Original Soil And Water Assessment Tool Research Articles

Streamflow response to land use/land cover change in the tropical Andes using multiple SWAT model variants

Study regionTropical watershed in the Colombian Andes, the Chico River (CR) watershed. Study focusHydrological models are widely used to project the impacts of LULC (Land Use/Land Cover) change on water budget. However, their ability to produce reliable predictions depends on how accurately they represent the role of vegetation in the watershed’s water balance. We analyze how different representations of Leaf Area Index (LAI) affect streamflow responses to LULC change using the Soil and Water Assessment Tool (SWAT) model. We also examine streamflow response to 100 % pasture cover (PAS), 100 % forest cover (FOR), and a control scenario using the original SWAT model, SWAT-T, and a proposed new variant (SWAT-Tb), which improves LAI bimodal representation for tropical regions. New hydrological insights for the regionSWAT-T and SWAT-Tb reproduce observed LAI and streamflow in the CR watershed. However, SWAT-T restricts LAI simulation to unimodal seasonality, while the original SWAT reproduces streamflow but not LAI seasonality. Results using SWAT-T (unimodal LAI) and SWAT-Tb (constant and bimodal LAI) show streamflow increases during the dry seasons for the FOR scenario and decreases for the PAS scenario. Conversely, the original SWAT, with its default LAI representation, tends to underestimate and overestimate streamflow changes in the FOR and PAS scenarios, respectively. Our results highlight that an unrealistic LAI representation can mislead LULC change impact assessments on streamflow in the tropical Andes.

Coupling the remote sensing data-enhanced SWAT model with the bidirectional long short-term memory model to improve daily streamflow simulations

Global climate change has led to an increase in the frequency and scale of extreme weather events worldwide, and there is an urgent need to develop better-performing hydrological models to improve the accuracy of streamflow simulations and to facilitate water resource planning and management. The Soil and Water Assessment Tool (SWAT) has a notable physical foundation and is widely used in hydrological research. However, it uses a simplified vegetation growth model, introducing uncertainty into the simulation results. This study focused on improving the model based on remotely sensed phenological and leaf area index (LAI) data. Phenological data were used to define vegetation dormancy, and the LAI data replaced the corresponding data simulated by the original model. This approach improved the accuracy of the model in describing vegetation dynamics. Then, the enhanced SWAT model was coupled with the bidirectional long short-term memory (BiLSTM) model to validate the simulation of hydrological processes upstream of the Hei River. During model validation, the performance of the enhanced SWAT model in simulating streamflow (R2 = 0.835, NSE = 0.819) was better than that of the original SWAT model (R2 = 0.821, NSE = 0.805). In terms of simulating evapotranspiration, the enhanced SWAT model demonstrated even greater advantages. During the verification period, compared to those of the SWAT model, the R2 and NSE values of the enhanced SWAT model for daily-scale simulations increased from 0.196 and −0.269 to 0.777 and 0.732, respectively. The R2 and NSE values for monthly-scale simulations increased from 0.782 and 0.678 to 0.906 and 0.851, respectively. Simultaneously, the performance levels of two coupling approaches in streamflow prediction were compared, i.e., direct coupling of the original SWAT and BiLSTM models (SWAT-BiLSTM) and coupling of the enhanced SWAT and BiLSTM models (enhanced SWAT-BiLSTM). The results showed that the enhanced SWAT-BiLSTM model always performed better than the SWAT-BiLSTM model during the entire simulation period, especially the enhanced SWAT-BiLSTM model, which could more accurately predict peak streamflow changes. This study demonstrated that coupling an improved physical model with deep learning models could improve the streamflow prediction accuracy.

Effect of water movement in frozen soil on long-term hydrological simulation of Soil and Water Assessment Tool in cold climate watershed in Hokkaido, Japan

