Impact of climate change on the magnitude and extent of riverine floods in a Peruvian Andean–Amazonian basin

Abstract

In the last two decades, approximately 2.15 million inhabitants in Peru were affected by floods, mainly concentrated in the Andean–Amazon basin, whose frequency has increased as a consequence of climate change. The objective of this research is to evaluate the potential impact of climate change on the magnitude of frequent (2-year) flood flows and the extent of flooded areas in the Huallaga River basin (HRB). We used the hydrological and hydrodynamic model MGB and an approach to select four CMIP6 global climate models (GCMs) representative of extreme conditions (Cold-dry, Cold-wet, Warm-dry, and Warm-wet) for the scenarios SSP2-4.5 and SSP5-8.5, analyzed in two future periods (2035–2065 and 2069–2099). The results show that the impacts on flood flows and flooded areas vary according to the selected GCM, with the largest increases concentrated in the headwater areas and along the main channel. Similarly, it was identified that the political provinces located in the north and the center-west of the HRB present the greatest agreement in the increase in flooded area (between 7 and 8 models of a total of 8 models). These findings are relevant for territorial planning, disaster risk management, and adaptation decision making in response to climate change.

1 Introduction

Floods impose a global risk that threatens approximately 1.81 billion people, or 23% of the world’s population, who are exposed to depths of at least 15 cm given a 100-year return period flood (Rentschler et al., 2022). This is because there is a relatively large population in areas close to current floodplains that is highly vulnerable even to modest increases in potential flood risk (Swain et al., 2020). In Peru, this problem is no less serious, as floods are one of the main risks associated with natural disasters (BID, 2015), which between 2003 and 2022 have affected 2,154,434 inhabitants, mainly concentrated in northern Peru in the departments of Loreto, San Martín, and Huánuco (INDECI, 2023).

Climate change is expected to intensify the hydrological cycle (O’Gorman, 2015), producing significant changes in human exposure to flood hazards (Arnell and Gosling, 2016). According to Arnell and Gosling (2016), by 2050, approximately 450 million people and 430,000 km² of farmland would be exposed to a doubling of flood frequency as a result of climate change, causing the total global flood risk to increase by 187% compared to the situation in the absence of climate change. These results are subject to considerable uncertainty (Chen et al., 2011; Clark et al., 2016) that is mainly related to: (i) selection of greenhouse gas emission scenarios, (ii) choice of climate model(s), (iii) specifications of the initial conditions of the climate model, (iii) choice of downscaling technique, and (iv) choice of the structure and parameters of the hydrological model(s). Among these options, the choice of representative concentration pathway or global climate model stands out as an important contributor to overall uncertainty in the projections (Chegwidden et al., 2019; Chen et al., 2011). On the other hand, identifying future flood-prone areas is key to implementing proactive or corrective risk management measures, both for the protection of the population and economic activities (Munar et al., 2023). An example of this is the study by Munar et al. (2023), who found an expected increase in the maximum extent of flooding, which would intensify the danger in the lowlands of the Magdalena River basin in Colombia.

Recent research conducted in South America, with a focus on the Amazon River basin (Sorribas et al., 2016; Brêda et al., 2023; Miranda et al., 2023), has examined how projections from climate models cascade into impacts on flood flows. Sorribas et al. (2016) found that there is no agreement among climate models on changes in total flood area and flow anomalies along the main channel, where models tend to agree better with wetter (drier) conditions in the western (eastern) Amazon. On the other hand, Brêda et al. (2023) concluded that projections for flood flow generation are influenced not only by extreme precipitation but also by antecedent soil moisture, because when extreme precipitation and antecedent soil moisture show different signs of change, the discharge for a 2-year return period is more likely to follow the same sign of change as soil moisture. In the case of South America, a generalized reduction in soil moisture has been identified (more pronounced with the RCP8.5 scenario), which has possibly led to the number of large basins (>1,000 km²) exhibiting a negative sign for flood discharge being greater than that for extreme precipitation, i.e., the fraction of basins showing positive/negative signs was 50%/50% for extreme precipitation, this ratio shifted to 30%/70% for flood discharge. In the Peruvian Andes and Amazon basin, there are few studies on the impact of climate change on flooding, and most have been conducted on larger spatial scales (e.g., Brêda et al., 2023; Hirabayashi et al., 2021; Sorribas et al., 2016). This highlights the need for specific research at the basin scale. In general, these studies agree that an increase in the frequency of flooding is expected in the Peruvian Amazon (Brêda et al., 2023; Hirabayashi et al., 2021).

There is evidence that climate change will impact the frequency and magnitude of flood flows, but frequency in particular is taking on an important role due to its negative repercussions, as any change in flood frequency is likely to have enormous economic, social, and environmental consequences, including loss of life (Iqbal et al., 2018; Rasmussen, 2016; Dong et al., 2018; Maghsood et al., 2019; Quintero et al., 2018; Arent et al., 2015). These impacts differ depending on the local climate and the characteristics of the river basin (Duan et al., 2017). For example, the Kabul River basin, as the largest tributary of the Indus River, experiences more frequent and intense flash floods in the lowlands of the basin, especially when heavy monsoon rains combine with snowmelt runoff during the summer months (Abbaspour, 2013). According to Milly et al. (2002), the frequency of floods with a 100-year return period increased substantially during the 20th century in basins larger than 200,000 square km worldwide, and there is a statistically significant positive trend in the risk of major floods. These findings support the importance of taking into account the impacts of climate change in assessing flood frequencies based on river discharge projections (Milly et al., 2002; Almasi and Soltani, 2017), despite their associated uncertainty and controversy (Rasmussen, 2016).

The objective of this study is to evaluate the impacts of climate change on the magnitude and extent of frequent river floods in the Huallaga River basin (HRB) based on CMIP6 climate projections. We used a hydrological and hydrodynamic model and applied an adapted climate model selection approach to represent the range of trends in precipitation and average temperature change. This article is organized into the following sections: (i) study area and materials, where the study area and data used are described; (ii) methods, which includes the description of the hydrologic and hydrodynamic model, the approach for GCM selection, bias correction, and impact assessment; (iii) results, presenting the evaluation of the hydrological model performance, the selection of GCMs and the impacts of climate change on precipitation, mean temperature, as well as on the magnitude of flood flows and the extent of flooded areas; and (iv) discussion, focusing on the analysis of climate change impacts, uncertainties, and limitations of the study.

2 Materials and methods

2.1 Study area and materials

The Huallaga River Basin (HRB), located in the Andean–Amazonian region of northern Peru, is one of the main headwater basins of the Amazon River system, which flows into the Atlantic Ocean (see Figure 1). The HRB has a drainage area of 90,232 km², representing approximately 1.5% of the Amazon basin and 7% of the national territory. Its territory mainly covers the political departments of San Martín, Huánuco, and Loreto, with an approximate share of 57, 21, and 15% of the total area of the basin, respectively. It has a varied topography with elevations between 100 and 6,100 m above sea level, an average annual rainfall of 1,700 mm, and an average temperature of 20 °C (Huamanchumo Sono et al., 2023).

Figure 1

Floods are one of the main natural hazards in Peru, especially between November and April, during the rainy season (BID, 2015). In this context, the HRB has been identified by the National Water Authority (ANA) and the National Center for Disaster Risk Estimation, Prevention, and Reduction (CENEPRED) as one of the basins with the highest risk of flooding, a risk that is greater during El Niño events (CENEPRED, 2023). These floods particularly affect primary economic activities, such as agriculture, on which a large part of the basin’s population depends, estimated at approximately 1.72 million people, concentrated mainly in Huánuco (34.0%), San Martín (48.9%), and Loreto (7.0%) (ANA, Engecorps, and Inclam, 2015).

2.2 Hydrometeorological data

Precipitation and climate data were obtained from the Peruvian Interpolated Data of SENAMHI’s Climatological and Hydrological Observations (PISCO) database. Precipitation data were used from PISCO’s operational daily gridded precipitation data (PISCOpd_Op; Millán-Arancibia and Lavado-Casimiro, 2023), which has a spatial resolution of 0.1° and provides data from 1981 to the present, with daily updates. This product was generated from 416 conventional rain gauges of the SENAMHI network and is based on the genRE interpolation method (van Osnabrugge et al., 2017).

