Assessing impact of climate change on streamflow: A case study in Ba River watershed, Gia Lai province, Vietnam

Surface water resources have been impacted by climate change. Due to the Coupled Model Intercomparison Project Phase 6 (CMIP6), the Intergovernmental Panel on Climate Change (IPCC) published its sixth assessment report in recent years. Therefore, this study aimed to evaluate the impact of the latest climate change scenarios on surface resources. The Ba River watershed’s streamflow was simulated under current and climate change conditions utilizing Soil and Water Assessment Tool (SWAT) model. Two shared Socio-economic Pathways 1-2.6 and 5-8.5 from ACCESS-CM2, CanESM5, and HadGEM3-GC31-LL General circulation models were used. The simulated annual discharge showed that surface water resources will increase during the 2030s, 2050s, 2080s, and entire periods for all the climate change scenarios in the watershed. Additionally, surface water resource maps of the watershed were created for each period of the climate change scenarios. The results from this study have indicated the effective application of SWAT model in assessing the impact of climate change on water resources.


Introduction
Surface water is defined as water bodies above ground, such as lakes, streams, rivers, wetlands, and so on.Because of the accessibility and ease of exploitation of surface water, it is used by humans for various purposes compared to groundwater.Vietnam is a country with a dense river system, in which the total annual surface water is between 830 and 840 billion m 3 [1].Surface waters play a key role in maintaining ecological systems and establishing habitats for living creatures all over the world.Also, they serve different human uses in agriculture, industry, domestic, aquaculture, irrigation, and others.Nowadays, along with industrialization and modernization, surface water resources are seriously decreasing in both quality and quantity [2].Vietnam is known as a country that spends the most on irrigation expenses in Southeast Asia [3].Polluting surface water has negative impacts on not only human health but also on animal species.Although water resources management is extremely important, the efficiency of the implementation mechanism of the Water Resources Law is inadequate based on many obstacles [4].Especially, the local-level management of the hydraulic structures in the Central Highlands of Vietnam has not yet adapted to the current market mechanism [5].
IOP Publishing doi:10.1088/1755-1315/1345/1/012025 2 Declining water resources in Vietnam are based on many reasons, such as climate change, excessive use and exploitation of water resources, and population growth.One of the primary reasons for environmental degradation is, notably, the effects of climate change [3].Vietnam is currently dealing with issues connected to water security, primarily caused by climate change and rising sea levels, and socioeconomic growth [4].Based on the report of the global long-term climate risk index 2019 released by Germanwatch, a German environmental organization, at the 24th Conference of the Parties to the United Nations Framework Convention on Climate Change (COP 24), Vietnam is in the group of 10 countries most heavily affected by extreme weather events [6].With its vast coastline, river deltas, and highlands that are prone to severe erosion and poor water retention, Vietnam is especially susceptible to the consequences of climate change [7].However, due to topographic factors, the impact of climate change on each region has impacts at different levels.The rising ambient temperature and the frequency of intense hydrological events are the primary climate change determinants affecting water quality [8].The impacts of climate change may tend to increase diffuse pollution through runoff from agriculture or urban areas.Hydrological modeling approaches, particularly the Soil and Water Assessment Tool (SWAT), have been used to tackle these shortcomings and to consider the impact of climate change on water discharge.Some research mentioned the wide application of the SWAT model for assessing the changes in water discharge under climate change conditions on a large watershed scale, such as Saade et al. [9], Li and Fang [10], and Ougahi et al. [11].As mentioned in the climate change scenarios, the Ministry of Natural Resources and Environment of Vietnam published and updated these scenarios in the past (2009,2012,2016) and most recently in 2020.Some research conducted in Southeast Asia and Vietnam has primarily concentrated on downscaling data from global climate models that are part of the Coupled Model Intercomparison Project Phase 3 (CMIP3) and Phase 5 (CMIP5) [12].Nevertheless, there remains a considerable lot of uncertainty concerning to the precipitation forecast.The most recent version of the scenarios was utilized for the Coupled Model Intercomparison Project Phase 6 (CMIP6) (2016-2021) and included in the IPCC Sixth Assessment Report (AR6) based on a set of shared Socioeconomic Pathways (SSPs).However, the resolution of global models is still coarse and unsuitable for application at regional and local scales.Therefore, it is required to downscale CMIP6 products with the most recent greenhouse gas scenarios [13].In recent years, research has established the climate change scenario SSPs for Vietnam, particularly Nguyen Duy et al. [14] and Thanh [12].
Therefore, this study's primary goal was to evaluate how surface water resources of a watershed in Vietnam may be impacted by latest climate change model.

