Assessment of Run-of-River and Hydropower Plants in Peru: Current and Potential Sites, Historical Variability (1981–2020), and Climate Change Projections (2035–2100)

Abstract

Hydropower is the main source of renewable energy and the most feasible for implementation in remote areas without access to conventional energy grids. Therefore, knowledge of actual, potential, and future perspectives of sustainable hydropower projects is decisive for their viability. This study aims to estimate the present and future potential capacity of Peru’s hydropower system and from the potential small hydroelectric plants, specifically Run-of-River class. First, we employed geospatial databases and hydroclimatological products to describe the current hydropower system and potential sites for Run-of-River projects. The findings identified 11,965 potential sites for Run-of-River plants. Second, we executed and validated a hydrological model to estimate historical daily streamflows (1981–2020) and hydropower parameters for actual and potential sites. It was determined there is an installed capacity of 5.2 GW in the current hydropower system and a total potential capacity of 29.1 GW for Run-of-River plants, mainly distributed in the northern and central Andes. Finally, we evaluated future changes driven by ten global climate models under three emission scenarios (SSP1-2.6, SSP3-7.0, and SSP5-8.5), compared with the baseline period of 1981–2010 with two future time slices. The main results about capacity indicated that operational hydroelectric plants (Run-of-River plants) are projected to decrease by 0.5 to −5.4% (−7.2 to −2.2%) during 2036–2065 and by −9.2 to 3.8% (1.8 to −11.9%) during 2071–2100. These outcomes provide relevant information to support policymakers in addressing sustainable development gaps in the coming decades and stakeholders involved in the implementation and mitigation of climate change impacts on hydropower projects in Peru.

1. Introduction

Access to electric energy is related to more significant opportunities for the socio-economic development of the population [1]. To this purpose, decentralizing energy to remote areas through off-grid systems is a viable alternative [2], and the utilization of renewal electric resources (RERs) available in remote areas is the most suitable route for the development and maintenance of these systems [3]. Hydropower is the most widely used RER (83.6%), constituting 16.4% of global energy production, due to the highest conversion efficiency (about 90%) [4,5,6]. Hydropower generated through hydropower plants (HPP) produces fewer greenhouse gases than other popular energy sources like fossil fuels. In this sense, it is necessary to bridge the gap between hydropower potential and installed hydropower; this value has a smaller difference in countries with greater technological development [7]. Currently, a value of 551.8 GW (40.6% of the total) is located in developing countries such as China, Brazil, and India [7]. The development of hydropower projects is therefore suggested to reduce the increasing energy demand gap and establish a country as a nation with sustainable economic growth under an energy regime with minimal CO₂ emissions [8,9].

Until the mid-20th century, large hydropower plants (LHP) were considered the most widely accepted renewable energy sources [10]. However, despite its socio-economic benefits and providing multiple services in addition to energy such as the regulation of peak events [9], water management, etc., it is still a very important energy source. LHPs cause disturbances in ecological streamflow by disrupting river connectivity [11], methane gas emissions involved in construction, and deteriorating riparian ecosystems [12,13,14]. Hence, the small hydropower plant (SHP) is considered a more environmentally friendly alternative and has gained greater notoriety among political decision makers and investors in recent decades [15]. The development of SHPs has increased their relevance due to their easy implementation compared to LHPs [16]. Globally, SHPs are defined as having an installed capacity above 1 MW and generally below 10 MW [17]. In the case of Peru, the upper threshold is 20 MW. In terms of SHPs, using the global threshold of 10 MW as a reference, the installed capacity in 2019 at the global, continental (South America excluding Peru), and Peru levels (considering the local definition) is 79 GW, 2462.6 MW, and 503.8 MW, respectively [18]. According to [19], the increasing development of SHPs in recent decades has shown to have a higher return on investments and achieve greater use of hydropower. SHPs, like type Run-of-River plants (RoR), are suitable RER options to supply electricity to populated centers in remote areas [5]. On the other hand, according to Jung et al. [20], SHPs are sensitive to climatic conditions since they only depend on the streamflow and net head. Therefore, it is essential to assess the impact of climate change on the hydro-energy, specially in SHP, as a result of changes on precipitation changes and the disruption in streamflow regimes caused by the increased frequency of extreme hydrological events [21,22].

On a global scale, there is an increase in the vulnerability of water resource availability as a result of the following changes: increase in population demand, changes in land use, increase in food production, expansion of industrialization based on water use, poor water quality, and the impacts of climate change [23,24]. According to COP21 in 2015, the objective is to limit the emission of CO₂ from energy so that the increase in temperature does not exceed 1.5 °C concerning pre-industrial levels [25,26]. Furthermore, the Intergovernmental Panel on Climate Change Special Report about the impacts of global warming assesses the hydrological impacts due to climate change [23,27], indicating, in summary, the following: as a result of the increase in greenhouse gases, it is estimated that for each degree of global warming, approximately 7% of the population of the world is projected to be exposed to a decrease in renewable water resources of at least 20%. For this reason, water scarcity has intensified in recent years, and in the coming decades, its availability is expected to decline further, particularly in developing regions such as South America [28,29,30]. The growing agricultural demand and rising population density in urban areas with limited water sources, such as the Pacific coast of Peru, exacerbate future prospects [31]. Furthermore, the consequences of climate change decrease the operating capacity and capacity of the hydroelectric plant [31,32]. The previous research evaluated the state of hydroelectricity in future scenarios using various regional or global models on a national scale, such as in the United States and China [33,34]. Moreover, recent studies [20,35] examine the impact of climate change on the hydropower potential of SHPs through a percentage comparison between future projections and a historical reference period, based on the general circulation models (GCMs) from the Coupled Model Intercomparison Project (CMIP). This study aims to fill a gap in the literature by estimating these projections specifically for Peru at a regional level, where no such research has yet been developed.

The water–energy nexus, which highlights the interdependence between hydrology and energy systems, has garnered increasing policy attention in recent years [36,37]. This study aims to contribute to sustainable hydropower planning in Peru by integrating socio-environmental constraints and future projections. The approach involves estimating the water–energy relationship of rivers suitable for RoR projects, characterizing their hydropower parameters, and assessing their future potential. First, we create an inventory of hydropower plants by reviewing national databases and developing a model to identify potential RoR sites, evaluating hydropower potential with technical and socio-environmental constraints. Next, we enhance the simulation of daily streamflows using a high-resolution hydrological model, addressing the primary source of uncertainty in hydropower assessments [38]. Finally, we present future hydropower projections based on GCMs under various scenarios. This comprehensive framework provides valuable insights into the spatio-temporal and future dynamics of hydropower potential across Peru.

2. Materials and Methods

2.1. Study Area

The study covered the national territory of Peru, which has an area of 1285 million km² and is located between 0°02′ N–17°50.2′ S and 68°10.2′ W–81°90.2′ W. Peru is characterized by high topographic variability from sea level to 6500 m above sea level, multiple climatic zones, interannual variations in precipitation, and events such as ENSO that alter climatic variability, affecting the coast through floods and droughts in the Andes [39,40,41,42,43]. Therefore, a variety of climate regimes are present, from arid to glacial to rainy tropical [44,45]. The north–south Andes divide Peru into three hydrographic basins: Pacific, Titicaca, and Amazon. The areas of these basins in Peru contain 21.7%, 3.8%, and 74.5% and include 62, 13, and 84 of the 159 hydrographic basins, respectively [46]. However, there is an inverse relationship between water availability and demographic distribution, with approximately 70% of the population concentrated in the Pacific basin, where access to water is only 1.8%, while the Amazon basin, where 26% of the population lives, has 97.7% of the water resources [46]. Furthermore, to identify the heterogeneity of the results in the climatic characteristics, the results have been analyzed by six regions [47]: South Pacific (PFS), North Pacific (PFN), North Atlantic (ALN), Central Atlantic (ALC), South Atlantic (ALS), and Titicaca (TIC) (Figure 1a); the Pacific, Atlantic, and Titicaca regions correspond to the Pacific, Amazon, and Titicaca basins.

