Flood Inundation and Streamflow Changes in the Kabul River Basin under Climate Change

Abstract

The Kabul basin yields around 16% of the total annual water availability in Pakistan. Changing climate will alter the precipitation regime in terms of intensity and frequency, which will affect the water yield and cause flood hazards. Against this background, this study aims to quantify the impacts of changing climate on the water yield, its timings, and, more importantly, the associated flood hazards in the transboundary Kabul basin. For this, we used a rainfall-runoff inundation (RRI) model coupled with the snow and glacier melt routines and drove it for historical and future climates simulated by the atmosphere-only general circulation model (AGCM) at 20 km spatial resolution. The model simulations reveal that rainfall runoff contributes around 50% of the annual flows, and the rest is contributed by glaciers and snow melts. Annual precipitation is projected to increase by 14% from 535 mm, whereas temperatures will rise by 4.7 °C. In turn, the Kabul River flows will only increase by 4% to 1158 m³s⁻¹ from 1117 m³s⁻¹, mainly due to an increase in winter flows. In contrast to a minute increase in the mean river flows, the maximum flood inundation area is projected to increase by 37%, whereas its depth will rise between 5 and 20 cm.

1. Introduction

Hindukush-Karakoram-Himalaya (HKH) is a complex mountainous chain in South Asia with towering peaks, large glaciers, and raging rivers. It provides crucial water supplies and ecosystem services to the upstream and downstream communities for their sustainable social and economic development and well-being [1]. In addition to serving as a lifeline, the HKH region is also exposed to natural hazards, such as floods, glacier surges and retreats, flash floods, and glacier lake outburst floods, therefore posing serious risks to the life and infrastructure of the dependent communities [2]. Changing climate will alter such exposures to natural hazards. For instance, a change in climate will alter the precipitation regime in terms of its intensity, frequency, and magnitude, which altogether will affect not only the overall water yield or its timings but also how it is received, particularly in the form of floods [3]. Since high-altitude regions are more vulnerable to climate change owing to the well-established elevation-dependent warming, the intensity and frequency of these natural hazards are also expected to be severe under changing climate [2,4].

HKH regions are neither equally exposed to the same natural hazards nor have the same level of resilience. The subsequent regionally heterogeneous hazard level for each individual natural hazard needs to be addressed on a representative spatial scale, such as a river basin, to adequately contribute to policy measures and action plans. This is particularly true for the Kabul River basin, which is the least studied among high-Asian counterparts owing to conflicts, its remoteness, and lack of data. However, there are some studies that have discussed the climatic and hydrologic trends in the basin. The studies have used GCM projections and estimated that river flows will increase in the coming decades because of increased precipitation and enhanced melt [5,6]. Studies also reported higher warming for the basin than the global average, as warming is projected to increase by 2–3 °C in the middle and by 5–6 °C by the end of the 21st century [7,8]. Such warming is reported to be higher for the northern region than for the southern region [9]. In contrast to warming, precipitation projections feature more distinct responses on both spatial and temporal scales. The ensemble of 21 Coupled Model Intercomparison Project Phase-5 (CMIP5) experiments suggests that the monsoonal precipitation will increase gradually from the mid to late century in contrast to a decrease in the westerly (winter) precipitation [7]. Spatial distribution exhibits higher drying in the western region than in the eastern region, which will also feature a slight increase in summer precipitation under the influence of monsoons [9]. The hydrologic simulations suggest that summer flows will continue to drop in the future, whereas spring and winter flows will increase [3,10].

Apart from projected warming and subsequent mean water availability, the flood hazard assessment remains an important issue within the basin, which has been less studied so far. This is further emphasized by the fact that the basin has recently been exposed to a series of massive floods [11]. For instance, an unprecedented flood in 2010 ravaged downstream regions due to intense summer precipitation and recorded a peak of 13,592 m³/s, which is almost twice the 100-year return flow [2]. The flood inundation in the downstream regions was up to 3 m deep [12]. The 2010 flood was followed by a series of severe floods in 2011 and 2014, raising the alarm for the water managers because the lack of studies on changing hydro-climatology and subsequent flood hazards has left gaps in the water management framework for the basin [2]. Therefore, there is a need to understand the evolution of floods on spatial and temporal scales, which will facilitate determining the potential hazard zones in the basin.