Brief introduction of the study area
Ba River watershed is located in Gia Lai, Dak Lak, Binh Dinh and Phu Yen provinces, Vietnam [15].This watershed has just enough water resources with an annual river discharge module of 25 l/s.km 2 [16].The watershed has very complex topography with average altitude ranging from 0 to 1,756 m above sea level and gradually decreases in the north-south direction [17].The mainstream of the Ba river is 396 km long, originating from the Ngoc Ro mountain of the Annamite mountains [18].The watershed has three main tributaries including Ayun, Krong Nang and Hinh rivers which enlarge the riverbed as it moves downstream [18].Rivers and streams in the watershed are narrow and deep with steep slopes, so the Ba River watershed has great potential for hydroelectric [18].The basin has a total annual water discharge of over 10 billion m 3 , which is sufficient to meet demand of socioeconomic development across the whole basin [16].The Ba River watershed's rainy season rainfall tends to increase in the context of climate change, boosting flood flow and escalating the likelihood of severe floods in the downstream area [16].Conversely, rainfall during the dry season tends to decline, which lowers dry season flow and allows saline to seep deeper into the river [16].The effects of climate change combined with the requirement of socioeconomic developments may raise the demand for water usage [16].As a result, there may be a rising risk of a water shortage in some areas of the Ba River watershed.
Due to data limitation, only a watershed part has an area of about 8,334.19km 2 in the Gia Lai province (see Figure 1), was chosen as the study area of this research.According to the 2020 land use map of Gia Lai Province, land use types in the watershed were classified into 5 main groups including (i) forest, (ii) agriculture, (iii) landscape, (iv) water surface and (v) unused land [17].Of which, forestry land occupies around 51.05% of the watershed area [17].The main soil types in the watershed include (i) Arenosolssoil, (ii) Fluvisols, (iii) Gleysols, (iv) Ferrasols, (v) Acrisols, (vi) Luvisols, (vii) Lixisols [32].Among them, Ferrasols has the largest area with 38.59% of the watershed area.The Ba River watershed located in the tropical monsoon highland climate zone with an average annual rainfall in the range of 1,200-1,750 mm [17].The climate is divided into two seasons: rainy (May to November) and dry (December to April).The average annual temperature ranges from 22°C to 25°C [17].

Assessing impact of climate change on surface water resources
For assessing the impact of climate change on streamflow in the study area, Soil and Water Assessment Tool (SWAT) was applied.SWAT is a semi-distributed hydrological model which is developed for predicting and evaluating the impacts of land use management practices on water and sedimentation for watershed scale [19,20].SWAT model was developed based on some physical natural phenomena and uses correlation and regression equations to describe the relationship between input parameters (e.g.land use/cover, soil, topography and climate) and output variables (e.g.streamflow, sedimentation, water quality).In this study, SWAT model was used to predict water discharge of the Ba River watershed under historical climate (1980-2020) and climate change (2021-2099) data.For climate change data, two climate change scenarios including Shared Socioeconomic Pathways (SSPs) 1-2.6 and 5-8.5 of three General circulation models (GCMs) including ACCESS-CM2, CanESM5 and HadGEM3-GC31-LL were used.These climate change data were selected corresponding to sustainable development (1-2.6) and fossil fuel-driven development (5-8.5)scenarios, respectively.More details about the climate change can be found elsewhere [13].These data were downloaded and processed into SWAT input format requirements for weather data (precipitation and temperature) [19].
This study utilized a developed SWAT model of the Ba River watershed from a previous study [21].Calibration and validation of the model was performed with three indices including (i) Nash-Sutcliffe Efficiency (NSE), (ii) Ratio of standard deviation of the observation to the root mean square error (RSR) and (iii) Percent bias (Percent Bias -PBIAS) [22,23].The calibration and validation periods were 1990-2000 and 2001-2011, respectively.The set of input parameters that have been calibrated was applied to the other sub-basins of the Ba River watershed.The calculated value of the evaluation indices showed that the SWAT model of the Ba River watershed has good performance in simulating the streamflow under historical climate condition [21].
The annual streamflow was used to evaluate the change of streamflow.The change of annual streamflow was simply estimated following equation 1 [24].Meanwhile, surface water resources map can be done based on a river discharge module, as shown in equation 2 [25].The river discharge module was chosen to assess surface water resource under climate change condition because it is highly representative and can effectively utilize the results from SWAT model.Table 1 shows the categories of surface water resources based on the river discharge module value [25].

∆𝑄 = (
Where ΔQ is the average change of all the difference annual streamflow of the period, Qy is annual streamflow of a year of the period, and Qy+1 is annual streamflow of the next year of the period. =   × 1000 (2) Where M is the river discharge module (l/s.km 2 ); Q is water discharge rate (m 3 /s); F is the watershed area (km 2 ).

