You are on page 1of 13

Applied Energy 224 (2018) 309–321

Contents lists available at ScienceDirect

Applied Energy
journal homepage: www.elsevier.com/locate/apenergy

A systematic decision support tool for robust hydropower site selection in T


poorly gauged basins

Abdul Moiza, , Akiyuki Kawasakia, Toshio Koikeb, Maheswor Shresthac
a
Department of Civil Engineering, The University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-8656, Japan
b
International Centre for Water Hazard and Risk Management, 1-6 Minamihara, Tsukuba, Ibaraki, Japan
c
Water and Energy Commission Secretariat, Singhdurbar Western Gate Rd, Kathmandu 44600, Nepal

H I GH L IG H T S

• Systematic site selection of small hydropower schemes in poorly gauged basins.


• Integrating a distributed hydrological model with a hydropower site selection tool.
• Selection of more economic sites in a hydropower system considering basin hydrology.

A R T I C LE I N FO A B S T R A C T

Keywords: The power output of run-of-river small hydropower (SHP) developments is very site sensitive and poses several
Decision support complex challenges, such as inaccessible terrain and numerous possible hydropower scheme alternatives. We
Run-of-river developed a geographic information system-based decision support tool that systematically evaluates all possible
Site selection hydropower scheme alternatives to assist decision makers in assessing the hydropower potential of large data-
Distributed hydrological model
scarce regions more objectively. A water and energy budget – based distributed hydrological model and a
preference ranking organization method for enrichment evaluations are integrated and employed in the de-
veloped tool. This approach enables the inclusion of both topographical and hydrological factors in the site
selection process, allowing the use of hydropower potential as a maximizing criterion during site selection. This
paper explores the consideration of topographical factors with and without hydrological factors as approaches
for optimized site selection. An application of the tool in the case of Kunhar River Basin in Pakistan demonstrates
its robustness. For equivalent criteria weights, site selection considering both factors identified a significantly
lower number of sites with a shorter waterway length compared to when only topographic factors were con-
sidered. However, both approaches identified essentially the same hydropower potential. Notably, the in-
tegration with a distributed hydrological model and the incorporation of hydropower potential as a maximizing
criterion for site selection revealed more economically attractive SHP sites. This approach enables a flexible and
rigorous evaluation of the hydropower potential of large poorly gauged basins, which is particularly useful for
developing countries.

1. Introduction hydropower capacity was estimated at 1246 GW in 2016, with 31.5 GW


installed in the same year [3]. However, it was also estimated that in
With the continuous depletion of fossil fuel reserves, energy devel- 2000, approximately 70% of the economically feasible hydropower
opment trends have shifted more toward renewable energy sources. By potential had yet to be developed [4]. In this context, the importance of
the end of 2015, hydropower was still the primary contributor to the hydropower cannot be denied, and this importance is further high-
global renewable energy mix, with 16.6% of global renewable elec- lighted by the fact that the expansion of the electricity sector in de-
tricity generated by hydropower [1]. Despite being a relatively mature veloping countries plays a significant role in enhancing their economic
technology, future projections indicate continual growth in the hydro- growth [5]. The goal of the study presented herein is therefore to devise
power sector. This trend is particularly true for small hydropower (SHP) a systematic tool for decision support during the planning phase SHP
developments [2]. According to the hydropower status report, global site selection process for promoting the development of untapped


Corresponding author.
E-mail address: abdul.moiz@hydra.t.u-tokyo.ac.jp (A. Moiz).

https://doi.org/10.1016/j.apenergy.2018.04.070
Received 16 January 2018; Received in revised form 23 March 2018; Accepted 26 April 2018
0306-2619/ © 2018 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/BY-NC-ND/4.0/).
A. Moiz et al. Applied Energy 224 (2018) 309–321

hydropower potential, particularly in data-scarce developing countries. of the site to be known before the discharge is simulated. To tackle this
Over the past decade, SHP has gained substantial popularity be- issue, Soulis et al. suggested the use of distributed hydrological models
cause of its significantly lower adverse impacts compared with its larger to estimate streamflow values at sites that are not known during the
counterparts. Being run-of-river in most cases, these SHP developments hydrological model setup [28]. A few researchers have also developed
generally do not possess a storage component, and as such, they have rather complicated tools and algorithms that evaluate a number of al-
lower socio-environmental impacts [6]. In contrast, the absence of a ternative sites and suggest the best alternatives based on an overlay
storage component makes hydropower potential very sensitive to the analysis [29,30]. However, these too rely upon the availability of re-
location of the project site as it is a function of the gross head and the gional FDCs.
available flow, both of which vary spatially. In a conventional site se- Recent advances in the area of distributed hydrological modeling
lection approach, it is rather cumbersome and time consuming to have led to the development of a Water-and-Energy-Budget-based
consider all the possible site alternatives in a spatial domain; hence, the Distributed Hydrological Model (WEB-DHM) that is capable of giving
site selection process may be adversely affected by the subjectivity of distributed representations of the spatial variation and physical de-
the decision maker, and the optimal project site may not be selected. scriptions of runoff generation and routing in river channels at basin to
Hydropower potential may be evaluated at four different levels de- continental scales [31,32]. The snow physics of the model were further
pending on the constraints applied to power development [7]. The improved by the incorporation of the three-layered snow scheme of
hydropower potential evaluated without the application of any con- Simplified Simple Biosphere model version 3 [33,34] with the albedo
straints is known as the theoretical potential. Not all theoretical po- scheme of the biosphere atmosphere transfer scheme [35,36]. With this
tential can be utilized to generate hydropower; the exclusion of re- improved snow process representation, WEB-DHM is referred to as
stricted or protected areas and the consideration of environmental flows WEB-DHM-S. The strength of the improved model lies in its ability to
reduce the theoretical potential to yield planning potential. Planning simulate the long-term continuous spatial distribution of the snow
potential is then further constrained by the consideration of turbine variables in addition to other hydrological processes [37–39]. Since
efficiencies and energy losses in the waterway to give technical po- potential hydropower sites are often located in river basins pre-
tential [8]. Economically feasible potential can then finally be eval- dominantly fed by snow, accurate modeling of snow hydrology of the
uated by considering the site-specific cost analysis of the hydropower river basin has profound significance for hydropower potential esti-
project. mation. As the use of distributed hydrological model provides dis-
The development of geographic information systems (GISs) has tributed representations of simulated runoff, it is also possible to derive
motivated several renewable energy resource potential and site selec- distributed environmental flows based on these physically based si-
tion studies in the past, such as for solar energy [9–11], biomass energy mulations, which are often neglected in such GIS-based tools or are
[12,13], and wind energy [14,15] site selections. However, in all these simply parameterized based on the morphological characteristics of the
cases, the energy potential is distributed over a uniform grid, which is basin [7].
in contrast to the case of hydropower, which has the energy potential Since the selection of a suitable SHP site involves several different
distributed over an intricate network of streams and in which the se- criteria (many of which can be quite contrary to each other), it is rather
lection of one hydropower generation site may interfere with the se- difficult to achieve a solution that is truly optimal for every criterion. In
lection of another site. To tackle this challenge, researchers over the such a case, Multi-Criterion Decision Making (MCDM) techniques can
years have developed tools of varied complexity through the creative be used to find a compromise solution by considering many different
use of GIS technology. However, these tools are often developed for criteria. One such family of MCDM techniques called the Preference
specific locations and have limited applicability in other areas owing to Ranking Organization METHod for Enrichment Evaluations (PROME-
their dependence upon certain databases that are unique to a particular THEE), developed by Brans, has gained particular attention because of
locality [16]. Ballance et al. used the coefficient of variation and low its flexible design and the fact that it only requires information that is
flow index to identify areas that were less susceptible to flow variability easily understood by decision makers [40–42]. This approach can be
and estimated the hydropower potential using derived slope maps and contrasted with other MCDM techniques such as the ELimination Et
regional mean annual runoff datasets for micro- and macro-hydropower Choix Traduisant la REalité (ELECTRE), the process and outcome of
schemes [17]. The general approach to evaluating the hydropower which may be difficult to understand from a decision maker’s point of
potential of a river basin consists of (i) specifying a number of points on view [41]. Other famous MCDM techniques, such as the Analytic
the stream network separated by a fixed interval, (ii) assuming these Hierarchy Process (AHP), become increasingly complex as the number
points to be potential sites, and (iii) evaluating the hydropower po- of alternatives and criteria increases [42]. In addition, these PROME-
tential at these sites using digital elevation models (DEMs) to calculate THEE methods have significant utility in addressing multi-criteria water
the gross head and regional flow duration curves (FDCs) or catchment resource problems [43]. Mladineo et al. suggested the application of
area to estimate the available flow [18–22]. Rojanamon et al. further PROMETHEE methods for ranking SHP location alternatives largely
considered the environmental and social impacts based on local data- because of the flexibility they offer to the decision makers in expressing
sets and field questionnaire surveys in addition to engineering and their preferences and because any number of criteria can be accom-
economic criteria for selecting suitable hydropower sites [23]. Palo- modated using these methods [44].
mino et al. evaluated theoretical hydropower potential under global Considering the developments in the areas of GIS, distributed hy-
changes, such as population growth and economic development [24]. drological modeling, and MCDM techniques, this paper primarily fo-
The study by Garegnani et al. is one of the first works that explicitly cuses on the objective of developing a systematic decision support tool
calculates hydropower potential at all four levels (theoretical, planning, for robust hydropower site selection that integrates the spatial data
technical and economic) discussed earlier [7]. handling capabilities of GIS, the advanced distributed snow modeling
However, the abovementioned approaches have limited applic- capabilities of WEB-DHM-S, and the flexible multi-criteria ranking
ability in poorly gauged areas where regional FDCs and gridded mean capability of PROMETHEE methods to rigorously evaluate all the pos-
annual discharge datasets are not available and hence are not applic- sible SHP site alternatives and suggest optimal potential sites for SHP
able globally. A number of studies have suggested the use of hydro- development. Second, the paper highlights two different approaches for
logical models to simulate the discharge at sites screened by a GIS optimized SHP site selection, namely, (i) considering only topo-
analysis [25–27]. However, the hydrological models employed in these graphical factors and (ii) considering both topographical and hydro-
studies, such as the Soil and Water Assessment Tool (SWAT) and Hy- logical factors and drawing a comprehensive comparison between
drologic Engineering Center’s Hydrologic Modeling System (HEC- them. Lastly, the capability of the developed tool is demonstrated by its
HMS), are of a pseudo-distributed nature and require the exact location application to the Kunhar River Basin, a snow-fed basin in the northern