ABSTRACT Despite the unique hydrological properties exhibited by frozen soil, its characteristics are often oversimplified in most distributed hydrological models, resulting in major errors in material cycle predictions in cold climate regions. This study aims to modify the Soil and Water Assessment Tool (SWAT), a semi-distributed hydrological model that originally assumes no water movement in frozen soil layers, by incorporating the dynamic change of soil permeability based on the degree of soil freezing, and to understand the effect of water movement in frozen soil on watershed hydrology. SWAT with frozen soil permeability (SWAT-FSP) was developed by adding two modules to the original SWAT: a heat conduction equation that considers the phase change of ice and liquid water and a frozen-soil percolation model that considers dynamic decrease in soil permeability due to soil freezing. The original SWAT and SWAT-FSP were set up in the Tokoro River watershed in northern Japan. The river flow and soil water dynamics during the snowmelt period (March–May) simulated using the original SWAT and SWAT-FSP were compared. The modified model significantly alters the soil water dynamics simulation to generate earlier soil water drainage and reduces the soil water content during the snowmelt period in years featuring weak soil freezing. Consequently, the SWAT-FSP estimated an earlier increase in river flow at the beginning of the snowmelt period and a smaller river flow in the middle of the snowmelt period as compared with the original SWAT. The modified model improved the overall performance in river flow simulation for the snowmelt period (March–May) of 2012–2019. The Nash-Sutcliffe efficiency obtained during the calibration period by the original SWAT was 0.602, and during the validation period it was 0.367. For the SWAT-FSP, it was 0.659 during calibration, and 0.639 during validation. Meanwhile, the SWAT-FSP showed unsatisfactory performance in 2014 and 2015, with underestimated snow water equivalent, indicating that improvements in the snowmelt/accumulation module may further enhance its accuracy. Our results showed that the dynamic change of soil permeability governed by the degree of soil freezing is an important factor affecting the water cycle of cold climate watersheds with seasonally frozen soil.

Modeling hydrological processes under Multi-Model projections of climate change in a cold region of Hokkaido, Japan

Local hydrological systems are highly sensitive to global warming, especially in cold-climate-dominated basins. The integrated climate-hydrological model is a powerful tool to analyze and predict such effects. However, modeling hydrological processes contains considerable uncertainty in cold areas due to the complex dynamics of freeze–thaw cycles and requirements of lot inputs data. Here, we developed a fast and simple approach to modify the soil temperature module in soil and water assessment tool (SWAT) model in a cold region of the Ishikari River basin (IRB), Hokkaido, Japan. Water balance components and sediment yield simulations were explored by the modified SWAT model, outperforming the original SWAT model by PBIAS less than 15 % and NSE higher than 0.60 in monthly streamflow and sediment yield during both calibration (2011–2015) and validation phases (2016–2019). The future prediction was integrated with future climate datasets under four Shared Socioeconomic Pathways (SSP) scenarios (SSP126, SSP245, SSP370, and SSP585) derived from two Global Circulation Models (GCMs), in two future stages (2045–2069 and 2075–2099) with relative to the baseline stage (1990–2014). The varying climate with warmer temperatures (+3.05℃) and less precipitation (-7%) in the future would lead to that evapotranspiration (ET, 67 %) increased. Water yield (WY, −9%), surface runoff (SURQ, −18 %), groundwater flow (GWQ, −10 %), lateral flow (LATQ, −7%), percolation (PERC, −10 %), streamflow (-7%) and sediment yield (-14 %) decreased across the whole basin. However, due to climatic spatial variation, at the subbasin scale, hydrological components (e.g., WY, GWQ, LATQ, and PERC) would increase in the northern region, at the same time, SURQ and ET rise significantly in the central and southern areas. Our findings could offer evidence for future water and soil protection policymaking across the IRB.

Enhancing SWAT model with modified method to improve Eco-hydrological simulation in arid region

The Soil and Water Assessment Tool (SWAT) model is diffusely applied to areas with uniform precipitation distribution and gentle slopes. However, the SWAT has apparent defects in the process of ecohydrological simulation in mid-high latitudes and arid regions. This model can merely be used in agricultural catchments with gentle slopes and uniform rainfall, and the simulation is not suitable for high-altitude alpine catchment. To this point, this paper makes appropriate improvements to the runoff module, and specifically corrects the key sensitive parameters. The global sensitivity and performance assessment of the model's parameters were carried out before and after the improvement using the monthly data of runoff and evaporation from 1990 to 2018 in the Kaidu River basin in northwestern China. It is indicated that the Modified SWAT simulates the performance of monthly runoff and evapotranspiration better than the original SWAT, which overestimates the runoff volume throughout hydrological changes in different periods. Compared with the monthly water production under different land use/cover conditions, the highest water production was found in forest and grassland. By modifying the SWAT algorithm, the structural non-determinacy of the Modified SWAT is significantly reduced, and the adaptability of the model is improved. Thus, the Modified SWAT shows a better simulation performance in arid areas with considerable terrain variation and scarce rainfall data.

