Applied Research in Science and Technology 5.
: 155Ae165 2025 Contents lists available at openscie.
E-ISSN: 2776-7205
Applied Research in Science and Technology DOI: 10.
33292/areste.
Journal homepage: https://areste.
org/index.
php/oai Sub-Watershed Prioritization for Sustainable Sediment Management in the Upper Cisokan Hydropower Catchment Using SWAT Laella Pusparinda1*.
Mariana Marselina1 Department of Environmental Engineering.
Institut Teknologi Bandung.
Bandung.
Indonesia *Correspondence: E-mail:
laellapuspa@gmail.
ARTICLE INFO
ABSTRACT
Article History:
Received 31 May 2025 Revised 2 July 2025 Accepted 12 July 2025 Published 7 October 2025 Background: Sedimentation poses a critical threat to hydropower sustainability, particularly in pumped storage systems such as the Upper Cisokan Pumped Storage (UCPS) plant in West Java.
Indonesia.
Aims and Methods: This study assesses the spatio-temporal dynamics of sediment yield in the Cisokan Watershed using the SWAT model, incorporating historical simulations .
3 and 2.
and a 2038 projection under a Business-As-Usual (BAU) scenario developed through supervised classification in Google Earth Engine (GEE).
Result: Model calibration based on observed discharge data yielded satisfactory results (NSE = 0.
80 in 2013, 0.
65 in 2.
, validating its suitability for sediment analysis.
Results reveal a nearly fourfold increase in average sediment yield from 0.
61 to 2.
25 tons/ha/year between 2013 and 2023, with a projected rise to 5.
57 tons/ha/year by A composite prioritization index, integrating current sediment output, decadal change, and sub-watershed area, identified SW-23.
SW16, and SW-5 as the highest priority areas for erosion mitigation.
These findings were validated against future projections, confirming their persistent erosion risk.
The study emphasizes the importance of scenario-based watershed planning in safeguarding hydropower By integrating sediment modeling with scenario-based land use projection via supervised classification in Google Earth Engine (GEE), this study provides a replicable framework for proactive watershed management and hydropower sustainability planning.
Keywords:
Cisokan hydropower.
Land use change.
Sediment yield.
SWAT .
Watershed management.
To cite this article: Pusparinda.
Marselina.
Sub-watershed prioritization for sustainable sediment management in the upper cisokan hydropower catchment using swat .
Applied Research in Science and Technology, 5.
, 155Ae165.
This article is under a Creative Commons Attribution-ShareAlike 4.
0 International (CC BY-SA 4.
License.
Creative Commons Attribution-ShareAlike 4.
0 International License Copyright A2025 by author/s Introduction IndonesiaAos renewable energy roadmap emphasizes the critical role of hydropower in achieving national climate targets, particularly as the country accelerates its transition toward a sustainable energy mix.
Among key infrastructure projects, the Upper Cisokan Pumped Storage (UCPS) hydropower plant stands out as the first large-scale pumped storage system in the country, designed to stabilize the JavaAeBali electricity grid by balancing peak and off-peak demand cycles.
However, the effectiveness of this system is strongly influenced by sedimentation risks originating from the Cisokan Watershed.
Changes in land use, especially deforestation and agricultural expansion, have been shown to significantly increase sediment yield (Sadhwani et al.
,2022.
Megersa et al.
,2.
, with direct consequences on reservoir storage capacity and erosion damage to turbines (Rodriguez et al.
,2.
underscoring the critical need for erosion management to preserve hydropower infrastructure.
Hydrological models such as SWAT have proven effective in simulating complex watershed processes, including erosion and sediment transport (Berhanu et al.
,2020.
Noora et al.
,2.
Previous studies .
Pusparinda & Marselina, 2.
have documented land-use impacts on sediment yield between 2013 and 2023.
This study builds upon that baseline by adding a 2038 scenario projection under a Business-As-Usual (BAU) assumption using Google Earth Engine (GEE) based classification.