For the first time, we aim to assess the flood inundation for the Kabul River basin in response to hydroclimatic extremes simulated under changing climate for the end-of-century climate. For this, we set up a two-dimensional rainfall-runoff inundation (RRI) model coupled with the snow and glacier melt routines to simulate the depth and extent of the flood inundation area. The calibrated and validated RRI model was ingested with climatic projections from a high-resolution (0.1875°, ~20 km) AGCM of the Meteorological Research Institute, Ibaraki, Japan (MRI-AGCM). The results of the study will demonstrate the seasonal contribution of rain, snow, and glacier melt in the river flows for the present and future. Furthermore, the changes in the river flows in different seasons are projected along with changes in the flood inundation.

2. Methodology

2.1. Study Area

Kabul is a transboundary river basin located within the western Karakoram and eastern Hindukush Mountain ranges, sharing its total area of 88,000 km² with Afghanistan and Pakistan (Figure 1). Featuring an elevation range of 284–6450 masl, the basin encompasses steep slopes in the north, flood plains in the south, and diverse land uses and covers. The basin serves nine million people for irrigation, hydropower generation, and domestic water supplies across the two countries and contributes around 16% of the total annual water availability in Pakistan [9,13,14].

Owing to a war conflict zone, data availability and accessibility from the basin have remained major issues until today. A few stations have been analyzed from the eastern side for a short period in the past, whereas few high-altitude stations are available within Pakistan, although these are unsuitable for presenting a comprehensive hydroclimatic analysis. From the available record, it is noted that the complex terrain and the confluence of distinct climatic regimes are responsible for substantial variability in the spatial and temporal distribution of precipitation across the basin [9,13,15,16]. Ref. [15] reported positive altitudinal and east-to-west gradients for precipitation. For instance, summer precipitation over the eastern regions is between 50 and 100 mm, which gradually increases to 600 mm in the west [9]. Ref. [17] reported annual precipitation of 330 mm from the western basin for the period 1957–1977. Overall, the mean annual precipitation for the basin is around 500 to 600 mm [15,18]. According to the available station records, the temperature varies from subfreezing −10 °C to 25 °C from winter to summer in the northern mountainous subbasins. The average temperature in the downstream region is relatively higher, with 10 °C in winter and up to 35 °C in summer.

The Kabul River’s total length is 700 km, of which 560 km is located in Afghanistan [13]. River flows are being measured at two locations (Figure 1) by the Water and Power Development Authority (WAPDA), namely, at Chitral (upstream), after which the river enters Afghanistan, and at Nowshera (downstream), which is before the confluence to the Indus River. At Nowshera, the hydrograph starts rising in March and reaches the maximum in July. Instead, it rises in April at Chitral because of generally colder climatic conditions at higher altitudes. Chitral annually drains 288 m³s⁻¹ discharge downstream, which becomes around 608 m³s⁻¹ before entering Pakistan and around 800 m³s⁻¹ at Nowshera [18]. The main source of these flows is the meltwater from snow and glaciers that start contributing to the onset of the melting season in spring [13,17]. The glaciers cover an area of 1650 km², which mostly lies in the northeastern subbasin of Chitral [19].

2.2. Datasets Used

We obtained the daily temperature and precipitation from the APHRODITE datasets available at a finer spatial resolution of 0.25 ͦ [20,21]. To calibrate and validate the RRI model, we obtained the daily observed river discharge at Nowshera station from the WAPDA for the period of 1980–2013. The glacier cover of the basin is extracted from [19], which was mapped from the Landsat imagery around 2000.