Informing the SWAT model with remote sensing detected vegetation phenology for improved modeling of ecohydrological processes

The Soil and Water Assessment Tool (SWAT) model has been widely applied for simulating the water cycle and quantifying the influence of climate change and anthropogenic activities on hydrological processes. A major uncertainty of SWAT stems from the poor representation of vegetation dynamics due to the use of a simplistic vegetation growth and development module. Using long-term remote sensing-based phenological data, the SWAT model’s vegetation module was improved by adding a dynamic growth start date and the dynamic heat requirement for vegetation growth rather than using constant values. The new SWAT model was verified in the Han River basin, China, and found its performance was much improved in comparison with that of the original SWAT model. Specifically, the accuracy of the leaf area index (LAI) simulation improved notably (coefficient of determination (R2) increased by 0.193, Nash–Sutcliffe Efficiency (NSE) increased by 0.846, and percent bias decreased by 42.18 %), and that of runoff simulation improved modestly (R2 increased by 0.05 and NSE was similar). Additionally, it is found that the original SWAT model substantially underestimated evapotranspiration (Penman-Monteith method) in comparison with the new SWAT model (65.09 mm (or 22.17 %) for forests, 92.27 mm (or 32 %) for orchards, and 96.16 mm (or 36.4 %) for farmland), primarily due to the inaccurate representation of LAI dynamics. Our results suggest that an accurate representation of phenological dates in the vegetation growth module is important for improving the SWAT model performance in terms of estimating terrestrial water and energy balance.

A modified hydrologic model for examining the capability of global gridded PET products in improving hydrological simulation accuracy of surface runoff, streamflow and baseflow

• SWAT model was modified to examine the capability of PET products in improving hydrological simulation accuracy. • Distributed input and higher spatial resolution of PET products are more conducive to hydrological modeling. • The input of PET products has a great impact on streamflow and baseflow, followed by surface runoff. • The accuracy of PET products should be improved from both improving PET products and considering water budget closure. Terrestrial evapotranspiration (ET), which links water and energy exchanges between atmosphere, land surface and underground water, significantly affects hydrological modelling. However, accurate ET estimation is still a great challenge for hydrological models. Currently, most hydrological models obtain ET mainly from their built-in potential evapotranspiration (PET) modules due to lack of observations. Note that the ET obtained in this way cannot be verified by observations and inevitably brings large model uncertainty. Fortunately, several PET products have been produced based on satellite remote sensing and/or reanalysis climate model output with global coverage and high spatiotemporal resolution, providing a great opportunity for tunning hydrological models. However, the accuracy of global gridded PET products varies in different regions. It is still unknown whether PET products can improve the accuracy of hydrological simulation, and how different resolutions and input types (lumped/distributed) of PET products affect the performance of hydrological models. The question about how hydrological models will be further developed in combination with PET products is also not well answered, given the complexity in interaction between ET and other water cycle variables. In this study, a modified Soil and Water Assessment Tool (SWAT) was chosen to examine the capability of PET products in improving hydrological simulation in terms of surface runoff, streamflow and baseflow, together with the Shaying River Basin in China as a case study. Two Moderate Resolution Imaging Spectroradiometer (MODIS) PET products were employed to form various PET scenarios. Results indicate that PET products are significantly different from the PET simulated by the SWAT built-in PET module in terms of spatial distribution and magnitudes. According to statistical metrics, the use of PET products to replace the SWAT built-in PET module can improve the accuracy of hydrological simulation compared with the original SWAT using its built-in PET module. The distributed input and higher spatial resolution of PET products are more conducive to improve the accuracy of hydrological simulation. Our research shows that both improving the accuracy of PET products and balancing water budget closure are critical for accurate hydrological modeling. These conclusions obtained based on MODIS products were further verified against a hourly PET data product (hPET). This study provides insights into understanding the use of PET products in improving the accuracy of hydrological simulation. The study findings can provide a beneficial reference for the further development of hydrological models in combination with PET products.

