The study focused on analysing the eutrophication indicators of the Krishnagiri Reservoir Project (KRP) using the Water Quality Analysis Simulation Program (WASP). The reservoir was divided into six segments to simulate the indicators, and field measurements from 2015 to 2018 were used to calibrate the model while 2019 to 2022 were used to validate it. The results showed that the model predictions were in good agreement with the measured values, indicating the reliability of the model. The study assessed the impact of nutrient loads on dissolved oxygen, nitrate, phosphate, and chlorophyll-a. Four nutrient loading scenarios were simulated, and the most effective scenario (Scenario 4) involved a 75% reduction in nutrient load, which increased the reaeration rate by 1.47% and decreased chlorophyll-a concentrations by 88%. The study concluded that maintaining nitrates below 2.5 mg/L and phosphates below 0.75 mg/L could help restore the KRP reservoir's trophic status from hypereutrophic to mesotrophic. Overall, the study demonstrated the use of the WASP model in developing nutrient loading scenarios to manage reservoir water quality effectively. The findings could help policymakers and managers make informed decisions about reducing nutrient loads and restoring the trophic status of eutrophicated reservoirs.

  • The study focused on the eutrophication management of the Krishnagiri Reservoir Project (KRP) using the Water Quality Analysis Stimulation Program (WASP) model with different nutrient load scenarios.

  • WASP is used to simulate the eutrophication indicators such as dissolved oxygen, chlorophyll-a, nitrate, and phosphate concentrations in the KRP reservoir.

  • Calibrated Eutro-model was used to develop scenarios for eutrophication management.

Water quality models are important tools in the management of water resources. Since a wide variety of models are available, it is necessary to choose the appropriate one for the intended use. A ScoRE approach has been proposed (Ejigu 2021) to select the most appropriate model based on the scope, record, and user experience among the eight water quality models (CE-QUAL-W2, MIKE HYDRO River, MOHID Water, SIMCAT, SisBaHIA, TOMCAT, QUAL2KW, and WASP7) for the selected problem. The Water Quality Analysis Simulation Program (WASP) model ranked as the second-best model with a ScoRE value of 3.6 (Mateus et al. 2018). Another comparison among the multiple regression, remote sensing methods and WASP model for water quality prediction (salinity) in Akkulam-Veli Lake (Albert Moses et al. 2016) found that the WASP model is the most efficient model among them with an efficiency of 0.83.

WASP (version 8) is an open-source dynamic compartment model for aquatic systems that can simulate both the water column and the underlying benthos. WASP's eutrophication sub-model is a process-based, mechanistic or deterministic water quality model that predicts phytoplankton growth patterns. Nutrients, phytoplankton, and dissolved oxygen (DO) concentrations are the key water quality elements studied in eutrophication modelling research. The availability of nutrients, temperature, and light all influence phytoplankton development in the WASP model (Moses et al. 2015).

WASP has been of interest to several researchers in the recent past. For example, the WASP model has been applied in El Pañe Reservoir (Mamani Larico & Zúñiga Medina 2019) to assess the eutrophication process by dividing the reservoir into 11 superficial segments representing one-dimensional horizontal distribution and found that aquaculture activities and the benthic flux are the major nutrient sources to control the phytoplankton community. WASP is also capable of capturing the seasonality and spatial variability of under-ice coupled river-lake system's water quality (Akomeah et al. 2020).

A similar study with WASP in Lake Obili showed that improved DO concentration levels in the reservoir can improve aquatic life (Ajeagah et al. 2017). There have been several studies (Narasimhan et al. 2010; Park et al. 2013; Mbuh et al. 2019; Obin et al. 2021; Shabani et al. 2021; Wimordi et al. 2021; Żelazny et al. 2023) that have investigated the effectiveness of WASP in water quality assessment and eutrophication control.

Krishnagiri Reservoir Project (KRP), located in South India is hyper-eutrophicated for the past two decades (Elangovan & Murali 2020). The research study (Sudha & Ambujam 2012; Ambujam & Sudha 2016) on the reservoir found that the nutrient loads resulting from the catchment processes significantly alter the trophic status of the reservoir. Total phosphorus from the cropland is the major contributor to the KRP (Sasitharan & Arunbabu 2020) leading to the present eutrophic condition of the reservoir. Remote sensing techniques deployed for the rapid assessment of reservoir trophic status (Elangovan & Murali 2020; Abdul Wahid & Arunbabu 2022) mapped the concentration of chlorophyll-a (chl-a), one of the indicators for eutrophication assessment emphasized the need for monitoring and managing the reservoir before it reaches an anoxic condition and loses its storage capacity. However, earlier work in Krishnagiri Reservoir does not attempt to develop process-based water quality models. The present study focused on the eutrophication management of the KRP using a process-based mechanistic WASP model with different nutrient load scenarios. To achieve this objective, WASP is used to simulate the eutrophication indicators such as DO, chl-a, nitrate, and phosphate concentrations in the KRP reservoir. The calibrated model was then used to develop scenarios for eutrophication management.

Study area

The KRP was constructed in 1958 across the Ponnaiyar River near Periyamuthur village is about 10 km from Krishnagiri town in Krishnagiri district, Tamil Nadu. It is located at a latitude of 12°28′ North and a longitude of 78°11′ East (Figure 1). The hydraulic structure is 1.003 km long with eight spillways, three river canals, and two canal sluices which irrigate 3,642 ha. The reservoir serves as the lifeline for the region serving multiple uses of water from irrigation to fish culture. The average annual rainfall in the KRP dam is 870 mm. The year is split into four seasons: the dry season (January to March), the summer season (SE) (April to May), the southwest monsoon (SEM) season (June to September), and the northeast monsoon (NEM) season (October to December). The mean daily maximum temperature recorded during summer is about 37 °C and the mean daily minimum temperature is about 25 °C.
Figure 1

Index map of the Krishnagiri Reservoir Project.

Figure 1

Index map of the Krishnagiri Reservoir Project.

Close modal

WASP model governing equations

The WASP model is based on the conservation of mass, which accounts for all material entering and exiting the water body through direct material addition (runoff and loading), advective and dispersive transport modes, as well as physical, chemical, and biological changes.

