Assessment of small hydropower potential in the Ciwidey subwatershed , Indonesia : a GIS and hydrological modeling approach

Generally, the remoteness of potential sites for small hydropower (SHP) which are mostly located in mountainous regions, and complex hydrological phenomena, remain significant barriers for SHP development. However, hydrological modeling together with the advancement of remote sensing and geospatial technology can be used to assess SHP potential. This study combined geographic information system (GIS) methods with the Soil and Water Assessment Tool (SWAT) hydrological model to assess the potential for SHP development in the Ciwidey subwatershed, Indonesia. Nine potential sites for SHP were identified according to criteria such as head/elevation drop, stream order, and distance between each potential site. The SWAT model reproduced the observed discharge in the watershed accurately producing an acceptable coefficient of determination (R2 = 0.75) and Nash-Sutcliffe Efficiency (NSE = 0.67). According to Flow Duration Curve (FDC) analysis at 60, 75, and 90% dependability threshold, a maximum SHP potential total of 1.72 MW can be harnessed in the Ciwidey subwatershed. This study is expected to boost the initiative of promoting renewable energy, mainly SHP, in Indonesia. Based on these results and the goal of increasing renewable energy resources to bolster national energy security, we recommend an initiative promoting SHP in Indonesia.


INTRODUCTION
As a developing country, Indonesia faces an acute energy crisis caused by rapidly increasing energy demand surpassing a limited supply availability, as well as low penetration of electricity into rural areas (Perusahaan Listrik Negara/ PLN, 2015).Along with economic development, electricity demand has been growing rapidly and is predicted to double from 201.5 in 2014 to 424 TWh in 2024 (PLN, 2015).Such a situation will cause Indonesia to struggle to meet its energy needs.Indeed, the Indonesian power system has been reported to have an undersupply issue and is no longer considered secure.The national electricity supply reserve (reserve margin) remains low at 24% in the main regions of Java, Madura and Bali, with deficits in other regions (Vithayasrichareon et al., 2012;PLN, 2015).At the same time, rural Indonesia also has reported lack of electricity access particularly due to logistical constraints and challenges in extending the power grid through difficult terrains (hilly mountainous areas, islands, ocean, etc.).This forces rural people to rely on fossil fuel (diesel), for which they have to bear the additional costs of fuel transport.There are 16.8 million households, equivalent to 63 million Indonesian people that have no access to modern electricity (PricewaterhouseCoopers Indonesia, 2013).Hence, a more effective strategy to meet energy needs has become an urgent agenda for the country.Nevertheless, considering fossil fuel resource depletion, high import prices, and the importance of clean energy and global concern related to climate change mitigation, the optimization of low carbon electricity through renewable energy (RE) should be considered.Thus, the Government of Indonesia is now struggling through its policy to improve development efforts and attract investments into RE.
Among RE sources, hydroelectric power plants (HPP) have been among the earliest and most robust RE generation methods adopted (Patil et al., 2013).Often the cheapest energy generation method (International Renewable Energy Agency/IRENA, 2015), hydropower has been a backbone of the energy economy in many countries, including China, India, and Brazil (Zhao and Zhu, 2004;Khan, 2015).However, it is undeniable that large HPP associated with dams, often cause socio-environmental problems, such as land inundation and resettlement of residents (Cernea, 2004;Sovacool and Bulan, 2012).Therefore, rather than large scale infrastructure, this study intends to emphasize the utilization of small hydropower (SHP).The United Nations Industrial Development Organization categorizes SHP as hydropower with an installed capacity of 1-10 MW (Liu et al., 2013), typically employing a run-of-river design.Meanwhile, hydropower with installed capacities below 1 MW is often called mini or micro hydropower.Instead of large dams, SHP only utilizes small weirs (European Small Hydropower Association/ESHA, 2004), thus it minimizes environmental tradeoff along with providing a clean and reliable energy solution.