310
A. Moiz et al. Applied Energy 224 (2018) 309–321

Fig. 1. Kunhar River Basin, Pakistan. (a) Map shows the location of the meteorological sites and discharge gauges used in this study and the basin elevation. (b)
Average precipitation (mm) over the year. (c) Average temperature (°C) over the year. (d) Average discharge (m3/s) over the year. (1971–2009).

region of Pakistan, as a case study to evaluate its hydropower potential. drastically from 18.6 °C at Balakot to approximately 7.3 °C at Naran.
The scope of this study is limited to the level of planning potential. These estimates are based on the climatological data collected from the
To the best of the authors’ knowledge, this is the first work that com- Pakistan Meteorological Department (PMD) and the Water And Power
pletely integrates a physically based distributed hydrological model Development Authority of Pakistan (WAPDA) for the period 1971–2010
with a GIS-based site selection and potential evaluation tool. The novel (Fig. 1b–d). Two peaks can be observed in the average precipitation
tool developed in this study can aid hydropower planners and decision over the year, the first occurring in March as a result of western dis-
makers in systematically and objectively surveying large poorly gauged turbances and the second in July, particularly in the southern region of
river basins for potential SHP locations and in evaluating the associated the basin, driven by the summer monsoon. The precipitation caused by
hydropower potential in a short time and a robust manner. As will be western disturbances mainly accumulates in the form of snow in the
demonstrated in the subsequent sections, the tool relies heavily upon northern areas and contributes to a peak in the annual hydrograph
datasets that are available at a global scale, thus limiting its reliance during June–July as snowmelt [45].
upon local datasets.
3. Methodology
2. Study area
This paper discusses the development of a systematic SHP site se-
The study area selected for this study is the Kunhar River Basin, a lection tool that makes the planning phase of site selection process less
sub-basin of the transboundary Jhelum River Basin, located in the time consuming, more objective, and more robust. A run-of-river-type
northern region of Pakistan. The basin lies between the longitudes of hydropower development is generally established by the construction
73°17′E and 74°08′E and the latitudes of 34°11′N and 35°10′N, of a weir that diverts the water through a waterway and utilizes the
stretching over an area of 2632 km2 (Fig. 1a). Originating from Lulusar natural head of the waterfall to generate electricity at the hydropower
Lake in Kaghan Valley, the Kunhar River drains the southern slopes of station. The gross natural head is dictated by the location of the di-
the Greater Himalayas, and approximately 65% of the discharge is version site and the hydropower station, which are connected by a
generated from snowmelt [45]. A number of settlements line the waterway. The catchment area draining into the diversion site, the
Kunhar River, e.g., Patrind, Balakot, Paras, Jared, Mahandri, Khannian, magnitude and availability of flow at the diversion site, the gross head
Kaghan, Kinari, Paludran, Naran, and Batakundi from downstream to between the diversion site and the hydropower station, and the length
upstream. The elevation of the basin changes precipitously from 636 to of the waterway connecting the diversion site to the hydropower station
5216 m above sea level, making it one of the most promising basins for have been identified as the most significant factors that should be
hydropower development in Pakistan [46]. considered during the site selection process [47].
The basin exhibits a humid subtropical climatology, with the Fig. 2 shows the detailed analytical framework of the developed
average annual precipitation ranging from 1619 mm at Naran to tool. At first, WEB-DHM-S is set up for the target basin to simulate the
1610 mm at Balakot. The average annual temperature decreases basin hydrology and to generate the FDCs in each hydrological unit.

311
A. Moiz et al. Applied Energy 224 (2018) 309–321

Fig. 2. Analytical framework of the integrated tool.