A mass balance in Cartesian coordinates for some arbitrary water quality ingredient with concentration C across a control volume of arbitrary size yields the three-dimensional advection–diffusion equation (Equation (1)) (Wool et al. 2020):
(1)
where C is the concentration of the water quality constituent (M/L3), t is time (T), Ux, Uy, and Uz are the longitudinal, lateral, and vertical advective velocities (L/T), Ex, Ey, and Ez are the longitudinal, lateral, and vertical diffusion coefficients (L2/T), SL is the direct and diffuse loading rate (M/L3 T), SB is the boundary loading rate (including upstream, downstream, benthic, and atmospheric) (M/L3 T), SK is the total kinetic transformation rate in which a positive value denotes the source and a negative value denotes a sink (M/L3 T). The parameters simulated using the model are DO, nitrate, phosphate, and chl-a. The constituent kinetic reaction (SK) included in the source and sink are as follows (Wool & Epa 2015).

Constituent kinetic reactions (SK)

Dissolved oxygen

(a)
where DO increases due to plant photosynthesis. It is lost via fast carbonaceous biochemical oxygen demand (CBOD) oxidation, nitrification, and plant respiration. Depending on whether the water is undersaturated or oversaturated, it is gained or lost by reaeration.

Nitrate-nitrogen

Nitrate-nitrogen increases due to the nitrification of ammonia. It is lost via denitrification and plant photosynthesis. The dissolved inorganic nutrients are simulated using the following processes:
(b)

Inorganic phosphorus

Inorganic phosphorous increases due to organic phosphorus hydrolysis, phytoplankton respiration/excretion, and bottom plant excretion. It is lost via plant photosynthesis and the settling of sorbed forms. Dissolved organic nutrient PO4 dynamics is governed by the following expression:
(c)

The parameters in Equations (b) and (c) are explained below.

S is dissolved inorganic nutrient concentration (M/L3), S′ is another inorganic form of the nutrient which decays to the form S (e.g., NH3 NO3), (M/L3), Sorg is the dissolved organic nutrient concentration (M/L3), Sdet is the suspended particulate organic nutrient concentration (M/L3), Ssed is the organic sediment nutrient concentration (M/L3), K1 is the transformation rate of S′ into S (T−1), K2 is the transformation rate of S into some other dissolved inorganic form of the nutrient (T−1), Korg is the hydrolysis rate of dissolved organic nutrient (T−1), Kdet is the decomposition rate of particulate organic nutrient (T−1), Ksed is the decomposition rate of organic sediment nutrient (T−1), Ks is the settling rate for particulate organic nutrient (T−1), Vs is the photosynthetic uptake rate for nutrient (M/L3 T), es is the soluble excretion rate of nutrient by all organisms (M/L3 T), f1 is the fraction of soluble excretions which are inorganic, f2 is the fraction of detritus decomposition products which are immediately available for algal uptake, and f3 is the fraction of sediment decomposition products which are immediately available for algal uptake.

Phytoplankton

Phytoplankton increase due to photosynthesis and are lost via respiration, death, and settling. The general model equation for attached algae and macrophytes can be expressed as follows:
(d)
where A is the phytoplankton biomass or concentration (dry weight biomass, chlorophyll, or equivalent mass of carbon, nitrogen, or phosphorus), mass or mass/volume, μ is the gross growth rate (T−1), r is the respiration rate (T−1), ex is the excretion rate (T−1), s is the settling rate (T−1), m is the non-predatory mortality (or decomposition) rate (T−1), and G is the loss rate due to grazing (M/T).

Model segmentation

Segments are the units that the reservoir is split for which the input data is provided for each segment and results are generated segmentally. The transit rates of water quality components are evaluated at the junction of neighbouring segments. Krishnagiri Reservoir is split into six one-dimensional surface segments (Figure 2), with segments 6 and 5 (S6 and S5) representing the inflow zone, segments 4 and 3 (S4 and S3) representing the middle portion of the reservoir, and segments 2 and 1 (S2 and S1) representing the zone near the dam structure. Preliminary sample analysis at all segments reveals that concentrations of the parameters showed less variation within the segments in the inflow (S6 and S5), middle (S4 and S3), and near dam structure (S2 and S1). Hence, segments six, three, and one are selected for further analysis. The kinematic wave flow is implemented from S6 to S2 to provide a realistic simulation of flow dynamics in one-dimensional transport controlled by bottom slope and bottom roughness. Ponded flow is implemented along with the kinematic flow for S1 controlled by the height of the dam structure.
Figure 2

Horizontal one-dimensional segmentation of Krishnagiri Reservoir for WASP model (segment 6 corresponds to the inflow region, segment 3 corresponds to the middle section, and segment 1 corresponds to the zone near the dam structure).

Figure 2

Horizontal one-dimensional segmentation of Krishnagiri Reservoir for WASP model (segment 6 corresponds to the inflow region, segment 3 corresponds to the middle section, and segment 1 corresponds to the zone near the dam structure).

Close modal

The geometry of reservoir segments

The reservoir geometry information is derived from the bathymetry study carried out in 2012 (Arunbabu et al. 2014). The depth information for each segment is collected from the bathymetry data from which the length, breadth, and associated volume of each segment were computed in the QGIS environment. The slope of each segment is computed using the reservoir's longitudinal profile from the bathymetry data. The continuity equation is used to calculate the flow velocity (Table 1) for each section:
where A1 is the area of segment 1 and V1 is the corresponding flow velocity of the segment.
Table 1

Reservoir geometry details used for the model segments

SegmentVolume (Mm3)Length (m)Breadth (m)Depth (m)Average velocity (m/s)Slope
S6 0.55 1,452 766 0.5 0.2064 0.00048 
S5 3.39 3,175 883 1.21 0.028 0.00137 
S4 7.871 3,172 880 2.82 0.012 0.00320 
S3 9.171 2,990 676 4.537 0.008 0.00671 
S2 9.45 1,774 932 5.72 0.010 0.00610 
S1 4.31 1,074 696 5.767 0.017 – 
SegmentVolume (Mm3)Length (m)Breadth (m)Depth (m)Average velocity (m/s)Slope
S6 0.55 1,452 766 0.5 0.2064 0.00048 
S5 3.39 3,175 883 1.21 0.028 0.00137 
S4 7.871 3,172 880 2.82 0.012 0.00320 
S3 9.171 2,990 676 4.537 0.008 0.00671 
S2 9.45 1,774 932 5.72 0.010 0.00610 
S1 4.31 1,074 696 5.767 0.017 – 

For eutrophication management in the Krishnagiri Reservoir, the EUTRO sub-model has been selected to simulate the water quality state variables. WASP requires datasets such as (i) time series data of meteorological parameters, (ii) inflow data, (iii) initial and boundary conditions of water quality parameters, (iv) environmental constant, and (v) channel geometry of reservoir segment.