Modifying the SWAT Model to Simulate Eco-Hydrological Processes in an Arid Grassland Dominated Watershed

Grasslands are the main land cover type and one of the most important ecosystems in arid and alpine endorheic basins. The vegetation coverage of grasslands is spatially heterogeneous in arid and alpine areas and it may lead to variations in water allocation. The Soil and Water Assessment Tool (SWAT) is one of the most widely used semi-distributed catchment-scale eco-hydrological models. The leaf area index (LAI) is one of the vegetation coverage indexes and is incorporated in the SWAT model. However, in SWAT, the LAI accumulation is controlled by heat, and neglects other relevant factors such as precipitation and terrain. To address the drawbacks of the SWAT in simulating vegetation coverage and plant patterns, several studies have focused on improving LAI estimation. However, they still have been limited to arid and alpine grasslands with different vegetation coverages. In this study, we modified the SWAT model using remotely sensed LAI data with high temporal and spatial resolution. We used this to better simulate eco-hydrological processes in grassland basins with different vegetation coverages in the upper reaches of the Bayin River Basin. Results showed that for the original SWAT model, the simulated LAI was homogeneous within each land use/cover type, whereas the remotely sensed LAI was spatially heterogeneous and better captured the vegetation coverage of the entire basin. The proper estimation of the LAI was reflected in the improved simulation of the monthly streamflow and sediment yield at the basin outlet and the monthly ET. These findings indicate that the modified SWAT could better simulate hydrological processes in arid and alpine grasslands with different vegetation coverages.

Hydrological Modeling of Karst Watershed Containing Subterranean River Using a Modified SWAT Model: A Case Study of the Daotian River Basin, Southwest China

Karst watershed refers to the total range of surface and underground recharge areas of rivers (including subterranean rivers and surface rivers) in karst areas. Karst water resources, as the primary source of domestic water supply in southwest China, are vital for the social and economic development of these regions. It is greatly significant to establish a high-precision hydrological model of karst watershed for guiding water resources management in karst areas. Choosing the Daotian river basin in the Wumeng Mountains of Southwest China as the study area, this paper proposed a method for simplifying karst subterranean rivers into surface rivers by modifying the digital elevation model (DEM) based on a field survey and tracer test. This method aims to solve the inconsistency between the topographical drainage divides and actual catchment boundaries in karst areas. The Soil and Water Assessment Tool (SWAT) model was modified by replacing the single-reservoir model in the groundwater module with a three-reservoir model to depict the constraints of multiple media on groundwater discharge in the karst system. The results show that the catchment areas beyond topographic watershed were effectively identified after simplifying subterranean rivers to surface rivers based on the modified DEM data, which ensured the accuracy of the basic model. For the calibration and two validation periods, the Nash–Sutcliffe efficiencies (NSE) of the modified SWAT model were 0.87, 0.83, and 0.85, respectively, and R2 were 0.88, 0.84, and 0.86, respectively. The NSE of the modified SWAT model was 0.09 higher than that of the original SWAT model in simulating baseflow, which effectively improved the simulation accuracy of daily runoff. In addition, the modified SWAT model had a lower uncertainty within the same parameter ranges than the original one. Therefore, the modified SWAT model is more applicable to karst watersheds.

Adaptability of machine learning methods and hydrological models to discharge simulations in data-sparse glaciated watersheds

The accurate simulation and prediction of runoff in alpine glaciated watersheds is of increasing importance for the comprehensive management and utilization of water resources. In this study, long short-term memory (LSTM), a state-of-the-art artificial neural network algorithm, is applied to simulate the daily discharge of two data-sparse glaciated watersheds in the Tianshan Mountains in Central Asia. Two other classic machine learning methods, namely extreme gradient boosting (XGBoost) and support vector regression (SVR), along with a distributed hydrological model (Soil and Water Assessment Tool (SWAT) and an extended SWAT model (SWAT_Glacier) are also employed for comparison. This paper aims to provide an efficient and reliable method for simulating discharge in glaciated alpine regions that have insufficient observed meteorological data. The two typical basins in this study are the main tributaries (the Kumaric and Toxkan rivers) of the Aksu River in the south Tianshan Mountains, which are dominated by snow and glacier meltwater and precipitation. Our comparative analysis indicates that simulations from the LSTM shows the best agreement with the observations. The performance metrics Nash-Sutcliffe efficiency coefficient (NS) and correlation coefficient (R2) of LSTM are higher than 0.90 in both the training and testing periods in the Kumaric River Basin, and NS and R2 are also higher than 0.70 in the Toxkan River Basin. Compared to classic machine learning algorithms, LSTM shows significant advantages over most evaluating indices. XGBoost also has high NS value in the training period, but is prone to overfitting the discharge. Compared with the widely used hydrological models, LSTM has advantages in predicting accuracy, despite having fewer data inputs. Moreover, LSTM only requires meteorological data rather than physical characteristics of underlying data. As an extension of SWAT, the SWAT_Glacier model shows good adaptability in discharge simulation, outperforming the original SWAT model, but at the cost of increasing the complexity of the model. Compared with the oftentimes complex semi-distributed physical hydrological models, the LSTM method not only eliminates the tedious calibration process of hydrological parameters, but also significantly reduces the calculation time and costs. Overall, LSTM shows immense promise in dealing with scarce meteorological data in glaciated catchments.