Restriction Zones

The development of hydro-energy projects such as SHPs is subject to environmental, social, and ecological restrictions that must be considered in their planning and development to ensure the protection of protected socio-cultural areas, aquatic ecosystems, and the water quality of rivers [3,48]. Therefore, to prevent negative impacts on these protected sites and to make it easier for SHP projects, these areas were restricted from the identification of sites with hydropower potential. In this sense, a geographic information systems layer was defined, presented in Figure 1b, from the union of the five groups of restricted zones: (i) urban; (ii) mining-energy; (iii) socio-cultural such as archaeological sites, Native communities, and PIACI (Indigenous Peoples in Isolation and Initial Contact); (iv) protected natural areas (ANP), such as reserved zones, national biosphere reserve, and RAMSAR; and (v) Amazonian ecological systems such as peneplain forests, bamboo forests, and floodable forests.

In

Table A1. Restriction zones.

Group	Type	Area (km2)	Source
Urban	Actual	2455.5	https://geoservidor.minam.gob.pe/recursos/intercambio-de-datos/ (accessed on 22 August 2024)
	Future	8490.5	https://gee-community-catalog.org/projects/urban_projection/ (accessed on 20 November 2022)
Mineral-energetic	Mining sites	803.5	https://geoservidor.minam.gob.pe/recursos/intercambio-de-datos (accessed on 22 August 2024)
Socio-Cultural	PIACI	30,226.5	https://visor.geoperu.gob.pe (accessed on 22 August 2024)
	Native Communities	10,715.3
	Archaeological Sites	15,780.1
	Definitive	251,679
Protected	Private conservation	3878.5
Natural	Regional conservation	35991	https://geo.sernanp.gob.pe/visorsernanp/ (accessed on 22 August 2024)
Areas	Reserved areas	5878.2
	Biosphere Reserve	72,629.4
	RAMSAR	69,435.8	https://geoservidor.minam.gob.pe/recursos/intercambio-de-datos/ (accessed on 22 August 2024)
Amazonian	Flooded forest	136,808.3
Biological	Penillanura forest	244,129.7	https://geoservidor.minam.gob.pe/recursos/intercambio-de-datos/ (accessed on 22 August 2024)
Systems	Bamboo forest	711,84.4

and Figure A2, the groups and types of restricted areas are shown, as well as their extension and source. In the case of future urban areas, the gridded data at 1 km spatial resolution of the global urban projections for the SSP5 scenario of CMIP6 to 2100 [49] was used; this information was complemented with the current urban areas. The floodable ecosystems and Amazonian ecological systems were extracted from the Ecosystem Map of Peru [50], selecting the groups of floodable forests and vegetation and the Amazonian peneplain ecosystems since they are located in depressions in the terrain and are exposed to periodic flooding for weeks or months. The reviewed bibliography suggests not carrying out hydroelectric projects in these restricted areas to preserve their functionality. In addition, it is recommended to add a buffer to these protected refuges and urban areas [3,19]. In this sense, an additional buffer of 500 m was considered for protected natural areas and 200 m for urban, private areas, wetlands, lakes, and other biological systems previously included. In total, 32.1% of the area of Peru is restricted for the development of hydropower projects of all kinds; most of these sites are located in eastern and northern Peru (Figure 1b).

2.2. Overview

The methodology of this study consists of three main stages (Figure 2): site identification, estimation of historical parameters, and their future projections. First, we obtained databases to identify operational and planned HPPs. Additionally, we located potential RoR sites in non-excluded areas through geo-processing of geospatial and hydrological data. Next, we generated daily streamflow series for the sub-basins where sites of interest were identified. Subsequently, based on the estimation of historically representative streamflows and the net head previously described, we defined the hydropower potential for historical period (1981–2020) for HPPs and RoRs. At last, we generated a series of future projections for hydropower potential, based on the percentage change between the parameters of a reference period (1981–2010) and those of medium-term (2036–2065) and long-term (2071–2100) future periods from an ensemble of statistically downscaled CMIP6 climate scenarios. The following sections provide detailed information and specifications regarding these outputs, including the methodologies employed.

2.3. Datasets

2.3.1. Historical Hydrological and Climatological Data

For the historical analysis, three key variables were necessary: precipitation, temperature, and streamflow. In this sense, the database prepared by the National Meteorological and Hydrological Service of Peru (SENAMHI), called Peruvian Interpolated data of the SENAMHI’s Climatological and Hydrological Observations (PISCO), was used. PISCO includes gridded precipitation and air temperature (PISCOp and PISCOt; 0.1° × 0.1°; 1981–2016; [51,52]) but also the simulated streamflow series based on hydrological model ARNOVIC, considered in this study as observed streamflow (PISCO_HyD_ARNOVIC; 1981–2021); more details on calibration and validation can be found in the respective study [53]). PISCOp and PISCOt were built from satellite information and weather station data that went through quality control, reconstruction of missing data, and spatial interpolation processes to have an extensive database of high temporal and spatial resolution for all of Peru. Meanwhile, PISCO_HyD_ARNOVIC used information from PISCOp and PISCOt to estimate daily streamflow series using a hybrid modeling framework that integrates sub-basin rainfall-runoff and river routing models calibrated with observed hydrological data. The selection of PISCO datasets is supported by their application in recent regional hydroclimatological studies [54,55] and is necessary given the scarcity and low quality of climatic data available in Peru [56,57,58]. Finally, historical daily streamflow simulations based on the PISCOp product for a 40-year period (1981–2020) were considered for this study.

2.3.2. Future Climatological Data

To generate the simulated future streamflows using the ARNOVIC hydrological model, we employed the precipitation and evapotranspiration, obtained using the temperature and Hargreaves–Samani equation, from the forcing data downscaling from the projections of the ten CMIP6 GCMs (CanESM5, IPSL–CM6A–LR, UKESM1–0–LL, CNRM–CM6–1, CNRM–ESM2–1, MIROC6, GFDL–ESM4, MRI–ESM2–0, MPI–ESM1–2–HR and EC–Earth3) of the BASD-CMIP6-PE product [59]. Three future emission scenarios (SSP1, SSP3, and SSP5) are considered to limit the uncertainties in the projection due to the different emission scenarios. The generated data have a daily time scale and a spatial resolution of 0.1° (compatible with the PISCO product grid configuration). Subsequently, daily gridded data were extracted in the study area for the period 1981–2010 under the historical scenario and for the periods 2036–2065 and 2071–2100 (mid-term and long-term) under SSP1-2.6, SSP3-7.0, and SSP5-8.5. In this study, the 20th, 50th, and 80th percentiles of the simulated streamflow ensemble in the SSP combinations are used.

2.3.3. Hydropower Plants Databases

The current HPPs belong to the National Interconnected Electrical System (SEIN), managed by the SEIN Economic Operation Committee (COES) and are composed of the electrical lines and substations that connect the energy generating units, such as the HPPs, with the consumption points, allowing electrical transfer in Peru. To inventory the main parameters, we considered (i) national documents, like supervision of contracts for generating plants in operation report (Compendium of Projects in Execution) [60], for operational (planned by future years) HPPs and (ii) COES operating records (2010 to present) about power and energy production of the operational HPPs. It should be noted that the number of years with information is not homogeneous across the HPPs.