The developed tool analyzes all the possible alternative hydropower static and dynamic datasets. Static datasets include the DEM (Fig. 1a),
schemes and finally selects the finite set of non-conflicting alternatives soil hydraulic characteristics (Fig. 3b), and static vegetation parameters
that gives the highest hydropower potential for the shortest length of (Fig. 3c). Dynamic datasets include meteorological forcings (precipita-
waterway to attain economic efficiency. To achieve this, PROMETHEE tion, air temperature, wind speed, relative humidity, downward
II with complete ranking was integrated into the developed tool. The shortwave radiation, downward longwave radiation, and air pressure)
tool was developed as a user-friendly Python script tool in ArcGIS with and dynamic vegetation parameters (leaf area index (LAI) and fraction
an interface similar to that of other tools in ArcGIS (see supporting of photosynthetically active radiation (FPAR)). The model was set up at
information Fig. S1). The subsequent sections discuss each component a grid resolution of 300 m. The basin is discretized into seventy-three
in detail. sub-basins based on the Pfafstetter codification scheme [48] (Fig. 3a) so
that each sub-basin has at most one stream on which the SHPs can be
3.1. WEB-DHM-S setup located, each of which is further sub-divided into flow intervals de-
pending on the time of concentration. Each flow interval is 1.5–2 times
To set up WEB-DHM-S, two sets of data must be prepared, namely, the model grid size. The runoff from each model grid is accumulated

Fig. 3. Static inputs for WEB-DHM-S. (a) Basin is sub-divided into seventy-three sub-basins using second-level Pfafstetter codification scheme with unique colors
indicating the first level Pfafstetter codification. (b) Soil type. (c) Land cover type.

312
A. Moiz et al. Applied Energy 224 (2018) 309–321

Table 1
Inputs for WEB-DHM-S setup and evaluation.
Data Spatial Resolution Temporal Resolution (Extent) Source

Static Inputs
DEM 30 m Static SRTMa
Soil type 5 arc minutes Static FAOb
Land use 1000 m Static USGSc

Dynamic Inputs
Precipitation Point Daily (1977–1997; 2007–2009) PMD; WAPDA
Meteorological data (precipitation, air temperature, wind speed, relative humidity, downward 0.5625° 3-hourly (1977–1997; JRA55d
shortwave radiation, downward longwave radiation, and air pressure) 2007–2009)
Leaf Area Index (LAI) 1000 m 8-day composite (2007–2009) MODIS Terra
(MOD15A2)e
Fraction of Photosynthetically Active Radiation (FPAR) 1000 m 8-day composite (2007–2009) MODIS Terra
(MOD15A2)e

Evaluation Data
Discharge Point Daily (1978–1997; 2008–2009) WAPDA
Snow cover 500 m 8-day composite (maximum MODIS Terra
snow extent) (MOD10A2)e

a
Shuttle Radar Topography Mission https://earthexplorer.usgs.gov/.
b
Food and Agriculture Organization [49].
c
United States Geological Survey. https://lta.cr.usgs.gov/GLCC.
d
Japanese Reanalysis 55-year. http://jra.kishou.go.jp/JRA-55/index_en.html.
e
MODerate resolution Imaging Spectroradiometer. https://reverb.echo.nasa.gov/.

into a flow interval and routed into a virtual channel toward the basin 3.2. SHP site selection tool
outlet. For the purposes of this study, the simulated discharge at each
flow interval is stored in a database that can be conveniently referred to The tool has been developed in such a way that it can be used with
using the unique sub-basin and flow interval IDs. All the necessary two different approaches for site selection. Site selection can be con-
static and dynamic datasets (specific to Kunhar River Basin) required to ducted based on either (i) only topographical factors or (ii) both to-
set up and evaluate WEB-DHM-S are given in Table 1. Point pre- pographical and hydrological factors. In approach (i), sites are selected
cipitation data were spatially interpolated using inverse distance by maximizing the gross head between the diversion site and the hy-
weighted interpolation. For detailed model physics, please refer to dropower plant and by maximizing the catchment area at the diversion
[31,32,37–39]. site, whereas the length of the waterway is minimized. The hydropower
The model is initialized on 29 August 2007 (when the snow cover is potential is then evaluated at the selected sites using the FDCs gener-
minimal) and is allowed to spin up for 1-year multiple times until hy- ated from the runoff simulated by distributed hydrological modeling.
drological equilibrium is reached. The model is then calibrated for the However, in approach (ii), the hydropower potential of each alternative
year 2008 and 2009 using the observed discharge at Talhata as well as is evaluated first by using the simulated runoff, and then, sites are se-
the MODIS snow cover. Since the MODIS dataset is available only after lected by maximizing the hydropower potential and minimizing the
the year 2000, calibration of the model is conducted for a period fol- length of the waterway. Approach (ii) should identify a better set of
lowing the validation period. The performance of the model was eval- alternatives because in this case, a complete redistribution of the sites is
uated using two quantitative statistics: the Nash – Sutcliffe efficiency achieved based on the actual hydropower potential calculated from the
(NS) and percent bias (PBIAS): distributed FDCs. In addition, this approach also makes it possible to
n control the scale of hydropower development, which is an obvious
∑i = 1 (Qiobs−Qisim)2
NS = 1− n
advantage over approach (i).
∑i = 1 (Qiobs−Qmean )2 (1)
n 3.2.1. Pre-processing and pre-screening search algorithm
∑i = 1 (Qiobs−Qisim) × 100
PBIAS = n , The basin under consideration is first delineated using the DEM and
∑i = 1 Qiobs (2)
the location of the basin outlet through the development of the flow
where Qiobs
is the observed discharge, Qisim
is the simulated discharge, direction and a flow accumulation grid. Following basin delineation,
Qmean is the mean of the observed discharge time series over the period the stream network grid is generated from the flow accumulation grid
of simulation, and n is the total number of observations. The spatially based on a minimum stream definition threshold that is specified as an
distributed snow cover simulation is compared with the MODIS snow input parameter in the tool. The stream network grid is then codified
cover to further evaluate model accuracy. Several sensitive soil, land using the Strahler stream order codification scheme [50]. The algorithm
and snow parameters that have been calibrated in this study are listed then loops through each stream in a given stream order. In the case of
in Tables 2 and 3. Default values have been adopted for the remaining the Kunhar River Basin, streams up to the fourth order were generated.
parameters; these are given in [31,32,37–39]. Each stream is then converted to a polyline and densified into points
However, to generate more reliable FDCs for evaluating hydro- using a polyline densification scheme (adopted from a custom ArcGIS
power potential, long-term data are required. Therefore, after calibra- tool at ianbroad.com). The stream densification interval is specified as
tion, the model is initialized on 29 August 1977 assuming minimal an input parameter in the tool, and it primarily controls the computing
snow cover and is validated for 21 years (1978–1998). Owing to the time required by the tool since each point generated will be processed
unavailability of the MODIS LAI and FPAR product (MOD15A2) during along with its alternatives. This interval also serves as the minimum
this period, the average for 2001–2009 is used as a substitute assuming distance between two consecutive hydropower schemes along the
that the long-term changes in vegetation dynamics do not significantly stream. The search algorithm begins by selecting the most downstream
affect the simulated discharge. The discharge was calibrated and vali- point and then moves upstream. The selected point is assumed to be a
dated at Talhata. hydropower station, and then, an annular search zone is established

313
A. Moiz et al. Applied Energy 224 (2018) 309–321

Table 2
Soil, land and snow calibration parameters in WEB-DHM-S.
Symbol Parameters Unit Soil Type Source

Cambisols, Fine Cambisols, Medium Lithosols

Ksat1 Hydraulic conductivity of 1st layer of unsaturated zone mm h−1 18.22 42.61 61.31 Optimization/FAO
Ksat2 Hydraulic conductivity of 2nd layer of unsaturated zone mm h−1 3.64 8.52 12.26 Optimization/FAO
Kg Groundwater hydraulic conductivity mm h−1 0.36 0.85 1.23 Optimization/FAO

anik Anisotropic hydraulic conductivity ratio 1.4 Optimization


SSTmax Maximum surface water detention mm 35.0 Optimization

Γlapse Air temperature lapse rate K m−1 -8.5 Optimization/Observation


∝vis Fresh snow albedo in visible band 0.75 Optimization
∝nir Fresh snow albedo in near-infrared band 0.65 Optimization
CF Calibration parameter for snow correction factor m−1 0.00026 Optimization
Tth Snow/rain threshold temperature °C 2.0 Optimization