Water quality data

DO, nitrate, phosphate, and chlorophyll-a are the water quality parameters modelled in WASP. The historical information on the KRP reservoir's water quality parameters from 2015 to 2020, as reported by researchers (Table 2) (Ambujam & Sudha 2016; Elangovan & Murali 2020; Abdul Wahid & Arunbabu 2022), was compiled into a database in the wet chemistry laboratory of the Centre for Water Resources and made available for model set-up and calibration. From 2021 to 2022, a field programme was set up to collect data monthly frequency on the water quality of the KRP reservoir. Chlorophyll sondes and multiparameter probes were employed to make in-situ measurements on chl-a and DO. Surface water samples were collected, tested in a laboratory for nitrate and phosphate, and used to validate the model.

Table 2

WASP Eutro-model state variables data sources for KRP

Model set-upDataset periodWork from Wet Chemistry Laboratory CWRState variables simulated
Initial conditions 2015 Ambujam & Sudha (2016)  DO, chl-a, nitrate, and phosphate 
Boundary condition and calibration 2016 
2018 Elangovan & Murali (2020)  
2019 Abdul Wahid & Arunbabu (2022)  
2020 
Validation 2021 In-situ measurements and sample testing in the laboratory 
2022 
Model set-upDataset periodWork from Wet Chemistry Laboratory CWRState variables simulated
Initial conditions 2015 Ambujam & Sudha (2016)  DO, chl-a, nitrate, and phosphate 
Boundary condition and calibration 2016 
2018 Elangovan & Murali (2020)  
2019 Abdul Wahid & Arunbabu (2022)  
2020 
Validation 2021 In-situ measurements and sample testing in the laboratory 
2022 

Meteorological data

Eutro-model in WASP requires temperature, precipitation, wind speed, relative humidity, and solar radiation to characterize the weather conditions of the study region. Solar radiation was acquired from NASA's POWER Data Access View while meteorological data such as temperature, precipitation, wind speed, and relative humidity were obtained from the State Surface and Ground Data Centre, Chennai.

Inflow data

The transport of the pollutant is dependent of the inflow into the reservoir. The inflow data was obtained from the Krishnagiri Reservoir, Water Resources Department.

Environmental constants

Each water quality metric has a predefined set of constants from which appropriate, and constants have to be chosen. For moderately deep to deep channels with heights 0–10 m and stream velocity 0.15–0.5 m/s, WASP simulates DO with a reaeration coefficient derived using O'Connor & Dobbins (1958) with depth-averaged values ranging from 0 to 10 per day. Temperature coefficient values between 1.022 and 1.024 are commonly employed in most modelling applications to represent the effects of temperature on reaeration (Downing & Truesdale 2007). Abiotic and biotic nutrients are both present in the aquatic system in different ways. The abiotic nutrient concentrations of dissolved inorganic nutrient NO3, dissolved organic nutrient PO4, and biotic nutrient phytoplankton with chlorophyll as its algal unit are all simulated using the WASP model. To model nutrients, Thomann & Fitzpatrick's (1982) equations are used to fix the denitrification and phosphorus transformation rate coefficient values. Phytoplankton dynamics were modelled using a maximum growth rate constant of 0.05 per day at 20 °C and a growth temperature coefficient that ranged from 1.02 to 1.1. Additional environmental constants, such as the phosphorus-to-carbon ratio, nitrogen-to-carbon ratio, carbon and chlorophyll ratio, nitrogen-to-chlorophyll ratio, and saturating light intensity, are needed to model the dynamics of phytoplankton (O'Connor & Dobbins 1958). The model is calibrated by adjusting the constants using the trial-and-error approach to acquire site-specific constants for the model. The environment constants used in the present study are presented in Table 4.

Model performance indicators

WASP model performance was evaluated using the Nash–Sutcliffe efficiency (NSE) and coefficient of determination (R2). NSE is used to verify the correctness of the model in comparison with the observed data. It has a range between −∞ and 1.0 (1 inclusive). In general, values between 0.0 and 1.0 are regarded as acceptable performance levels, while values less than 0 imply that the mean observed value is a better predictor than the simulated value, which denotes unacceptable performance (Yang et al. 2018). It is given by Equation (2):
(2)
where is the mean of observed values, M is the model value, O is the observed value at time t, and n is the number of observations.
The degree of collinearity between simulated and measured data is described by the coefficient of determination (R2) given by Equation (3):
(3)
where is the model simulated values, is the mean model simulated value, is the field-measured values, and is the mean field-measured value.

The degree of linear association between observed and simulated data is indicated by the correlation coefficient, which has a range of −1 to +1. There is no linear connection if r is equal to zero. A complete positive or negative linear connection is obtained when r is −1 or 1, respectively (Moriasi et al. 2007). Similarly, R2 represents the fraction of the variation in measured data that the model explains. R2 has a value between 0 and 1, with higher values suggesting less error variation; normally, values above 0.5 are regarded as acceptable (Shalabh & Dhar 2021).

Eutrophication control scenarios

Various scenarios have been considered to control the eutrophication problem in the reservoir. Nutrients enter water bodies via point and non-point source pollution. Scenarios are defined in a way to control this pollution by adapting to better management practices and controlling the loads entering the reservoir. The scenarios for which the model has been run are defined below.