Subsequently, we manually validated five parameters for our study, location (latitude and longitude), status, installed capacity, firm energy (EF), and energy production (EP). To validate the location of (i) operational HPPs, we used satellite images from Google Earth to visualize the location of the reservoir or the water catchment point of the corresponding water source; while for (ii) planned HPPs, the Compendium of Projects [60] was used, which indicates the possible locations for their location. Likewise, the current status of the HPP projects was collected from the most recent National Documents, where they are classified as operational, planned, and under construction, this last class was analyzed in our methodology as a planned HPP. In the case of the installed capacity of the HPPs, it was found in the national documents and the COES records. For the EF and EP information, the multiannual average of the series available in the COES records during the period 2010–2023 was considered. Finally, Figure A1 presents the parameters of the HPPs grouped by their classification according to their system: Storage and RoR; this classification was defined as indicated in the national documents.

The distribution of the operational HPPs is presented in Figure 1c, the total installed capacity (minimum and maximum per HPP) is 5233.5 MW (0.3 to 798 MW), EF of 25,341 GWh (0.78 to 4068 GWh), and EP of 29,897 GWh (0.68 to 5278 GWh); the parameters of the HPPs according to the region to which they belong are displayed in Figure A1. On the other hand, 73 HPPs are planned to be executed over the next decade, this projects will add a total of 8780 MW to the installed capacity and 32,554 GWh to EP [60]. We consider these planned HPPs for evaluating the projections of future hydro-energy parameters (Figure A1).

2.3.4. Potential Sites Definition Databases

Geographic information systems tools were used to select potentially exploitable sites for RoR development. To define these sites, restrictions in urban, cultural, and environmental areas defined in Table A1 were considered. Subsequently, technical feasibility requirements were defined based on similar studies Rojanamon et al. [61], Kouadio et al. [62], compiled in

Table 1. Technical restriction factors considered for the identification of sites for RoR projects.

Description	Valor
Length of stream network	≥500 m
Average slope	≥2%
Distance to urban-cultural areas	200 m
Distance to environmental conservation areas	500 m
Spacing between consecutive RoRs	≥600 m
Net head	≥10 m

.

Regional or continental scale studies used the longitudinal river profile by catchment using digital elevation modeling (DEM) [3,63]. The DEM spatial resolution will define the accuracy in the delineation of basins and their respective river profile, as a coarser resolution will provide a lower representativeness of the height and the basin [64,65,66]. To represent the terrain elevation and define the longitudinal profile to identify potential sites, we employed the MERIT-Hydro DEM [67], a product used in studies to assess hydro-energy potential and with attributes comparable to those used as input for the generation of PISCO-HyD-ARNOVIC. MERIT-Hydro DEM has a spatial resolution of 3 arc s (∼90 m at the equator) and was hydrologically conditioned to estimate the hydrographic networks with quality control based on water body data, global surface water occurrence, and OpenStreetMap.

To perform the identification of potential sites, some studies used a minimum net height of 10 m [65,68], and regarding the slope of the water network, it is recommended that it be greater than or equal to 2% [19], both to guarantee a minimum hydropower generation. Several authors consider selecting the river minimum accumulated streamflow of ${10}^4$ grid cells, based in DEM with spatial resolution of 30 m, used to establish ordering using the Strahler method with order 2 as minimum to ensure sufficient streamflow availability for effective energy production [3,62,65,69,70,71].

2.4. Methods

2.4.1. Definition Sites

Locations for hydropower potential along rivers from no excluded areas (Figure 1b) were selected using criteria according to Table 1. Four criteria were examined in the technical assessment: minimum head, slope, stream length, and consecutive spacing. To avoid crowds in definition sites is based on the distance from the river stream using a separation of 600 m between sites. The minimum separation ensures that the backwater of the discharge pipe would not affect the continuous RoR in the same river, since a minimum length of 500 m was considered between the point of water intake to the turbine and a separation between two sections of the project of 100 m [72]. Furthermore, a division of the river reach into shorter segments does not significantly (<10%) increase the accuracy of the capacity estimate [69,73].

2.4.2. Historical Parameters

Hydro-energy is the energy that can be obtained from the potential and kinetic energy of rivers; therefore, the hydro-energy potential is the amount of energy in the entire river basin [74]. The hydro-energy potential of HPPs is analyzed for the historical period between 1981 and 2020 and expressed in terms of capacity (power) and energy.

Capacity is divided into these main types: theoretical (CR) and technical (CT) [20,75]. CR is the total streamflow energy at the basin surface due to its altitude difference and with an efficiency of 100% [75,76], while CT (various authors [3,4,19,20,64,69,72,77]) considers the efficiency due to technical limitations; represented by

Table 2. Equations for hydro-energy parameters.

Acronyms	Name	Units	Equation
CR	Capacity theoretical	MW	$\gamma \times Q_{80}\times H_n\times {10}^{-6}$
CT	Capacity technical	MW	$\gamma \times Q_{80}\times H_n\times {10}^{-6}\times \eta$
$Q_{avg}$	Average streamflow	${\mathrm{m}}^3/\mathrm{s}$	${\displaystyle \frac{Q_{100}+Q_{90}+Q_{80}+Q_{70}+Q_{60}+(5\times Q_{50})}{10}}$
EP	Energy Production	GWh/yr	$\gamma \times Q_{avg}\times H_n\times \eta \times {10}^{-6}\times 8.76$
EF	Firm Energy	GWh/yr	$\gamma \times Q_{95}\times H_n\times \eta \times {10}^{-6}\times 8.76$

equations. Both equations are based on three components: (i) specific weight of water, product by water density, and gravity acceleration; (ii) net height, considering hydraulic losses, but to simplify computational calculations, gross height ($H_g$) is assumed [77]; and (iii) flow, defined as the streamflow between 80 and 85% of the flow duration curve ($Q_{80}$ to $Q_{85}$ from Figure A2) to design the streamflow of SHPs [19,48,78]. In addition, the overall or system efficiency in the conversion of the water–energy ratio ($\eta$) is a product of turbine and generator efficiency [77,79]; finally, from the overall efficiencies reviewed was defined a $\eta$ of 85% [76].

Where $\gamma$ is the specific weight of water (9810 Nm³), $\eta$ is the system efficiency, and $H_n$ is the net head (m).

For a better description, we also classified the HPPs by their CR, based on the local definition compiled by UNIDO [18], which is described in

Table 3. Classification of HPPs by size.

Size	Nomenclature	Capacity (MW)
Micro		<0.1
Mini	SHP	0.1–1
Small		1–20
Median	LHP	20–100
Large		>100

. Moreover, [69] considers 0.1 MW as minimum CR to analyze the production of SHPs, since the contribution of SHPs with power less than this threshold is insignificant with respect to the total. This classification was also used to characterize the CR by hydrographic regions. For simplification purposes, mini or small HPPs in this classification will be called SHPs.

Secondly, in the estimation of hydroelectric production, the hydro-energy parameters are estimated: EP and EF [4,77], which are defined in Table 2 equations, for the total estimation of hydroelectric production throughout the year under normal and very dry hydrological conditions, respectively. Essentially, they are composed of the CR equation, multiplied by a conversion factor to transform the power produced in one hour (MW) to one year (GWh), assuming that the EP has a daily operation of 24 h 365 days a year. It should be noted that the streamflow rate used in the EP and EF equation is equal to $Q_{avg}$ and $Q_{95}$. The first is for a closer approximation to the average daily streamflow rate through the turbines throughout the year [4], while the second is the streamflow rate for the EF with reference to the minimum streamflow rate of $Q_{95}$.