Table 3 accumulation grid. Moreover, the ID of the sub-basin and flow interval
Inputs for the SHP site selection tool. grid is also assigned to the diversion site within which it falls. In ap-
Input Value/Type Source
proach (ii), the FDCs are constructed for every diversion site to obtain
the design discharge against a specified percentage of flow exceedance.
Elevation Raster Grid (30 m) SRTM Potential candidates are then screened out based on specified
Target Basin Outlet Point – thresholds, namely, minimum gross head, minimum waterway slope,
Stream Densification Interval 500 m [26]
and minimum catchment area. This screening is done to ensure that
Maximum Waterway Length 5000 m Decision Maker
Minimum Waterway Length 2000 m Decision Maker only sites with some significant gross head and discharge are con-
Minimum Allowable Gross Head 30 m (for medium & high [52] sidered in the analysis to reduce the computational costs. In the case of
head) approach (i), the catchment area draining into the diversion site is used
Minimum Allowable Slope 2% [26]; Decision
as an indicator of flow availability. For approach (ii), the hydropower
Maker
Minimum Allowable Flow 50 km 2
Decision Maker potential for each screened alternative is evaluated using the power
Accumulation equation:
Exclude Sites in Protected Areas True (Optional) –
P = ηρgQd H (3)
Protected Areas Polygon Shapefile [53]
(Optional)
Buffer around Protected Areas 5000 m (Optional) Decision Maker Qd = (Qsim−Qe )FE (4)
Approach (i) Topographic Factors –
(ii) Topographic and Qe = minQsim × e (5)
Hydrological Factors
where P is power, η is system efficiency, ρ = 1000 kg m−3 is the density
Pfafstetter Sub-basin Grid Grid (300 m) (Optional) WEB-DHM-S
Pre-processing of water, g = 9.81 m s−2 is the acceleration due to gravity, H is gross
WEB-DHM Flow Interval Grid Grid (300 m) (Optional) WEB-DHM-S head, Qd is the design discharge, Qsim is the simulated discharge, and Qe
Pre-processing is the environmental flow expressed as a fraction e of the minimum
System Efficiency 80.0% (Optional) [25] daily historical discharge. According to the recent practice of WAPDA,
Flow Exceedance 70.0% (Optional) Decision maker
the value of e can range from 0.2 to 0.3. Qd is taken at a particular flow
Environmental Flow (as fraction 0.25 WAPDA
of minimum daily historical exceedance FE to take into account the seasonal variation of the river
discharge) flow. FE depends on the number of days that the hydropower genera-
Minimum Allowable 2 MW (for SHP) (Optional) [52] tion system is expected to run at full capacity in a year, which in turn is
Hydropower
governed by the energy mix of the country’s power generation system.
Maximum Allowable 25 MW (for SHP) [52]
Hydropower (Optional) The scale of hydropower development is controlled by specifying the
minimum and maximum target hydropower potentials.

around it, with the inner and outer radii representing the minimum and 3.2.2. PROMETHEE compromise site selection
maximum allowable lengths of the waterway, which can be specified as To rank the screened alternatives, a PROMETHEE II (complete
inputs to the tool. This is done to allow the user control over the length ranking) approach is adopted at each search point to allow a pairwise
of the waterway and to avoid the development of unrealistically long or comparison and a complete ranking of the alternatives [40]. PROME-
short waterways. All the remaining points that fall within the annular THEE II is selected because compared with other MCDM techniques, it
search zone are assumed to be diversion site alternatives, and a number requires limited and easily understood inputs from the decision maker
of alternative schemes are constructed, which connect each diversion [41,42]. The only inputs required are the evaluation criteria, the pre-
site alternative with the hydropower station under consideration ference function and the criteria weights. Moreover, it also reflects the
(Fig. 4a). preferences of the decision maker for each criterion, which can vary
Alternative schemes are only constructed for those diversion site/ significantly from region to region. The objective is to maximize the
hydropower station combinations that do not fall in the buffer zone hydropower potential and minimize the length of the waterway, as
created around any environmentally protected areas. Gross head is shown below.
calculated as the difference between the elevation of the diversion site
Approach (i ) max{c1 (s ),c2 (s ),−c3 (s )|s ∈ S } (6a)
and hydropower station, and slope is calculated by dividing this ele-
vation difference by the distance between the diversion site and hy- Approach (ii ) max{c4 (s ),−c3 (s )|s ∈ S }, (6b)
dropower station. The catchment area at the diversion site is evaluated
based on the number of cells draining into that point using the flow where c1,c2,c3, and c4 are the evaluation criteria for gross head, catch-
ment area, waterway length, and hydropower potential (evaluated

314
A. Moiz et al. Applied Energy 224 (2018) 309–321

Fig. 4. Functionality of the SHP site selection tool. (a) Pre-screening search algorithm. (b) PROMETHEE II site selection. (c) Cut-off tests.

using Eq. (3)), respectively. S is a set of all screened hydropower adopted in this study is given by
scheme alternatives {s1,s2,…,sn} . It should be noted that besides these
four criteria, other criteria should also be included in Eq. (6) depending ⎧0 dk (si,sj ) ⩽ q
⎪ dk (si,sj) − q
upon the level at which hydropower potential is to be evaluated. For Pk [dk (si,sj )] = q < dk (si,sj ) ⩽ p
⎨ p−q
example, in the present case, the only measure of cost used is the length ⎪1 dk (si,sj ) > p (8)
of the waterway. However, in the case of economically feasible po- ⎩
tential, site-specific cost factors such as the distance to the electric grid where P is the preference function, d is the deviation, q is the in-
(or the closest consumer), length of access roads, turbine technology difference threshold, and p is the strict preference threshold. The pre-
(depending upon flow characteristics), diameter of the waterway tunnel ference values for all the criteria are then aggregated into a single index
(depending upon flow characteristics), and compensation to land as follows:
owners, if estimated appropriately, can be included as criteria in Eq. (6)
N
for site selection.
Each alternative is compared with every other alternative by cal-
π (si,sj ) = ∑ Pk [dk (si,sj )] wk
k=1 (9a)
culating the deviation between the performance of each alternative and
N
each criterion considered. For example, for two screened alternatives, si
and sj , and for the k th criterion, the deviation is calculated as
π (sj,si ) = ∑ Pk [dk (sj,si )] wk,
k=1 (9b)
dk (si,sj ) = ck (si )−ck (sj ). (7) where wk is the relative weight for the k th
criterion and π (si,sj ) and
π (sj,si ) indicate the degree to which si is preferred over sj and vice versa,
These deviations are then translated into preferences using a pre-
respectively.
ference function defined by the decision maker. For the sake of sim-
The outranking flows are calculated by comparing each alternative
plicity, in this study, a V-shaped preference function is adopted.
si against the (n−1) other screened alternatives in S .
However, depending on the way that these preferences reflect reality,
the decision maker may assign a preference function other than V- 1
shaped, such as Usual, U-shaped, Level, V-shaped with indifference
Positive Outranking Flow Φ+ (si ) =
n−1
∑ π (si,s )
(10a)
criterion and Gaussian preference functions [51,40]. The preference
1
function allows the decision maker to specify thresholds below which Negative Outranking Flow Φ− (si ) =
n−1
∑ π (s,si ).
(10b)
the deviation is insignificant and above which the deviations are so
significant that a strict preference is established for one alternative over The positive outranking flow indicates how the alternative under
another. The mathematical representation of the preference function consideration outranks the other (n−1) alternatives, whereas the