Climate data were obtained from PISCOeo_pm (Huerta et al., 2022), which provides daily and monthly average data from 1981 to 2016 with a spatial resolution of 0.01°. PISCOeo_pm was generated from more than 300 weather stations of the SENAMHI network. The spatial interpolation is based on regression Kriging using satellite-derived predictors such as land surface temperature, cloudiness, terrain elevation, and wind speed from WorldClim v2.1 (Fick and Hijmans, 2017). In this study, only long-term monthly averages of temperature, sunshine hours, and wind speed from PISCOeo_pm were used, solely for the estimation of climatic variables required to compute potential evapotranspiration. Relative humidity was derived from temperature, dew-point temperature, and vapor pressure following Shuttleworth (1993).

Discharge records were obtained from the hydrological station network maintained by SENAMHI. A total of 13 gauging stations with adequate temporal coverage were used, located at various points across the basin and mainly along the main river (see Figure 1). Detailed information for these stations is provided in Table 1.

2.3 Hydrologic–hydrodynamic model

The Modelo de Grandes Bacias (MGB; Collischonn et al., 2007) was selected for this study due to its history of development focused on hydrological processes of large tropical basins, mainly in South America (Collischonn et al., 2007; De Paiva et al., 2013; Siqueira et al., 2018), and previous applications in climate change studies covering the Amazon basin (Sorribas et al., 2016; Brêda et al., 2023; Brêda et al., 2020).

MGB is a large-scale, semi-distributed conceptual hydrological model that simulates the processes of the terrestrial hydrological cycle. The model divides the basin into small elements called unit-catchments, each one subdivided into Hydrological Response Units (HRUs) defined by similar land cover, soil, or topographic features. In these HRUs, runoff generation and evapotranspiration are calculated at a daily time step. Groundwater, subsurface, and surface runoff are routed to the main channel of the unit-catchment using a linear reservoir approach, while flow propagation through the river network is computed using an explicit inertial approximation of the full 1D hydrodynamic equations (Pontes et al., 2017). Thus, the model also enables the estimation of other hydraulic variables than streamflow, such as water levels and flood inundation areas.

For this study, we used the same MGB model for the HRB presented in Saavedra et al. (2022). This model was constructed based on the MERIT-Hydro dataset (Yamazaki et al., 2019) (http://hydro.iis.u-tokyo.ac.jp/~yamadai/MERIT_Hydro/), a suite of products derived from the MERIT DEM (Yamazaki et al., 2017) that facilitates the preparation of large-scale hydrologic–hydrodynamic models. Unit catchments were delineated using a threshold of 100 km² for minimum upstream drainage area and a fixed river reach length of 8 km, yielding 1,168 unit catchments with an average area of 77 km². Land use map for the year 2015 was retrieved from the fourth collection of MapBiomas Amazonía (Turpo Cayo et al., 2022) (https://amazonia.mapbiomas.org/) and was reclassified into five main land cover types, including forest (66.07%), grassland (12.93%), agriculture (19.85%), semi-impermeable (0.21%), and water (0.95%). HRUs were delineated by combining the reclassified land cover with categorical maps of DEM slope and Height Above the Nearest Drainage (HAND).

River channel width (w) and depth (d) were estimated by developing hydraulic geometry relationships between field measurements of these parameters, obtained from ADCP surveys at seven river cross sections, and their corresponding upstream drainage areas. River widths were further refined manually by introducing additional cross sections from Google Earth imagery and by visually comparing observed and simulated flow timing. The Manning’s coefficient was uniformly set to 0.035, except in the Biavo sub-basin, where it was reduced to 0.03 to prevent excessive flow attenuation (De Paiva et al., 2013). The model calibration was performed manually from January 1994 to December 2022, considering the Nash-Sutcliffe efficiency coefficient (NSE), the NSE of the logarithm of flows (NSElog), the Kling-Gupta efficiency (KGE), and the correlation coefficient (r). The reason for calibrating the MGB with the entire data period is that this strategy leads to a more robust set of parameters, as suggested by recent works (Arsenault et al., 2018; Shen et al., 2022). For further details on the MGB model setup, the reader is referred to Saavedra et al. (2022).

2.4 GCM selection approach

The future projections of monthly precipitation and air temperature were obtained from the GCMs of the Coupled Model Intercomparison Project phase 6 (CMIP6) available at the following link: https://cds.climate.copernicus.eu/datasets/projections-cmip6?tab=download. CMIP6 is a climate dataset project that provides five shared socioeconomic pathways (SSPs) that were developed taking into account socioeconomic factors such as urbanization, economic growth, population, education, and the rate of technological development (Riahi et al., 2017). This dataset was developed by the effort of different climate research institutions for the understanding of climate change (Meresa et al., 2022). In this study, we extract data for the scenarios SSP2-4.5 (low level) and SSP5-8.5 (high level) for the period from 2030 to 2099. Only GCMs with available precipitation and mean air temperature data were considered in the analysis. As a result of this selection process, a total of 17 GCMs were obtained for both scenarios.

We then selected a subset of the GCMs by using an approach adapted from Nigussie et al. (2024), Dhaubanjar et al. (2023), and Lutz et al. (2016), which consisted of weighting the range of projections in the GCM ensemble and the ability of the individual model to simulate historical climate. For each SSP, four GCMs were selected based on the precipitation and temperature corners. This process was carried out at the basin level, where the main steps consisted of:

• Determine the 10th and 90th percentiles based on projections of changes in precipitation (∆P) and temperature (∆T) between 1984–2014 (historical period) and 2069–2099 (future period). Using the values obtained for ∆P and ∆T, the four corners of the range of temperature and precipitation projections representing future scenarios have been established, which are categorized as follows:

• “Cold-dry”: 10th percentile values ∆T and ∆P.

• “Cold-wet”: 10th and 90th percentiles of ∆T and ∆P values, respectively.

• “Warm-dry”: 90th and 10th percentiles of ∆T and ∆P values, respectively.

• “Warm-wet”: 90th percentile values ∆T and ∆P.

• The range score is calculated for each GCM as the Euclidean distance to each vertex (Equation 1), which is then normalized for the entire range of projections ΔT and ΔP.

where represents the Euclidean distance between model i and corner j, defined by the values ΔP and ΔT. Here, ( y ) correspond to the 10th and/or 90th percentiles of the set of models (see previous vignette).

• Calculate the skill score for each GCM using the mean absolute error (MAE) for the historical time series of precipitation and temperature variables from the GCMs with respect to the PISCO database. The MAE values calculated for precipitation (Equation 2) and mean temperature (Equation 3) are multiplied to obtain a combined skill score (Equation 4).

where and represent the mean absolute error of model i for the monthly mean values of precipitation and mean temperature, respectively; and are the observed values, which in this case were obtained from the PISCO database, while and are the values simulated by model i for sample z of monthly mean observations (i.e., 1, 2, …, 12), and N indicates the number of months equal to 12. The value corresponds to the product of both errors.

• The final score for each GCM model is obtained by averaging the skill and range scores (previously calculated) and inverting them so that higher values reflect a better score (see Equation 5). At each corner, the GCM with the highest score is selected in order to obtain the most representative model.

2.5 Bias correction

GCMs model terrestrial processes in coarse grid cells, which is not suitable for regional or local studies due to their existing biases (Maraun, 2016; Navarro-Racines et al., 2020; Brêda et al., 2020). For this reason, it is necessary to apply some type of bias correction to adjust selected statistics from a climate model simulation to better match statistics observed during a current reference period (Maraun, 2016). Although current bias correction techniques have limited capacity (Maraun, 2016), they are still necessary in hydrological modeling because a bias in precipitation may largely affect the simulated flows (Brêda et al., 2020).

The monthly delta change method was used because it is simple and robust, and has been used in various climate change impact studies (Brêda et al., 2020; Mendez et al., 2020; Mejía et al., 2023). This method does not exactly correct the bias of a climate model, but rather adds a signal of change derived from a climate model to the current climate (Maraun, 2016). In this study, data were extracted at the level of ten sub-basins (Figure 1), considering a historical period (1984–2014) and two future periods (2030–2060 and 2069–2099). Based on this information, monthly change factors were estimated, relative to precipitation and additive for temperature, used in the correction of future scenarios (see Equation 6):

where and are the simulated monthly averages of precipitation from climate models for the present and future periods, respectively, and are the simulated monthly averages of temperature from climate models for the present and future periods, respectively, and correspond to the observed average precipitation and temperatures in the present period, obtained from PISCOpd_Op (0.1°) and PISCOeo_pm (0.01°), respectively. and are the bias-corrected mean precipitation and temperatures for the future period.