In regards to SHP development, good planning and design play an important role.Yet, generally, complex hydrological phenomena and the remoteness of SHP potential sites which are mostly in mountainous regions, remain barriers that complicate the assessment effort and make it time consuming, and lead to the requirement for large capital budgets.However, with the advancement of remote sensing and geospatial mapping technology together with hydrological modeling, it is possible to optimize the assessment of SHP potential.GIS can help with topography and spatial analysis, and recent hydrological modeling (Moradkhani et al., 2009on Devi et al., 2015) can be used to predict the basin system's behavior and to understand complex hydrological processes within various climate possibilities.Studies of water resources assessment and climate change impact are indeed the most popular studies of hydrological modeling particularly for planning and disaster risk purposes, e.g., HEC-HMS for simulating flood inundation in Bago Basin, Myanmar (Zin et al., 2015) and Soil Water Assessment Tools (SWAT) for assessing climate change impact in Laos (Sayasane et al., 2016) .Yet, in this study, GIS and hydrological modeling were used to assess the HPP potential.Previous researchers have either introduced the use GIS alone or GIS and hydrological modeling together as powerful tools to provide a potential assessment study of hydropower.Dealing with the energy security in the country, some Indian researchers have shown big interest as discussed within Kusre et al. (2010), Patil et al. (2013), andGoyal et al. (2015) who developed a similar methodology to SWAT to estimate the SHP potential in India in four river catchment areas within Kopili, Bennihalla, and Mahanadi basin.On the other hand, another hydrological modeling approach of HYDREEMS was also used by Toyoda et al. (2015) to evaluate the potential of SHP generation in Japan.Using the GIS and manual Soil Conservation Service-Curve Number (SCS-CN) measurement, Das and Paul (2006), Feizizadeh and Haslauer (2012) and Setiawan (2015) worked on SHP development in Himalayan Region of India, Tabriz Basin of Iran and Kalimantan, Indonesia, respectively.
Departing from the above discussion, it can be seen that the SHP potential assessment through GIS and hydrological modeling becomes necessary considering the urgency of the increasing energy demand and the need to overcome the low energy access of the Indonesian rural areas.The utilization of water resources for energy needs with minimal ecosystem interruption also has been an advantage of SHP.Here, the tools of GIS and SWAT were chosen as the main methodologies.There are several objectives of this study as follows: (a) identifying SHP potential sites using GIS; (b) assessing and estimating the potential power generation through SHP in study areas combining GIS with hydrological modeling.In this study, the formula of SHP potential (p E-SHP ) (kW) is outlined in Equation ( 1).
where ρ is the water density (kg m -3 ), g is the acceleration due to gravity (9.8 m s -2 ), Q is the dependable discharge (m 3 s -1 ), H is the head (height difference) or elevation drop (m), and η is the efficiency constant (0.8) (Nasir, 2014).
Through GIS and hydrological modeling, this study aims to determine the potential sites of SHP according to H and Q.
Head is the distance that a given water source has to fall before the point where power is generated while Q represents the dependable amount of water to be converted into electricity.Some literatures, e.g., Tokyo Electric Power Company (2005) and Pravin (2012) provide a basic, simple, and understandable introduction to SHP.