On the other hand, several studies focused on the quantification of CR in RoR at a regional level recommend analyzing the available CR in future scenarios, since these types of HPP are vulnerable to streamflow variability under extreme hydrological conditions. For example, Sterl et al. [63] recommends analyzing the available CR, using the streamflows $Q_5$ and $Q_{95}$ in very humid and very dry periods. A more regulated proposal by Zaidi and Khan [72] proposes hydro-energy parameter scenarios with the streamflows $Q_{40}$, $Q_{50}$, and $Q_{60}$. Therefore, in this study, the streamflow rate $Q_{80}$ will be used to estimate the CR, while for the analysis of extreme hydrological conditions, the streamflow rates of the 20th and 80th percentiles of the GCM streamflow rate sample were used; the streamflow rates at each identified site are the streamflow rates estimated in their respective sub-basin. Subsequently, $H_n$, determined from a DEM, was derived from the river reach’s longitudinal profile, identifying the maximum difference in level between (i) an average distance range of 500 m (450 m to 635 m, attributable to the DEM’s resolution) at a point level and (ii) ≥500 m at the sub-basin level (depending on the length of the river reach).

The analysis of parameter annual series provides information on the rates and significance of change over a period of time, in this case, decades. Therefore, the non-parametric Mann–Kendall (MK) test [80,81] was used; it is widely used to identify monotonic trends in hydro-meteorological series, such as hydro-energy parameters that are a function of streamflows. In addition, MK shows greater robustness against outliers and imposes no requirement for data to follow a normal distribution [51]. Otherwise, the Sen method was employed to estimate the magnitude of the trend [82,83], where a negative (positive) value indicates a decrease (increase) in the trend. Subsequently, the trends in the sites of interest were classified based on the probability value (p-value) defined as 0.90, with the magnitude annual trend expressed on a 10-year scale.

This study has an uncertainty analysis that concentrates on assessing the method used to estimate the CR, EF, and EP, aiming to determine its reliability. To validate the generated results, the historical hydro-energy parameters of the existing HPPs will be used as a reference. Therefore, a mix of quantitative and qualitative criteria were applied to evaluate the accuracy of the simulated results Tamm and Tamm [64]. Regarding the statistical indicators, the correlation coefficient (r) was used due to the difference between the magnitudes of hydro-energy parameters in the study regions. The coefficient r is a measure of the strength of relationship between two variables where a result of 1 indicates a perfect relationship with a positive slope, while −1 indicates a relationship with a negative slope.

2.4.3. Future Projection

To evaluate the different future projections of the CMIP6 GCMs on the hydro-energy potential using the BASD-CMIP6-PE product, we used the climatological database of temperature and precipitation simulations for Peru and Ecuador, based on CMIP6 models, with bias correction adjustments and statistical downscaling based on observed data from PISCO product to improve the accuracy of the projections [59]. The time period covers a historical period from 1850 to 2014 and with future projections from 2015 to 2100, we took as reference the base period 1981–2010. The precipitation and evapotranspiration of the BASD_CMIP6_PE product were used to obtain the simulated streamflow series using the ARNOVIC model, following the same methodology and calibrated parameters generated for the PISCO-HyD-ARNOVIC product. In total, 30 streamflow series were obtained, coming from the 10 GCMs and three scenarios in the range 1981–2100.

To assess the agreement of future climate projections between selected GCMs from BASD_CMIP6_PE, we use an analysis based on projected changes in mean air temperature (Temp), annual precipitation sums (Prec), and considering two ETCCDI indices [84] for both air temperature and precipitation for three SSPs. For characterization by the GCMs, changes in air temperature extremes, in the warm spell duration index (WSDI), and in the cold spell duration index (CSDI) are analyzed. For the characterization of changes in precipitation extremes, the precipitation due to extremely wet days (R99P) and the number of consecutive dry days (CDD) are considered. To evaluate those GCMs, the ETCCDI indices were calculated using same the procedures as the recent studies [85,86], based on compared future projections between 2036–2065 and 2071–2100 to refer to the baseline period, thus indicating the projected change over 30 year. Changes in Prec, Temp, WSDI, CSDI, R99P, and CDD for the selected GCMs and SSPs are provided in Figure A3.

To achieve a greater representation of the high amount of data, recent studies have simplified the GCM series by using percentiles based on extreme hydrological scenarios [87]. Therefore, the outputs of the 10 GCMs were summarized in three series of hydrological references, taking into account the extreme scenarios proposed by COES. This choice is based on the 20th and 80th percentiles % of the 10 streamflow series grouped by each scenario and period to refer to extremely wet and dry years, respectively. For the analysis of future projections, two future climatic slices of 30 years each were considered: 2036–2065 and 2071–2100, called the 2050s and 2080s, respectively.

The future projection is based on the percentage of change [87,88] in hydro-energy parameters between the future period (2036–2065 and 2071–2100) and the reference period (1981–2010); for both periods, the parameters are averaged over a period of 30 years [20,35,89,90,91]. Emphasis was placed on exposing the percentage of change grouped by regions, as well as expressing the results obtained through boxplots to estimate the uncertainty of the GCMs using the interquartile ranges [92]. Finally, the percentage of positive or negative change obtained for each hydrological condition and scenario was applied to observed series of hydro-energy parameters and the climatologies of precipitation, temperature, and streamflow.

3. Results

3.1. Potential Hydropower Sites

The results show that there are a total of 2523 sub-basins corresponding to 96 hydrographic units, located outside the restricted zones. The river sections located in these sub-basins add up to a potential theoretical capacity of 91.7 GW; these values make up the hydro-energy potential of Peru (

Table 4. Descriptive statistics of potential Theoretical Capacity.

Statistics	Total (a)	RoR (b)	RoR (c) (Only Mini)	RoR (d) (Only Small)
Sum (GW)	91.68	29.07	2.6	26.46
Mean (MW)	36.34	2.43	0.43	4.51
Median (MW)	4.17	0.96	0.37	2.86
Min (MW)	0.1	0.1	0.1	1
Max (MW)	5021.84	19.98	1	19.98
Standard Error	2.95	0.03	0	0.05
Standard Desv.	148.42	3.55	0.25	4.14
Coefficient of variation (%)	606.2	5.19	0.15	3.79
N° sub-basins	2523	1735	1158	1023
N° sites	-	11,965	6103	5862

). Potential (exploitable) generation production is 133 TWh/yr according to Zhou et al. [93], in the top 15 in the world. This value is 50% higher than that estimated by Alemana et al. [75], and the difference is due to three factors: the number of sub-basins, the use of $Q_{80}$ instead of $Q_{50}$, and the semi-distributed hydrological model to generate the PISCO-HyD-ARNOVIC product.

The sub-basins with hydropower potential, described in Table 4, belong to 96 hydrographic units as basins and inter-basins. The Upper Huallaga Inter-basin has the highest total capacity with approximately 10.4 GW, followed by the Middle Lower Huallaga Inter-basin, Inter-basin 49877 and Inambari Basin with an estimated 8.9, 6.2, and 5.8 GW, respectively. On the other hand, the Coata Basin has the lowest capacity of 0.14 MW with only one sub-basin in total. In the estimation of hydro-energy potential in general, 2523 sub-basins were considered; however, for the identification of sites, with the technical considerations of Table 1 and a superior capacity of 20 MW, the number of sub-basins is reduced to 1735, with a mean of 6.9 potential RoR sites per sub-basin (Table 4).