For climate forcing, we obtained the daily MRI-AGCM temperature and precipitation for the historical period of 1979–2003 and for the end-of-century climate of 2075–2099, which has a planar resolution of 0.1875 ͦ (~20 km). The MRI-AGCM model reliably simulates monthly mean precipitation, cyclones in the tropical Western Pacific, and the seasonal march of East Asian summer monsoons, among other key features [22]. The model outputs have also been analyzed for the South Asian summer monsoon over Bangladesh [23]. The end-of-century climate was available from four experiments (c0, c1, c2, and c3) simulated under an extreme climate change scenario of the representative concentration pathway 8.5 [22].

2.3. Climatic Changes

First, we analyzed the APHRODITE observational datasets to assess the temporal and spatial distribution of temperature and precipitation across the basin. For this, we calculated long-term averages of annual total precipitation and mean daily temperatures. Similarly, long-term mean monthly temperature and total precipitation were calculated to prepare climatological annual cycles. It is to be noted that the APHRODITE precipitation contains biases along the elevation ranges. Ref. [24] have corrected these biases based on the altitudinal gradients of precipitation approximated from the glacier mass balance estimates. We employed the same altitudinal gradients to correct the APHRODITE precipitation dataset before its analysis using the following:

$$P_{cor}=P_{Aphro} \left(1+\left(H-HREF\right)\right) PG\ast 0.01 for HREF<H \le HMAX$$

(1)

$$P_{cor}=P_{Aphro}\left(1+\left((HMAX-HREF\right)\right)+(HMAX-H)) PG\ast 0.01) for H \ge HMAX$$

(2)

where HREF is the reference elevation, H is the elevation of the cell, PG is the precipitation gradient, HMAX is the maximum elevation, P_cor is the corrected precipitation, and P_APHRO is the Aphrodite precipitation.

Climate modeling experiments are usually biased, particularly over regions of complex mountainous terrain, and need to be adjusted before their use in impact assessment studies. [25] reported that climate modeling experiments generally feature substantial cold and wet biases over the HKH region. First, we estimated the biases in the MRI-AGCM historical experiments against the APHRODITE observational datasets and then adjusted them for future experiments using the delta-change method (Equations (3) and (4)), which corrects mean biases between the two datasets [26].

$$T_{adj, fut}=T_{sim,fut}+\left({\overline{T}}_{obs,hist}-{\overline{T}}_{sim,hist}\right)$$

(3)

$$P_{adj, fut}=P_{sim, fut} \frac{{\overline{P}}_{obs,hist}}{ {\overline{P}}_{sim,hist}}$$

(4)

where $\overline{T}$ and $\overline{ P}$ are temperatures and precipitation climatology, sim stands for simulated, hist and fut means historical and future climates, and adj means adjusted datasets.

We assessed changes in annual precipitation, 5-day maximum precipitation, 90th percentile, and 15 and 25-year return periods based on biased adjusted climate projections for the historical climate and for the end-of-century climate for four experiments. For seasons, changes are calculated for January–March (extended winter), April–June (snowmelt season), July–September (glacier melt and monsoon season), and October–December (post-monsoon, dry season).

2.4. Rainfall-Runoff Inundation Model Setup

We employ the RRI model for the flood inundation modeling for the Kabul River basin as the model has already been satisfactorily applied to simulate the rainfall-inundation relation for the whole Indus basin for the unprecedented flood of 2010. The model is a two-dimensional model capable of simulating rainfall-runoff and flood inundation simultaneously [12].

The RRI model mainly requires precipitation, evaporation, and topographic information to simulate streamflow and inundation [27]. At a grid cell on which a river channel is located, the model assumes that both slope and river are positioned within the same grid cell. A channel is discretized as a single vector along the center line of the overlying slope grid cell. The channel represents an extra flow path between grid cells lying over the actual river course. Lateral flows are simulated on slope cells on a (two-dimensional) 2D basis. Slope grid cells on the river channel have two water depths: one for the channel and the other for the slope (or floodplain) itself. The inflow–outflow interaction between the slope and river is calculated based on different overflowing formulae depending on water level and levee height conditions [12]. On the other hand, the vertical infiltration flow is estimated by the Green–Ampt model [28].