In addition, a composite sub-watershed prioritization framework is proposed that combines three dimensions: .
sediment magnitude, .
decade-long trend, and .
contributing area size.
This multicriteria approach offers a more dynamic perspective than single-metric sediment mapping and aligns with modern risk-based watershed planning frameworks implemented in other highland basins (Berhanu et al.
,2020.
Lypez-Pyrez et al.
,2024.
Noora et al.
,2.
The aim of this study is to assess the spatio-temporal dynamics of sediment yield in the Cisokan Watershed using the SWAT model.
Methods This research was conducted in the Cisokan Watershed.
West Java, covering the upstream Citarum Basin.
It spans approximately 374 kmA across West Bandung and Cianjur Regencies, characterized by steep slopes and heterogeneous land cover.
Seven categories of data were used:
Table 1.
Data Used.
Data
DEM
Soil type
Land use Future land use Climate Future Climate Observed river discharge Year
2013, 2023
Source PLN UIP JBT (FAO-clas.
Ministry of Environment and Forestry (KLH) Supervised classification via GEE MSWEP.
ERA5
ISIMIP database Cirata Hydropower Unit All spatial layers were projected to UTM zone 48S.
Land use for 2038 was derived using GEE with classifier training based on historic LUCC transitions and calibrated accuracy via visual interpretation (Gorelick et al.
,2.
Due to the limited availability of mid-range climate projections in the ISIMIP platform, this study utilized SSP3-7.
0 as a conservative representation of BAU conditions.
While SSP2-4.
5 would more closely resemble a "middle-of-the-road" scenario (OAoNeill et al.
,2.
, the SSP3-7.
0 pathway reflects a plausible outcome under continued land use pressure and moderate mitigation efforts (Riahi et ,2.
Watershed delineation was conducted using QSWAT , resulting in 25 sub-watersheds and 3,830 Hydrological Response Units (HRU.
Modeling scenarios were run for 2013, 2023, and projected Observed discharge .
3Ae2.
was used for model calibration.
The performance was assessed using Nash-Sutcliffe Efficiency (NSE) and Kling-Gupta Efficiency (KGE).
The NashAeSutcliffe Efficiency (NSE) is used to assess the predictive skill of hydrological models and is formulated as follows (Moriasi et al.
,2007.
Furqani et al.
, 2.
ycAycIya = 1 Oe Ocycuycn=1.
cEycuycayc,ycn Oe ycEycycnyco,ycn ) Ocycuycn=1.
cEycuycayc,ycn Oe ycEIycuycayc ) Where ycEycuycayc,ycn refers to observed discharge or sediment yield, ycEycycnyco,ycn is the simulated discharge or sediment yield, and ycEIycuycayc is the mean value of observed data.
The KlingAeGupta Efficiency (KGE) further evaluates model robustness by combining correlation, bias, and variability components into a single metric (Gupta et al.
,2.
yayaya = Oo.
c Oe .
u Oe .
u Oe .
2 yuN Where r is the linear correlation coefficient between observed and simulated values, = yuNycycnyco is the yaycOycycnyco bias ratio, yu = yaycO ycuycayc ycuycayc is the variability ratio.
Sub-watersheds were prioritized using a composite score based on sediment yield in 2023, change from 2013Ae2023, and area share, with 2038 projections used to validate future consistency.
Scoring followed a multi-criteria framework (Halder et al.
,2.
The total priority score was calculated as ycEycycnycuycycnycyc ycIycaycuycyceycn = ycIycIycU2023 ycIOIycIycU ycIyaycyceyca Where ycIycIycU2023 is the rank score for sediment yield in 2023, ycIOIycIycU is the rank score for sediment yield change from 2013 to 2023, ycIyaycyceyca is the rank score based on sub-watershed area proportion.
The workflow of sediment yield modeling using SWAT is illustrated in Figure 1.
