Environmental models for ecosystem management must be calibrated before they are applied in the development of alternative scenarios. The initial stage of this study focused on calibrating the PSETLR parameter, which affects TP-C in lake water. The RWP 0 scenario was used as the baseline model for developing alternative scenarios. The RWP 0 calibration results show that at PSETLR = 269, Pbias was found to be 45.7%, RMSE was 0.0005 mg/L, and SMAPE was 18.6%. The calibrated PSETLR value serves as the baseline setting for developing all scenarios in the study. The validation results using the RWP 0 scenario show Pbias of 4.7%, RMSE of 0.0042 mg/L, and SMAPE of 6.8%.

RWP 1–3 represent scenarios with existing land use in a mesotrophic–eutrophic state, whereas RWP 4–7 represent modified land use in a mesotrophic state. Reduced fertilizer use on rice fields resulted in an annual decrease in TP-C in the lake under the RWP 1 scenario to 0.029 mg/L, or 0.003 mg/L (12.1%) less than the baseline. RWP 3 had the highest TP-C value of 0.040 mg/L, which was 33.3% above the baseline (Figure 4). An agroforestry scenario (RWP 4) reduced TP-C in lakes to 0.020 mg/L (-39.4%) relative to the baseline, or 33.3% below the eutrophic threshold. The deepening of the lake under the same catchment management conditions increased the TP-C in the RWP 2 scenario to 0.035 mg/L or 6.1% relative to the baseline. The RWP 7 scenario showed a TP concentration value of 0.028 mg/L with a 15.2% decrease in TP concentration relative to the baseline.

The TP-C in the water body was quantified and the sources of P-load in the Lake Rawapening catchment area were mapped. Table 6 presents the annual P-load entering the lake resulting from land use change interventions and fertilization activities. In RWP Scenario 1, the P-load under conditions without fertilization was 275 tons P/year. The most significant contributors to P loading into the lake in this scenario included RNGE 92.5 tons P/year (33.6%), URHD 58.1 tons P/year (21.1%), and AGRR 46.6 tons P/year (16.9%). In the scenario, agricultural groups contributed a P-load value of 148.5 tons P/year (54%). The paddy field groups (RCEE and RICE) produced a P-load of 56.4 tons P/year (20.5%). Interventions to transform agricultural land to agroforestry (ORCD) reduced the total P-load entering the lake (RWP4) to 182.9 tons P/year (-33.5%). In the agricultural group, the shift to agroforestry resulted in a 62% decrease in the P-load, from 148.5 tons P/year to 56.5 tons P/year.

In the RWP 4 scenario, the most significant P-load came from settlement (URHD) land use at 58.1 tons P/year (31.8%), and ORCD at 56.3 tons P/year (30.8%). Figure 5 shows the effect of agricultural land use and fertilizer management interventions on paddy fields. Agricultural land is predominantly located in the Panjang, Torong, Galeh, Legi, Parat and Sraten sub-catchments (Figure 5a). The majority of agricultural land is in the P-load class of 6.8–13.5 kg P/ha/year and the class of 13.5–20.3 kg P/ha/year in parts of the Panjang, Parat and Sraten areas. In addition to reducing TP concentrations in lakes, changes in agricultural land use also affected the P load entering lakes. Figures 5b and 5c show changes in phosphorus load classes in agricultural areas from 13.5–20.3 tons P/year to 6.8–13.5 tons P/year. The paddy fields are located around the lake, and some areas extend into Lake Rawapening in the Torong, Panjang and Rengas sub-catchments (Figure 5d). Residential areas are also scattered around the paddy field areas, with the highest concentration in Sraten and Ringis sub-catchment areas.

The application of fertilizers in paddy fields influences the influx of phosphorus into lakes. Fertilization of irrigated paddy fields (RICE) increased the annual P-load from 38.6 tons P/year to 60.3 tons P/year, or 56.2% (Figure 8). On the other hand, in rainfed paddy fields (RCEE), there was an increase in annual P load from 17.9 tons P/year to 27.9 tons P/year (55.9%). The total area of paddy fields (RICE + RCEE) without fertilization contributed 56.5 tons of P/year (20.5%) to the total P load in the RWP 0 scenario and 30.9% in the RWP 4 scenario. Meanwhile, when fertilization was applied to the paddy fields, the P load became 88.2 tons P/year (28.7%) of the total P load in scenario RWP 1 and 41.1% in RWP 5 (Table 7).