Since snow and glacier melt significantly contribute to the Kabul River flows, we coupled the RRI model with snow and glacier melt routines. The snow melt, rainfall, and glacier melt are calculated for each grid cell of the input grid. The precipitation is partitioned into snow and rain by using a temperature threshold of 0 °C, which implies that all the precipitation above this threshold falls as rain and below this threshold as snow. The snowfall accumulates in winter and starts to melt in spring, followed by the glacier melt. The description of the snow and glacier melt routines employed is given in the Supplementary Material.

In this study, the future glacier area was extracted from the estimates given by [29]. They simulated the glacier area in the future using an ensemble of GCMs for South Asia, where the Kabul River basin is located. According to their study, the region will lose approximately 75% of its glacier mass by the last quarter of this century. Therefore, we assumed that the current glacier area will reduce to 25% of the present values. In terms of the spatial variation of the glacier cover under future climate conditions, we also theorize that the glacier area will mainly be lost from the lowest elevation areas due to the higher temperatures; this pattern of glacier evolution is confirmed by [30].

We calibrated the RRI model for the first 17 years (1980–1996) and validated it for the last 17 years (1997–2013) of the available record against the observed discharge at the Nowshera gauging station. The model performance has been assessed against the statistical parameters of R², root mean square error (RMSE), percentage bias (PBIAS), and volumetric error or difference. The validated RRI model was then ingested with the bias-adjusted end-of-century future climates from four experiments. In addition to assessing changes in the mean and extreme availability and its timings, we assessed spatial changes in the maximum flood inundation extent and its depths. Flow duration curves are additionally presented for two potential dam sites at Ayun and Kedam.

3. Results and Discussions

3.1. Calibration and Validation

The RRI model is calibrated for the 17 years from 1980 to 1996 and validated for 17 years from 1997 to 2013. The comparison of monthly average river flows from 1980–2013 shows some inaccuracies in the simulations. The spring flows are overestimated, particularly in April and May. However, average annual river flows are fairly simulated, as shown in Figure 2. The model assumes that precipitation below 0 °C is snow and the rest is rain, which has caused some inconsistencies in the spring river flow simulations. Moreover, the coarse resolution of the APHRODITE is not an ideal choice for the basin, which has steep slopes and high altitudes. The performance is judged against some statistical parameters like R², Root mean square, PBIAS, and volumetric difference. The indexes reveal the overestimation in the simulated discharges, which can be improved by undertaking measures like finer resolution climate data and improved physical modeling [31] (

Table 1. The statistical indices of the model performance for calibration and validation periods.

Period	R2	Root Mean Square	PBIAS	Volume Difference
Calibration 1980–1996	0.25	1.05	0.1%	18%
Validation 1997–2013	0.12	1.41	18%	6.7%

while R² = 1 − ∑ (Q_obs − Q_sim)²/∑ (Q_obs − Q_avg)²; Root mean Square =$\sqrt{\frac{1}{n}\sum _i{Q}_{sim}^2}$; PBIAS = 100 ${\sum }_{i=1}^N (Q_{sim}-Q_{obs})$/${\sum }_{i=1}^N{Q}_{obs}$.

).

In the upper Indus River basin, where the Kabul River basin is located, the snow and glaciers contribute 45% to the annual river flows [32]. This study shows that rainfall is the dominant source of runoff, followed by snow and glacier melt. In spring, the rainfall-runoff precedes snow melt runoff, which indicates that during spring, a large proportion of precipitation falls as rain. The glacier melt runoff at Nowshera is mainly contributed by the Chitral sub-basin, whose share is 78%. In August and September, the river flows mainly consist of glacier runoff. Rainfall-runoff often contributes in these months and can cause flooding.

3.2. Present Climate

The average annual precipitation in the basin is 510 mm; however, significant spatial and seasonal variations exist according to the APHRODITE dataset. The Aphrodite dataset is corrected using the technique developed for the region [24]. After the correction, the precipitation in the basin is found to be 535 mm annually, which is higher than the original APHRODITE estimate.