315
A. Moiz et al. Applied Energy 224 (2018) 309–321

negative outranking flow indicates how the alternative under con- assignment of a different snow correction factor for various elevation
sideration is being outranked by (n−1) alternatives. bands might reduce this discrepancy, as suggested by [37].
For PROMETHEE II (complete ranking), the net outranking flow is To further facilitate the calibration of parameters and evaluation of
calculated as the model, the model-simulated snow cover was compared with the
MODIS-derived snow cover and the bias between them was calculated.
Φ(si ) = Φ+ (si )−Φ− (si ) (11)
A pixel-by-pixel comparison suggested that the model demonstrates
All the screened alternatives within the annular search zone are high accuracy in simulating the snow cover (see supporting information
ranked based on their net outranking flow; ultimately, the scheme with Fig. S2.1-S2.2). On some days, however, the model underestimates the
the highest net flow is selected (Fig. 4b). snow cover due to the limited observation stations in high-elevations
areas, resulting in a poorly interpolated precipitation grid. This effect is
3.2.3. Cut-off tests particularly noticeable in the case of the Kunhar River Basin as the
Following pre-screening and PROMETHEE-based compromise site basin is dominated by a number of small-scale valleys; as a result, most
selection within the annular search zone, a number of sites are selected; precipitation events are driven by the local orography rather than large-
however, many of these sites cannot be developed simultaneously be- scale processes.
cause they are cut off by some other development, as shown in Fig. 4b. For the validation of long-term simulations (1978–1998), the results
A set of three cut-off tests is conducted to determine whether the se- are reasonable during 1988–1998; however, a significant under-
lected scheme is being interfered with by other schemes and to finally estimation is observed in 1978–1987, particularly during the snowmelt
generate a non-conflicting set of selected schemes, as shown in Fig. 4c. season (Fig. 6a). Since Naran is the only high-altitude observation
The net outranking flows for all the schemes are calculated again station, further investigation revealed that the precipitation measure-
using the same methodology described in Section 3.2.2, but this time, ment at Naran seems inconsistent as the average annual precipitation
each scheme is compared with all the other (n−1) schemes in the entire drastically changes from 1047 mm in the period 1978–1987 to
basin. The scheme with the highest net outranking flow is selected and 2277 mm in the period 1988–1998. This change in the average annual
tested for interferences. There are at most three possible cases of in- precipitation between these two periods was not as significant for Ba-
terference, namely, (i) the diversion site of another scheme is in the cut- lakot or Muzaffarabad (see supporting information Fig. S3). To address
off portion of the selected scheme, (ii) the hydropower station of an- this issue, a different snow correction factor is applied to the two per-
other scheme is in the cut-off portion of the selected scheme, or (iii) the iods. The value of the snow correction factor was increased from
selected scheme is cut off by another scheme. In this way, the inter- 0.00026 to 0.0008 m−1 to take into account this inconsistency between
fering schemes for every selected scheme are discarded until there are the two periods. The results show a significant improvement with an
no more interferences. The remaining schemes are stored as the final set increase in the NS coefficient from 0.65 to 0.74 and a reduction in the
of selected schemes. PBIAS from −21.67% to −7.2% (Fig. 6b and supporting information
Fig. S4).
3.2.4. Tool application
This section discusses the inputs used for the application of the tool
to the Kunhar River Basin, as summarized in Table 3. It should be noted 4.2. Distributed representation of FDCs
that all the spatial inputs have been re-projected to the UTM 43N WGS
1984 coordinate system. Each sub-basin has a number of flow intervals. In this study, the
The Kunhar River Basin possesses a relatively steep topography; as simulated discharge Qsim is archived for each sub-basin – flow interval
such, the minimum permissible head has been set so that only medium- combination. The environmental flows Qe are also evaluated for each
and high-head sites are analyzed. Although there is no generally agreed sub-basin – flow interval combination using Eq. (5) with e = 0.25 based
upon definition of medium- and high-head hydropower plants (as it on the minimum daily discharge simulated over 21 years. This is the
varies from region to region), medium- and high-head hydropower portion of discharge that cannot be used for hydropower generation and
plants are defined herein as having a gross head greater than 30 m. must be released to sustain the basin ecology in the cut-off region of the
Since the target scale of the development is SHP, a threshold of hydropower scheme. The spatial distribution of Qe for the whole basin is
2–25 MW has been set in this study; however, this threshold also varies shown in Fig. 7. The environmental flows are lower in the tributaries,
from region to region and should, in general, be decided by the hy- and they increase substantially in the downstream direction along the
dropower planner or decision maker [52]. A design discharge corre- main river, dictating the discharge available for hydropower genera-
sponding to 70% flow exceedance is adopted based on the assumption tion.
that the hydropower plant is only used for base load generation, which The resulting environmental flows are then subtracted from Qsim to
is reasonable since in the case of Pakistan, during periods of low flow, obtain the discharge available for hydropower generation at each grid.
hydroelectric power is supplemented by thermal power. FDCs are then plotted for each of these sub-basin – flow interval com-
binations, as per Eq. (12).
4. Results and discussion
M
FE = 100 ×
4.1. Calibration and validation of WEB-DHM-S D+1 (12)

Fig. 5 shows a comparison of the simulated discharge (resampled to where M is the rank of discharge value and D is the number of events in
daily temporal resolution) with the observed discharge at Talhata for a period.
the calibration period. The simulated discharge in general agrees well The distribution of the FDCs over the hydrological units (Fig. 8)
with the observed one with an NS coefficient of 0.9 and 0.96 for the gives a more explicit representation of the hydrological suitability of
years 2008 and 2009, respectively. The water budget is also well si- the site for hydropower generation. In addition, this result can also give
mulated by the model with a PBIAS (%) of −1.86% and −6.3% for the us an indication of how hydrologically sensitive the location of the
years 2008 and 2009, respectively. Some differences can be observed in project site is in a particular sub-basin. For example, the water avail-
the months of May and August, which may be because the air tem- ability in sub-basins 5 and 9 varies considerably depending upon the
perature lapse rate varies considerably throughout the year in the basin location in the stream; however, this variation is not significant in the
under consideration, but in this simulation, only a constant air tem- case of sub-basins 1, 3 and 7.
perature lapse rate is assumed for the entire year. Second, the

316
A. Moiz et al. Applied Energy 224 (2018) 309–321

Fig. 5. Comparing the simulated and observed discharge at Talhata for 2008–2009 (calibration period).

Fig. 6. Long-term validation of the discharge simulation (1978–1998). Comparison of the observed discharge with the simulated one for the (a) same value and for
(b) different values of the snow correction factor.