A New Physically-Based Spatially-Distributed Groundwater Flow Module for SWAT+

Watershed models are used worldwide to assist with water and nutrient management under conditions of changing climate, land use, and population. Of these models, the Soil and Water Assessment Tool (SWAT) and SWAT+ are the most widely used, although their performance in groundwater-driven watersheds can sometimes be poor due to a simplistic representation of groundwater processes. The purpose of this paper is to introduce a new physically-based spatially-distributed groundwater flow module called gwflow for the SWAT+ watershed model. The module is embedded in the SWAT+ modeling code and is intended to replace the current SWAT+ aquifer module. The model accounts for recharge from SWAT+ Hydrologic Response Units (HRUs), lateral flow within the aquifer, Evapotranspiration (ET) from shallow groundwater, groundwater pumping, groundwater–surface water interactions through the streambed, and saturation excess flow. Groundwater head and groundwater storage are solved throughout the watershed domain using a water balance equation for each grid cell. The modified SWAT+ modeling code is applied to the Little River Experimental Watershed (LREW) (327 km2) in southern Georgia, USA for demonstration purposes. Using the gwflow module for the LREW increased run-time by 20% compared to the original SWAT+ modeling code. Results from an uncalibrated model are compared against streamflow discharge and groundwater head time series. Although further calibration is required if the LREW model is to be used for scenario analysis, results highlight the capabilities of the new SWAT+ code to simulate both land surface and subsurface hydrological processes and represent the watershed-wide water balance. Using the modified SWAT+ model can provide physically realistic groundwater flow gradients, fluxes, and interactions with streams for modeling studies that assess water supply and conservation practices. This paper also serves as a tutorial on modeling groundwater flow for general watershed modelers.

Development of alternative SWAT-based models for simulating water budget components and streamflow for a karstic-influenced watershed

Most semi-distributed and distributed hydrological models can potentially be applied for comprehensive catchment simulations, but these models may require modifications to capture dominant hydrological processes. The Soil and Water Assessment Tool (SWAT) ecohydrological model, an open source program, provides flexibility for modification to capture different components of hydrological processes in a given catchment. In this study, two new approaches were tested to simulate the hydrological processes of a karst system in the Zagros Mountains, Iran. The major modifications used in these two methods included adjusting the percolation rates in karst hydrologic response units (HRUs) using SWAT-Maharlu Lake (SWAT-ML), and modification of the crack flow module using SWAT-Crack Flow (SWAT-CF). Hydrological datasets from 1980 to 2013 were used for calibration and validation of surface runoff, baseflow, crop yields, and actual evapotranspiration. Both modified models outperformed the standard SWAT model in simulating runoff. The respective average Nash Sutcliffe Efficiencies of the original SWAT, SWAT-ML and SWAT-CF were 0.64, 0.68 and 0.66. Also, the results showed that higher deep aquifer infiltration values were estimated in the modified models versus the standard SWAT for a karst HRU, which reflects more accurate levels of recharge to the deep aquifer. However, the results showed that karst conditions and geometry of sinkholes in karst zones did not play a major role in creation of runoff.