2.6 Evaluation of impacts

The evaluation of the impacts of climate change on flood flows and extent was carried out at the river section level throughout the HRB. For this purpose, we first determined separately the maximum annual flows and flooded areas for the historical period (1984–2014) and for two future periods: medium term (2035–2065) and long term (2069–2099). We then calculated the values associated with the 2-year return period (PR2) using the Weibull position formula (Hallinan, 1993) to estimate the projected changes in recurrent flood events.

In addition, we also used two measures of hydrological behavior signatures obtained from the flood duration curve (FDC) (Mendoza et al., 2015; Saavedra et al., 2022; Pokhrel et al., 2012) to assess the impact of climate change on the percentage bias in FDC mid-segment slope (%Bias FMS; see Equation 7):

where m1 = 0.2 and m2 = 0.7, while Q_m1 and Q_m2 are flows with exceedance probabilities of 0.2 and 0.7, respectively. A steep slope indicates a more immediate response in flow, while a gentler curve indicates a relatively dampened response and greater storage. We also evaluate the impact of climate change on the percent bias in FDC high-segment volume (%Bias FHV; see Equation 8):

where h = 1,2,…, H are the indices in the flow matrix with a probability of exceedance less than 0.02. FHV is a measure of the basin’s response to high rainfall.

3 Results

3.1 Evaluation of the hydrological model

The MGB hydrological model was calibrated in the HRB using available observed flow data obtained from hydrological stations (see Table 1). The results were evaluated by comparing simulated and observed flows using performance metrics such as NSE, NSElog, KGE, and PBIAS (see Figure 2 and Table 2). According to Moriasi et al. (2007), the performance of the hydrological model is considered ‘very good’ when the NSE, NSElog, and KGE values are between 0.75 and 1.00, and PBIAS is between ±10%; ‘good’ when the NSE, NSElog, and KGE values are between 0.65 and 0.75, and PBIAS is between ±10 and ±15%; ‘satisfactory’ when the NSE, NSElog, and KGE values are between 0.50 and 0.65, and PBIAS is between ±15 and ±25%.

Figure 2

The model showed satisfactory to very good performance along the main channel (Huallaga River). In the high-elevation part of the basin (south), there is good agreement between simulated and observed flows [e.g., Taruca (TAR)], although at Puente Higueras (PUH), it had difficulty capturing flow fluctuations in 2015–2016, and in Chalca (CHA), the total flows were slightly overestimated. In the south-central basin, including Tingo María (TIN) and Puente Tocache (PUT), the simulated discharge also performed relatively well, although TIN underestimated peak and low flows, particularly between 2005 and 2008. In the north-central basin [i.e., Picota (PIC) and Chazuta (CHZ)], the model showed good to very good performance, with PIC underestimating minimum flows between 2015 and 2018 and overestimating flows in 2017. On the other hand, the model showed variable performance in tributary stations. Huayabamba (HUA) showed satisfactory performance, although with difficulties in rapid peaks. Shanao (SHN) had low to satisfactory performance, with systematic underestimations. In Biavo (BIA), Cumbaza (CUM), and Shamboyacu (SHA), performance was poor, especially in small basins with high variability. Overall, the model showed satisfactory performance, with greater accuracy at the main stations along the Huallaga River, especially in the southern part of the basin, while limitations were observed in smaller tributaries and during periods of rapid flow changes.

The performance of the simulated flows is within expectations, compared to previous hydrological modeling studies in the Peruvian Amazon that used different precipitation datasets (Fernandez-Palomino et al., 2022; Zubieta et al., 2015, 2017). Also noteworthy is a recent application of the SWAT model in the Huallaga basin, within the framework of the ENANDES project, where the NSE along the main river (Tingo María, Puente Tocache, Picota, and Chazuta) varied between 0.53 and 0.65 (Ding et al., 2024).

3.2 Selected GCM

The set of models (17 GCMs) shows wide variability in the values of ∆P and ∆T, depending on the SSP scenario considered (see Figure 3). For precipitation, both increases and decreases were observed: in the SSP2-4.5 scenario, the delta values ranged from −7 to +11%, with 5 models projecting a decrease and 12 a positive trend; in the SSP5-8.5 scenario, the range widened from −15 to +20%, with 6 models showing a negative trend and 11 a positive trend. In terms of average temperature, ∆T values ranged from 1.3 °C to 3.5 °C for SSP2-4.5, and from 2.5 °C to 5.8 °C for SSP5-8.5.

Figure 3

The results obtained after applying the selection criterion at the basin scale are presented in Figure 3, where it can be seen that both the range and skill scores have influenced the choice of model in each of the four defined corners. For example, in the case of the Cold-wet corner, the selected model does not correspond exactly to the closest one in terms of Euclidean distance, which demonstrates the robustness of the approach used in prioritizing a balance between proximity and performance. This also reinforces the validity of the selection procedure applied.

Finally, the selected models are specified in Table 3, where it can be seen that for the Cold-dry and Cold-wet corners (i.e., a colder climate scenario from the range of projections, hereafter denoted as cold climate), the selected models are the same for both the SSP2-4.5 and SSP5-8.5 scenarios. In contrast, for the Warm-dry and Warm-wet corners (hereafter referred to as warm climate), the selected models are different for each SSP scenario. Based on these results, it can be seen that for this particular study area and for a cold climate, there is greater stability in the selection of models in different SSP scenarios than for a warm climate.

3.3 Projected changes in precipitation and average temperature in HRB

Based on the results of the selected models, we determined the monthly average values of precipitation and average temperature variables for the SSP2-4.5 and SSP5-8.5 scenarios and for the two study periods (see Figure 4). In this case, we only show the results for three sub-basins located in the north (sub-basin 1), center (sub-basin 6), and south (sub-basin 10) of the HRB.

Figure 4

Spatial and seasonal precipitation patterns vary significantly across the basin, with higher values in the northern and central regions (Amazonian zone) and lower values in the southern region (Andean zone). Based on the selected climate models, the following behaviors were identified:

• Cold-dry corner: This combination shows the most pronounced changes in the northern part of the basin, in all scenarios and periods analyzed. Specifically, a decrease in precipitation between November and January of up to −35% is projected for period 1 of the SSP2-4.5 scenario (January), followed by an increase between February and August of up to 88% for period 2 of the SSP5-8.5 scenario (June). In the central zone, the changes are less significant, with a decrease of up to −25% projected for period 2 of the SSP5-8.5 scenario (September). In the southern part, the variations are smaller, with a decrease of up to −23% for period 1 of the SSP5-8.5 scenario (August) and an increase of up to 10% for period 2 of the SSP5-8.5 scenario (June).

• Cold-wet corner: Increases in precipitation are projected throughout the basin and in almost all months for all scenarios and periods. For example, for the northern part of the basin, increases of up to 56% are expected for period 2 of the SSP5-8.5 scenario (June), and decreases of up to −27% for period 1 of the SSP2-4.5 scenario (August).

• Warm-dry corner: In the northern region, precipitation is projected to decrease in almost all months by up to −33% for period 2 of the SSP5-8.5 scenario (August), and to increase by up to 24% for period 2 of the SSP5-8.5 scenario (June). In the center of the basin, the decrease is mainly concentrated between September and November, with a reduction of up to −28% for period 2 of the SSP5-8.5 scenario (October), and increases of up to 42% are projected for period 2 of the SSP5-8.5 scenario (June). In the southern zone, increases are projected between December and June of up to 37% for period 2 of the SSP5-8.5 scenario (May), especially under the SSP5-8.5 scenario, and a decrease of up to −35% is projected for period 2 of the SSP5-8.5 scenario (August).

• Warm-wet corner: This combination projects mostly increases in precipitation throughout the basin and in almost all months, corresponding to all scenarios and periods. For example, for the northern part of the basin, increases of up to 31% are expected for period 2 of the SSP5-8.5 scenario (April), and decreases of up to −14% for period 1 of the SSP5-8.5 scenario (September).