Lake deepening intervention increased the depth from 2.6m to 4.6m. In a single-intervention scenario focusing solely on lake deepening, changing from baseline to RWP 3 increased the TP-C of the lake from 0.033 mg/L to 0.040 mg/L (21.2%). The impact of deepening on other scenarios was as follows: RWP 1 to RWP 2 increased from 0.029 mg/L to 0.035 mg/L (20.7%); RWP 4 to RWP 6 increased from 0.020 mg/L to 0.024 mg/L (20%); and RWP 5 to RWP 7 increased from 0.023 mg/L to 0.028 mg/L (21.7%). The simulation results show an increase in volume from 50.5 × 106 m³ to 95.9 × 106 m³, or a 90% increase, while the discharge remained stable at between 13.0 and 13.7 m³/s ( Table 8). Phosphorus residence time due to lake deepening increased from 43 days to 81–85 days (88–98%), while the lake retention values ranged from 88.7 to 89.2%.

The SWAT model used for the study needed to be calibrated with comparative AoEF observation data. (Luján et al., 2025) and (Mai, 2023) state that many parameters in environmental models are conceptual in nature and cannot be directly measured in the field, thereby introducing uncertainty. However, such uncertainty can still be used to determine the relative change in TP-C for each scenario in our study. (Han et al., 2025) and (Yu et al., 2022) argue that the calibration process should focus on parameters that are most relevant to the research objectives so that the model is not overly complex, while (Lopez et al., 2025) emphasize that models should remain simple and model limitations should be recognized in order to reduce overfitting. Therefore, this study focuses on the PSETLR (Phosphorus Settling Rate) parameter in the .LWQ file that affects the TP-C results in the lake. Calibration of the RWP 0 scenario yielded a PSETLR value of 269 m/day, which is the maximum limit specified in the SWAT manual for reservoirs. The results show a relatively high deviation from the calibration dataset, but with an absolute deviation of TP-C of 0.005 mg/L. Validation of TP-C in RWP 0 using the AoEF dataset for 2020–2024 yielded a PBIAS of 14.7% and an RMSE of 0.0042 mg/L, indicating that the relative and absolute deviations are acceptable compared to the validation dataset. The SMAPE of 6.8% indicated relatively low model error in following the validation dataset pattern. Based on the validation results, the RWP 0 scenario was considered feasible for use as a baseline for developing other scenarios.

This study developed seven scenarios (RWP 1 – RWP 7) for the rehabilitation of the Lake Rawapening ecosystem. Such rehabilitation requires an integrated approach and demands solutions tailored to specific ecological and socioeconomic contexts ((Gunacti & Cetinkaya, 2025); (Ruiz-Marín et al., 2024); (Zhan et al., 2023)). These scenarios integrate both catchment-level and in-lake interventions, referring to the pillar-based and object-based management concepts proposed by (Mardiatno et al., 2022). The proposed concept encompasses management scenarios and actions, good governance, and science-based information as its main management pillars. In addition, water bodies, watersheds and fertilization (N and P) are the main study objects. The development of various scenarios was designed to connect the concept of management with the operational management of the Lake Rawapening ecosystem. The SWAT model used for scenario simulation can be utilized as a science-based tool for providing information related to impact estimates and trade-offs between the scenarios developed. Figure 4 illustrates the impact of intervention on TP-C dynamics in Lake Rawapening. The results of RWP scenarios 1–7, with a TP-C value range of 0.020–0.040 mg/L, illustrate the dynamics of changes in the lake's trophic state resulting from the implemented interventions. Results from several scenarios can be classified into three categories: ideal scenarios, compromise scenarios with trade-offs, and scenarios to be avoided. The ideal scenario is the RWP 4 scenario, with a 33.3% decrease in TP-C below the eutrophic state threshold. This condition was reached by converting agricultural land to agroforestry, excluding fertilizing and lake deepening actions. This scenario was shown to effectively reduce the trophic status of the lake from eutrophic to mesotrophic. However, it is challenging to implement this approach, due to the established practice of fertilizing paddy fields in Lake Rawapening. A study of Lake Rotorua, in New Zealand revealed that less intensive farming can yield long-term economic benefits by enhancing ecosystem services (Mueller et al., 2019). In the context of Lake Rawapening management, RWP 4 represents the best solution ecologically; however, socioeconomically, achieving short-term agricultural yields is challenging.