Restricting livestock access to the reservoir, improving riparian buffer zones, and establishing built wetlands are the best management strategies for addressing pollution from both point and non-point sources (Zamparas & Zacharias 2014). To limit sediment yield, stone barriers, field bunding, mulching, and bio-fencing can be used in the catchment region (Arunbabu et al. 2014). Table 3 depicts situations that combine several best management techniques to minimize load from both point and non-point sources.

Table 3

Scenarios considering Best Management Practices (BMPs) for point and non-point sources

ScenarioStock access to the reservoir (1)Riparian buffer zone (2)Wetlands (3)Stonewall, field bunds, mulching, bio-fencing (4)Percentage of organic farming in the catchment area (5)Nutrient loading in percent (1 + 2 + 3 + 4 + 5)
Not restricted Not enhanced Not constructed Not practised 100 
Not restricted Not enhanced Not constructed Practised 25 75 
Restricted Enhanced Constructed Not practised 50 50 
Restricted Enhanced Constructed Practised 75 25 
ScenarioStock access to the reservoir (1)Riparian buffer zone (2)Wetlands (3)Stonewall, field bunds, mulching, bio-fencing (4)Percentage of organic farming in the catchment area (5)Nutrient loading in percent (1 + 2 + 3 + 4 + 5)
Not restricted Not enhanced Not constructed Not practised 100 
Not restricted Not enhanced Not constructed Practised 25 75 
Restricted Enhanced Constructed Not practised 50 50 
Restricted Enhanced Constructed Practised 75 25 
Table 4

Environmental constants used in the WASP model

ConstantsMinMaxTrial 1Trial 2Trial 3 (value fitted)
Phytoplankton maximum growth rate constant @20 °C (1/day) 0.01 0.05 
Phytoplankton growth temperature coefficient 1.07 0.02 0.5 
Phytoplankton carbon to chlorophyll ratio (mg C/mg Chl) 200 12 24 50 
Optimal temperature for growth (C) 16 20 
Shape parameter for below optimal temperatures 0.001 0.025 0.05 
Shape parameter for above optimal temperatures 0.002 0.03 0.05 
Phytoplankton respiration rate constant @20 °C (1/day) 0.5 0.01 0.05 0.1 
Phytoplankton respiration temperature coefficient 1.08 0.01 0.5 0.1 
Phytoplankton death rate constant (non-zoo predation) (1/day) 1.07 0.5 1.07 
Phytoplankton death rate due to salinity toxicity (1/day) 0.5 0.8 
Salinity at which algal mortality is half maximum value (g/L) 100 0.2 0.4 
Phytoplankton zooplankton grazing rate constant (1/day) 30 12 18 30 
Grazability (0 to 1) 0.5 0.8 
Nitrogen fixation option (0 = no, 1 = yes) 0.4 0.9 
Phytoplankton optimal light saturation as PAR (Watts/m2350 120 230 300 
Phytoplankton half-sat. for mineralization rate (mg Phyt C/L) 1.5 × 10−6 1.0 × 10−6 
Phytoplankton half-saturation constant for N uptake (mg N/L) 0.05 1.0 × 10−6 1.0 × 10−6 
Phytoplankton half-saturation constant for P uptake (mg P/L) 0.05 0.01 0.015 0.02 
Phytoplankton half-saturation constant for Si uptake (mg Si/L) 0.2 0.02 0.04 
Phytoplankton nitrogen to carbon ratio (mg N/mg C) 0.43 0.15 0.2 0.18 
Phytoplankton phosphorus to carbon ratio (mg P/mg C) 0.24 0.02 0.05 
Phytoplankton silica to carbon ratio (mg Si/mg C) 0.3 0.5 0.8 
Global reaeration rate constant @20 °C (1/day) 10 0.26 
Calc Reaeration Option (0 = Covar, 1 = O'Connor, 2 = Owens, 3 = Churchill, 4 = Tsivoglou) 
Minimum reaeration velocity (m/day) 24   
Maximum allowable calculated reaeration rate per day 100 24 24 24 
Use the total depth of water column for reaeration 0.5 
Waterbody type used for wind-driven reaeration rate 
Elevation above sea level (m) 15,000 492 492 492 
Oxygen to carbon stoichiometric ratio 2.67 2.5 2.55 2.67 
Theta – reaeration temperature correction 1.03 1.01 1.024 
Theta – SOD temperature correction 1.1 0.6 
Denitrification rate constant @20 °C (1/day) 0.09 0.08 0.09 
Denitrification temperature coefficient 1.04 1.04 
Half-saturation constant for denitrification oxygen limit (mg O2/L) 
Dissolved organic phosphorus mineralization rate constant @20 °C (1/day) 0.22 0.1 0.19 
Dissolved organic phosphorus mineralization temperature coefficient 1.08 
ConstantsMinMaxTrial 1Trial 2Trial 3 (value fitted)
Phytoplankton maximum growth rate constant @20 °C (1/day) 0.01 0.05 
Phytoplankton growth temperature coefficient 1.07 0.02 0.5 
Phytoplankton carbon to chlorophyll ratio (mg C/mg Chl) 200 12 24 50 
Optimal temperature for growth (C) 16 20 
Shape parameter for below optimal temperatures 0.001 0.025 0.05 
Shape parameter for above optimal temperatures 0.002 0.03 0.05 
Phytoplankton respiration rate constant @20 °C (1/day) 0.5 0.01 0.05 0.1 
Phytoplankton respiration temperature coefficient 1.08 0.01 0.5 0.1 
Phytoplankton death rate constant (non-zoo predation) (1/day) 1.07 0.5 1.07 
Phytoplankton death rate due to salinity toxicity (1/day) 0.5 0.8 
Salinity at which algal mortality is half maximum value (g/L) 100 0.2 0.4 
Phytoplankton zooplankton grazing rate constant (1/day) 30 12 18 30 
Grazability (0 to 1) 0.5 0.8 
Nitrogen fixation option (0 = no, 1 = yes) 0.4 0.9 
Phytoplankton optimal light saturation as PAR (Watts/m2350 120 230 300 
Phytoplankton half-sat. for mineralization rate (mg Phyt C/L) 1.5 × 10−6 1.0 × 10−6 
Phytoplankton half-saturation constant for N uptake (mg N/L) 0.05 1.0 × 10−6 1.0 × 10−6 
Phytoplankton half-saturation constant for P uptake (mg P/L) 0.05 0.01 0.015 0.02 
Phytoplankton half-saturation constant for Si uptake (mg Si/L) 0.2 0.02 0.04 
Phytoplankton nitrogen to carbon ratio (mg N/mg C) 0.43 0.15 0.2 0.18 
Phytoplankton phosphorus to carbon ratio (mg P/mg C) 0.24 0.02 0.05 
Phytoplankton silica to carbon ratio (mg Si/mg C) 0.3 0.5 0.8 
Global reaeration rate constant @20 °C (1/day) 10 0.26 
Calc Reaeration Option (0 = Covar, 1 = O'Connor, 2 = Owens, 3 = Churchill, 4 = Tsivoglou) 
Minimum reaeration velocity (m/day) 24   
Maximum allowable calculated reaeration rate per day 100 24 24 24 
Use the total depth of water column for reaeration 0.5 
Waterbody type used for wind-driven reaeration rate 
Elevation above sea level (m) 15,000 492 492 492 
Oxygen to carbon stoichiometric ratio 2.67 2.5 2.55 2.67 
Theta – reaeration temperature correction 1.03 1.01 1.024 
Theta – SOD temperature correction 1.1 0.6 
Denitrification rate constant @20 °C (1/day) 0.09 0.08 0.09 
Denitrification temperature coefficient 1.04 1.04 
Half-saturation constant for denitrification oxygen limit (mg O2/L) 
Dissolved organic phosphorus mineralization rate constant @20 °C (1/day) 0.22 0.1 0.19 
Dissolved organic phosphorus mineralization temperature coefficient 1.08 