In general, the abovementioned analysis shows that the greatest relative changes are expected to occur mainly in the wet season (in Peru from November to April) and in period 2 (2069–2099). Similarly, the models selected under the “wet” criterion show an increase in precipitation compared to the historical period, while the models selected under the ‘dry’ criterion mostly show a decrease in precipitation, with the exception of the model selected for “Cold-dry,” which shows increases in the north. On the other hand, the average temperature shows an increase for all selected climate models, with higher values in the northern part of the basin compared to the southern region. This increase is consistent throughout all months of the year and in both SSP scenarios and periods studied. However, the increases are more pronounced under the SSP5-8.5 scenario and during the second study period.

3.4 Impact of GCM projections on flood flows and flooded areas

Different impacts are observed on both the magnitude of flood flows (represented by the range of blue colors at the river section level) and the extent of flooded areas (represented by the range of red colors at the unit-catchment level) associated with the 2-year return period (PR2), which vary depending on the climate models selected (see Figure 5). The magnitude and distribution of these impacts vary depending on the climate corner used for model selection and tend to intensify during the second analysis period.

Figure 5

The models associated with drier projections (Cold-dry and Warm-dry) result in moderate increases in both flood flows and inundated areas, despite indicating a decrease in average annual precipitation (see Figure 3). On average across all river sections, flood flow increases of up to 6% and flood area increases of up to 4% are projected, both corresponding to the Warm-dry model for period 2 of the SSP5-8.5 scenario. These increases are mainly concentrated in the north of the basin for the Cold-dry model and in the northwest in the case of the Warm-dry model, being more notable under the SSP5-8.5 scenario.

In contrast, models linked to higher humidity in the future (Cold-wet and Warm-wet) show a more marked trend toward an increase in flood magnitudes for a 2-year return period, which translates into a significant expansion of flood-prone areas throughout the basin. On average across all river sections, flood flow increases of up to 71% and flood area increases of up to 50% are projected, both corresponding to the Cold-wet model for period 2 of the SSP5-8.5 scenario. Although a general increase in flood flows is projected in the different river sections, the most significant increases in the extent of flooded areas are mainly located in the headwaters and along the main channel (see thick line in Figure 5), with this effect being particularly pronounced under the “Cold-wet” scenario.

3.5 Impact of GCM projections on hydrological signatures

The HRB shows a percentage increase in FDC mid-segment slope (FMS) as a result of projected changes in precipitation and temperature (see Figure 6), which represents changes in medium-duration high flows, reflecting the basin’s responsiveness between exceedance probabilities of 0.2 and 0.7 (see Figure 6). Increases in FMS indicate a steeper middle segment of the FDC, which corresponds to a faster hydrological response, lower storage capacity, and consequently a greater propensity for short-to-medium-duration flood events. The magnitude of this increase varies depending on the climate model considered. In particular, models associated with higher humidity in the future project substantial increases in FMS, with greater intensity in the central-northern part of the basin and less in the southern sector. On average across all river sections, an increase in FMS of up to 13% is projected in the Cold-wet model for period 2 of the SSP5-8.5 scenario. This pattern suggests a faster hydrological response, which favors an increase in peak flows and, consequently, a greater magnitude of flood events. On the other hand, models linked to drier projections show moderate percentage increases in the FMS, mainly located in the central-western region of the basin. Exceptionally, the model selected under the “Cold-dry” criterion for the first period shows more notable increases in the central-southern zone, while the model associated with the “Warm-dry” criterion does not show significant changes under the SSP2-4.5 scenario in either of the two periods analyzed.

Figure 6

The impact of climate change on FDC high-segment volume (FHV), which represents long-duration high flows associated with exceedance probabilities below 0.02 (i.e., persistent high-flow conditions), was evaluated based on its percentage change (see Figure 7). Increases in FHV therefore indicate more prolonged periods of elevated discharge, reflecting a higher likelihood of long-duration flood events. Results show that models associated with wetter projections present the largest increases in FHV, particularly in the central area of the basin, while more moderate increases occur in northern and southern regions. On average across all river sections, an increase in FHV of up to 52% is projected in the Cold-wet model for period 2 of the SSP5-8.5 scenario. The increase in FHV suggests a higher frequency and intensity of flood events, which increases the risk of flooding. In addition, this increase may place additional pressure on existing hydraulic infrastructure, compromising its functionality and safety, and cause alterations in river dynamics, as well as negative impacts on riparian ecosystems and the ecosystem services they provide.

Figure 7

In general, it can be observed that the greatest impacts on both hydrological signatures (FMS and FHV) are concentrated in the tributary river sections, while in the main sections, the projected effects are comparatively minor. An exception to this pattern is the “Warm-dry” model under the SSP2-4.5 scenario, which does not show notable increases in either of the two hydrological indicators evaluated (FMS and FHV).

4 Discussion

4.1 Impacts on precipitation and riverine floods

Precipitation showed heterogeneous spatial and seasonal patterns throughout the basin, with the highest values in the north and the lowest in the south. Similarly, the climate models evaluated show contrasting behaviors in relation to monthly and annual average values, where models selected under the “wet” criterion project increases in precipitation compared to the historical period, while those selected under the ‘dry’ criterion mostly show decreases, with the exception of the “Cold-dry” model, which projects increases in the north of the basin. Lavado Casimiro et al. (2011) also identified variable projected changes in precipitation for the HRB, depending on the scenario and the period analyzed, with general trends ranging from −1.6 to +2.1%. For their part, Brêda et al. (2020), using the average of 20 climate models for the RCP 4.5 and RCP 8.5 scenarios, found precipitation changes mainly between −5 and +5%, with decreasing trends in the south of the HRB, more pronounced under the RCP 8.5 scenario. In a more recent study, Brêda et al. (2023) reported that, for events with a return period of 2 years, decreasing trends in precipitation predominate, while for events with a return period of 22 years, these trends are reversed, showing increases in certain river sections. Finally, in terms of average temperature, there is a general increase in all months compared to the historical period, which is more pronounced during the second analysis period and under the SSP5-8.5 scenario. These results are consistent with those reported by Brêda et al. (2020) and reflect the robust and spatially coherent warming signal characteristic of climate projections in the tropical Andes.

The combined effects of changes in precipitation and temperature are particularly relevant for floods, as hydrological responses are known to be nonlinear and sensitive to interactions between these drivers (Benestad and Haugen, 2007). Increases in temperature intensify potential evapotranspiration, modify soil moisture dynamics, and can reduce antecedent soil storage, thereby influencing the translation of precipitation into runoff. Conversely, increases in precipitation (especially in high-intensity events) typically dominate the flood response, particularly in humid tropical catchments where soils may already be close to saturation. These mechanisms help explain why some sub-regions exhibit increased flood magnitude even under “dry” climate models, reflecting the strong coupling between climate forcings and hydrological behavior in the basin.

Regarding the flood flows (determined based on a 2-year return period), it can be observed that, like precipitation, this varies according to the scenario and the period analyzed. The models selected under the “wet” criterion show a greater number of sections with increased flows compared to the historical period. In contrast, the “dry” models show fewer sections with increases in flows. Similarly, Brêda et al. (2023) found that, for the upper limit of the climate model ensemble, flood intensity increases in a greater number of sections, while the lower limit reflects smaller increases.

The impacts of climate change on flooding show mixed results, but are consistent with the changes observed in FMS and FHV. Under the SSP2-4.5 scenario, increases in maximum flows and the extent of flood-prone areas are mainly concentrated in the central and northern parts of the basin, especially in the “wet” models, while the “dry” models show more limited increases. In SSP5-8.5, these responses intensify across the board, with significant increases in FHV indicating greater persistence of high flows and, consequently, greater flood extent. Even in models classified as “dry,” the projected increases can be attributed to a faster hydrological response (higher FMS) or high-intensity events concentrated in the wet season. Taken together, these results show that the basin is highly sensitive to changes in the temporal structure of precipitation and scenarios of greater warming systematically lead to more extreme hydrological conditions.

In general, the selected models do not show a uniform trend of increase or decrease in the magnitude of flood flows compared to the historical period. This lack of consistency is explained by the wide variability of change signals observed during the climate model selection process (see Figure 3). This variability is reflected in a range of fluctuation between −7 and +11% for SSP2-4.5, and between −15 and +20% for SSP5-8.5.