For a more detailed description of the hydropower potential, the following cases were analyzed (Table 4): (a) All sub-basins excluding restricted zones, (b) sites identified with capacity up to 20 MW, (c) sites identified with capacity up to 1 MW (mini-HPP), and (d) sites identified with capacity from 1 to 20 MW (small-HPP). Sites with capacity below 0.1 MW (micro-HPP based on Table 3) were excluded, because their total capacity is insignificant with respect to the national total (less than 0.2%).

(a)

The total rivers belonging to the 2523 sub-basins have a total capacity of 91.7 GW. The average of these sub-basins is 36.3 MW with a maximum of 5022 MW. Other estimates differ by approximately 30%, such as 58.9–69.4 GW [75,94]. In contrast, the study [94] indicates that the theoretical capacity is 170 GW.

(b)

If we limit this threshold to the one allocated to SHPs, the potential capacity decreases to 29.1 GW with an energy of 484.5 TWh/yr. Furthermore, over 305 MW are currently installed in SHPs [60]. As a result, the potential CR is approximately 28.8 GW.

(c)

When selecting only the mini SHPs (Table 2), the potential CR is only 2600 MW with an energy of 54.2 TWh/yr, i.e., 10 and 12% of the total, respectively. More than 3.4 MW are currently installed in SHP [60]. As a result, the potential CR is approximately 2597 MW, distributed over 6103 sites with an average of 0.43 MW.

(d)

When selecting only small SHPs (Table 2), the potential CR is only 26,460 GW with an energy of 430.3 TWh/yr, i.e., 90 and 88% of the total, respectively. More than 302 MW are currently installed in SHPs [60]. As a result, the potential CR is approximately 26,198 MW, distributed over 5862 sites with an average of 4.5 MW.

A dataset of 80 operational HPPs and 11,965 potential sites for SHP development is presented in Figure 3 and Figure 4, respectively. About the second group, metrics by stream order are presented in

Table A2. Run-of-River hydro-energy potential classified by river order.

Order	Sum	Mean	Median	Minimum	Maximum	Units
	(GW)	(MW)	(MW)	(MW)	(MW)
1	8.6	1.4	0.6	0.1	19.9	6098
2	8.1	2.4	1.2	0.1	20.0	3383
3	6.3	3.7	2	0.1	19.9	1723
4	5.1	7.3	6.6	0.1	19.9	695
5	0.9	14.3	14.7	5.8	19.7	66
	29.1					11,965

, showing that the number of sites (sub-basins) decreases as the river order increases; however, these sites exhibit a higher CR, with a median of 14.7 MW (148 MW) for the 66 (106) rivers of order 5 and 0.6 MW (1.8 MW) for the 6098 (1136) rivers of order 1. The following basic characteristics can be identified in the web application: Type of plant (RoR or on-grid systems), basin and sub-basin name, coordinates, theoretical installed capacity, firm energy, energy production, streamflow, and head.

It should be noted that the GIS modeling identified sites with the maximum possible capacity within the river segment of the sub-basins excluded from the restricted zones and with technical feasibility. The optimal sites were selected on the basis of the highest potential capacity, the horizontal distance of the watercourse, and the minimum proximity between two consecutive sites. In addition, the CR of a sub-basin can be artificially increased, for example, if two consecutive sites are accumulated by 600 m for a single HPP of the same sub-basin depending on the technical characteristics of the project or if a derivation scheme is applied to increase the height [69]. Finally, a field study survey is necessary to determine the precise location of the identified SHPs.

Table 5. The numbers of Hydropower and Run-of-River plants by region.

Regions	HPP (Operational)	HPP (Planned)	RoR (Planned)
PFS	30	16	1002
PFN	14	5	871
ALN	7	24	4553
ALC	18	13	2779
ALS	11	15	2711
TIC	0	0	49
	80	73	11,965

presents the locations sites for planned RoRs, as well as operational and planned HPPs by analysis regions in the estimation of historical (operational) hydro-energy parameters and their future (planned) projections. The largest number of HPPs (51%) is highlighted in the PFS and ALC region, while the largest number of RoRs (62%) is in the ALN and ALC region. By the end of 2023, the 80 HPPs reach an installed capacity of 5.4 GW, constituting only 5.9% of the potential CR. It should be noted that the CR does not consider the efficiency level of the turbine and generator mechanisms.

3.2. Historical Hydropower Assessment

3.2.1. Evaluation of Hydropower Parameters

The historical analysis of the hydro-energy parameters of HPPs only includes the operational ones (80 HPPs). Figure 3 shows the average annual hydro-energy parameters of HPPs; the median (minimum and maximum) for CR is 7.1 MW (0.1 and 549.4), EF is 42.9 GWh/yr (0.1 and 3015.3 GWh/yr), and EP is 119.6 GWh/yr (0.01 and 9664 GWh/yr). The ALC region showed 1270 MW, following PFS and ALS regions with an approximate CR of 481 and 291 MW, respectively; in contrast, PFN region shows the lowest CR with approximately 224 GW (Figure 3a). Regarding the EP (EF) results shown in Figure 3b (Figure 3c), the ALC region has a total of 23.2 (6.7) TWh/yr, followed by PFS with 6.0 (3.5) TWh/yr, while the PFN region presents the lowest total with 3.5 (1.3) TWh/yr. One reason for these results is due to the number of HPPs presented in Table 5, with a total of 22.5% at the identified sites in ALC regions, respectively, while only 17.5% of the identified sites are in PFN region.

On the other hand, Figure 4 mapped the average annual for the RoRs with defined CR range (0.1–20 MW) and according to the restricted zones. In total 11965 potential sites were identified, their respective characteristics as a $Q_{80}$, net head, slope, and the altitude have a median of 2.5 m³/s, 37.8 m, 7.2%, and 1432.7 masl, respectively. Moreover, at unit level, RoRs have a median (minimum and maximum) for the CR is 0.95 MW (0.1 and 20), EF is 5 GWh/yr (0.1 and 164.3 GWh/yr) and EP is 16 GWh/yr (0.9 and 946.7 GWh/yr). The ALN region showed 12.9 GW, following ALC and ALS regions with an approximate CR of 6.8 and 6.8 GW, respectively; in contrast, TIC region shows the lowest CR with approximately 0.015 GW (Figure 4a). Regarding the EP (EF) results shown in Figure 4b (Figure 4c), the ALN region has a total of 198.6 (78.6) TWh/yr, followed by ALS with 128.1 (37.8) TWh/yr, while the TIC region presents the lowest total with 0.3 (0.1) TWh/yr. One reason for these results is due to the number of identified RoR presented in Table 5, with a total of 38 and 23.2% of the identified sites in ALN and ALC regions, respectively, while only 0.41% of the identified sites are in TIC region.

3.2.2. Trends

The trends analysis of the operational HPPs is presented in Figure 5. All the hydro-energy parameters of the five regions (defined in the Table 5) analyzed show positive trends; however, trends with a confidence level higher than 90% were only identified in the CR of PFS and ALN, with a positive decadal rate of 42 and 140 MW, respectively. Furthermore, it is observed that the ALN and PFS present significant trends in the EP and EF, with decadal increases of 2893 and 218.5 GWh, respectively. It should be noted that the ALC region is where the highest CR, EF, and EP are found, with an average of up to three times higher than that of the other regions. The RoR trend analysis presented in Figure 6 shows hydro-energy parameters with positive trends in the Pacific (Atlantic) region, with maximum decadal rates in the EP of 0.67 GWh (0.92 GWh) for PFS (ALN). On the other hand, only in the TIC region were negative trends identified with a maximum decadal decrease of −0.25 GWh in EP. However, positive trends with a confidence level above 90% were only identified in the ALN region at a decadal rate of 0.92 GWh of EP.