Scenario 1

Scenario 1 depicts that there is no management practice carried out to control the loads entering the reservoir. Farmers use commercial fertilizers and no other alternative practices like organic farming or the construction of check dams to control sediment yield. The load entering the reservoir is 100%, without any conservation or management practices.

Scenario 2

In this scenario, the load entering the reservoir is reduced by 25%, due to the management practices mentioned above to reduce the point and non-point source pollution. A certain number of farmers in the upper catchment are also opting for organic farming which also helps in controlling the nutrients entering the reservoir. The amount of load entering the reservoir is 75%.

Scenario 3

In this scenario, the load entering the reservoir is reduced by 50%, due to the increase in the number of farmers practising organic farming compared to Scenario 2 which also helps in controlling the nutrients entering the reservoir. The amount of load entering the reservoir is 50%.

Scenario 4

In this scenario, the load entering the reservoir is reduced by 75%, if the majority of the farmers opted for organic farming, which helped in reducing the nutrients coming into the reservoir. The amount of load entering the reservoir is 25%, which is reduced due to the management practices and if a wastewater treatment system is also implemented to reduce the effluent load.

The four scenarios used in the present study are presented in Table 3 and the overall workflow of the study is presented in Figure 3.
Figure 3

Workflow for eutrophication management through the WASP model in Krishnagiri Reservoir Project.

Figure 3

Workflow for eutrophication management through the WASP model in Krishnagiri Reservoir Project.

Close modal

Model calibration and validation

The WASP model was calibrated for the water quality parameters such as DO, nitrate, phosphate, and chl-a. The model period has been set from 1/1/2015 to 31/12/2022 with daily time steps for simulation. Measured water quality data from the KRP reservoir from 2015 to 2020 was used for calibration and the measured data from 2021 to 2022 were used for validating the model. The environmental constants and global kinetic constants for each parameter were adjusted to match the measured data during the calibration process (Table 4).

The model performance is evaluated using the coefficient of determination (R2) and NSE (Table 5). The time series graph between the model simulated and observed levels of DO (Figure 4), chl-a (Figure 5), nitrate (Figure 6), and phosphate (Figure 7) indicates a good match between the observed concentration and simulated concentration of key indicators. NSE values for the four key indicators (DO, chl-a, N, P) were greater than 0.6 during both the calibration and the validation periods. The model performance in the present study based on NSE for both the calibration and validation periods can be rated as ‘very good’ (Moriasi et al. 2007) for DO, N, and P, whereas the model performance is rated as ‘satisfactory’ for chl-a during the calibration process and ‘very good’ during validation. Similarly, the R2 value for the four key indicators (DO, chl-a, N, P) was greater than 0.75 during both the calibration and the validation periods.
Table 5

Summary of model performance during calibration and validation

Model parameterCalibration
Validation
R2NSER2NSE
DO 0.794 0.900 0.904 0.860 
Chl-a 0.770 0.636 0.774 0.712 
Nitrate 0.894 0.909 0.890 0.891 
Phosphate 0.826 0.864 0.843 0.760 
Model parameterCalibration
Validation
R2NSER2NSE
DO 0.794 0.900 0.904 0.860 
Chl-a 0.770 0.636 0.774 0.712 
Nitrate 0.894 0.909 0.890 0.891 
Phosphate 0.826 0.864 0.843 0.760 
Figure 4

Model simulated time series graph from 2015 to 2022 with measured values for state variable DO.

Figure 4

Model simulated time series graph from 2015 to 2022 with measured values for state variable DO.

Close modal
Figure 5

Model simulated time series graph from 2015 to 2022 with measured values for state variable chl-a.

Figure 5

Model simulated time series graph from 2015 to 2022 with measured values for state variable chl-a.

Close modal
Figure 6

Model simulated time series graph from 2015 to 2022 with measured values for state variable nitrate.

Figure 6

Model simulated time series graph from 2015 to 2022 with measured values for state variable nitrate.

Close modal
Figure 7

Model simulated time series graph from 2015 to 2022 with measured values for state variable phosphate.

Figure 7

Model simulated time series graph from 2015 to 2022 with measured values for state variable phosphate.