Given this uncertainty, for each study period, trends in the increase in flood areas at the provincial level were analyzed, considering four selected climate models and the SSP2-4.5 and SSP5-8.5 scenarios (8 GCMs per period). The purpose of this analysis was to evaluate the degree of agreement in the projections of flood area increases (see Figure 8). The results highlight that provinces 15, 17, and 18 (Huallaga, Rioja, and Moyobamba, respectively) have a higher level of agreement in terms of the increase in flood area (between 7 and 8 models), which means that even under models selected with “dry” criteria, sections with an increase in the extent of flood-prone areas can be observed. This finding is consistent with the records of the National Information System for Response and Rehabilitation (SINPAD 2.0) (http://sinpad2.indeci.gob.pe/sinpad2/faces/public/portal.html), which document the occurrence of emergencies throughout the country. In addition, additional provinces in the northern and central-western regions were identified that show a high level of concordance in the increase in flooded area, reinforcing the importance of incorporating multiple scenarios into territorial planning processes and disaster risk management strategies.

Figure 8

4.2 Importance of this investigation

The HRB was identified as one of the areas with the highest risk of flooding in Peru (CENEPRED, 2023). However, there are few studies evaluating its hydrology (e.g., Lavado Casimiro et al., 2011) and, in particular, there is little research assessing the impact of climate change on flood flows. Most existing assessments are carried out at spatial scales larger than that of the HRB (e.g., Brêda et al., 2023; Sorribas et al., 2016). This gap was addressed in this study, but in this case, we focused on assessing the impact of climate change on 2-year floods, as populations settled in floodplains are particularly affected by more frequent events (Iqbal et al., 2018; Swain et al., 2020; Wasko et al., 2021) and that these impacts are intensified under climate change scenarios (Rasmussen, 2016).

To this end, we selected four climate models representative of the “corners” associated with the 10th and 90th percentiles of delta changes in precipitation and mean temperature, from an initial set of 17 CMIP6 climate models. The comparison between the historical period (1984–2014) and the future period (2069–2099) shows projected ranges of change in precipitation from −7 to +11% for the SSP2-4.5 scenario and from −15 to +20% for SSP5-8.5, which shows a high degree of uncertainty in the projections for this basin. Despite this uncertainty, our results provide robust scientific evidence and enable the exploration of a broad and realistic range of possible climate futures, consistent with values reported in recent large-scale studies (Sorribas et al., 2016; Brêda et al., 2023; Miranda et al., 2023; Hirabayashi et al., 2021; Lavado Casimiro et al., 2011).

It is important to note that identifying provinces with high concordance between the selected models (corners) with regard to increased flooding provides valuable information for policy-making and risk management. Local and regional authorities can use these results to prioritize areas for strengthening early warning systems, designing protective works, updating zoning regulations, and developing adaptation plans aligned with Peru’s National Disaster Risk Management System (SINAGERD). The methodological approach used and the spatially explicit results provide a technical basis that can be easily integrated into ongoing initiatives led by CENEPRED, ANA, and regional governments. By identifying areas where flood risk could increase even under drier climatic conditions, this study supports evidence-based decision-making and contributes to more efficient resource allocation to the most vulnerable sectors of the Huallaga basin.

4.3 Uncertainty and limitations

The scarcity of meteorological and hydrological data represents one of the main limitations for assessing the impact of climate change on riverine floods, especially in complex regions such as the Peruvian Andean–Amazon basin (Lavado Casimiro et al., 2011; Llauca et al., 2021; Saavedra et al., 2022). To address this limitation, the PISCO database was used, whose operational version of gridded precipitation (Millán-Arancibia and Lavado-Casimiro, 2023) was developed from 416 rainfall stations distributed throughout Peru. However, this database has a lower density of stations on the eastern Andean slope, precisely where high volumes of precipitation with high spatial variability are recorded. According to Bárdossy and Anwar (2022), the greatest underestimations of peaks occur in situations of low observation density, with adequate representation only possible through very high-density observation networks. In this study, we use the PISCO operational precipitation product (PISCOpd_Op), which uses reference climatologies (long-term PISCO averages) to guide interpolation in areas with sparse station coverage. While this approach allows for an acceptable representation of total volume and average spatial variability, it can introduce significant biases in daily patterns, affecting the accuracy of the hydrological simulation. In this regard, the results obtained are considered consistent with expectations, given the inherent complexity of hydrological modeling in Andean–Amazonian environments.

Although the MGB performs satisfactorily in the main stem of HRB, some stations in tributary basins show low to moderate values of NSE and KGE. However, this effect is not directly proportional in determining flood-prone areas (Munar et al., 2023), since the extent of flooding also depends on the geometric representation of river channels in the model, which influences the volume of water transferred to floodplains (Pontes et al., 2017). The absence of systematic observations of flooded areas for typical floods in the Huallaga basin limits the direct validation of the simulated flood extent. It is also important to note that metrics like NSE and KGE are sensitive to daily fluctuations and do not necessarily reflect performance in the simulation of flows associated with recurrent floods, such as those with a 2-year return period.

We acknowledge that the uncertainties associated with the structure and parameterization (runoff generation, river channel geometry, and floodplain topography) of the hydrologic–hydrodynamic model may affect the assessment of climate change impact on flows (Mendoza et al., 2015, 2016) and primarily flood extent areas (Yamazaki et al., 2017; De Paiva et al., 2013; Fleischmann et al., 2019), although decisions related to the selection of representative concentration pathways (SSPs) and global climate models (GCMs) still constitute a major source of uncertainty within the modeling chain (Chegwidden et al., 2019; Krysanova et al., 2017; Chen et al., 2011). This challenge has been addressed by the scientific community through the development of criteria for selecting climate models. One such approach is to select models that adequately reproduce the historical climate in the study area in order to increase the reliability of their future projections (Bağçaci et al., 2021; Lun et al., 2021). Another approach is based on statistical methods of coupling and representing the variability of the entire set of climate models (Ficklin et al., 2013; Dhakal et al., 2018; Goyburo et al., 2023).

The delta-change method was used to adjust the climate projections used as input to the hydrological model. Although this approach is widely used due to its simplicity and robustness, it has significant limitations (Mendez et al., 2020), as it does not reproduce changes in interannual variability or climate extremes, since it applies average monthly increments and does not modify the temporal structure of the historical series. More advanced methods, such as quantile mapping, allow the complete distribution of data to be corrected and alterations in variability to be captured; however, their application to extreme events involves considerable statistical extrapolation, and there is ongoing debate about their reliability for projecting changes in low-probability events (Mendez et al., 2020; Maraun, 2016). In this study, the choice of the delta-change method is considered appropriate given that the analysis focuses on recurrent floods associated with 2-year floods, whose magnitude depends mainly on seasonal mean changes and not on low-probability extremes. However, it is recognized that the simplicity of the method introduces uncertainties that must be considered when interpreting the results, and that future assessments targeting more severe events could benefit from the use of bias correction techniques based on quantile mapping, delta quantile mapping, and others.

In this study, we adopt a selection approach adapted from Nigussie et al. (2024), Dhaubanjar et al. (2023), and Lutz et al. (2016), which combines the individual ability of models to reproduce historical climate with their position within the range of projections of the GCM ensemble, allowing us to select four climate models representative of the extremes (10th and 90th percentiles) of projected changes in precipitation and temperature. Seventeen CMIP6 climate models were used, with projected ranges of change in precipitation between −7 and +11% for the SSP2-4.5 scenario, and between −15 and +20% for SSP5-8.5. We consider these results to be useful, as they allow us to explore a broad and realistic range of possible climate changes in the region. Furthermore, these results are consistent with the values reported by other recent studies in the Huallaga basin (Sorribas et al., 2016; Brêda et al., 2023; Miranda et al., 2023; Hirabayashi et al., 2021; Lavado Casimiro et al., 2011).

Although this study does not include a formal analysis of uncertainty, we believe that the methodology used provides an explicit representation of it through the selection of the four corners. These models correspond to extreme combinations of ∆P and ∆T within the available ensemble, acting as a conceptual analogue of the 1st and 3rd quartiles of a distribution, although without including the median. This approach allows us to capture the extremes of the range of projected climate changes and, therefore, explore a set of plausible hydrological futures. Consequently, the estimated changes in the magnitude of floods should be interpreted as possible scenarios within which the hydrological response of the Huallaga River basin could evolve. This perspective underscores the importance of considering multiple scenarios in land-use planning and disaster risk management, particularly in regions with high climatic and hydrological variability.