In the eastern region, the influence of monsoons, which are the cause of flooding, is quite evident. In the north, the winter precipitation is higher than that in summer [17]. The eastern region receives approximately 1000 mm annually, while in the mountains, the precipitation is ~700 mm. Basin-wise, the precipitation regime is dominated by winter and spring precipitation, which falls as both rain and snow (Figure 3).

The temperature in the basin decreases with the elevation (Figure 3). The low-lying flood plains are warmer, with an average annual temperature of 17~18 °C, while in the mountainous north, the temperature remains close to 0–6 °C (Figure 3).

3.3. Climate Change

The future climate will see significant changes, which will likely impact the hydrologic cycle in space and time. The average annual temperature will increase by 4.7 °C in the future, while summer and winter will observe a higher increase than spring (Figure S2). From October to March, the mountainous northern regions will experience temperature increases between 5–5.6 °C (Figure 4). An important aspect of warming is that the northern mountainous region will grow warmer at a higher rate than low-elevation zones. However, in June–August, the downstream regions will become warmer than the mountainous zones (Figure 4).

The precipitation will also increase by 14% from 535 to 609 mm per year. Bias-corrected projections show precipitation increases in every season in the future, however, with different spatial patterns. In winter, the northern mountain will receive more precipitation than the other regions, which indicates more snowfall and higher spring melt. The central zone and flood plains downstream will have a modest increase in precipitation (Figure 4). The future precipitation regime will be wetter and more intense because of the annual increase in precipitation. The maximum 5-day precipitation will increase, as will the 90th quartile of precipitation, causing an increase in the flood intensity. Moreover, the intensity of precipitation events for the given return period will also increase, indicating (

Table 2. Different precipitation indices in present and future climate in the basin. The values in the brackets show % change. The C0, C1, C2, and C3 are the different experiments.

Parameter	Present	MRI-AGCM c0	MRI-AGCM c1	MRI-AGCM c2	MRI-AGCM c3
Precipitation (mm/y)	535	633 (18)	576 (7.5)	584 (9)	646 (21)
P5-day max (mm)	48	46 (−5)	63 (30)	81 (68)	64 (33)
PQ90 (mm)	3.77	4.32 (14)	4.03 (7)	4.00 (6)	4.54 (20)
15-yr return period (mm)	660	804 (21)	715 (8)	700 (6)	817 (23)
25-yr return period (mm)	695	850 (22)	755 (8)	732 (5)	866 (24)

).

3.4. Chamge in the Hydrologic Cycle

Average monthly runoff components and total runoff are shown in Figure 5. In the future, the average annual river flow at Nowshera will experience a slight change in quantity, increasing from 1117 m³.s⁻¹ to 1158 m³/s⁻¹ (

Table 3. The summary of changes in precipitation, temperature, and runoff components (m³/s) at Nowshera and Chitral.

Station	Period	P (mm)	T (°C)	Qrain	Qsnow	Qglaicer	Qall
Nowshera	Present	535	6.59	450	232	255	1117
	Future (2075–2099)	635	11.3	915	85	92	1158
Chitral	Present	558	4.9	90	52	201	385
	Future (2075–2099)	670	10	143	32	86	308

). However, notable changes in the seasonal river flows are projected by the simulations. Winter and spring will register higher seasonal flows, while summer will have fewer supplies (Figure 5). The river flows from November to April will increase from 550 m³/s to 1347 m³/s. On the other hand, the summer flows (May to October) will drop from 1642 m³/s to 970 m³/s. On the other hand, [6,33] have projected that the river flows at the Nowshera will continue to increase in the coming decades because of an increase in precipitation and enhanced melt.