Start Data Collection DEM Data (Digital Elevation Mode.
Soil Type Data
Land Cover Data from Indonesia RBI
Climate Data
Soil classification Land use and cover
Rainfall, temperature, humidity, solar radiation, wind speed
talEevio
Slope percentage SOFTWARE: QGIS
Data Processing SOFTWARE: SWAT
Watershed Delineation HRU Generation SOFTWARE: SWAT EDITOR
Running the SWAT Model
SOFTWARE: SWAT TOOLBOX
Model Calibration and Validation Running Final SWAT Simulation Model Output Data Analysis End Figure 1.
Research Workflow Results and Discussion 1 Model Performance Model calibration was conducted using observed streamflow data, as direct sediment measurements were unavailable.
The SWAT model showed satisfactory performance (Table .
, with NSE and KGE values of 0.
80 and 0.
81 in 2013, and 0.
65 and 0.
72 in 2023, respectively.
These metrics indicate that the model captures the hydrological behavior of the Cisokan Watershed with reasonable accuracy, particularly during the baseline year (Figure .
Streamflow-based calibration is a common approach in data-scarce regions, where sediment records are limited.
Similar methods have been applied in the Xinjiang River Basin (Yuan & Forshay, 2.
and the Cantareira System in Brazil (Pontes et al.
,2.
, with reliable sediment predictions derived from calibrated flow outputs.
These precedents strengthen the methodological validity of this study.
Table 1.
Model performance statistics (NSE and KGE) for the calibration years.
Year
NSE
KGE
Figure 2.
Comparison of observed and simulated monthly streamflow for the years 2013 and 2023 These results indicate that the calibrated model can be confidently used for further simulation of sediment yield dynamics, including long-term changes and future projections, as discussed in the following sections.
2 Spatio-Temporal Sediment Yield .
3Ae2.
Changes in land use between 2013 and 2023 significantly affected sediment yield patterns across the Cisokan Watershed.
Based on SWAT simulation outputs, the average annual sediment yield increased from 0.
61 tons/ha/year in 2013 to 2.
25 tons/ha/year in 2023.
Based on spatial distribution maps (Figure .
, sub-watersheds located in the south-central region, particularly sub-watersheds 18, 19, and 23, experienced the most significant escalation in sediment These areas coincide with zones of extensive forest conversion into dryland agriculture and settlements, often occurring on steep slopes, which accelerates soil detachment and transport.
Figure 3 presents a comparative visual of the sediment yield distribution in 2013 and 2023.
noticeable expansion of localized high-yield zones (Ou60 Mg/ha/yea.
is observed in 2023, particularly within critical sub-watersheds.
It should be noted that these high-yield values represent HRU-level outputs and do not necessarily reflect sub-watershed average values, as presented in Table 2, which lists the ten sub-watersheds with the highest increase in sediment yield between 2013 and 2023.
These findings emphasize the intensifying erosion risk and reinforce the importance of spatially targeted soil and water conservation interventions (Serryo et al.
,2021.
Wang et al.
,2.
Figure 3.
Spatial distribution of simulated sediment yield in the Cisokan Watershed for the years 2013 .
and 2023 .
, highlighting the emergence of critical erosion zones.
Table 2.
Sub-watersheds ranked by the largest absolute increase in sediment yield between 2013 and 2023, based on SWAT simulation outputs.
Sub Landuse 2013 frse_suhf.
frse_suhf.
frse_suhf.
frse_suhf.
frse_suhf.
frse_suhf.
frse_suhf.
frse_suhf.
Landuse 2023 frse_suhf.
frse_suhf.
frst_sums.
frse_suhf.
frst_sums.
frse_suhf.
frst_sums.
frse_suhf.
frst_sums.
frse_suhf.
frse_suhf.
frst_sums.
frse_suhf.
frse_suhf.