Therefore, a scenario is needed that can provide a reasonable trade-off between ecological and socioeconomic recovery for the community around Lake Rawapening. The scenarios classified as acceptable compromises include RWP 5, which has a more realistic equilibrium between ecological and socioeconomic trade-offs, although the results are not as significant as RWP 4. The RWP 5 scenario effectively reduced TP-C to 0.023 mg/L, which is 22.7% beneath the eutrophic threshold. This condition suggests that agroforestry within the lake catchment can reduce TP-C without compromising crop productivity, provided that fertilizer application is restricted in paddy fields. The RWP 3 scenario, which focuses only on lake deepening, represents a scenario that should be avoided. The related findings resulted in an elevation of TP-C to 0.040 mg/L, representing a 33.3% increase above the eutrophic threshold, alongside expensive lake dredging. Further discussion on the impact of interventions in agricultural land use change, fertilization management in paddy fields, and deepening will be discussed in the following sub-sections.

Scenarios were developed for converting the agricultural land use group (AGRL, AGRR and RNGE) to agroforestry (ORCD). Table 6 shows that moving from seasonal agriculture (RWP 1) to agroforestry (RWP 4) resulted in a 33.5% reduction in the P-load in the catchment, from 275.0 tons P/year to 182.9 tons P/year. In scenario RWP 1, the most significant contributors are RNGE at 92.5 tons P/year (33.6%), URHD at 58.1 tons P/year (21.1%), and AGRR at 46.6 tons P/year (16.9%). The primary contributor to the P-load by land use classification area is RCEE, with a value of 0.015 tons P/ha/year, followed by RICE at 0.013 tons P/ha/year. The transition of seasonal agricultural groups to agroforestry (RWP 4) resulted in a 62% decrease in P-load, from 148.5 tons P/year to 56.4 tons P/year. The P-load contribution of ORCD was 0.004 tons P/ha/year, which was inferior to that of the agricultural groups AGRR, AGRL and RNGE, which reached 0.014, 0.012 and 0.010 tons P/ha/year respectively. Nayanarangani et al., (2025) state that actively- managed perennial agricultural land has eleven times lower P availability than seasonal agriculture with the same cultivation practices. Furthermore, (Good et al., 2025) confirm that agriculture using deep root systems and minimal intervention is more effective in P retention. These findings suggests that converting seasonal agricultural to agroforestry effectively reduces the P-load from the catchment area.

Alongside agricultural land, settlement areas (URHD) emerged as major contributors across all scenarios. The phosphorus load in the URHD class was 58.1 tons P/year (21.1%), rendering it the second-largest contributor in the RWP 1 scenario. In the RWP 4 scenario, it was the predominant contributor, accounting for 31.8%. The condition relates to the ratio of residential zones, encompassing 6,734.8 hectares or 25.5% of the overall catchment area. However, settlement areas have a comparatively low specific P load (0.009 tons P/ha/year); their extensive area leads to a continual significant accumulation of P loads. The development of residential areas puts pressure on the balance of the ecosystem. Research at Phewa Lake, Nepal, reveals that settlement expansion has altered the lake's status from oligotrophic to eutrophic over 20 years (Subedi et al., 2025). while research at Poyang Lake, China, indicates that agricultural and urbanization activities exert pressure on the phosphorus influx into the lake (Li & Shen, 2025). Figure 5a shows that agricultural land use is concentrated in the Panjang, Torong, Galeh, Legi, Parat and Sraten sub-catchments.

The P load in these regions typically ranges from 6.8 to 13.5 kg P/ha/year, whereas in certain sections of the Parat, Panjang and Sraten sub-catchments, it varies from 13.5 to 20.3 kg P/ha/year. The conversion of agricultural land to agroforestry reduced the P load from 6.8–13.5 kg P/ha/year and 13.5–20.3 kg P/ha/year to less than 6.8 kg P/ha/year (Figure 5c). In scenario RWP 4, paddy fields are situated around the lake, with specific fields positioned within the lake itself, as observed in the Panjang, Torong and Rengas sub-catchment areas. This situation highlights that the management of paddy fields is a crucial issue. The distribution of residential areas throughout all the sub-catchments requires regulation, particularly in highly populated regions such as the Sraten and Ringis areas. The transition from seasonal agricultural patterns (RWP 1) to agroforestry (RWP 4) consistently reduced the lake's trophic state, as discussed in subsection 4.2. These scenario results have the potential to inform the design of long-term intervention priorities for the rehabilitation of Lake Rawapening.