3.2.3. Statistical Assessment

The evaluation of the methodology applied for the estimation of the hydro-energy parameters is shown in Figure 7. The streamflow with persistence at 80% and $\eta$ of 85% was used to evaluate CT from forty-one selected RoR, with CT lower than 75 MW from operational HPPs. The energy production presents a correlation of 0.57, technical potential 0.59, and firm energy 0.65. All values are greater than 0.57, which would indicate an acceptable historical representation in the estimation of the hydro-energy parameters with the method described in this study. It should be noted that some HHPs were omitted (PCH Chaglla and MCH Cerro del Águila), since their productions are regulated based on energy demands. Moreover, the results are influenced by the simplification of technical specifications such as efficiency, pressure losses, turbine type, etc. It is noted that the assessment results are less reliable for hydropower plants with CT (or CR) above 40 MW. In addition, Figure 7 shows a greater underestimation in TC and EF compared to EP; this could be explained by using a composite reference streamflow for this parameter ($Q_{avg}$).

3.3. Future Hydropower Assessment

Future projections for the SEIN hydro-energy parameters in the 2050s and 2080s were assessed in relation to the reference period of 1981–2010.

Table 6. Median percentage of total change in SEIN HPPs classified by hydro-energy parameter, scenario, and future period. Highlighted cells red show a negative change in future projections.

Hydro-Energy Parameters	Period	Scenery SSP	PFS	PFN	ALN	ALC	ALS	Total
CR	2050s	1–2.6	7.1	3.9	−3	0	3.9	0.5
		3–7.0	4.6	−2.8	−7	−6.7	−5.2	−5.4
		5–8.5	8.1	0.6	−5.5	−2.2	−3.6	-3.1
	2080s	1–2.6	6.5	3.9	1.5	5.6	5.1	3.8
		3–7.0	8.4	−1.1	−12	−8.3	−11.6	−9.2
		5–8.5	16.6	−0.6	−16.2	−7.5	−6.5	−8.7
EF	2050s	1–2.6	5.9	0.2	−6.2	−3.8	−1.5	−3.1
		3–7.0	2.7	−5.2	−12.1	−11.3	−13.4	−10.7
		5–8.5	6.7	−4.3	−10.6	−8	−9.6	−8
	2080s	1–2.6	5.5	1.3	−2.9	−2.7	−2.7	−1.9
		3–7.0	6.4	−6.7	−17.3	−13.8	−17.8	−14.3
		5–8.5	14.4	−5.6	−22.6	−13.3	−14	−14.6
EP	2050s	1–2.6	10.2	4.6	2	5.8	5.1	4.3
		3–7.0	7.7	2.1	−0.9	0.4	2	0.8
		5–8.5	12.4	6.3	0.5	4.4	3.1	3
	2080s	1–2.6	9.9	7.5	5.5	7.5	7.7	6.9
		3–7.0	13.9	7.9	−1.5	4.3	3	2.3
		5–8.5	26.1	12.8	−1.2	9.9	4.9	5.2

shows the median percentage change for the hydro-energy parameters classified by scenarios, future periods, and regions. It is highlighted that the median total change for the EF is negative for all periods, regions, and scenarios; on the contrary, the EP shows an increase; both hydro-energy parameters do not exceed 15% absolute change. Only the CR shows discrepancies in the change signal; only the PFS shows positive changes in the range of 4 to 17%; for ALC, ALS, ALN, and PFN, most of the change is negative; the magnitude of this signal is accentuated as a function of the period, scenario, and in the order of the PFN, ALS, ALC, and ALN regions. Secondly, it can be mentioned that the change signal of the SEIN HPPs varies in magnitude by hydro-energy parameters, but the signal is reduced in the period 2080s and scenario SSP5.8.5 (2050s and SSP1-2.6), following the order of regions PFS, PFN, ALS, ALC, and ALN from 16.6 to −6.5% (7.1 to 3.9%), 14.4 to −14% (5.9 to −1.5%), and 26.1 to 4.9% (10.2 to 5.1%), for CR, EF, and EP, respectively.

On the other hand, the amplitude of the change found in the SEIN HPPs is visually expressed in Figure 8, which also includes the changes obtained with the 20th and 80th percentiles. On the other hand, complementing the results shown in Table 6, it is observed that the range of change projections widens (reduces) during 2080s—SSP5 scenario (2050s—SSP1), with values from −30 to 40% (−5 to 15%), −30 to 40% (−5 to 10%), and −10 to 60% (−5 to 20%), for CR, EF, and EP, respectively. The results presented in Figure 8, exclusively for the 50th percentile, disaggregated at the level of the SEIN HPPs, are shown in Figure 9. Considering these results, it is projected that the positive change is generally positive in the HPPs of the extreme south of the South Pacific region and in two HPPs of the North Pacific; for the other regions, the change per HPP is neutral (−10 to 10%) or negative (<−10%). It is relevant to mention that the Central Atlantic region, where the set of HPPs with the highest CR is located (52.7 and 59.2% of the total operational and planned, respectively), is the one that shows the greatest number of HPPs, with signs of reduction in CR and EF, which is accentuated from the 2050s to the 2080s and from the SSP1-2.6 to SSP5-8.5 scenarios. In general, it was found that the change signal tends to increase in the extreme north/south zones of the Pacific slopes and decrease in the ALC region. The change in the average precipitation and streamflows leads to an increase/decrease in the hydro-energy parameters.

The future projections for the 2050s and 2080s in the identified RoRs were evaluated with respect to the reference period 1981–2010.

Table 7. Median percentage of total change in identified RoR classified by hydro-energy parameter, scenario, and future periods. Highlighted cells red show a negative change in future projections.

Hydro-Energy Parameters	Period	Scenery SSP	PFS	PFN	ALN	ALC	ALS	TIC	Total
CR	2050s	1–2.6	10.4	1.5	−3.6	−4.2	−0.3	0.3	−2.2
		3–7.0	3.2	−4	−7	−11.2	−5.8	−5	−7.2
		5–8.5	9.8	0.8	−5.4	−8.8	−3.7	−1.4	−5
	2080s	1–2.6	8.1	4	1.7	0.4	2.3	1.3	1.8
		3–7.0	10.8	0.1	−11.8	−18.7	−8.9	−9.1	−11.5
		5–8.5	24	2.6	−14	−20.3	−7.2	−5.9	−11.9
EF	2050s	1–2.6	8.6	−1.1	−6.2	−6.4	−3.4	−1.1	−4.8
		3–7.0	1.4	−5.7	−10.6	−14.3	−12.7	−10.3	−11.4
		5–8.5	9.2	−3	−9.4	−12.4	−9.5	−3.9	−9.2
	2080s	1–2.6	7.1	2.5	−1.4	−6.1	−5.5	−5	−3.1
		3–7.0	8.4	−3.7	−15.9	−23	−15.9	−16.5	−16.1
		5–8.5	21.7	−0.9	−18.2	−26.2	−14.5	−17	−16.9
EP	2050s	1–2.6	13.1	4.7	1.9	3.2	3.2	3.4	2.9
		3–7.0	6.3	2.5	−1.2	−2.8	−0.6	0.4	−1.1
		5–8.5	12.9	6.1	1.4	−1.2	1.3	4.3	1.2
	2080s	1–2.6	11	7.9	6.2	4.4	3.5	5.8	5.2
		3–7.0	16.6	10.6	−1.1	−5.7	0.7	1.3	−0.9
		5–8.5	33.3	16	1.4	−4.8	3.3	10.2	1.6