SY 2013
(Mg/ha/year SY 2023 (Mg/ha/year Yield Note: Land use abbreviations, agrl: dryland agriculture.
agrr: mixed dryland agriculture.
frse_suhf:
secondary dryland forest.
frst_sums: mixed forest.
orcd: orchard.
rice: paddy field.
urml: low-density barr: barren land.
shrb: shrub.
3 Projection for 2038 (BAU) The 2038 land use/land cover (LULC) projection was developed using a supervised classification approach in Google Earth Engine (GEE), leveraging a Random Forest algorithm trained on historical land cover transitions between 2013 and 2023 (Figure .
The model incorporated explanatory variables such as elevation (SRTM), slope, distance from roads, and temporal change patterns (Pande et al.
,2024.
Sheeba et al.
,2.
Training and validation were conducted using stratified random sampling, achieving an overall accuracy of 99.
7% and a Kappa coefficient of 0.
The projection adopted a Business-As-Usual (BAU) scenario consistent with IndonesiaAos SSP2 climate development pathway, assuming current land use trends continue without significant policy shifts.
Several postclassification refinements were applied, including spatial masking for reservoir inundation, partial retention of rice fields, and slope-based adjustment for forest and agricultural areas.
Additionally, the expansion of water bodies was adjusted to account for the planned reservoirs in the Upper Cisokan Pumped Storage (UCPS) project, with the Lower Dam covering approximately 260 hectares and the Upper Dam around 80 hectares (PT PLN (Perser.
, 2.
Figure 4.
LULC comparison of the Cisokan watershed in 2023 (KLH) and projected 2038 (BAU/SSP2 scenario using Random Forest model in GEE) The resulting LULC projection provides the basis for subsequent sediment yield modeling, revealing substantial shifts in erosion risk across the watershed.
Model results show that average sediment yield in 2038 is expected to reach 5.
57 tons/ha/year, more than double the 2023 estimate of 25 tons/ha/year.
This increase is strongly associated with continued expansion of mixed dryland agriculture and build-up area .
rban settlement.
at the expense of forest cover, especially in the eastern parts of the watershed.
These dynamics are further illustrated in Table 3, which summarizes the area changes for each land use class across 2013, 2023, and the projected 2038 scenario.
Table 3.
LULC in the Cisokan Watershed in 2013, 2023, and Projected 2038 under the BAU Scenario.
Class Water Body Plantation Forest Secondary Dryland Forest Estate Crops Dryland Agriculture Mixed Dryland Agriculture Paddy Field Shrubland Open Area Built-up Area 2013 .
Total 2023 .
16,739.
3,156.
8,127.
9,405.
37,577
15,083.
8,963.
5,016.
4,900.
2,147.
37,577
2038/BAU .
14,814.
4,279.
3,428.
6,170.
2,796.
5,040.
37,577
4 Sub-Watershed Prioritization The prioritization analysis revealed significant spatial variability in erosion contribution across the 25 sub-watersheds.
Based on the composite scoring system, which integrates sediment yield in 2023, change from 2013 to 2023, and relative area size, three sub-watersheds (SW-23.
SW-16, and SW-.
were classified as Very High Priority (Table .
These sub-watersheds consistently exhibited high sediment yield, notable positive trends over the past decade, and considerable areal coverage.
These findings underscore their critical role in the overall sediment load entering the UCPS reservoir.
Table 4.
Composite prioritization of sub-watersheds based on sediment yield, change over time, and area proportion.
Sub-Watershed SW-23
SW-16
SW-5
SW-9
SW-7
SW-22
SW-19
SW-18
SW-13
SW-10
SW-12
SW-8
SW-6
SW-20
SW-21
SW-4
SW-15
SW-11
SW-2
SW-14
SW-24
SW-25
SW-17
SW-1
SW-3
SY 2023
i SY Area (%) Total Score Priority Class Very High Very High Very High High High High High High High High High Medium Medium Medium Medium Medium Medium Low Low Low Low Low Very Low Very Low Very Low Figure 5.