Close modal

Dissolved oxygen

The highest levels of DO detected in KRP during winter was 8.3 mg/L, and in summer, it was 8.2 mg/L whereas the DO during the SEM was 7.2 mg/L and during the NEM was 7.4 mg/L (Figure 8). Summertime high levels of DO are caused by higher grazing and remineralization rates caused by hot temperatures. This mechanism provides nutrients for algae development while also boosting DO during the summer months (Ansari & Gill 2014). DO levels are greater near the dam structure (S1) and in the middle zone of KRP (S3) than in the inflow zone (S6) (Figure 9). The high level of DO near the dam structure is due to the eddy formation and enhanced aeration caused by the operation of the river sluice and canal outlet used for the release of irrigation water. Active organic matter decomposition occurs at deeper depths near the dam structure where the water column is higher, whereas breakdown occurs at the surface level in shallow depths. The inflow zone is shallow, and breakdown occurs at the surface level, resulting in lower DO (Yan et al. 2013). Segment 6 is the KRP's littoral zone, which is populated by grasses and shrubs on land. During the rainy season, this area is submerged, supplying the reservoir with a plentiful supply of organic materials. The anoxic conditions in the reservoir can be caused by the microbial biodegradation of phytoplankton biomass and moist substances, which can drastically lower DO concentrations. Despite being the input point that gets large nutrient loads, segment 6's modest depth and natural shape lead to a drop in DO. Eutrophication and reaeration have a significant impact on segment 1 and segment 3 DO level fluctuations, but decomposition has less of an impact. While it is true that segment 6's chl-a levels are greater year-round (Figure 11), active decomposition may be preventing this segment's chl-a from contributing to segment 6's ability to increase the DO value through photosynthesis (Figure 10). Again, this is due to the geometrical characteristics and shallow depth of the KRP reservoir.
Figure 8

Seasonal average of DO concentrations from 2015 to 2022 for S6, S3, and S1.

Figure 8

Seasonal average of DO concentrations from 2015 to 2022 for S6, S3, and S1.

Close modal
Figure 9

Model simulated DO concentrations from 2015 to 2022 for S6, S3, and S1.

Figure 9

Model simulated DO concentrations from 2015 to 2022 for S6, S3, and S1.

Close modal
Figure 10

Average monthly variation of DO for S6, S3, and S1.

Figure 10

Average monthly variation of DO for S6, S3, and S1.

Close modal
Figure 11

Seasonal average of chl-a concentrations from 2015–2022 for S6, S3, and S1.

Figure 11

Seasonal average of chl-a concentrations from 2015–2022 for S6, S3, and S1.

Close modal

Chlorophyll-a

Chl-a is more concentrated at the inflow region (segment 6) of the KRP, as evidenced by the concentration data from the three segments (Figure 11). Also, it is clear from Figure 11 that inflow segment 6 in all three seasons has predominant quantities of chl-a where concentrations of the nutrients nitrate (Figure 13) and phosphate (Figure 14) are greater than other segments. The availability of nutrients in the reservoir affects the quantity of chl-a (Wang et al. 2015). A greater level of chl-a is seen in the reservoir during the monsoon and winter months due to the addition of nutrients to the inflow. The greatest concentration of chl-a was discovered to be in the winter (38.4 μg/L), while the second-highest concentration was found in the NEM (36.7 μg/L) (Figure 11). Nutrients have an impact on the development and operation of all living organisms, including phytoplankton. The concentration of chl-a is influenced by the nutrient contents of phosphorus (P) and nitrogen (N), either jointly or separately (Trommer et al. 2013). It is evident from the KRP's spatiotemporal study that chl-a, nutrients, and the influx into the reservoir are related. The reservoir's trophic status may be determined using the chl-a (Carlson 1977). From the analysis, it could be inferred that the Krishnagiri Reservoir comes under hypereutrophic conditions as the highest value of chl-a observed is 57.1 μg/L in September (Figure 12).
Figure 12

Average monthly variation of chl-a concentrations for S6, S3, and S1.

Figure 12

Average monthly variation of chl-a concentrations for S6, S3, and S1.

Close modal
Figure 13

Seasonal average of nitrate concentrations from 2015–2022 for S6, S3, and S1.

Figure 13

Seasonal average of nitrate concentrations from 2015–2022 for S6, S3, and S1.

Close modal
Figure 14

Average monthly variation of phosphate for S6, S3, and S1.

Figure 14

Average monthly variation of phosphate for S6, S3, and S1.

Close modal

Nitrate and phosphate

The two main nutrients that contribute to the enrichment of eutrophication processes are nitrate and phosphate. Figures 13 and 14 show the concentrations of these nutrients nitrate and phosphate, which are said to be higher in the NEM and the winter than at other times of the year. Nitrate and phosphate monthly averages in segments 6 (Figure 15) and 1 (Figure 16) show spatial fluctuation, indicating that the reservoir's contents were low during periods of low input and high during periods of strong inflow. The observed pattern shows an exact correlation between the inflow into the reservoir and the nutrient load.
Figure 15

Inflow into the KRP reservoir, nitrate and phosphate concentration for segment 1.

Figure 15

Inflow into the KRP reservoir, nitrate and phosphate concentration for segment 1.

Close modal
Figure 16

Inflow into the KRP reservoir, nitrate and phosphate concentration for segment 6.

Figure 16

Inflow into the KRP reservoir, nitrate and phosphate concentration for segment 6.

Close modal
Under the influence of the altered input volume and nutrient content, a notable shift in chl-a concentration was also noticed. The concentration of chl-a is observed to be at its highest during monsoon months with a high inflow volume (September 57.1 μg/L, October 57 μg/L) (Figure 12). The waterbody's chl-a content rises with substantial nutrient enrichment throughout the monsoon and winter seasons and eventually declines with decreased inflow (Figure 17). It is evident from the in-depth temporal study that there is a direct association between inflow and chl-a, inflow and nutrient, as well as nutrient and chl-a. Contrary to DO, chl-a in the inflow point (segment 6) is much greater than in the midsection (segment 3) and near the dam construction (segment 1) regions (Figure 11). This is due to greater concentrations of nitrate (Figure 13) and phosphate (Figure 14) in the inflow zone (segment 6) and lower concentrations in the intermediate region (segment 3) and near the dam structure (segment 1). Eddy creation and improved aeration caused by the sluice gate's operation, as well as the short retention time, are the causes for reduced bloom proliferation, and therefore, the concentration of chl-a was found to be lower near the dam structure (segment 1). Whereas in segment 6, the geometry of the reservoir (shallow depth and wide area) naturally traps a significant portion of the nutrients like nitrate and phosphate along with the sediments and finds a greater retention period on either side of the inflow stream, leading to higher concentrations of chl-a.
Figure 17

Chl-a, nitrate, and phosphate concentration for segment 1, segment 3, and segment 6.

