Assessment of StreamFlow Using SWAT Hydrological Model

Transcription

1 Assessment of StreamFlow Using SWAT Hydrological Model Sreelakshmi C.M. 1 Dr.K.Varija 2 1 M.Tech student, Applied Mechanics and Hydraulics, NITK, Surathkal,India 2 Associate Professor, Department of Applied Mechanics and Hydraulics, NITK, Surathkal,India ABSTRACT Estimation of stream flow in a catchment which is ungauged still poses a challenge to hydrologists. Many methods have been developed to find the discharge in ungauged basins. Hydrological models are one of the commonly used techniques for the discharge simulation in ungauged stations. In order to get the discharge at each outlet point in a catchment we can make use of the hydrological models. The objective of this study is to assess the stream flow and various hydrological processes in the Harangi catchment using Soil and Water Assessment Tool which is commonly known as SWAT hydrological model. SWAT is a semi-distributed physical model. The study is conducted on the catchment area of Harangi River which is a tributary of the river Cauvery. Apart from the rainfall data Digital Elevation Model (DEM), soil map and landuse map are the other major inputs required for running the model. Landuse map is prepared using ERDAS.The results after model calibration and validation showed a good agreement between observed and simulated stream flow. Sensitivity analysis was performed.cn2; GW_DELAY, ESCO, EPCO and SOL_K were the most sensitive parameters. Keywords: SWAT, Stream flow, Hydrological model, Simulation, Calibration, Sensitivity analysis, Validation 1. INTRODUCTION Water is emerging as one of the most critical natural resource issues facing humanity. It is the source of life on earth but also turns out to be source of catastrophes. As we progress into 21st century the world s population is expanding rapidly. Poor availability of hydrological studies leads to lack of planning and inadequate water resources management strategies in the catchment. Importance of efficient and sustainable water management has gained interest today in the scenario of rapid urbanization and development.so hydrologistscommonly use rainfall runoff models to extrapolate stream flow time series in time and space. A rainfall runoff (RR) model is a mathematical formulation describing the processes through which rainfall is converted into runoff from a catchment or a drainage basin. It is a representation of natural system. Soil and Water Assessment Tool (SWAT) is a most commonly used model for hydrological modeling which gives reliable results. The soil and water assessment tool (SWAT) is a physicallybased, continuous, long-term, semi distributed hydrologic model, developed by the U.S. Department of Agriculture (Neitsch et al. 2005; Zhang et al. 2008). It is a conceptual model that works both on daily and sub daily time steps (Arnold and Fohrer 2005). It can simulate surface and subsurface flow, soil erosion, nutrient cycling and transport, and sediment deposition, and has been applied worldwide in hydrologic and water quality modeling (Zhang et al. 2008). It has also been applied extensively over a wide range of spatial scales. Alejandra S. (2008) attempted to model the rainfall runoff processes in the Vergara River basin which is a sub basin of Biobío River in Chili with Root mean squared error (RRMSE); mean absolute error (ABSERR); the Nash Sutcliffemodeling efficiency index (EF); the coefficient of determination (R 2 ) and the % of deviation from observed stream flow (PBIAS) as the model performance indicators. The model was calibrated for four control points of which the observed data was available. The impact of landuse changes on River Nzoia catchment, Kenya was studied by Faith G. et. al. (2009) using SWAT. CLUE-S model was used in this study to obtain landuse change scenario. Zeyuan Qiu and Lizhong Wang (2014) used SWAT model for the simulation of stream flow and water quality assessment of Neshanic River watershed. Comparison between simulated and observed values of TSS, TN and TP showed a good correspondence between observed and simulated data. CH-K2, CN2 and ALPHA-BF were the most sensitive parameters in this study. Manish K. et. 566

2 al.(2014) simulated the hydrological process in Nuevo watershed and got NSE coefficient 0.76 for calibration period and 0.5 for validation period. Presence of karstic aquifer in the watershed resulted in underestimation of some of the peak flow events during calibration period. Water balance components of a humid tropical watershed was studied (Sandra George and SathianK.K., 2015) using SWAT model. The NSE and R 2 values obtained for both calibration and validation period was above This study also brought into light the importance of baseflow in the study area. The applicability of SWAT in Iranian mountainous region for prediction of stream flow and sediment yield was evaluated by RokhsareRostamian et al.(2008) for two basin. The aim of this study was to assess the applicability of SWAT model in Harangi river basin and to simulate the stream flow. 2. Description of the study area The river Harangi which is a tributary of the river Cauvery originates in the Pushpagiri hills in Coorg,(Kodagu) district. Coorg is one of the most popular hill station in India. Harangiriver flows in Karnataka through a length of 50 km and joins with Cauvery river near Kudige in the Somwarpettaluk in Kodagu district. Heavy rainfall from the south-west monsoon is received in the catchment area of Harangi. The study area is the catchment of Harangiriver taken at Kudige. The catchement area is km 2. The study area consists of a reservoir, Harangi reservoir which has a catchment area of km 2.Thick forests, paddy, coffee and pepper are the major vegetation in Harangi. The rainfall in the river basin varies from 127 cm (50 ) to 381 cm (150 ) per annum.harangi is having a moderate climate with an average temperature ranging from 11 to 28 C (52 to 82 F), with the highest temperatures occurring in April and May. The study area is given below. Figure : 1 Study Area 3. Model Description 3.1 SWAT model The Soil and Water Assessment Tool (SWAT) model is a conceptual, continuous, distributed and deterministic hydrologic model. It can evaluate the impact of management practices on water, sediment and 567

