 1 .
CN(III) is the final Curve Number value applied in further analysis, and CN represents the Curve Number for each different land cover type.After applying the correction, the calculated Curve Number for the study area is 86.53.The computed values of the Curve Number are presented in Table 1.Composite Curve Number

Table 3 .
Hourly Rainfall Distribution

Table 4 .
(2).Based on equation(2), the obtained value of Cd for the Leprak river channel is Since this value is less than 0.3, a value of 0.3 is used for the concentration of the moving debris flow.Channel Parameters for Calculating Sediment Concentration