shows the median percentage change in the parameters grouped by scenarios, future periods, and regions. It is highlighted that the total change in EF and CR is negative in all periods, regions, and scenarios (except for CR in the 2080s period and scenario SSP1-2.6); on the contrary, energy production shows an increase (without considering the reduction in the 2080s period and scenario SSP3-7.0). In general, the average change in the parameters does not exceed the lower and upper thresholds of −16.9 and 5.2%. The only discrepancies found in the CR and EP change signals are due to the median signal found in the Atlantic regions, which differs only in the 2080s period and SSP1-2.6 or SSP3-7.0 scenarios. At the regional level, only in the PFS are there positive changes in the range of 1.4 to 33.3% in all hydro-energy parameters; for the other regions most of the change is negative in the CR and EF, and the magnitude of this signal is accentuated as a function of the period, scenario, and generally following the order of the PFN, TIC, ALS, ALN, and ALC regions. Secondly, it can be mentioned that the change signal of the identified RoRs grouped by regions varies in magnitude by hydro-energy parameters, finding a pattern of reducing change, in the period 2080s and scenario SSP5–8.5 (2050s and SSP1–2.6), following the following regional order: PFS, PFN, TIC, ALS, ALN, and ALC, in the range of 24 to −20.3% (10.4 to −4.2%), 21.7 to −26.2% (8.6 to −6.4%), and 33.3 to −4.8% (13.1 to 3.2%) for CR, EF, and EP, respectively.

On the other hand, the range of change percentages found in the identified RoRs is visually expressed by box plots in Figure 10, which also includes the changes obtained with the 20th and 80th percentiles. It is observed that the range of the signals is expanded for the value groups for the 2080s period and SSP5-8.5 scenarios, compared to what was found in the signals for the 2050s period and SSP1-2.6 scenario. In the first case (second case), the minimum to maximum signals are within the range of −30 to 40% (−5 to 15%), −40 to 40% (−20 to 20%), and −20 to 60% (−5 to 15%) for the hydro-energy parameters CR, EF, and EP, respectively.

At the level of RoRs (50th percentile), we found that (Figure 11) it is projected that the positive change in the hydro-energy parameters is generally positive in the RoRs of the PFS and ALS regions and only concerning the EP during the 2080s period in the PFN; for the other regions, the change in the RoRs is generally neutral (−10 to 10%) or negative (<−10%). It is relevant to mention that the ALN and ALC regions where the largest amount of RoRs are found (61.3% of the total) mostly show signs of reduction in CR and EF, which are accentuated from lower to higher from the 2050s to the 2080s period and from the SSP1-2.6 to the SSP5-8.5 scenario. In general, it was found that the change signal tends to increase (decrease) in the Pacific (Atlantic) regions. The change in the averages of precipitation and streamflows leads to a change with the same positive or negative signal of the hydro-energy parameters analyzed.

4. Discussion

4.1. Historical Variability Hydro-Energy Parameters Validation

The presented research used 40 continuous years of data, as it is recommended to use extensive flow records (30 years or over) to estimate the hydro-energy potential [72]. Considering the results of Table 4, we compare our estimates of historical hydroelectric potential with those from the few existing studies on hydroelectric potential in Peru. Our theoretical potential agrees with the global assessment estimates by Hoes et al. [95] and Alemana et al. [75], which utilized similar input data resolution and theoretical capacity, respectively. Furthermore, our estimates are more precise than last Peru’s Atlas [94], which were based on regionalized streamflows and nonlinear equations derived from areal precipitation. The differences between this study and the previous ones [75,94,95] can be explained as follows: the first difference stems from the semi-distributed hydrological model based on sub-basins, offering a better representation of streamflows through the regionally calibrated hydrological model. For the second case, the primary distinction lies in the use of high-resolution digital elevation models, which more accurately depict the elevation gradients of water networks. Regarding the third study, the key difference is the application of a hydrological model with flood routing correction that accounts for streamflow variability over the past decade. Additionally, greater uncertainties arise when comparing resource evaluation models for hydroelectric potential using runoff-area models, as potential evaluations based on regression models are increasingly rare [38].

Improving the DEM used in this study is important because it is indicated that it is not the same as the one used to determine the sub-basins and river reaches of the PISCO-HyD-ARNOVIC product, but it was the closest to determine the longitudinal profiles of the rivers compared to other publicly available 90 m DEMs. Moreover, the accuracy of the DEM has important implications for the identification of potential RoR sites, as it affects the net head estimation and the river delimitation procedure [48,64,96]. This study implemented a 90m DEM corresponding to the globally available MERIT-Hydro DEM, therefore, field measurements are required to verify the reliability of the identified locations. Normally, a coarser DEM gives a decreased representation of the subbasin area and their net head, which causes possible limitations for the estimation of hydropower potential, especially on flat areas [72,96]. The present research showed most potential sites are located in the mountainous Andean regions (ALN, ALC and ALS) with a high slope, so it is expected that the limitations of the DEM used to estimate net head will not be a determining issue.

The lack of a clear definition in the selection of exceedance streamflow and standardized evaluation methods for hydroelectric potential classes, alongside the selection of hydroelectric design parameters and spatio–temporal resolution, amplifies the discrepancies in results from national or regional studies [97]. In this context, the estimates from this study can be considered reasonable compared to the referenced previous studies. A comparison of our approach at the local or basin scale, using a compilation of digital elevation models with higher spatial resolution, would provide greater validation of the results.

4.2. Comparison for Future Projections with Other Studies

The pattern and spatial direction of future changes in hydropower potential found in our study are consistent with different spatial scales, scenarios, or models, with their respective limitations and uncertainties associated with the base data or methodologies used. However, the changes we identify are generally higher. Recent studies [98,99] found changes of 6 to 9% in the CR of a group of SEIN HPPs located in the South Pacific and Central/South Atlantic regions, applying 21 CMIP5 GCMs under RCP4.5 and RCP8.5 scenarios, of a similar scope to SSP3-7.0 and SSP5-8.5, respectively [98]. The decrease in CR change in the operational and planned HPPs of Peru using a set of CMIP5 GCMs was also found by Paltán et al. [99], with signals of up to −25% in the Atlantic regions due to the change in the streamflow regime accentuated by (i) the global trend in the number of rainy days and (ii) the increase in the frequency of droughts with greater intensity and duration. This decrease projection is also supported by Paltán et al. [99], based on the analysis of three GCMs of the CMIP5 RCP8.5 scenario, where a reduction in CR is estimated in the Atlantic regions due to the reduction in precipitation that would affect the hydroelectric performance of the region, particularly in the period 2070–2099. Even at a regional level in South America, future projections of CR are estimated by Zhang et al. [100] in Peru, indicating a likely decrease −20 to 0% in the most southern parts and a low increase up to 20% in the remaining parts, for the period 2080–2099 compared to 1971–2000, using CMIP5 GCMs under the RCP8.5 scenario, mainly due to streamflow reduction.