Improved hydrologic modeling for depression-dominated areas

Modeling of overland flow for a depression-dominated area is a difficult task due to the spatial distribution of depressions. Particularly, the hierarchical relationships of depressions and their variable contributing areas make the hydrologic modeling more complicated. In traditional hydrologic modeling, however, digital elevation models (DEMs) that are used to represent surface topography are often edited by filling depressions before delineation, and the impacts of depressions are considered by using certain simplified and lumped approaches. Due to such flaws, these hydrologic models fail to account for the actual dynamic influences of surface depressions. The objective of this study is to improve hydrologic modeling for depression-dominated regions by developing a new modeling framework to quantify the influence of spatially distributed depressions. To achieve this objective, a new puddle-based unit (PBU) probability distributed model (PBU-PDM) is developed and coupled with the existing Soil and Water Assessment Tool (SWAT). The PBU-PDM facilitates separate simulations for non-depressional and depressional areas, and accounts for the hierarchical relationships of depressions, their filling and spilling processes, and the dynamic variations in their contributing areas. The PBU-PDM enhanced SWAT was applied to the Upper Maple River watershed in North Dakota, and calibrated and validated using the observed data. In particular, it was also compared with the original SWAT model, which demonstrated the abilities of PBU-PDM to mimic the filling-spilling overland flow dynamics and the threshold behavior. The PBU-PDM provides improved watershed-scale hydrologic modeling, especially for depression-dominated areas.

Freeze-Thaw cycle representation alters response of watershed hydrology to future climate change

Hydrologic models are widely used for projecting influences of changing climate on water resources. In this study, we compared the original Soil and Water Assessment Tool (SWAT) model and an enhanced version of SWAT model with physically based Freeze-Thaw cycle representation (SWAT-FT) for simulating future annual ET, stream flow, water yield, surface runoff, and subsurface runoff in the Upper Mississippi River Basin (UMRB). SWAT-FT projected fewer frozen days than the original SWAT model due to its better representation of snow cover insulation effects. Both models derived declining trends in annual streamflow and terrestrial water yield in the late 21st century due to increased ET under warmer climate. However, these two models exhibited contrasting mechanisms underlying the streamflow decline. For original SWAT model, the decrease in surface runoff was the major driver, while for SWAT-FT, reduced subsurface runoff was the main cause. In general, the original SWAT model predicted more surface runoff and less subsurface runoff than SWAT-FT. Further geospatial inspection shows large discrepancies between these two models, particularly in the northern colder parts of the UMRB, where the maximum differences in annual surface and subsurface runoff reached 130 mm yr−1 and 140 mm yr−1, respectively. Collectively, the results demonstrate the importance of accounting for Freeze-Thaw cycles for reliable projection of future water resources.

Optimization of SWAT-Paddy for modeling hydrology and diffuse pollution of large rice paddy fields

The Soil and Water Assessment Tool (SWAT) simplified the hydrological processes in large patches of paddy fields. Based on the original SWAT code, the main equations are modified and validated with the long-term field observations in freezing-thawing watershed. The original SWAT and SWAT with paddy module (SWAT-P) both perform well in the simulation of stream flow at the basin scale (NSE ≥ 0.7). SWAT-P performs better in the soil water simulation (RMSE = 24.22 mm, smaller than 104.63 mm of SWAT). The monthly diffuse nitrogen loadings simulated by SWAT-P are within a reasonable range and close to the monitored nitrate loads in 2011 and 2012. SWAT-P represents the ponding water depth dynamics during the rice growing period. The original SWAT overestimates the diffuse pollution loads in the freeze-thaw period. SWAT-P has better performance for water cycles and diffuse pollution load simulations in watersheds with large rice paddy fields.

Using an improved SWAT model to simulate hydrological responses to land use change: A case study of a catchment in tropical Australia