4.3. Comparative analysis of approach (i) and approach (ii) the category of SHP are also identified as promising sites. However, in
the case of approach (ii), all these sites are filtered out from the very
When using only the topographic factors for site selection, a total of beginning of the analysis. Second, approach (i) tended to select schemes
40 schemes are identified with total basin hydropower potentials of based only on the gross head and flow accumulation, both of which are
1023, 388, 239 and 149 MW at 30%, 50%, 70% and 95% flow ex- topographical factors. As a result, many sites are identified in the upper
ceedances, respectively. However, the schemes identified on a first- mountainous reaches of the basin. However, in the case of approach
order Strahler stream segment have been excluded from further ana- (ii), these sites are not identified since in this case, schemes are iden-
lyses due to exceedingly low runoff in these reaches and because the tified based on the actual hydropower potential, which is a function of
second-level Pfafstetter delineation only covers the Strahler stream the discharge in addition to the gross head.
orders higher than one in the model developed, and hence, there are no The significance of approach (ii) is quite evident because under
flow intervals that represent first-order streams (for detailed results for equivalent criteria weights, this approach can select a combination of
each individual selected site, please see supporting information Table run-of-river schemes capable of generating roughly the same hydro-
S1). power as approach (i) (235 MW at 70% flow exceedance) with a sig-
Following this exclusion, approach (i) was able to identify 36 sites, nificantly lower number of hydropower sites (Fig. 10). A smaller
whereas only 26 sites were identified by approach (ii), as shown in number of hydropower sites could mean a lower cost for achieving the
Fig. 9. Using approach (i), the total basin hydropower potential was same generating capacity in the basin. Moreover, the total length of the
evaluated to be 1004, 382, 235 and 147 MW compared with 1013, 388, waterway is also significantly reduced from 85 km when using ap-
238 and 149 MW when using approach (ii) at 30%, 50%, 70% and 95% proach (i) to 72 km when using approach (ii), thus achieving a higher
flow exceedances, respectively. Equivalent criteria weights were used economic efficiency for basin-scale development.
for both approaches. There are two main reasons for the different The novelty of the developed tool lies in the notion that by using
numbers of sites identified using both approaches. One reason is that approach (ii), a complete redistribution of the hydropower schemes is
approach (ii) makes it possible to control the scale of hydropower de- achieved through the consideration of the spatially heterogeneous hy-
velopment (in terms of generation capacity); however, this is not pos- drology of the basin and by selecting a non-conflicting set of hydro-
sible in the case of approach (i), so a number of sites that do not fall in power schemes with essentially the same hydropower potential but at a

317
A. Moiz et al. Applied Energy 224 (2018) 309–321

historical flows without the need to parameterize it based on catchment


area. Additionally, the physical basis of WEB-DHM-S requires only a
minimal effort to calibrate it in comparison to conceptual hydrological
models. The capability of the tool is demonstrated in its application to
the Kunhar River Basin in Pakistan. The tool is globally applicable as it
relies on datasets that are available with global coverage (except for the
gauge precipitation required as an input and the gauge discharge re-
quired for model validation). This feature makes the tool especially
suitable for poorly gauged basins.
However, it should be noted that the tool is only meant to aid the
decision maker during the preliminary site selection process and is in no
way intended to replace the decision maker or the detailed site feasi-
bility studies carried out at a later stage. The scope of the tool devel-
oped in this study is limited to planning potential. The tool should be
extended in the future to consider technical and economic potential.
Moreover, planning potential not only includes the consideration of
environmental factors but also social factors. However, due to the lack
of data in the target basin to support the estimation of flows required
for social needs, such factors have not been considered in the present
study. The availability of high-resolution population, land use and ir-
rigation network datasets in the future could facilitate the estimation of
flows needed for water supply and irrigation. For the sake of simplicity,
in this study, equivalent criteria weights were assigned to each criterion
in both of the approaches discussed. However, in order to reflect re-
gional differences and priorities, the decision maker can assign different
weights for more effective planning. One of the drawbacks of using
Fig. 7. A distributed representation of environmental flows Qe . PROMETHEE is the difficulty in deciding the weights for the criteria.
The combination of PROMETHEE with other MCDM techniques may be
explored to overcome this issue. Another limitation of the tool algo-
lower cost. The developed tool focuses on objective-oriented decision
rithm developed in this study is that every alternative waterway scheme
making and robustly evaluates all possible combinations of non-con-
has a diversion site and hydropower station on the same stream seg-
flicting hydropower schemes, thus reducing the element of subjectivity
ment. In reality, however, it is possible that a waterway may divert
in the decision-making processes. In addition, being based on a more
water from one stream segment to another. Another limitation is that
systematic approach, the tool allows large areas to be analyzed in re-
the tool assumes that all waterways are tunnel type, represented by a
latively shorter times. The strength of the developed tool primarily lies
straight line. Future research should be directed toward the develop-
in its ability to utilize a distributed hydrological model to generate
ment of more complex algorithms that can represent waterways be-
distributed FDCs, which make it possible to locate a potential site at any
tween two different stream segments and open channel waterways that
location in the basin irrespective of the availability of regional FDCs.
cannot be represented by straight lines.
This study is the first that attempts to overcome this barrier.
In real-world applications, this tool could be the key for the pre-
The use of a distributed hydrological model also facilitates the
paration of basin-wide hydropower development master plans in a
generation of spatially distributed environmental flows based on

Fig. 8. A distributed representation of FDCs. (a) Level 1 group of Pfafstetter sub-basins with the even-numbered sub-basins (tributaries) grayed-out. (b) FDCs for the
odd-numbered sub-basins (main river). The range shows the FDCs associated with all the flow intervals in a sub-basin. The solid line shows the average of all the flow
intervals in a sub-basin.

318
A. Moiz et al. Applied Energy 224 (2018) 309–321

Fig. 9. Comparing the spatial distribution of selected run-of-river schemes using (a) approach (i) considering topographical factors and (b) approach (ii) considering
hydrological and topographical factors.

short time and with minimal effort. The developed plans can provide an tool could make a significant contribution to studying the impact of
overview of the hydropower potential in the basin to the decision climate change on SHP potential to facilitate long-term planning.
maker to support the screening of sites for a detailed feasibility study.
As mentioned earlier, gauge precipitation is the only local dataset used 5. Conclusions
as an input for the hydrological model, with all other inputs being
derived from global datasets. The use of global satellite precipitation In this study, the development of a systematic GIS-based hydro-
products (after bias correction) could be used in the future to make the
power site selection tool was presented for the estimation of basin-wide
tool presented in this study completely independent of local datasets. planning potential. The developed tool considered all possible sets of
Moreover, hydrological regime of many basins across the globe will
hydropower scheme alternatives and further employed multi-criteria
likely be affected by climate change, thus causing either an increase or decision making methods to obtain a complete ranking of the alter-
decrease in hydropower potential. Since the tool developed in this work
natives. Integration of the tool with a water and energy budget – based
is able to include the spatially distributed hydrology of the basin in site distributed hydrological model made it possible to include hydropower
selection, as demonstrated using approach (ii) using WEB-DHM-S, the
potential as a criterion to be maximized in the objective function in

Fig. 10. Comparing SHP potential at 70% flow exceedance using approach (i) considering topographical factors and approach (ii) considering hydrological and
topographical factors. The dashed lines represent the number of selected sites.