The study assesses the impact of fertilizer application on paddy fields in the Lake Rawapening catchments. The development of this scenario is grounded in established paddy cultivating procedures. Table 7 shows that the total area of rice fields is 4,221.6 ha, or 16.0% of the total catchment area. Irrigated paddy fields before fertilization have a P load of 38.6 tons P/year, with a specific P load of 0.013 tons P/ha/year. On the other hand, rainfed paddy fields exhibit a P load of 17.9 tons P/year, with a specific P load of 0.015 tons P/ha/year. (Yang et al., 2026) argue that, mathematically, an increase in nutrients from a level of 0.4 to 0.65 (63%) results in a reduction of tipping time from 9 to 3 (67%) units of time. An elevated nutrient load from fertilization influences the rate at which lakes transition to a more adverse trophic state. The findings of this study indicate that the elimination of fertilization treatments resulted in a reduction of TP-C from 0.033 mg/L to 0.029 mg/L, consequently altering the lake's trophic status from eutrophic to mesotrophic.

After fertilization, Irrigated paddy fields experienced a 56.4% increase in P-load, and rainfed paddy fields experienced a 55.8% increase in P-load. These conditions suggest that applying fertilizer to paddy fields results in a substantial overall increase in P loads in the lake catchments by 31.7 tons P/year (10 – 15%). Tables 3 and 4 show the fertilizer dosages and schedules for Irrigated paddy fields and rainfed paddy fields. The SP-36 fertilizer application was 100 kg/ha, equivalent to 15.7 kg P/ha, while the phosphorus requirement for the Inpari 32 and Inpari Nutrizinc rice varieties is around 10–30 kg P/ha ((Sembiring et al., 2025); (Sitaresmi et al., 2023)). Even though the fertilizer dose was 52% below the recommendation, there was still an increase in P-load and TP-C. A meta-analysis indicated that crop yield can be sustained with a 30% reduction in fertilizer usage (Jin et al., 2023). In addition, (Zhang et al., 2025) found that 20.2–25.8% of P-load was lost during the high-risk period of 0–20 days after fertilization. This intervention scenario demonstrates that, although a fertilizer dose is below the recommended level, it does not necessarily ensure complete safety for the ecological rehabilitation process of a lake. (Zhang et al., 2025) indicated that a mixture of organic fertilizer (40%) and chemical fertilizer (60%) applied to paddy plants could decrease phosphorus loss by 13%.

The Irrigated paddy fields are situated directly along the lake's border, with some areas extending into the water body of Lake Rawapening, particularly in the Torong, Panjang and Rengas sub-catchments. This situation makes Lake Rawapening highly vulnerable to an influx of phosphorus that instantly enters the lake. Based on spatial distribution, Irrigated paddy fields are located around the lake bodies in the Galeh, Legi, Parat, Ringis and Kedungringin sub-catchments. In contrast, rainfed paddy fields are located in the upstream areas of the Rengas, Ringis, Parat and Legi sub-catchments. These differences in position present distinct challenges and necessitate different management strategies for paddy fields. Generally, Irrigated paddy fields are situated in the downstream area, whereas the wet zone of Lake Rawapening receives fertilizer runoff from upstream, posing a direct risk of water pollution. Meanwhile, the rainfed paddy fields area located upstream requires adjustments to the dosage and fertilization schedule. Based on the conditions at Lake Rawapening, a best management practice strategy needs to be developed, it should 1) gradually reduce fertilizer doses on rainfed paddy fields by 20% of the recommended dose; 2) reduce fertilizer doses for Irrigated paddy fields around the lake border by 30% of the recommended dose; 3) reduce the use of fertilizer on Irrigated paddy fields within the lake's water body more aggressively or even eliminate it; and 4) make improvements to the timing of P fertilizer application to prevent accumulation at a specific time. Alongside fertilizer management approaches, engineering interventions such as the establishment of ecological ditches and temporary storage ponds have demonstrated a 37.9% reduction in TP (Yang et al., 2024). (Nada et al., 2023) found that managing riparian and buffer zones in Lake Rawapening to a distance of 30 meters could diminish the P-load by 45.5%. In the context of the lake, this strategy could be implemented in the form of technical civil engineering in areas with high P load, such as in the Galeh, Parat, Ringis and Kedungringin sub-catchments. Ecological ditches could be constructed in proximity to the banks of the rivers that discharge into the lake to reduce phosphorus loads. Irrigated paddy fields land could function similarly to the wetland concept, so that rice nutrient requirements can be met from nutrients carried by the flow. With an adaptive management strategy, the trade-off between ecological rehabilitation and agricultural productivity could be more balanced and feasible.