5 Conclusion

This study presents an adapted framework for assessing the impacts of climate change on the magnitude and extent of frequent (2-year) river floods in the Huallaga River basin (HRB), using a coupled hydrologic-hydrodynamic model driven by selected GCM projections. The GCM selections correspond to an adapted approach based on the weighting of the range of projections in the ensemble and the ability of the individual model to simulate historical climate in order to obtain four extreme GCMs related to the Cold-dry, Cold-wet, Warm-dry, and Warm-wet corners, for the SSP2-4.5 and SSP5-8.5 scenarios, analyzed in two future periods (Period 1: 2035–2065, Period 2: 2069–2099).

The models selected varied according to the selection criteria and climate change scenario. The GCMs for colder projections (cold-dry and cold-wet) were consistent in both scenarios (SSP2-4.5 and SSP5-8.5), while for relatively warmer projections (warm-dry and warm-wet), the models differed, highlighting the influence of temperature and precipitation projections on model selection.

Mean annual precipitation presented a wide variability of change projections in the application of the climate model selection approach. For the SSP2-4.5 scenario, delta values ranged from −7 to +11% (i.e., 5 models projected a decrease and 12 an increase), whereas for the SSP5-8.5 scenario, the range was extended from −15 to +20% (i.e., 6 models projected a decrease and 11 an increase). In general, the spatial and seasonal patterns vary significantly throughout the basin, with higher values in the northern and central regions (Amazonian zone) and lower values in the southern region (Andean zone). On the other hand, the mean annual temperature showed an evident increase, varying between 1.3 °C and 3.5 °C for SSP2-4.5, and between 2.5 °C and 5.8 °C for SSP5-8.5.

The magnitude of flows and flooded areas is projected to have different impacts as a result of climate change, which varies according to the climate model selected. The climate models associated with higher humidity in the future show marked tendencies of increase, mainly in the headwater areas of the basin and along the main channel. However, the climate models associated with drier projections, in spite of exhibiting a tendency of decrease in mean annual precipitation, also show increases mainly in the northwest of the basin. These results are consistent with the changes observed in the hydrological signatures FMS and FHV.

In general, we found that the provinces located in the north and center-west of the HRB (mainly in the provinces of Huallaga, Rioja and Moyobamba) will present a greater agreement in the increase in flood area (between 7 and 8 models of a total of 8 models), which implies that even under models selected with “dry” criteria we can expect river reaches with an increase in the extent of floodable areas. Although future climatic and hydrological conditions are uncertain, these results will be of importance for decision makers, territorial planning, and disaster risk management in the HRB.

References

• 1

  AbbaspourK. C. (2013). SWAT-CUP 2012: SWAT calibration and uncertainty programs—a user manual. Dübendorf, Switzerland: Eawag, 103.

  • Google Scholar

• 2

  AlmasiP.SoltaniS. (2017). Assessment of the climate change impacts on flood frequency (case study: Bazoft Basin, Iran). Stoch. Environ. Res. Risk Assess.31, 1171–1182. doi: 10.1007/s00477-016-1263-1

  • CrossRef
  • Google Scholar

• 3

  ANA, Engecorps, and Inclam (2015). Evaluación de recursos hídricos en la cuenca de Huallaga. Autoridad Nacional del Agua. Available online at: https://repositorio.ana.gob.pe/handle/20.500.12543/19 (Accessed August 15, 2025).

  • Google Scholar

• 4

  ArentD. J.TolR. S. J.FaustE.HellaJ. P.KumarS.StrzepekK. M.et al. (2015). Key economic sectors and services. In Climate Change 2014 Impacts, Adaptation and Vulnerability: Part A: Global and Sectoral Aspects. Cambridge University Press. 659–708. doi: 10.1017/CBO9781107415379.015

  • CrossRef
  • Google Scholar

• 5

  ArnellN. W.GoslingS. N. (2016). The impacts of climate change on river flood risk at the global scale. Clim. Chang.134, 387–401. doi: 10.1007/s10584-014-1084-5

  • CrossRef
  • Google Scholar

• 6

  ArsenaultR.BrissetteF.MartelJ.-L. (2018). The hazards of split-sample validation in hydrological model calibration. J. Hydrol.566, 346–362. doi: 10.1016/j.jhydrol.2018.09.027

  • CrossRef
  • Google Scholar

• 7

  BağçaciS. Ç.YucelI.DuzenliE.YilmazM. T. (2021). Intercomparison of the expected change in the temperature and the precipitation retrieved from CMIP6 and CMIP5 climate projections: a Mediterranean hot spot case, Turkey. Atmos. Res.256:105576. doi: 10.1016/j.atmosres.2021.105576

  • CrossRef
  • Google Scholar

• 8

  BárdossyA.AnwarF. (2022). Why our rainfall-runoff models keep underestimating the peak flows?Hydrol. Earth Syst. Sci. Discuss.2022, 1–30. doi: 10.5194/hess-27-1987-2023

  • CrossRef
  • Google Scholar

• 9

  BenestadR. E.HaugenJ. E. (2007). On complex extremes: flood hazards and combined high spring-time precipitation and temperature in Norway. Clim. Chang.85, 381–406. doi: 10.1007/s10584-007-9263-2

  • CrossRef
  • Google Scholar

• 10

  BID (2015). Perfil de riesgo por inundaciones en Perú. Lima: BID, 191.

  • Google Scholar

• 11

  BrêdaJ. P. L. F.de PaivaR. C. D.CollischonW.BravoJ. M.SiqueiraV. A.SteinkeE. B. (2020). Climate change impacts on south American water balance from a continental-scale hydrological model driven by CMIP5 projections. Clim. Chang.159, 503–522. doi: 10.1007/s10584-020-02667-9

  • CrossRef
  • Google Scholar

• 12

  BrêdaJ. P. L.de PaivaR. C. D.SiqueiraV. A.CollischonnW. (2023). Assessing climate change impact on flood discharge in South America and the influence of its main drivers. J. Hydrol.619:129284. doi: 10.1016/j.jhydrol.2023.129284

  • CrossRef
  • Google Scholar

• 13

  CENEPRED. 2023. Escenario de riesgo por inundaciones de la cuenca Alto Huallaga. Available online at: https://sigrid.cenepred.gob.pe/sigridv3/storage/biblioteca/17126_escenario-de-riesgo-por-inundacion-de-la-cuenca-alto-huallaga.pdf (Accessed August 15, 2025).

  • Google Scholar

• 14

  ChegwiddenO. S.NijssenB.RuppD. E.ArnoldJ. R.ClarkM. P.HammanJ. J.et al. (2019). How do modeling decisions affect the spread among hydrologic climate change projections? Exploring a large ensemble of simulations across a diversity of hydroclimates. Earths Future7, 623–637. doi: 10.1029/2018EF001047

  • CrossRef
  • Google Scholar

• 15

  ChenJ.BrissetteF. P.PoulinA.LeconteR. (2011). Overall uncertainty study of the hydrological impacts of climate change for a Canadian watershed. Water Resour. Res.47:W12509. doi: 10.1029/2011WR010602

  • CrossRef
  • Google Scholar

• 16

  ClarkM. P.WilbyR. L.GutmannE. D.VanoJ. A.GangopadhyayS.WoodA. W.et al. (2016). Characterizing uncertainty of the hydrologic impacts of climate change. Curr. Clim. Chang. Rep.2, 55–64. doi: 10.1007/s40641-016-0034-x

  • CrossRef
  • Google Scholar

• 17

  CollischonnW.AllasiaD.Da SilvaB. C.TucciC. E. (2007). The MGB-IPH model for large-scale rainfall—runoff modelling. Hydrol. Sci. J.52, 878–895. doi: 10.1623/hysj.52.5.878

  • CrossRef
  • Google Scholar

• 18

  De PaivaR. C. D.BuarqueD. C.CollischonnW.BonnetM. P.FrappartF.CalmantS.et al. (2013). Large-scale hydrologic and hydrodynamic modeling of the Amazon River basin. Water Resour. Res.49, 1226–1243. doi: 10.1002/wrcr.20067

  • CrossRef
  • Google Scholar

• 19

  DhakalK.KakaniV. G.LindeE. (2018). Climate change impact on wheat production in the southern Great Plains of the US using downscaled climate data. Atmos. Clim. Sci.08:2. doi: 10.4236/acs.2018.82011