To complement future projections of hydro-energy parameters, we rely on signals of change in hydrological variables such as precipitation. In Peru, some studies at regional or national level [101,102,103] project an agreement in the annual signal of wet (dry) patterns based on the precipitation regime over most of the Pacific slope and the Andes region (extreme south and jungle regions) across a set of CMIP6 GCMs under the SSP2. 4-5 and SSP5-8.5 scenarios; it is worth noting that seasonally during June–November, the decrease (increase) in wet (dry) patterns intensifies. Additionally, recently SSP3-7.0 received increased attention due to leading to the warmest climate scenarios [104]. Our results coincide with these observations, since, in the regions analyzed and mainly for the period 2050s, higher rates of decrease and lower increase rates are observed for all hydro-energy parameters (Table 6 and Table 7). Moreover, various studies evaluating the performance of GCMs in representing historical climate patterns indicate a higher accuracy in capturing temperature variations compared to precipitation; consequently, future projections of annual precipitation cycles and subsequent hydrological products, such as runoff estimates, exhibit greater uncertainty than temperature trends [105,106,107,108,109].

In this sense, studies focused on South America suggest that projected increases or decreases in precipitation vary significantly across models due to internal fluctuations in circulation patterns; these projections tend to show greater consistency during the mid-future period (2040–2070), whereas uncertainty increases substantially in the late future (2070–2100), where model dispersion is more pronounced [59,106,109]. This is observed in our results of level of agreement of changes between GCMs; the range in the projection for Temp and Prec is much larger for ensemble results for the SSP5 model ensemble than SSP1, and the same pattern is observed in the 2080s result set, with respect to 2050s. Furthermore, the ranges of changes are higher in the SSP5 dataset with a range between 7.6 and 15.8% in Temp and −12.2 and 8.5% in Prec, while in SSP1 they are 4.6 and 9.8% and −8.5 to 7.4%, respectively (Figure A3). Furthermore, with respect to changes in the ETCCDI indices by GCMs, it is generally observed that models projecting high negative (positive) changes in Prec also show high positive CDD (R99P) changes; while Temp models projecting high changes also show high changes in WSDI and CSDI. Although an increase in potential capacity is projected due to a rise in annual average streamflows on a global scale, a large part of the HPPs are located in sub-basins where decreases in annual average streamflows are expected, with consequences in the decrease of their hydroelectric generation [100,110]. The differences in the future capacity or generation projections of the studies analyzed are mainly due to the uncertainties associated with the different climate models, greenhouse gas emission scenarios, reference periods, as well as the input data and hydrological models used for the generation of streamflows [111].

4.3. Implications for Energy Security and Future Perspectives

The results of this study can be considered in the development of public policies for National Energy Planning [112], aimed at closing the energy deficit gaps and quantifying the economic viability of hydro-energy projects identified at the local scale in the first instance, especially in regions that do not have access to the SEIN energy grid [5,61]. Furthermore, the level at which energy security can be achieved through sustainable hydropower depends on spatio-temporal changes in future energy demand and the availability of other RERs [86]. In 2023, annual energy consumption in South American countries followed an increasing pattern, with Peru having the lowest electricity consumption compared to Brazil and Chile at 1.81, 2.73, and 4.36 MWh per capita, respectively [113,114]. In Peru, hydroelectricity currently represents about 50% of national energy generation and 90% if only RER production is considered [114]. According to studies on the future of electricity, to meet growing energy demand under scenarios of extreme hydrological conditions, it is estimated that Peru will have to add to its current installed hydroelectric capacity in 2030 (2050) by 11.1 GW (32.6 GW) and to energy production in 2030 (2050) by 88.4 TWh (176.4 TWh) [115]. For this reason, the sustainable portfolio of projects composed of the potential sites for RoR development identified in this study would cover 88.3% of projected hydropower demand in Peru by 2050.

This research updates the understanding of the potential hydropower resources and their future projections for Peru, information key to socioeconomic development, and improvement of climate adaptation policies in terms of water resource management [98,116]. In this context, the improvement of our capacities, like (i) developing a hydrological model of higher temporal and spatial resolution, (ii) inclusion of regional GCMs for the analysis of future projections, and the (iii) improvement of DEMs to increase the representativeness of the net head and longitudinal profile of rivers, is important to improve our understanding of hydrological processes in the overall assessment and development of our hydropower resources [96,109]. Furthermore, national energy planning frameworks must also adapt hydropower development in systems that include the nexus water–energy–food sectors under future projection scenarios, as strategies need to be developed for the design of equitable structures among the stakeholders, to achieve SGD-6 and SGD-7 [97,100,117].

5. Conclusions

The rivers in Peru are not yet fully developed for hydropower generation, with only 5.7% (5.2 out of 92.7 GW) of the gross potential in current use, according to the historical period analyzed, setting a major challenge for the identification of sustainable pathways for hydropower development. The values of hydropower potential estimated were compared with previous studies, some differences were found, and the reasons are given in the discussions. Our study showed a method using GIS tools and requiring only open-access streamflow and remote sensing data to sizing and identifying suitable sites for RoR with high theoretical hydropower potential. Furthermore, these results combine CMIP6 climate projections with hydropower modeling to assess the projected future hydropower potential of existing hydropower plants and potential RoR in Peru.

Based on the siting method, it is concluded that Peru has potential in 1735 sub-basins, mainly located in the northern and central Andes region, for suitable RoR with 5862 (6103) sites with unit installed capacity between 1 to 20 MW (0.1 to 1 MW), with availability estimates for CR at 26.5 GW (2.6 GW) and EP at 430.3 TWh/year (54.2 TWh/year). The location of these 11,965 sites is given in non-excluded areas and takes into account technical feasibility requirements. It should be noted that the above methodology cannot replace field work, but it can save time and give an accurate initial idea of potential sites and provide baseline information for follow-up pre-feasibility studies.

We show that the main parameters of hydropower potential are projected to decrease under most climate change scenarios, with regional and sub-basin variations in terms of magnitude. According to changes in 2050, the historical on-grid (SEIN) theoretical capacity of 5.2 GW changes by −5.4 to 0.5%, while the energy production of 29.9 TWh/year increases by 0.8 to 4.3%, and the firm energy of 25.3 TWh/year decreases by −10.7 to −3.1% (adding the projections of the 73 planned HPPs). On the other hand, for potential RoR, their historical theoretical capacity of 29.1 GW decreases by −2.2 to −7.2%, while the energy production of 484.5 TWh/year changes by −1.1 to 2.9% and the firm energy of 172.7 TWh/year by −11.4 to −4.8%. Only SSP1-2.6 shows some increase in projections, and SSP3-7.0 shows the largest decrease. Some spatial differences highlight the importance of regional assessment of hydropower potential, as the PFS region has a different signal than other regions. In this regard, we present datasets to improve hydropower planning by using RoR that are better adapted to the effects of climate change impacts due to their location, under different scenarios and periods.

Consequently, these results will help private promoters to supply the energy demands for their investment projects through potential RoRs in areas with low risk of environmental and/or social conflicts. Besides, this information is useful for national policy makers in the energy sector to plan a decentralised electrification with an accent on off-grid systems in the remote localities in the Andean regions, with the objective of achieving future energy security based on renewable and sustainable energy. Therefore, this methodology can be replicated at regional level in other areas of interest with scarce data to obtain a brief visualisation of the hydro-energy potential of RoRs. Finally, for understanding the limitations related to the selected datasets, it is recommended the development on studies with a wider set of verification data for estimating effect of spatial resolution from DEMs in the determination head net and subsequent hydro-energy potential of hydrographic units. Also, we recommend investigating the economic feasibility of potential RoRs and validate it through in-situ surveys.