The rainfall is going to increase in the future because of the change in climate and the phase of precipitation. On an annual basis, the rainfall hydrograph will increase two-fold at the Nowshera. However, most of this change will take place in the spring. In August, however, the rainfall-runoff will increase by 32%, which will be vital because of the drop in glacier runoff. The snow hydrograph, because of higher temperatures and phase change in precipitation, will diminish at Nowshera from 232 m³.s⁻¹ to 85 m³.s⁻¹. An early rise in the snow hydrograph will be observed because higher temperatures in March will initiate snow melt earlier. In the glacier hydrograph, a major reduction will take place because of their retreat. It is projected that almost two-thirds of the glacier hydrograph will disappear in the future at Nowshera. Ref. [10] performed glacio-hydrological simulations of the Kabul River basin for the last quarter of the 21st century and reported that rainfall runoff and glacier melt would increase, but snow melt would decrease.

Chitral’s annual river flows will respond differently from Nowshera. The annual river flows will decrease to 308 m³.s⁻¹ from 385 m³.s⁻¹ at this station. This drop can be attributed to the glacier retreat because this subbasin hosts the largest mass of glaciers in the entire Kabul River basin. Similarly, the Chitral’s glacier flows will drop from 201 m³.s⁻¹ to 86 m³.s⁻¹. The rainfall and snow hydrographs will follow the trends of their corresponding runoff components at Nowshera (Table 3).

At Nowshera, both increasing and decreasing trends in the annual river flows have been reported by [10,33].

3.5. Change in Flood Hazards

The increase in precipitation in the future will translate into higher flood intensities (Figure S3). To evaluate the change in the inundation extent, the maximum, 90th, 50th^, and 10th percentiles of flood extents for the present and future are compared in Figure 6. The bottom panel of the figure shows the differences between the future and present according to the corresponding percentiles. All four experiments have projected mixed trends of increase and decrease in the future. For example, the maximum percentile has shown an increase in the depth (up to 0.1 m) across the basin except MRI-AGCM c0, which has shown a decrease in the depth by 0.05 m. The 50th and 90th percentiles have projected a decrease in the inundation depth in Chitral across the four scenarios while projecting an increase in the western region and floodplains upstream of Nowshera up to 0.25 m. Unlike the 50th and 90th percentiles, the 10th percentile inundation shows an increase in the depth of floods in the Chitral. This increase in the Chitral is more variable in magnitude, i.e., 0.05 m to 0.25 m. It can be implied from the patterns that the mountainous northern zones, i.e., Chitral, will experience high-frequency but low-intensity floods, whereas the southern zones will face low-frequency but high-intensity flood events in the future. The increase in flood magnitude up to 18–40% and occurrences of river flows above 5000 m³.s⁻¹ will be more frequent in the Kabul River basin [33].

This study also highlights a significant shift in the occurrences of the flood peaks, i.e., the probability of occurrence of inundation will be higher in May than in July–August because of higher runoff generation in spring. A similar trend is reported by [5], where the return period of the 50-year period will reduce to 3 to 24 years in the mid and late century. Furthermore, [6] also reported an increase in the frequency of flood events in the region.

Due to the abundant river flows, steep slopes, and large population, the Kabul River basin has always been an ideal spot for hydropower plants. Several reservoirs and run-of-river (RoR) power stations are already located in the basin [15,34]. In this study, two stations are selected for the evaluation of hydropower potential. The sites (Ayun and Kedam) are up and downstream of the Chitral sub-basin (Figure 2). Ayun has an estimated discharge of 20 m³.s⁻¹ and a head of 187 m, while Kedam has an average discharge of 3 m³.s⁻¹ and a head of 580 m [34]. An assumption is made that the sites are RoR because of the low discharges and heads. For the quantification of the hydropower, the following equation is used:

$$\mathrm{P}=\mathsf{\gamma }·\mathrm{Q}·\mathrm{H}$$

(5)

where P is the power (watts), γ is the specific weight of the water (9.81 kN/m³), Q is the discharge (m³.s⁻¹), and H is the head (m) [35]. Figure 7 shows the daily duration curves at both sites; the low flows at the sites will remain above the present values, which indicates that hydropower potential will be higher in seasons of low flows, i.e., winter. At Ayun, the hydropower will increase to 88 × 10⁶ watts from 86 × 10⁶ watts, while in February–May, it will change by +25% and in June–September by −53%. At Kedam, the power will increase to 16 × 10⁶ from 12 × 10⁶, while changes will be +46% and −60% in the spring and summer, respectively. [15] projected that hydropower in the basin will increase by 1.4% to 1.7% in the mid- to late century. However, because of uncertain precipitation trends, significant variations in the projections have also been reported.