  • CrossRef
  • Google Scholar

• 20

  DhaubanjarS.LutzA. F.SmolenaarsW. J.KhanalS.JamilM. K.BiemansH.et al. (2023). Quantification of run-of-river hydropower potential in the upper Indus basin under climate change. Front. Water5:1256249. doi: 10.3389/frwa.2023.1256249

  • CrossRef
  • Google Scholar

• 21

  DingZ.AngaritaH.Albert Montesinos CáceresC.Lavado-CasimiroW.GoldsteinJ. A.BatistaN.et al. (2024). Sustainable land and irrigation management to limit loss of hydropower in the Andes-Amazon headwaters. Commun. Earth Environ.5:648. doi: 10.1038/s43247-024-01738-4

  • CrossRef
  • Google Scholar

• 22

  DongN. D.JayakumarK. V.AgilanV. (2018). Impact of climate change on flood frequency of the Trian reservoir in Vietnam using RCMs. J. Hydrol. Eng.23:05017032. doi: 10.1061/(ASCE)HE.1943-5584.0001609

  • CrossRef
  • Google Scholar

• 23

  DuanJ. G.BaiY.DominguezF.RiveraE.MeixnerT. (2017). Framework for incorporating climate change on flood magnitude and frequency analysis in the upper Santa Cruz River. J. Hydrol.549, 194–207. doi: 10.1016/j.jhydrol.2017.03.042

  • CrossRef
  • Google Scholar

• 24

  Fernandez-PalominoC. A.HattermannF. F.KrysanovaV.LobanovaA.Vega-JácomeF.LavadoW.et al. (2022). A novel high-resolution gridded precipitation dataset for Peruvian and Ecuadorian watersheds: development and hydrological evaluation. J. Hydrometeorol.23, 309–336. doi: 10.1175/JHM-D-20-0285.1

  • CrossRef
  • Google Scholar

• 25

  FickS. E.HijmansR. J. (2017). WorldClim 2: new 1-km spatial resolution climate surfaces for global land areas. Int. J. Climatol.37, 4302–4315. doi: 10.1002/joc.5086

  • CrossRef
  • Google Scholar

• 26

  FicklinD. L.StewartI. T.MaurerE. P. (2013). Climate change impacts on streamflow and subbasin-scale hydrology in the upper Colorado River basin. PLoS One8:e71297. doi: 10.1371/journal.pone.0071297,

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 27

  FleischmannA.PaivaR.CollischonnW. (2019). Can regional to continental river hydrodynamic models be locally relevant? A cross-scale comparison. J. Hydrol. X3:100027. doi: 10.1016/j.hydroa.2019.100027

  • CrossRef
  • Google Scholar

• 28

  GoyburoA.RauP.Lavado-CasimiroW.BuytaertW.Cuadros-AdriazolaJ.HornaD. (2023). Assessment of present and future water security under anthropogenic and climate changes using WEAP model in the Vilcanota-Urubamba catchment, Cusco, Perú. Water15:1439. doi: 10.3390/w15071439

  • CrossRef
  • Google Scholar

• 29

  HallinanA. J.Jr. (1993). A review of the Weibull distribution. J. Qual. Technol.25, 85–93. doi: 10.1080/00224065.1993.11979431

  • CrossRef
  • Google Scholar

• 30

  HirabayashiY.TanoueM.SasakiO.ZhouX.YamazakiD. (2021). Global exposure to flooding from the new CMIP6 climate model projections. Sci. Rep.11:3740. doi: 10.1038/s41598-021-83279-w,

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 31

  Huamanchumo SonoD.Alvarez MamaniL.Quiroz JiménezK. (2023). Hydrological modeling using satellite precipitation in the Huallaga river basin. LACCEI1. doi: 10.18687/LACCEI2023.1.1.1406

  • CrossRef
  • Google Scholar

• 32

  HuertaA.BonnesoeurV.Cuadros-AdriazolaJ.GutierrezL.Ochoa-TocachiB. F.Román-DañobeytiaF.et al. (2022). PISCOeo_pm, a reference evapotranspiration gridded database based on FAO penman-Monteith in Peru. Sci Data9:328. doi: 10.1038/s41597-022-01373-8

  • CrossRef
  • Google Scholar

• 33

  INDECI (2023). Compendio estadístico del INDECI 2023 de la gestión reactiva. Perú: INDECI.

  • Google Scholar

• 34

  IqbalM. S.DahriZ. H.QuernerE. P.KhanA.HofstraN. (2018). Impact of climate change on flood frequency and intensity in the Kabul River basin. Geosciences8:114. doi: 10.3390/geosciences8040114

  • CrossRef
  • Google Scholar

• 35

  KrysanovaV.VetterT.EisnerS.HuangS.PechlivanidisI.StrauchM.et al. (2017). Intercomparison of regional-scale hydrological models and climate change impacts projected for 12 large river basins worldwide—a synthesis. Environ. Res. Lett.12:105002. doi: 10.1088/1748-9326/aa8359

  • CrossRef
  • Google Scholar

• 36

  Lavado CasimiroW. S.LabatD.GuyotJ. L.Ardoin-BardinS. (2011). Assessment of climate change impacts on the hydrology of the Peruvian Amazon–Andes basin. Hydrol. Process.25, 3721–3734. doi: 10.1002/hyp.8097

  • CrossRef
  • Google Scholar

• 37

  LlaucaH.Lavado-CasimiroW.MontesinosC.SantiniW.RauP. (2021). PISCO_HyM_GR2M: a model of monthly water balance in Peru (1981–2020). Water13:1048. doi: 10.3390/w13081048

  • CrossRef
  • Google Scholar

• 38

  LunY.LiuL.ChengL.LiX.LiH.XuZ. (2021). Assessment of GCMs simulation performance for precipitation and temperature from CMIP5 to CMIP6 over the Tibetan plateau. Int. J. Climatol.41, 3994–4018. doi: 10.1002/joc.7055

  • CrossRef
  • Google Scholar

• 39

  LutzA. F.ter MaatH. W.BiemansH.ShresthaA. B.WesterP.ImmerzeelW. W. (2016). Selecting representative climate models for climate change impact studies: an advanced envelope-based selection approach. Int. J. Climatol.36, 3988–4005. doi: 10.1002/joc.4608

  • CrossRef
  • Google Scholar

• 40

  MaghsoodF. F.MoradiH.Massah BavaniA. R.PanahiM.BerndtssonR.HashemiH. (2019). Climate change impact on flood frequency and source area in northern Iran under CMIP5 scenarios. Water11:273. doi: 10.3390/w11020273

  • CrossRef
  • Google Scholar

• 41

  MaraunD. (2016). Bias correcting climate change simulations-a critical review. Curr. Clim. Chang. Rep.2, 211–220. doi: 10.1007/s40641-016-0050-x

  • CrossRef
  • Google Scholar

• 42

  MejíaS. X. N.PazS. M.TabariH.WillemsP. (2023). Climate change impacts on hydrometeorological and river hydrological extremes in Quito. Ecuador. J. Hydrol.49:101522. doi: 10.1016/j.ejrh.2023.101522

  • CrossRef
  • Google Scholar

• 43

  MendezM.MaathuisB.Hein-GriggsD.Alvarado-GamboaL. F. (2020). Performance evaluation of bias correction methods for climate change monthly precipitation projections over Costa Rica. Water12:482. doi: 10.3390/w12020482

  • CrossRef
  • Google Scholar

• 44

  MendozaP. A.ClarkM. P.MizukamiN.GutmannE. D.ArnoldJ. R.BrekkeL. D.et al. (2016). How do hydrologic modeling decisions affect the portrayal of climate change impacts?Hydrol. Process.30, 1071–1095. doi: 10.1002/hyp.10684

  • CrossRef
  • Google Scholar

• 45

  MendozaP. A.ClarkM. P.MizukamiN.NewmanA. J.BarlageM.GutmannE. D.et al. (2015). Effects of hydrologic model choice and calibration on the portrayal of climate change impacts. J. Hydrometeorol.16, 762–780. doi: 10.1175/JHM-D-14-0104.1

  • CrossRef
  • Google Scholar

• 46

  MeresaH.TischbeinB.MekonnenT. (2022). Climate change impact on extreme precipitation and peak flood magnitude and frequency: observations from CMIP6 and hydrological models. Nat. Hazards111, 2649–2679. doi: 10.1007/s11069-021-05152-3