Spatial distribution of sub-watershed prioritization based on composite sediment yield scoring in Cisokan Watershed To validate the robustness of the prioritization results, sediment yield projections for the year 2038 under the BAU scenario were examined for the three sub-watersheds classified as Very High Priority:
SW-23.
SW-26, and SW-5.
All three sub-watersheds demonstrated a consistent or increasing trend in sediment yield (Table .
These findings suggest that the prioritization framework not only reflects current erosion pressures but also aligns with projected long-term sediment dynamics, thereby supporting its applicability for forward-looking watershed management strategies.
Table 5.
Sediment yield validation for Very High Priority sub-watersheds .
3Ae2038.
BAU).
Sub-Watershed SW-23
SW-16
SW-5
SY 2013 (Mg/ha/y.
SY 2023 (Mg/ha/y.
SY 2038 (Mg/ha/y.
5 Implications for UCPS Sustainability and Conservation Planning The increasing trend of sediment yield observed between 2013 and 2023, as well as the projected escalation under the 2038 Business-As-Usual (BAU/SSP.
scenario, presents a tangible risk to the long-term sustainability of the Upper Cisokan Pumped Storage (UCPS) hydropower plant.
As a closedloop facility dependent on dual reservoirs.
UCPS is particularly vulnerable to sediment accumulation, which can reduce effective storage volume, compromise turbine efficiency, and necessitate costly maintenance or dredging (Shrestha & Shrestha, 2019.
Noon & Kim, 2.
The simulation results underscore the need to treat sediment control not only as an environmental imperative but also as a fundamental operational safeguard for large-scale energy infrastructure.
The prioritization framework developed in this study provides a targeted strategy for erosion mitigation, highlighting sub-watersheds that contribute disproportionately to current and future sediment yield.
Sub-watersheds such as SW-23 and SW-16 emerged as consistent high-priority areas across both historical and projected scenarios, indicating persistent erosion pressure.
Integrating these insights into watershed management planning enables stakeholders to allocate conservation resources more effectively, focusing on upstream interventions such as agroforestry, land cover restoration, and slope stabilization (Halder et al.
,2.
Furthermore, the use of temporal and scenario-based prioritization strengthens the adaptive capacity of erosion control measures, supporting the long-term reliability and efficiency of UCPS operations under dynamic land use conditions.
Conclusions This study assessed sediment yield dynamics in the Cisokan Watershed using SWAT simulations for the years 2013, 2023, and a 2038 projection under the Business-As-Usual (BAU/SSP.
The results show a significant increase in sediment yield driven by land use change, particularly the expansion of dryland agriculture and settlements.
A composite prioritization index incorporating sediment magnitude, decadal change, and sub-watershed area successfully identified critical erosion zones .
SW-23.
SW-16.
SW-.
for targeted intervention.
These findings underscore the importance of integrating sediment management into long-term hydropower planning, especially for the Upper Cisokan Pumped Storage (UCPS) system.
Future research should explore alternative land use scenarios, such as afforestation and sustainable land management, to compare their mitigation potential, and extend the modeling scope to reservoir sedimentation and routing processes.
Importantly, future model validation should utilize observed sedimentation data, including sediment trap measurements, turbidity-based monitoring, or bathymetric surveys, to enable direct comparison with simulated outputs and improve the modelAos accuracy and operational reliability.
This study contributes a replicable, forward-looking framework for erosion risk assessment under dynamic land use pressures, supporting more adaptive and evidence-based watershed management in tropical upland Acknowledgment This research was supported by a scholarship from PT PLN (Perser.
under the AuProgram Belajar Jarak Jauh Ae Climate Change SolutionsAy.
Technical data, land cover maps, and hydrological reports used in this study were provided by PT PLN (Perser.
Unit Induk Pembangunan Jawa Bagian Tengah (UIP JBT), particularly through the UCPS project team.
References