4. Conclusions

The Kabul River basin is a lifeline for millions of people in Pakistan and Afghanistan. The runoff generated by the rain, snow, and glacier melt has provided vital services to the ecology and for the survivability of humans. Due to the remoteness of the region and conflicts, enough information on water availability and disaster risks is not available. Some efforts have been made in the recent past to understand hydro-climatology and flood risks. This study explains the vital hydrologic characteristics along with the flood hazards and hydropower potential in more detail.

The study has incorporated global climate datasets with temperature-index snow and glacier melt models to quantify the runoff generation in the basin. Furthermore, hydrologic and flood simulations have been performed using the RRI model. The simulation results have been reliable; however, some overestimation in the spring river flows have also been noticed. This shortcoming of the simulations is likely due to the coarse resolution of the climate data and unavailability of key datasets like river channel depths, width, and the assumption that precipitation can be divided between rain and snow by using a 0 °C threshold, etc. The inclusion of glacier mass balance models and fine-resolution digital elevation models will yield more accurate estimates of runoff components. Furthermore, the calibration of discharges at additional locations will enable robust models.

Despite the shortcomings in the simulations, the annual river balance is well simulated, and contributions of rain, snow, and glaciers are well in range, as reported by other studies. The results state that rain is the main source of runoff in the river basin, followed by snow and glacier melt. Most of the glacier is contributed by the Chitral sub-basin owing to its large glacier cover. The downstream regions are vulnerable to flood inundation in summer.

The climate projections in the last quarter of the century project a warmer climate all over the basin, specifically in the mountainous north. The precipitation will increase in all seasons, but there are some variabilities as well. The MRI-AGCM is preferred owing to its fine spatial resolution, which has enabled a detailed study of climate patterns and extremes.

The future river flow regime will be significantly altered because of climate change and glacier retreat. The glacier hydrograph is expected to lose up to two-thirds of its present quantity. However, because of the increase in the precipitation and particularly due to the increase in the rainfall contribution, the annual river flows will increase by a slim margin of 3.5%. However, this is not the case with the Chitral sub-basin, where river flows will decrease by 20% because of the retreat of glaciers.

The flood inundation simulation and its projection for the future is an important finding of the study. The floods usually occur in July–August in the basin, and inundation often takes place just upstream of the Nowshera in the flood plans. Present trends in intensity and timing will change in the future because of the changing patterns in temperature and precipitation. The present flood peak is expected to shift in May because of the combined effect of precipitation and excess melt. Additionally, the inundation hazard will expand to the western region as well. The hydropower potential is projected to increase with a slender margin because of the higher precipitation. Most of the power will be generated in the spring because of the higher runoff generation. In addition to that, low flows will enable the production of power in winter as well.

This study presents scenarios of water availability, flood intensities, and power potential in the Kabul River basin. The overall results show a mix of opportunities and challenges for the sustainability of the environment and the survivability of the human ecosystem. Hydropower will play a role in replenishing the higher electricity demand and provide an alternative to fossil-fuel-dependent power. The summer will be a stressful season because of the depleted river flows. Therefore, serious efforts are necessary to conserve the surface and groundwater resources to adapt to this situation. The identification of the potential hazard zones of inundation could facilitate the process of risk reduction. The construction of dykes, storage reservoirs, etc., can be planned for the inundation-prone zones. Furthermore, the installation of high-altitude climate stations will enable better monitoring of the climate and hydrologic cycle. The future evolution of flood hazards and water availability has necessitated a water management treaty between Pakistan and Afghanistan; otherwise, sustainable development in the region will be negatively impacted.