Land use change is one of the dominant driving factors of watershed hydrological change. Thus, hydrological responses to land use changes require detailed assessments to ensure sustainable management of both water resources and natural ecosystems. The Soil and Water Assessment Tool (SWAT) model has been widely used to simulate the impacts of land use change on water balance. However, the original SWAT model has poor performance in estimating the leaf area index (LAI) of different vegetation types for tropical areas. The objective of this study was to simulate the impact of different land use change scenarios (deforestation, afforestation and urbanization) on the water balance, using an improved SWAT model with vegetation growth calibrated from MODIS LAI data. The North Johnstone River catchment in wet tropical eastern Australia was selected as the case study area. Results showed that the modified SWAT model was able to reproduce smoothed MODIS LAI with NSE ≥ 0.59 (NSE < 0 for default SWAT), R2 ≥ 0.70 (R2 ≤ 0.66 for default SWAT), and |PBIAS| ≤ 2.5% (|PBIAS| ≥ 42% for default SWAT), and to predict monthly streamflow well with NSE ≥ 0.92 (NSE ≥ 0.90 for default SWAT), R2 ≥ 0.94 (R2 ≥ 0.90 for default SWAT). It is noted that SWAT-T had |PBIAS| ≤ 10% while |PBIAS| ≤ 5% for default SWAT. Land use change impacted all hydrological variables, with the impact on surface runoff being the most notable at yearly scale (8.9%, 5.7%, −9.5% and 15.9% for scenario 1, 2, 3 and 4, respectively). Absolute changes of surface runoff under land use change scenarios differed across months, with the most notable absolute change occurring during the wet season (December to May) (1.2 ~ 6.6 mm, 1.0 ~ 3.5 mm, −7.3 ~ -1.1 mm and 3.0 ~ 9.0 mm for scenario 1, 2, 3 and 4, respectively). Urbanization increased surface runoff (5.7% and 15.9% for scenario 2 and 4, respectively) and decreased lateral runoff (−0.7% and −1.3%) and groundwater (-0.9% and −3.5%), but produced no clear change in total runoff (0.2% and 0.2%), actual evapotranspiration (−0.3% and −0.3%), and soil water (0.5% and 0.7%) at the annual time scale. Furthermore, afforestation could decrease surface runoff (−9.5% for scenario 3) and soil water (−2.0%), increase evapotranspiration (1.7%), and lead to slight changes (absolute values ≤0.8%) in other hydrological variables at the annual time scale. A strong positive correlation (r ≥ 0.94) was observed between annual rainfall and total runoff for forest-evergreen, range-grasses, and urban land use. Forest-evergreen generally produced less total runoff than range-grasses and urban land use under conditions of the same rainfall, terrain slope, and soil texture. In addition, urban land use generally produced more surface runoff and less lateral runoff and groundwater than forest-evergreen and range-grasses under the same conditions. These results contribute important information for development of effective adaptation strategies and future policy plans for sustainable water management in tropical eastern Australia.

Modification and validation of the SWAT model based on multi-plant growth mode, a case study of the Meijiang River Basin, China

With the aim of simulating non-point source pollution of the agroforestry system, we established a forest growth model featuring variable density and mixed vegetation types to address the drawbacks of the Soil and Water Assessment Tool (SWAT) in estimating the accumulated biomass, based on the average forest vegetation density and single plant growth pattern. In addition, a dominant vegetation abundance remote sensing inversion model and leaf area index (LAI) and extinction coefficient remote sensing inversion model were set up to obtain relevant parameters of the forest growth model. Moreover, based on radiant energy under intercropping, we used the Keating equation with the intercropping index as the variable to modify the original SWAT single biomass accumulation model. This was done to explore the intercropping growth model as well as the remote sensing inversion method, considering the multiple cropping and intercropping indexes. We selected the Meijiang River basin as the study area, a sub-basin of the Poyang Lake basin in China in a subtropical humid monsoon climate zone and red soil region, to explore the validity of the modified SWAT based on field data. The results show that the modified SWAT outperformed the original model in terms of simulated flow and nutrient load, using variables such as vegetation coverage, different forest components (e.g., coniferous forest, broadleaf forest, and sparse shrubs), and LAI. As for simulated flow, the validity of the modified SWAT was increased by 7.8%, and simulation of the average, maximum, minimum, and total flow rates were closer to the measured values, better reflecting the actual surface water storage. In the simulation of nutrient load, validity was increased by 6.4% (total phosphorus) and 6.1% (total nitrogen), better representing macroscopic and microscopic characteristics of the forest vegetation landscape.

Application of SWAT Model with a Modified Groundwater Module to the Semi-Arid Hailiutu River Catchment, Northwest China