Figure 17

Chl-a, nitrate, and phosphate concentration for segment 1, segment 3, and segment 6.

Close modal

Model simulated scenarios for eutrophication management

The WASP model's extensive spatiotemporal analysis of water quality metrics such as DO, chl--a, nitrate, and phosphate showed that nutrient content entering the reservoir serves as a source of eutrophication. Eutrophication at the surface causes the active breakdown of organic materials, which influences the benthic zone's DO depletion and improves the surface DO through photosynthesis and aeration. While nutrient loading is the ineluctable cause of the process of water quality degradation, this study proposes the creation of control scenarios for eutrophication management under various loading situations.

The WASP model was run under four distinct scenarios for nitrate and phosphate at 100, 75, 50, and 25% loads entering the reservoir. The corresponding DO and chl-a were simulated for each load. The model simulated results for the decreasing loading scenario depict an increase in DO (Figure 18) and a decrease in chl-a (Figure 19). With a decreasing load scenario, the increasing rate of DO (1.47%) is less than the decreasing rate of chl-a (43.3%) clears that there is a substantial association of nutrient concentration with eutrophication.
Figure 18

The model simulated DO concentrations for different scenarios.

Figure 18

The model simulated DO concentrations for different scenarios.

Close modal
Figure 19

The model simulated chl-a concentrations for different scenarios.

Figure 19

The model simulated chl-a concentrations for different scenarios.

Close modal

The breakdown of organic matter from the benthic to the surface, which depletes DO, and reaeration during photosynthetic activity both have an impact on the processes of nutrient loading and eutrophication.

The average chl-a value was found to be 39 g/L in scenario 2, which still maintains the reservoir in the hypereutrophic condition (>22 μg/L) (Carlson 1977), even when watershed conservation activities are successfully implemented to prevent erosion in the catchment region. Under scenario 3 (25 μg/L), there may be a modest shift from a hypereutrophic to a eutrophic condition when the livestock access to the reservoir is controlled and a wetland is created at the inflow zone. However, scenario 4 with stonewalls, mulching, bio-fence, and field bunds used in the agricultural fields in the catchment together with the riparian buffer zone, limited livestock access, and constructed wetlands leads to a reservoir that is a near mesotrophic state as the average chl-a concentration is lowered to 6 μg/L. According to research on water quality forecasting conducted at the Krishnagiri Reservoir (Abdul Wahid & Arunbabu 2022), the reservoir is degrading quickly, and it is predicted that the water quality will deteriorate to a level which cannot be used for agriculture after 10 years. In general, a rising trend with DO and a lowering trend with chl-a are shown when the native nutrient load is restricted to 75%. For effective management of the KRP reservoir using WASP, scenario 4 with optimal management techniques to control point and non-point sources is recommended.

The investigation of the water quality metrics showed that the current eutrophic condition of the reservoir is what is changing the DO value. The nutritional load and chl-a value both grow in response to an increase in intake. The average concentration of water quality parameters for the years 2015 to 2022 is relatively high using the USEPA criteria (0.1 mg/L for phosphate, 10 mg/L for nitrate) to limit eutrophication. There is a direct correlation between the high chl-a level and the high phosphate and nitrate levels in the reservoir. The reservoir's average chl-a level is over 22 μg/L, which denotes hypereutrophic conditions that harm the water quality and kill fish. The WASP water quality model successfully predicted the water quality variables and offered the ideal loading scenario for the eutrophication issue. Out of all the scenarios put out, scenario 4 with 75% load reduction, resulted in the appropriate concentrations of DO and chl-a. The dissolved oxygen levels were found to be relatively greater than the other loading situations, and the chl-a concentrations were reduced by 88.2%. The construction of check dams in the appropriate location along the river lengths and a gradual transition to organic farming are the watershed management strategies suggested in the current study. As a result, the nutrient loads (nitrate and phosphate) entering the reservoir will be reduced, which will raise the reservoir's DO levels. Through this work, it has been shown that the WASP model is a reliable tool for forecasting monthly water quality variables and that it can successfully simulate eutrophication control factors. This information aids environmental managers and policymakers in developing better strategies for managing and monitoring water quality.

The authors thank the Officials of the Water Resources Department and Fisheries Department at KRP dam for their support during the field measurement campaign. The authors also acknowledge the technical support rendered by Dr Sudha in the Wet Chemistry Laboratory, Centre for Water Resources, Anna University.

A.B.E. conceived the original idea of the project and planned the experiments. D.V.K.C. and A.W.A. carried out the field campaign, collected water quality samples, and performed laboratory tests. A.K.J. carried out data curation and reviewed the draft manuscript. P.N. also performed data curation. D.V.K.C. performed WASP simulations. A.W.A. helped in the interpretation of the results. All authors contributed to writing the manuscript. A.B.E. provided critical feedback and helped shape the research, analysis, and manuscript.

All relevant data are included in the paper.

The authors declare there is no conflict.