STUDY LOCATIONS
The Ciwidey subwatershed, Upper Citarum has been selected as the analogical study location illustrating the use of GIS and hydrological modeling to assess SHP potential.
The study region covers a catchment area of around 204 km 2 with elevations ranging from 624 to 2,425 m a.s.l dominated by agricultural land and rice farms alongside some lowdensity residential areas (Figure 1).The river originates at Wayang Windu Mountain and flows through Bandung City, a major regional city.The Ciwidey subregion has a tropical climate with a mean temperature of 18-25°C.The average annual precipitation is 2,000-2,500 mm.Records show that more than 85% of the precipitation is concentrated in the rainy season (September-December) and transition periods (January-April) leaving the dry season of June, July, and August with little rainfall.In regards to SHP planning, empirical records show that during the years 2006-2013, the precipitation amount in the study location is considered large enough.The number of wet months (with monthly rainfall of more than 100 mm) is 9 months and average monthly precipitation is greater than 150 mm.Such high precipitation, which also identically occurred in the tropical Indonesian regions, implies a high discharge particularly for SHP generation.

DATA AND METHODOLOGIES
In accordance with GIS tools, particularly ArcGIS or QGIS, SWAT comes as an interface program.SWAT is a physically based long-term hydrological simulation model Numbers denote each subbasin which has proven tremendous applicability in various hydrological studies (Abbaspour et al., 2007;Kusre et al., 2009;Abbaspour, 2015;Goyal et al., 2015).The SWAT model also has complete ability to quantify the impact of land management practices in large and complex catchments.The main components or inputs of SWAT are topographical data or Digital Elevation Model (DEM), soil type data, land use/ land cover data, and weather data such as temperature and precipitation.Table I lists the necessary datasets in this study.SWAT divides a watershed into stream network sub-basins, which are later distributed again spatially into multiple HRU (hydrological response unit) which have homogenous land use, slope, and soil type characteristics.The HRU are distributed into sub-basins spatially within SWAT (Neitsch et al., 2001).Theoretically, SWAT is based on the formula of hydrological cycle, known as the water balance equation (Arnold et al., 1998;Goyal et al., 2015), shown here as Equation ( 2): where SW t is the final soil water content (mmH 2 O), SW 0 is the initial soil water content (mm H 2 O), t is the day, R day is the amount of precipitation on day t (mmH 2 O), Q surf is the amount of surface runoff on day t (mmH 2 O), E a is the amount of evaporation and transpiration (ET) on day t (mmH 2 O), W seep is the amount of percolation and bypass through the bottom of the soil profile on day t (mmH 2 O), and Q gw is the amount of return flow on day t (mmH 2 O).

Potential site identification
Figure 2 illustrates the working flow of SHP assessment in this study.Using spatial analyst tools in ArcGIS, the Digital Elevation Model (DEM) with a 30 m resolution from USGS were first extracted to produce delineated stream segments with a minimum threshold of 500 meters.Later, the DEM and stream segment map were overlaid and analyzed to generate the sub-basins distribution.Then, to select potential SHP sites, some requirements were applied during the analysis as follows: (a) potential sites should be located in stream order higher than 2nd order since smaller stream segments (≤ 2nd order) are not considered good sites for SHP due to intermittent flow; (b) the elevation drop or height difference (H) should be higher than 10 meters; (c) distance between each potential site should be at least 1 km.

Discharge assessment in SWAT and model calibration
The SWAT model was set to produce monthly discharge during 2009-2012 to obtain dependable discharge for the purpose of SHP generation.Following this, discharge model calibration was performed by changing the hydrological  parameters to minimize model-data misfits (see Table II).
Monthly discharge data from one available gauging station was used to calibrate the model.To be a good model, the model results should satisfy the requirements of established indices, e.g., coefficient of determination (R 2 ) and Nash-Sutcliffe Efficiency (NSE) as recommended by the American Society of Civil Engineers (Ahl et al., 2008).

Dependable discharge through Flow Duration Curve
In the case of hydropower, a Flow Duration Curve (FDC) describes the water flow resource and is a critical piece of information that feeds into the design of a hydropower project.The FDC is a cumulative frequency curve that shows the percent of time during specified units (e.g., discharge) equaled or exceeded in a given period (Searcy, 1959).FDC was used to obtain dependable discharge (Q) for SHP generation.In this study, FDC at dependability levels of 60%, 75%, and 90% were undertaken accommodating uncertainty level due to seasonal variability.Together with H, the dependable discharges (Q) obtained were input into equation (1) to estimate the potential power output of SHP (p E-SHP ).

RESULTS
Results of this study consist of (a) result of potential sites; (b) output of SWAT model and discharge calibration; (c) results of FDC and power estimation.

Result of potential sites
Nine potential sites were identified that meet the criteria with head of 11-19 m, in streams higher than 2nd order, and where horizontal distance between each site is equal to or greater than 1 km.The potential sites were distributed in five subbasins (see Figure 5 for overall results).

Output of SWAT model and calibration
Using SWAT, we modeled the discharge of each subbasin within the subwatershed.To calibrate the model, observed and modeled discharge data in subbasin 6 were compared and analyzed, as shown in Figure 3.Although (R 2 ) was 0.76, NSE was quite low, at -0.16.Therefore, calibration of the hydrological parameters was necessary to improve the accuracy of model outputs.Calibration was performed for the period of 2009-2012.The sensitivity of each parameter was also evaluated carefully (Patil et al., 2013).The model's performance in reproducing observed discharge improved after the calibration of the hydrological parameter values, with (R 2 ) = 0.75 and NSE = 0.67 (Figure 3; Table II).