In the original soil and water assessment Tool (SWAT) model (SWAT-O), the contributions of shallow aquifers and deep aquifers to streamflow are simulated using the linear reservoir method. The movement of groundwater was limited in the hydrological response unit which is a minimum calculation unit in the SWAT. However, this computational method may not be suitable for the areas where a groundwater system is complicated, and the river is predominately recharged by groundwater. In this paper, we proposed an enhanced groundwater module which divides shallow aquifers into upper and lower aquifers, integrates all the deep aquifers of a sub-basin into a regional aquifer, and simulates interactive water amount between lower aquifer and deep aquifer using water depth difference. The modified groundwater module was introduced to the original SWAT model, hereby referred to as SWAT-MG. The SWAT-MG and SWAT-O models were applied to the Hailiutu River catchment, which is a semi-arid wind sandy grass shoal catchment. Results showed that both models underestimated streamflow in peak flow, while the simulated streamflow of SWAT-MG was closer the observed values than that of SWAT-O. Three evaluation criteria (NSE, RSR, PBIAS) were applied to evaluate the performance of the models and the results showed that SWAT-MG had a better performance than SWAT-O. The baseflow index of Hailiutu River which was calculated by the results of SWAT-MG was 96.78%, which means the streamflow is predominately recharged by groundwater, and this conforms to the actual situation of Hailiutu River catchment. This indicates that a SWAT model with a modified groundwater module could better represent the groundwater flow behavior in the study area.

Multi-Objective Calibration of a Distributed Hydrological Model in a Highly Glacierized Watershed in Central Asia

Understanding glacio-hydrological processes is crucial to water resources management, especially under increasing global warming. However, data scarcity makes it challenging to quantify the contribution of glacial melt to streamflow in highly glacierized catchments such as those in the Tienshan Mountains. This study aims to investigate the glacio-hydrological processes in the SaryDjaz-Kumaric River (SDKR) basin in Central Asia by integrating a degree-day glacier melt algorithm into the macro-scale hydrological Soil and Water Assessment Tool (SWAT) model. To deal with data scarcity in the alpine area, a multi-objective sensitivity analysis and a multi-objective calibration procedure were used to take advantage of all aspects of streamflow. Three objective functions, i.e., the Nash–Sutcliffe efficiency coefficient of logarithms (LogNS), the water balance index (WBI), and the mean absolute relative difference (MARD), were considered. Results show that glacier and snow melt-related parameters are generally sensitive to all three objective functions. Compared to the original SWAT model, simulations with a glacier module match fairly well to the observed streamflow, with the Nash–Sutcliffe efficiency coefficient (NS) and R2 approaching 0.82 and an absolute percentage bias less than 1%. Glacier melt contribution to runoff is 30–48% during the simulation period. The approach of combining multi-objective sensitivity analysis and optimization is an efficient way to identify important hydrological processes and recharge characteristics in highly glacierized catchments.

Enhancing SWAT with remotely sensed LAI for improved modelling of ecohydrological process in subtropics

Vegetation growth in Soil and Water Assessment Tool (SWAT) is a crucial process for quantifying ecohydrological modelling, as it influences evapotranspiration, interception, soil erosion and biomass production. The simplified version of Environmental Policy Integrated Climate (EPIC) in SWAT was originally designed for temperate regions and naturally based on temperature to simulate growth cycles of vegetation. However, tropical or subtropical vegetation growth is mainly controlled by rainfall. Due to this limitation, current SWAT simulations in tropics and subtropics have been facing a series of problems on vegetation dormancy, water balance and sediment yield. Therefore, we proposed an approach to enhance the modelling of SWAT vegetation dynamics with remotely sensed leaf area index (LAI), to finally increase the applicability of SWAT in tropical or subtropical areas. Spatially and temporally continuous LAI products (1 day, 500 m) from Moderate Resolution Imaging Spectroradiometer (MODIS) observations were integrated into SWAT to replace the LAI simulated by built-in EPIC module. Two advanced filter algorithms were employed to derive a downscaled LAI (30 m) to keep a consistent spatial scale with the size of Hydrological Response Units (HRU) and open data (i.e. SRTM, 30 m), and the source code of the plant growth module were correspondingly modified to incorporate the downscaled LAI into SWAT. To examine the performance of our proposed approach, a case study was conducted in a representative middle-scale (6384 km2) subtropical watershed of Meichuan basin, China, and detailed analysis was performed to investigate its ecohydrological effects, such as streamflow, sediment yield and LAI dynamics from 2001 to 2014. Model performances were compared among three scenarios: (1) original SWAT, (2) SWAT with a corrected plant dormancy function, and (3) modified SWAT after integration of MODIS LAI (our proposed method). Results showed that the modified SWAT took advantage of downscaled MODIS LAI and produced more reasonable seasonal curves of vegetation cover factor (C) of plants than the original model. Correspondingly, the modified SWAT substantially improved streamflow and sediment simulations. The findings demonstrated that SWAT model can be a useful tool for simulating ecohydrological process for subtropical ecosystems when integrated with our proposed method.