  • CrossRef
  • Google Scholar

• 47

  Millán-ArancibiaC.Lavado-CasimiroW. (2023). Rainfall thresholds estimation for shallow landslides in Peru from gridded daily data. Nat. Hazards Earth Syst. Sci.23, 1191–1206. doi: 10.5194/nhess-23-1191-2023

  • CrossRef
  • Google Scholar

• 48

  MillyP. C. D.WetheraldR. T.DunneK. A.DelworthT. L. (2002). Increasing risk of great floods in a changing climate. Nature415, 514–517. doi: 10.1038/415514a,

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 49

  MirandaP. T.PaivaR. C. D. D.GamaC. H. D. A.BrêdaJ. P. L. F. (2023). River discharge in South America: agreement and contradictions between recent alteration and projected changes. RBRH28:e18. doi: 10.1590/2318-0331.282320220085

  • CrossRef
  • Google Scholar

• 50

  MoriasiD. N.ArnoldJ. G.Van LiewM. W.BingnerR. L.HarmelR. D.VeithT. L.et al. (2007). Model evaluation guidelines for systematic quantification of accuracy in watershed simulations. Transactions of the ASABE.50, 885–900. doi: 10.13031/2013.23153

  • CrossRef
  • Google Scholar

• 51

  MunarA. M.MendezN.NarvaezG.Campo ZambranoF.Motta-MarquesD.Lyra Fialho BrêdaJ. P.et al. (2023). Modelling the climate change impacts on river discharge and inundation extent in the Magdalena River basin–Colombia. Hydrol. Sci. J.68, 1286–1300. doi: 10.1080/02626667.2023.2215932

  • CrossRef
  • Google Scholar

• 52

  Navarro-RacinesC.TarapuesJ.ThorntonP.JarvisA.Ramirez-VillegasJ. (2020). High-resolution and bias-corrected CMIP5 projections for climate change impact assessments. Sci Data7:7. doi: 10.1038/s41597-019-0343-8,

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 53

  NigussieA. B.TenfieH. W.EndalewA. (2024). Selection of multiple ensemble representative CMIP5 climate models for climate change study in developing river basin: the case of Awash River basin, Ethiopia. Arab. J. Geosci.17:160. doi: 10.1007/s12517-024-11959-8

  • CrossRef
  • Google Scholar

• 54

  O’GormanP. A. (2015). Precipitation extremes under climate change. Curr. Clim. Chang. Rep.1, 49–59. doi: 10.1007/s40641-015-0009-3,

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 55

  PokhrelP.YilmazK. K.GuptaH. V. (2012). Multiple-criteria calibration of a distributed watershed model using spatial regularization and response signatures. J. Hydrol.418, 49–60. doi: 10.1016/j.jhydrol.2008.12.004

  • CrossRef
  • Google Scholar

• 56

  PontesP. R. M.FanF. M.FleischmannA. S.de PaivaR. C. D.BuarqueD. C.SiqueiraV. A.et al. (2017). MGB-IPH model for hydrological and hydraulic simulation of large floodplain river systems coupled with open source GIS. Environ. Model. Softw.94, 1–20. doi: 10.1016/j.envsoft.2017.03.029

  • CrossRef
  • Google Scholar

• 57

  QuinteroF.MantillaR.AndersonC.ClamanD.KrajewskiW. (2018). Assessment of changes in flood frequency due to the effects of climate change: implications for engineering design. Hydrology5:19. doi: 10.3390/hydrology5010019

  • CrossRef
  • Google Scholar

• 58

  RasmussenP. F. (2016). Assessing the impact of climate change on the frequency of floods in the Red River basin. Canadian Water Resources J.41, 331–342. doi: 10.1080/07011784.2015.1025101

  • CrossRef
  • Google Scholar

• 59

  RentschlerJ.SalhabM.JafinoB. A. (2022). Flood exposure and poverty in 188 countries. Nat. Commun.13:3527. doi: 10.1038/s41467-022-30727-4,

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 60

  RiahiK.Van VuurenD. P.KrieglerE.EdmondsJ.O’neillB. C.FujimoriS.et al. (2017). The shared socioeconomic pathways and their energy, land use, and greenhouse gas emissions implications: an overview. Glob. Environ. Change42, 153–168. doi: 10.1016/j.gloenvcha.2016.05.009

  • CrossRef
  • Google Scholar

• 61

  SaavedraD.MendozaP. A.AddorN.LlaucaH.VargasX. (2022). A multiobjective approach to select hydrological models and constrain structural uncertainties for climate impact assessments. Hydrol. Process.36:e14446. doi: 10.1002/hyp.14446

  • CrossRef
  • Google Scholar

• 62

  ShenH.TolsonB. A.MaiJ. (2022). Time to update the split-sample approach in hydrological model calibration. Water Resour. Res.58:e2021WR031523. doi: 10.1029/2021WR031523

  • CrossRef
  • Google Scholar

• 63

  ShuttleworthW. J. (1993). “Evaporation” in Handbook of hydrology. ed. MaidmentD. R. (New York: McGraw-Hill).

  • Google Scholar

• 64

  SiqueiraV. A.PaivaR. C.FleischmannA. S.FanF. M.RuhoffA. L.PontesP. R.et al. (2018). Toward continental hydrologic–hydrodynamic modeling in South America. Hydrol. Earth Syst. Sci.22, 4815–4842. doi: 10.5194/hess-22-4815-2018

  • CrossRef
  • Google Scholar

• 65

  SorribasM. V.PaivaR. C.MelackJ. M.BravoJ. M.JonesC.CarvalhoL.et al. (2016). Projections of climate change effects on discharge and inundation in the Amazon basin. Clim. Chang.136, 555–570. doi: 10.1007/s10584-016-1640-2

  • CrossRef
  • Google Scholar

• 66

  SwainD. L.WingO. E.BatesP. D.DoneJ. M.JohnsonK. A.CameronD. R. (2020). Increased flood exposure due to climate change and population growth in the United States. Earths Future8:e2020EF001778. doi: 10.1029/2020EF001778

  • CrossRef
  • Google Scholar

• 67

  Turpo CayoE. Y.BorjaM. O.Espinoza-VillarR.MorenoN.CamargoR.AlmeidaC.et al. (2022). Mapping three decades of changes in the tropical andean glaciers using landsat data processed in the earth engine. Remote Sens14:1974. doi: 10.3390/rs14091974

  • CrossRef
  • Google Scholar

• 68

  van OsnabruggeB.WeertsA. H.UijlenhoetR. (2017). genRE: a method to extend gridded precipitation climatology data sets in near real-time for hydrological forecasting purposes. Water Resour. Res.53, 9284–9303. doi: 10.1002/2017WR021201

  • CrossRef
  • Google Scholar

• 69

  WaskoC.NathanR.SteinL.O'SheaD. (2021). Evidence of shorter more extreme rainfalls and increased flood variability under climate change. J. Hydrol.603:126994. doi: 10.1016/j.jhydrol.2021.126994

  • CrossRef
  • Google Scholar

• 70

  YamazakiD.IkeshimaD.NealJ. C.O'LoughlinF.SampsonC. C.KanaeS.et al. (2017). MERIT DEM: a new high-accuracy global digital elevation model and its merit to global hydrodynamic modeling. AGU fall meeting abstracts. 2017, H12C-04).

  • Google Scholar

• 71

  YamazakiD.IkeshimaD.SosaJ.BatesP. D.AllenG. H.PavelskyT. M. (2019). Merit hydro: a high-resolution global hydrography map based on latest topography dataset. Water Resour. Res.55, 5053–5073. doi: 10.1029/2019WR024873

  • CrossRef
  • Google Scholar

• 72

  ZubietaR.GetiranaA.EspinozaJ. C.LavadoW. (2015). Impacts of satellite-based precipitation datasets on rainfall–runoff modeling of the Western Amazon basin of Peru and Ecuador. J. Hydrol.528, 599–612. doi: 10.1016/j.jhydrol.2015.06.064

  • CrossRef
  • Google Scholar

• 73

  ZubietaR.GetiranaA.EspinozaJ. C.Lavado-CasimiroW.AragonL. (2017). Hydrological modeling of the Peruvian–Ecuadorian Amazon Basin using GPM-IMERG satellite-based precipitation dataset. Hydrol. Earth Syst. Sci.21, 3543–3555. doi: 10.5194/hess-21-3543-2017,

  • Pubmed Abstract
  • CrossRef
  • Google Scholar