Akomeah
E.
,
Davies
J.-M.
&
Lindenschmidt
K.-E.
2020
Water quality modeling of phytoplankton and nutrient cycles of a complex cold-region river-lake system
.
Environmental Modeling & Assessment
25
(
3
),
293
306
.
Albert Moses
S.
,
Janaki
L.
,
Joseph
S.
&
Joseph
J.
2016
Evaluation of performance of water quality models for a tropical lake system
.
Lakes & Reservoirs: Research & Management
21
(
2
),
82
93
.
Ambujam
N. K.
&
Sudha
V.
2016
Environmental Impacts of Point and Non-Point Source Pollution in Krishnagiri Reservoir: A Case Study in South India
.
Available from: https://zenodo.org/record/1123903 (accessed September 4, 2022)
.
Ansari
A. A.
&
Gill
S. S.
2014
Eutrophication: Causes, Consequences and Control
.
Springer
,
Dordrecht
,
Netherlands
.
Available from: http://link.springer.com/10.1007/978-94-007-7814-6 (accessed October 26, 2022)
.
Arunbabu
E.
,
Ravichandran
S.
&
Sreeja
P.
2014
Sedimentation and internal phosphorus loads in Krishnagiri Reservoir, India
.
Lakes & Reservoirs: Research & Management
19
(
3
),
161
173
.
Carlson
R. E.
1977
A trophic state index for lakes 1: Trophic state index
.
Limnology and Oceanography
22
(
2
),
361
369
.
Downing
A. L.
&
Truesdale
G. A.
2007
Some factors affecting the rate of solution of oxygen in water
.
Journal of Applied Chemistry
5
(
10
),
570
581
.
Elangovan
A.
&
Murali
V.
2020
Mapping the chlorophyll-a concentrations in hypereutrophic Krishnagiri Reservoir (India) using Landsat 8 Operational Land Imager
.
Lakes & Reservoirs: Science, Policy and Management for Sustainable Use
25
(
4
),
377
387
.
Mamani Larico
A. J.
&
Zúñiga Medina
S. A.
2019
Application of WASP model for assessment of water quality for eutrophication control for a reservoir in the Peruvian Andes
.
Lakes & Reservoirs: Science, Policy and Management for Sustainable Use
24
(
1
),
37
47
.
Mateus
M.
,
Vieira
R.
,
Almeida
C.
,
Silva
M.
&
Reis
F.
2018
ScoRE – a simple approach to select a water quality model
.
Water
10
(
12
),
1811
.
Moriasi
D. N.
,
Arnold
J. G.
,
Van Liew
M. W.
,
Bingner
R. L.
,
Harmel
R. D.
&
Veith
T. L.
2007
Model evaluation guidelines for systematic quantification of accuracy in watershed simulations
.
Transactions of the ASABE
50
(
3
),
885
900
.
Moses
S. A.
,
Janaki
L.
,
Joseph
S.
&
Joseph
J.
2015
Water quality prediction capabilities of WASP model for a tropical lake system
.
Lakes & Reservoirs: Science, Policy and Management for Sustainable Use
20
(
4
),
285
299
.
Narasimhan
B.
,
Srinivasan
R.
,
Bednarz
S. T.
,
Ernst
M. R.
&
Allen
P. M.
2010
A comprehensive modeling approach for reservoir water quality assessment and management due to point and nonpoint source pollution
.
Transactions of the ASABE
53
(
5
),
1605
1617
.
O'Connor
D. J.
&
Dobbins
W. E.
1958
Mechanism of reaeration in natural streams
.
Transactions of the American Society of Civil Engineers
123
(
1
),
641
666
.
Park
J. Y.
,
Park
G. A.
&
Kim
S. J.
2013
Assessment of future climate change impact on water quality of Chungju Lake, South Korea, using WASP coupled with SWAT
.
JAWRA Journal of the American Water Resources Association
49
(
6
),
1225
1238
.
Sasitharan
S.
&
Arunbabu
E.
2020
Sediment and nutrient export modeling in a subwatershed of Krishnagiri Reservoir Project
.
Pollution Research
39
(
2
),
385
392
.
Shabani
A.
,
Woznicki
S. A.
,
Mehaffey
M.
,
Butcher
J.
,
Wool
T. A.
&
Whung
P.
2021
A coupled hydrodynamic (HEC-RAS 2D) and water quality model (WASP) for simulating flood-induced soil, sediment, and contaminant transport
.
Journal of Flood Risk Management
14
(
4
).
Available from: https://onlinelibrary.wiley.com/doi/10.1111/jfr3.12747 (accessed March 31, 2022)
.
Shalabh
&
Dhar
S. S.
2021
Goodness of fit in nonparametric regression modelling
.
Journal of Statistical Theory and Practice
15
(
1
),
18
.
https://doi.org/10.1007/s42519-020-00148-x
.
Thomann, R. & Fitzpatrick, J. 1982 Calibration and verification of a mathematical model of the eutrophication of the Potomac Estuary. (Technical Report) 10.13140/2.1.2533.1526.
Trommer
G.
,
Leynaert
A.
,
Klein
C.
,
Naegelen
A.
&
Beker
B.
2013
Phytoplankton phosphorus limitation in a North Atlantic coastal ecosystem not predicted by nutrient load
.
Journal of Plankton Research
35
(
6
),
1207
1219
.
Wang
Y.
,
Jiang
H.
,
Jin
J.
,
Zhang
X.
,
Lu
X.
&
Wang
Y.
2015
Spatial-temporal variations of chlorophyll-a in the adjacent sea area of the Yangtze River Estuary influenced by Yangtze River discharge
.
International Journal of Environmental Research and Public Health
12
(
5
),
5420
5438
.
Wimordi
C.
,
Yudianto
D.
&
Guan
Y.
2021
Application of WASP model to simulate water pollution control of Duriangkang Dam
.
Lakes & Reservoirs: Science, Policy and Management for Sustainable Use
26
(
1
),
23
32
.
Wool
T. A.
&
Epa
U. S.
2015
WASP8 Multiple Algae – Model Theory and User's Guide, p. 20
.
Yang
M.
,
Xiao
W.
,
Zhao
Y.
,
Huang
Y.
,
Li
B.
,
Jiang
D.
,
Lu
F.
&
Hou
B.
2018
Impact of changes in land use and climate on the runoff in the coastal areas of South China – A case study of the Nanliujiang catchment
.
MATEC Web of Conferences
246
,
02001
.
Zamparas
M.
&
Zacharias
I.
2014
Restoration of eutrophic freshwater by managing internal nutrient loads. A review
.
Science of The Total Environment
496
,
551
562
.
This is an Open Access article distributed under the terms of the Creative Commons Attribution Licence (CC BY 4.0), which permits copying, adaptation and redistribution, provided the original work is properly cited (http://creativecommons.org/licenses/by/4.0/).