319
A. Moiz et al. Applied Energy 224 (2018) 309–321

addition to the generation of distributed flow duration. This study is the [7] Garegnani G, Sacchelli S, Balest J, Zambelli P. GIS-based approach for assessing the
first to attempt to tackle this issue. Moreover, the tool was developed as energy potential and the financial feasibility of run-off-river hydro-power in Alpine
valleys. Appl Energy 2018;216:709–23. http://dx.doi.org/10.1016/j.apenergy.
a user-friendly and flexible ArcGIS script tool to allow users to tailor the 2018.02.043.
selection criteria per the preferences of the decision maker and the local [8] Price T, Probert D. Harnessing hydropower: a practical guide. Appl Energy
regulations. Finally, the integrated tool was applied to the Kunhar River 1997;57:175–251. http://dx.doi.org/10.1016/S0306-2619(97)00033-0.
[9] Wu Y, Geng S, Zhang H, Gao M. Decision framework of solar thermal power plant
Basin in Pakistan as a case study to demonstrate its capabilities. Two site selection based on linguistic Choquet operator. Appl Energy 2014;136:303–11.
approaches were explored for site selection, namely, (i) considering http://dx.doi.org/10.1016/j.apenergy.2014.09.032.
only topographic factors and (ii) considering both topographic and [10] Al Garni HZ, Awasthi A. Solar PV power plant site selection using a GIS-AHP based
approach with application in Saudi Arabia. Appl Energy 2017;206:1225–40. http://
hydrological factors. dx.doi.org/10.1016/j.apenergy.2017.10.024.
For equivalent factor criteria weights, the site selection using ap- [11] Assouline D, Mohajeri N, Scartezzini JL. Large-scale rooftop solar photovoltaic
proach (i) revealed 36 optimal SHP sites with a waterway length of technical potential estimation using Random Forests. Appl Energy
2018;217:189–211. http://dx.doi.org/10.1016/j.apenergy.2018.02.118.
85 km in the basin compared with only 26 sites with a waterway length
[12] Gonzalez-Salazar MA, Morini M, Pinelli M, Spina PR, Venturini M, Finkenrath M,
of 72 km identified when considering both factors but essentially the et al. Methodology for estimating biomass energy potential and its application to
same hydropower potential of 235 MW at 70% flow exceedance. Colombia. Appl Energy 2014;136:781–96. http://dx.doi.org/10.1016/j.apenergy.
Evidently, the generation of distributed flow duration curves using 2014.07.004.
[13] Chen X. Economic potential of biomass supply from crop residues in China. Appl
distributed hydrological modeling and the incorporation of hydropower Energy 2016;166:141–9. http://dx.doi.org/10.1016/j.apenergy.2016.01.034.
potential as a maximizing criterion for site selection revealed more [14] Sánchez-Lozano JM, García-Cascales MS, Lamata MT. GIS-based onshore wind farm
economically attractive small hydropower sites in basin-scale planning. site selection using Fuzzy Multi-Criteria Decision Making methods. Evaluating the
case of Southeastern Spain. Appl Energy 2016;171:86–102. http://dx.doi.org/10.
In addition, this method also allowed the decision maker to limit the 1016/j.apenergy.2016.03.030.
analysis to the target scale of hydropower development. [15] Ritter M, Deckert L. Site assessment, turbine selection, and local feed-in tariffs
The developed tool made the preliminary hydropower site selection through the wind energy index. Appl Energy 2017;185:1087–99. http://dx.doi.org/
10.1016/j.apenergy.2015.11.081.
process less time consuming, more robust, and more systematic, con- [16] Punys P, Dumbrauskas A, Kvaraciejus A, Vyciene G. Tools for small hydropower
sequently making it less susceptible to the possible subjectiveness or plant resource planning and development: a review of technology and applications.
bias introduced by the decision maker, both of which are due to the Energies 2011;4:1258–77.
[17] Ballance A, Stephenson D, Chapman RA, Muller J. A geographic information sys-
complexity of the problem when using a conventional approach. The tems analysis of hydropower potential in South Africa. J Hydroinformatics
new approach also reduced the number of suitable hydropower sites to 2000;2:247–54.
a finite set that can be investigated in detail during feasibility studies. [18] Arefiev N, Nikonova O, Badenko N, Ivanov T, Oleshko V. Development of auto-
mated approaches for hydropowerpotential estimations and prospective hydro-
Although the tool’s capability is demonstrated for the Kunhar River
power plants siting 2015; 2:41–50. doi:http://doi.org/10.17770/etr2015vol2.260.
Basin, it is not limited in application to a specific basin as it exhibits [19] Feizizadeh B, Haslauer EM. GIS-based procedures of hydropower potential for
reduced reliance upon local datasets and is thus especially well-suited tabriz basin, Iran. Int J 2012:495–502.
for applications in poorly gauged basins. [20] Gergeľová M, Kuzevičová Ž, Kuzevič Š. A GIS based assessment of hydropower
potential in Hornád basin. Acta Montan Slovaca 2013;18:91–100.
[21] Khan M, Zaidi AZ. Run-of-river hydropower potential of Kunhar River, Pakistan.
Acknowledgment Pakistan J Meteorol 2015. <http://www.pmd.gov.pk/rnd/rndweb/rnd_new/
journal/vol12_issue23_files/Run-of-River_Hydropower_Potential_of_Kunhar_River_
Pakistan.pdf> [accessed 10 March 2016].
The authors would also like to extend their appreciation to the [22] Müller MF, Thompson SE, Kelly MN. Bridging the information gap: a webGIS tool
Water And Power Development Authority (WAPDA) of Pakistan and for rural electrification in data-scarce regions. Appl Energy 2016;171:277–86.
Pakistan Meteorological Department (PMD) for sharing the necessary http://dx.doi.org/10.1016/j.apenergy.2016.03.052.
[23] Rojanamon P, Chaisomphob T, Bureekul T. Application of geographical information
meteorological and hydrometric datasets. system to site selection of small run-of-river hydropower project by considering
engineering/economic/environmental criteria and social impact. Renew Sustain
Funding Energy Rev 2009;13:2336–48. http://dx.doi.org/10.1016/j.rser.2009.07.003.
[24] Palomino Cuya DG, Brandimarte L, Popescu I, Alterach J, Peviani M. A GIS-based
assessment of maximum potential hydropower production in La Plata basin under
This research is partially supported by the DIAS project funded by global changes. Renew Energy 2013;50:103–14. http://dx.doi.org/10.1016/j.
the Japanese Ministry of Education, Culture, Sports, Science and renene.2012.06.019.
[25] Al-Juboori AM, Guven A. Hydropower plant site assessment by integrated hydro-
Technology (MEXT).
logical modeling, gene expression programming and visual basic programming.
Water Resour Manag 2016;30:2517–30. http://dx.doi.org/10.1007/s11269-016-
Appendix A. Supplementary material 1300-3.
[26] Kusre BC, Baruah DC, Bordoloi PK, Patra SC. Assessment of hydropower potential
using GIS and hydrological modeling technique in Kopili River basin in Assam
Supplementary data associated with this article can be found, in the (India). Appl Energy 2010;87:298–309. http://dx.doi.org/10.1016/j.apenergy.
online version, at http://dx.doi.org/10.1016/j.apenergy.2018.04.070. 2009.07.019.
[27] Rospriandana N, Fujii M. Assessment of small hydropower potential in the Ciwidey
subwatershed, Indonesia: a GIS and hydrological modeling approach. Hydrol Res
References Lett 2017;11:6–11. http://dx.doi.org/10.3178/hrl.11.6.
[28] Soulis KX, Manolakos D, Anagnostopoulos J. Development of a geo-information
[1] REN21. Renewables 2016 Global Status Report. Paris; 2016. doi:ISBN 978-3- system embedding a spatially distributed hydrological model for the preliminary
9818107-0-7. assessment of the hydropower potential of historical hydro sites in poorly gauged
[2] REN21. Renewables 2013 Global Futures Report. Paris; 2013. <http://www.ren21. areas. Renew Energy 2016;92:222–32.
net/Portals/0/documents/activities/gfr/REN21_GFR_2013.pdf> [accessed 10 [29] Larentis DG, Collischonn W, Olivera F, Tucci CEM. Gis-based procedures for hy-
March 2016]. dropower potential spotting. Energy 2010;35:4237–43.
[3] International Hydropower Association. Hydropower Status Report; 2017. p. 1–83. [30] Yi CS, Lee JH, Shim MP. Site location analysis for small hydropower using geo-
<https://www.hydropower.org/sites/default/files/publications-docs/ spatial information system. Renew Energy 2010;35:852–61. http://dx.doi.org/10.
2017%20Hydropower%20Status%20Report.pdf> [accessed 10 January 2018]. 1016/j.renene.2009.08.003.
[4] International Hydropower Association, International Comission on Large Dams, [31] Wang L, Koike T, Yang K, Jackson TJ, Bindlish R, Yang D. Development of a dis-
International Energy Agency, Canadian Hydropower Association. Hydropower and tributed biosphere hydrological model and its evaluation with the Southern Great
the World’s Energy Future; 2000. <http://www.ieahydro.org/media/ffab53b0/ Plains experiments (SGP97 and SGP99). J Geophys Res Atmos 2009;114:1–15.
Hydropower%20and%20the%20World's%20Energy%20Future%20.pdf> [ac- http://dx.doi.org/10.1029/2008JD010800.
cessed 10 March 2016]. [32] Wang L, Koike T, Yong K, Yeh PJF. Assessment of a distributed biosphere hydro-
[5] Chen ST, Kuo HI, Chen CC. The relationship between GDP and electricity con- logical model against streamflow and MODIS land surface temperature in the upper
sumption in 10 Asian countries. Energy Policy 2007;35:2611–21. http://dx.doi.org/ Tone River Basin. J Hydrol 2009; 377:21–34.
10.1016/j.enpol.2006.10.001. [33] Shufen S, Yongkang X. Implementing a new snow scheme in Simplified Simple
[6] Paish O. Small hydro power: Technology and current status. Renew Sustain Energy Biosphere Model. Adv Atmos Sci 2001;18:335–54. http://dx.doi.org/10.1007/
Rev 2002;6:537–56. http://dx.doi.org/10.1016/S1364-0321(02)00006-0. BF02919314.