In many studies, lake deepening through dredging is considered to be an effective alternative for controlling lake degradation, such as Lake Taihu and Lake Dianchi. However, the results of the deepening scenario show that lake deepening cannot stand alone. Without consistent control of phosphorus loading, an increase in TP-C is shown. Table 8 indicates that deepening under the same management conditions actually increased TP concentrations by 20.0–21.7%. Dredging increased the lake volume from 50.5 × 106 m³ to 95.9 × 106 m³ (89.1%), but the outflow remained almost unchanged, ranging between 13.0 m³/s and 13.7 m³/s. This situation resulted in an increase in P residence time from 43 days to 81 days (88.4%). Lake Rawapening consistently has a high P retention value, ranging between 88.7% and 89.2%. The RWP 1 scenario involves conditions closest to reality in Lake Rawapening. Deepening interventions directly increase TP-C by 21.2% in RWP 1 from 0.033 mg/L (eutrophic), to 0.040 mg/L (eutrophic) in RWP 3. This implies that deepening alone is counterproductive to the goal of rehabilitating lake ecosystem conditions. In lakes with high retention, an increase in residence time leads to a corresponding increase in P concentration in the water column (Liu et al., 2024). (Riza et al., 2023) found that the results of dredging for lake eutrophication control are limited by time and the external P loads entering the lake. (Xiang et al., 2024) report that dredging has a positive impact on water quality within 2–5 years after dredging, depending on the lake depth, initial water quality, and the dredging method. (Kato et al., 2025) also argue that dredging increases nutrient concentrations and disrupts the balance of the aquatic ecosystem that existed prior to dredging, a disruption that persists for several months. Lake deepening is optimal when performed in deep lakes with relatively calm hydrodynamics ((Wu et al., 2025); (Yang et al., 2023)).

The study presents alternatives that can be implemented to optimize lake deepening. In the RWP 6 scenario, a deepening scenario is combined with a change in land use from agriculture to agroforestry. The related results show a 27.3% decrease in TP concentration from 0.033 mg/L (RWP 1 - eutrophic) to 0.024 mg/L (RWP 6 - mesotrophic). Although the model results show improvement, the scenario has the potential to reduce rice field productivity due to the loss of fertilization. Trade-offs in deepening must consider short-term and long-term recovery as well as ecological improvement and socioeconomic sustainability. Scenario RWP 7 shows a more moderate trade-off, with a 15.2% improvement in TP concentration from 0.033 mg/L (RWP 1 – eutrophic) to 0.028 mg/L (RWP 7 – mesotrophic). Although not as good as RWP 6, RWP 7 is the most feasible intervention for Lake Rawapening. In various scenarios developed for the lake, combining land use changes could be the most significant factor for improvement. Agroforestry land use changes can reduce the impact of fertilization on rice fields and lake deepening. This implies that lake management policies involving deepening through dredging need to consider controlling the phosphorus load entering the lake from upstream.

The study is limited by the scarcity of continuous field monitoring data required to calibrate and validate Total Nitrogen (TN) and Chlorophyll-a (Chl-a) parameters. The SWAT model also has limitations in simulating sediment export in scenarios involving deepening. This implies that adjustments need to be made to the assumptions of morphological deepening and sediment dredging in the water body. The limited data on the types of agriculture in the Lake Rawapening catchment also pose a challenge to the accuracy of phosphorus loads entering the lake. Furthermore, measured discharge and sedimentation data at each of the 14 inlets and the single outlet are una vailable, introducing potential bias into the calculated P load balance. Further research should focus on collecting and analyzing observations of the TN, Chl-a and P content in sediments, accompanied by measurements of discharge and sedimentation, to provide more comprehensive simulation results.