Assessing change of streamflow
As for Figure 2   The results showed that, for all three GCMs models, the future streamflow is predicted to exhibit slightly increasing trends for the period of 2021-2099.The streamflow is expected to peak in 2042 at 677.2 m 3 /s, according to the CanESM5 model.According to the ACCESS-CM2 and HadGEM3-GC31-LL scenarios, they are expected to reach their greatest point in 2082 at 593.0 m 3 /s and in 2056 at 538.3 m 3 /s, respectively.Obviously, projected climate change between 2021 and 2099 will have an impact on the streamflow in the Ba River watershed.
Based on the SSP1-2.6 scenario, the ACCESS-CM2 model predicts that the annual streamflow will drop from 25.2% in the 2030s to 6.5% in the 2050s and then rebound to 20.2% in the 2080s (Table 2).The CanESM5 model foreseens that in the 2030s, 2050s, and 2080s, there will be decreases in trends of 13.7%, 8.6%, and 6.0%, respectively.The HadGEM3-GC31-LL model's streamflow change had a similar trend to that of the ACCESS-CM2 model, rising from 14.3% to 8.83% before leveling off at 17.5%.In the ACCESS-CM2, CanESM5, and HadGEM3-GC31-LL models, the changes in annual streamflow under the SSP1-2.6 scenario are predicted to be 17.3%, 9.4%, and 13.5% for the period 2021-2099.
Under the SSP5-8.5 scenario, the annual streamflow in the ACCESS-CM2 model was envisaged to increase between 7.5% in the 2030s to 14.5% in the 2050s and then decrease to 8.7% in the 2080s.Considered to the CanESM5 model, there is predicted to decrease trends by 15.0%, 13.9%, and 8.3% in 2030s, 2050s, and 2080s, respectively.The change of streamflow in the HadGEM3-GC31-LL model had the same tendency with the CanESM5 model, particularly decreasing from 16.1% to 12.0% and continuous dropping to 7.8%.During a 2021-2099 period, the changes of annual streamflow in three models under SSP1-2.6 scenario are foreseen by 10.2%, 12.4%, and 12.0% in the ACCESS-CM2, CanESM5, and HadGEM3-GC31-LL models.

Future distribution of surface water resources
Based on the results of calculating the river discharge module, the Ba River watershed has enough to relatively abundant water resources for all the scenarios (Figure 4 and 5).Especially, the HadGEM3-GC31-LL model predicts that the surface water resource will be enough in the 2030s in particular, while the two other models suggest that it will be relatively abundant.Based on Figure 5, however, as for SSPs 5-8.5, it is predicted that the surface water resource from the HadGEM3-GC31-LL model is scarcity at the subbasin 4 (398 km 2 ) during 2030s period whereas there will be relatively abundant at the same subbasin in the two remaining models (Figure 5a).The reason is due to the limited rainfall in this area 6 (~1,280 mm/year) which is 43.8% lower than other areas.According to the climate change scenarios, spatial distribution of surface water resources in the watershed may not be balanced between the western and eastern areas.While the subbasin in the western area may have more surface water resources (2-3 times) than the subbasins in the eastern area.Among the subbasins, subbasin number 6 may have the highest river discharge module range from 35 to 58 l/s.km 2 , respectively.On the other hand, subbasin number 4 may have lowest river discharge module range from 18 to 32 l/s.km 2 .In general, there will be big difference (-37.9%-140.9%) in the water resources between the climate change scenarios as compared to base scenarios.While the increasing or decreasing of water resource will be different depending on the subbasin and climate change scenarios, subbasin 6 showed that water resource will decrease for all the climate change scenarios.Also, climate change scenario SSPs 5-8.5 from HadGEM3-GC31-LL model showed that water resource will decrease at all the subbasin except for the subbasin number 8.Although surface water resources may increase or decrease following each the climate change scenario at the subbasins, it will increase following all the climate change scenarios at the watershed.

Conclusion
This research used the modelling approach with SWAT model to evaluate streamflow change for the Ba River watershed under climate change SSPs scenarios.Simulated results of streamflow showed that annual streamflow will increase during 2030s, 2050s, 2080s and whole periods for all the SSPs scenarios.Therefore, surface water resources of the watershed will increase in the future.At the subbasin level, however, surface water resources may be increased or decreased depending on the subbasin and climate change scenarios.In addition, distribution of surface water resources of the watershed corresponding to the periods of all the SSPs scenarios were created.The results from this study have demonstrated the effective of the SWAT model for simulating streamflow under climate change condition.Finally, the impact of climate change on streamflow and spatial distribution may not only be assessed annually but also considered seasonally.Thus, further research should increase the temporal scale of assessment.

Figure 1 .
Figure 1.The location and topography of the Ba River watershed, Gia Lai province.
and 3, the annual streamflow in the Ba River watershed is simulated to have increasing tendencies for all the climate change SSPs scenarios, in which there was a dramatic rise in the CanESM5 model compared to the two remaining models.The maximum streamflow values among several models are predicted to be 1,228.7 m 3 /s under the CanESM5 model in 2045, 485.3 m 3 /s under the ACCESS-CM2 model in 2095, and 590.5 m 3 /s under the HadGEM3-GC31-LL model in 2057.

Figure 2 .Figure 3 .
Figure 2. The changes of streamflow among different GCMs model in the SSPs1-2.6 scenario

Table 2 .
The changes of annual streamflow in the different periods, GCMs models and SSP scenarios.