Results of FDC and power estimation
The SHP potential assessment was performed through evaluating the dependable discharge at each potential site using FDC analysis.The FDC describes the water flow that can be relied upon as a resource and is a critical piece of information for the design of hydropower projects.Accommodating the impact of seasonal variability towards SHP generation throughout the year, the dependability rates of 60%, 75%, and 90% have been considered to represent  optimum, normal, and minimum discharge thresholds (particularly during dry season).In the example, we found that the dependable discharges of the 5-year modeling period (2009-2013) were ranging from 3.3 to 7.55 m 3 s -1 in subbasin 7. A similar approach was then applied to all nine SHP potential sites in five subbasins to find the dependable discharge of SHP (Figure 4).The SHP potentials were then calculated according to equation (1), using head data obtained from the DEM and dependable discharge based on the SWAT model (Figure 5).We found that overall total potential ranging from 762-1.722 kW can be harnessed in the Ciwidey subwatershed depending on seasonal variability.

CONCLUSION
The development of RE in Indonesia has become an urgent agenda.Nationally, an energy crisis has resulted from a rapid increase in electricity demand and a limited supply of power, together with low penetration of modern energy and electricity infrastructure across rural Indonesia.At the same time, fossil fuel resource depletion and the threat of climate change are also experienced globally.In terms of available resources, hydropower generation is one of the most promising RE sources in Indonesia.However, instead of large-scale hydropower, this study emphasized SHP due to its lower environmental burden and more direct effect on local communities.The traditional SHP planning, in which potential sites are relatively located in remote areas, is time and cost consuming and results remain inaccurate due to a failure to capture complex hydrological phenomena.Hence, with the advancement of geospatial mapping technology, spatial analysis and hydrological modeling can be used to assess SHP potential.For this purpose, we combined GIS-based topographical and spatial analyses with a SWAT model to characterize the hydrological properties of the Ciwidey subwatershed.
Considering the importance of study that can help accelerate Indonesian energy diversification, we suggest that GIS and SWAT should be a proper solution and capable tool for the study of SHP potential assessment.The method is expected to help promote renewable energy, mainly SHP, in Indonesia.However, our findings and experience show how the availability of hydrological data, e.g., gauged discharge station and daily precipitation, become one of the barriers to be overcome to produce a potential assessment study through this approach.For instance, the support of adequate hydrological data can be further used to study potential SHP generation in every Indonesian watershed.The appropriate assessment will help facilitate the development of SHP in the future towards national energy security and attract more investors to support the development of energy infrastructure in Indonesia.

Figure 1 .
Figure 1.Overlaid maps of land use and sub-basin distribution in Ciwidey subwatershed, Indonesia.Areas are covered by agriculture land, rice fields, and some low density urban settlements.29 subbasins were identified in the Ciwidey subwatershed areas through spatial analyst tools in ArcGIS.Numbers denote each subbasin

Figure 2 .
Figure2.Flowchart of small hydropower (SHP) potential assessment method using GIS and the SWAT hydrological model.The formula of SHP potential (p E-SHP ) (kW) is defined as p E-SHP = ρ water × g × Q × H × η where ρ is the water density (kg m -3 ), g is the acceleration due to gravity (9.8 m s -2 ), H is the elevation drop (m), and Q is the dependable discharge (m 3 s -1 )

Figure 3 .
Figure 3. Top figure (a) shows observed and modeled time-series of discharge in subbasin 6 from January 2009 to December 2012: observed (in blue), uncalibrated (with initial hydrological parameter values; in red), and calibrated (with optimized hydrological parameter values; in green) values are included.Bottom figure (b) presents scatter plot of observed vs. calibrated monthly discharge in subbasin 6, R 2 = 0.75

Figure 4 .Figure 5 .
Figure 4. Example of Flow Duration Curve to determine the dependable discharge identified as optimum, moderate, and minimum threshold (60-90% dependability rate) for SHP generation in subbasin 7

Table I .
List of necessary datasets in this study.All maps were projected to UTM/WGS 84 zone 48S

Table II .
Optimal values of hydrological parameters in the SWAT model