320
A. Moiz et al. Applied Energy 224 (2018) 309–321

[34] Xue Y, Sun S, Kaha DS, Jiao Y. Impact of parameterizations in snow physics and decision making methods for selection of building structural system. Am J Civ Eng
interface processes on the simulation of snow cover and runoff at several cold re- Archit 2014;2:149–59. http://dx.doi.org/10.12691/ajcea-2-5-1.
gion sites. J Geophys Res 2003;108:8859. http://dx.doi.org/10.1029/ [43] Hajkowicz S, Collins K. A review of multiple criteria analysis for water resource
2002JD003174. planning and management. Water Resour Manag 2007;21:1553–66. http://dx.doi.
[35] Dickinson E, Henderson-Sellers A, Kennedy J. Biosphere-atmosphere Transfer org/10.1007/s11269-006-9112-5.
Scheme (BATS) Version 1e as Coupled to the NCAR Community Climate Model. [44] Mladineo N, Margeta J, Brans JP, Mareschal B. Multicriteria ranking of alternative
NCAR Tech Rep NCAR/TN-3871STR 1993; 72:77. doi:http://doi.org/10.5065/ locations for small scale hydro plants. Eur J Oper Res 1987;31:215–22. http://dx.
D67W6959. doi.org/10.1016/0377-2217(87)90025-7.
[36] Yang ZL, Dickinson RE, Robock A, Vinnikov KY. Validation of the snow submodel of [45] Mahmood R, Jia S, Babel MS. Potential impacts of climate change on water re-
the biosphere-atmosphere transfer scheme with Russian snow cover and meteor- sources in Kunhar River Basin, Pakistan. Water 2016:1–24. http://dx.doi.org/10.
ological observational data. J Clim 1997;10:353–73. http://dx.doi.org/10.1175/ 3390/w8010023.
1520-0442(1997) 010<0353:VOTSSO>2.0.CO;2. [46] Private Power and Infrastructure Board. Hydropower Resources of Pakistan; 2011.
[37] Shrestha M, Koike T, Hirabayashi Y, Xue Y, Wang L, Rasul G, et al. Integrated <www.ppib.gov.pk/HYDRO.pdf> [accessed 15 March 2017].
simulation of snow and glacier melt in water and energy balance-based, distributed [47] Japan International Cooperation Agency. Guideline and Manual for Hydropower
hydrological modeling framework at Hunza River Basin of Pakistan Karakoram Development Vol. 2 Small Scale Hydropower; 2011. <open_jicareport.jica.go.jp/
region. J Geophys Res Atmos 2015; 120(10). doi:http://dx.doi.org/10.1002/ pdf/12024881_01.pdf> [accessed 15 March 2017].
2014JD022666. [48] Verdin KL, Verdin JP. A topological system for delineation and codification of the
[38] Shrestha M, Wang L, Koike T, Xue Y, Hirabayashi Y. Improving the snow physics of Earth’s river basins. J Hydrol 1999;218:1–12. http://dx.doi.org/10.1016/S0022-
WEB-DHM and its point evaluation at the SnowMIP sites. Hydrol Earth Syst Sci 1694(99)00011-6.
2010;14:2577–94. http://dx.doi.org/10.5194/hess-14-2577-2010. [49] Food and Agriculture Organization. Digital soil map of the world and derived soil
[39] Shrestha M, Wang L, Koike T, Xue Y, Hirabayashi Y. Modeling the Spatial properties. L Water Digit Media Ser Rev 1 2003. Available in CD-ROM.
Distribution of Snow Cover in the Dudhkoshi Region of the Nepal Himalayas. J [50] Strahler AN. Hypsometric (area-altitude) analysis of erosional topology. Geol Soc
Hydrometeorol 2012;13:204–22. http://dx.doi.org/10.1175/JHM-D-10-05027.1. Am Bull 1952;63:1117–42.
[40] Brans JP, Mareschal B. PROMETHEE methods. In: Figueira J, Greco S, Ehrgott M, [51] Brans JP. L’ingénièrie de la décision; Elaboration d’instruments d’aide à la décision.
editors. Multiple criteria decision analysis: state of the art surveys, Boston: Springer La méthode PROMETHEE. In: Nadeau R, Landry M, editors. L’aide à la décision
Science + Business Media Inc; 2005, p. 163–95. doi:https://doi.org/10.1007/ Nature. Quebec, Canada: Instruments Perspect. d’Avenir; 1982. p. 183–213.
b100605. [52] Gatte MT, Kadhim RA. Hydro Power, Energy Conservation, Dr. Azni Zain Ahmed
[41] Velasquez M, Hester PT. An analysis of multi-criteria decision making methods. Int (Ed.), ISBN: 978-953-51-0829-0, InTech, doi: 10.5772/52269.
J Oper Res 2013;10:56–66. [53] IUCN, UNEP-WCMC. The World Database on Protected Areas (WDPA). Cambridge,
[42] Balali V, Zahraie B, Roozbahani A. A comparison of AHP and PROMETHEE family UK UNEP-WCMC 2017. <https://www.protectedplanet.net/>.

321

You might also like