3 agricultural chemical yields in ungauged basins (Arnoldetal.,1998). SWAT model is computationally efficient and capable of continuous simulation over long time periods. The model s major components include weather, hydrology, erosion, soil temperature, plant growth, nutrients, pesticides, land management, channel and reservoir routing (Arnold et al., 1998).Water balance is the driving force behind all the processes in SWAT. Within the SWAT conceptual framework, the representation of the hydrology of a basin is divided into two major parts: (a) the land phase of the hydrological cycle; and (b) the routing of runoff through the river network. For modeling the land phase, the river basin is divided in sub-basins, each one of which is composed of one or several hydrological response units (HRUs) which are areas of relatively homogenous land use/land cover and soil types. The characteristics of the HRUs define the hydrological response of a sub-basin. For a given time step, the contributions to the discharge at each sub-basin outlet point is controlled by the HRU water balance calculations (land phase). The river network then connects then connects the different sub-basin outlets, and the routing phase determines movement of water through this network towards internal control points, and finally toward the basin outlet. 3.2 SWAT CUP SWAT Calibration and Uncertainty Programs (SWAT-CUP) is an interface that was developed for SWAT. This is a program for calibration of SWAT models. The. SWAT-CUP supports the program links SUFI2, PSO, GLUE, ParaSol, and MCMC procedures to SWAT. With this the user can do sensitivity analysis, calibration, validation, and uncertainty analysis of the models. In this study SUFI2 program is used for calibration and validation. 4. Data used SWAT requires landuse and soil typedata, and a digital elevation model (DEM) as input. SWAT also requires daily precipitation data, as well as daily maximum and minimum temperature data. A 30 m resolution Digital Elevation Model (DEM) was downloaded from for the Harangi river basin. The satellite image LANDSAT 7 is downloaded from U.S. Geological Survey(USGS).The downloaded image was processed and classified using ERDAS software. Soil map with a scale 1: was collected from NBSS & LUP (National Bureau of Soil Survey and Landuse Planning). Rainfall data was obtained from the District Statistical Office, Madikeri. Rainfall data for 6 stations for the years were used in the study. Data from Amathy, Harangi,Kudige,Somwarpet, Suntikoppa and Surlabbi rain gauge stations were used to perform the analysis. The daily maximum and minimum temperature for the Madikeri climatic station was obtained from Indian Meteorological Department. The daily stream flow data for Harangiriver at Kudige outlet was obtained from WRIS portal. 5. Results and discussion The data required for running the SWAT model was prepared according to the requirement. The model was run for the period A warm up period of 4 years was given ( ).After watershed delineation the whole catchment was divided into 29 sub basins and 236 HRUs. Calibration was done for the period 2005 to 2010 and were considered for validation process. The model performance was evaluated using the statistical indicators mentioned in the previous section. Sensitivity analysis Sensitivity analysis is the process of determining the rate of change in model output with respect to changes in model parameters (Arnold et. al, 2012). A parameter is considered sensitive when the change in that parameter causes a large change in output. With sensitivity analysis the key parameters required for model calibration can be identified. If the sensitive parameters are identified prior to calibration, it will helps to reduce the number of parameters that must be calibrated, which in turn helps in reducing the computational time required for model calibration Two types of sensitivity analysis can be performed using SWAT-CUP. They are Global sensitivity analysis and One at a time sensitivity analysis. One at a time sensitivity shows the sensitivity of a variable to the changes in a parameter if all other parameters are kept constant at some value. In this work 568

4 Simulated daily discharge (cumecs) Simulated daily discharge (cumecs) International Journal of Engineering Technology, Management and Applied Sciences global sensitivity analysis was used. The sensitive parameters were CN2, GW_DELAY, ESCO, EPCO and SOL_K. These parameters were changed in iteration for calibration and the SWAT model was run again with the new values of parameters. In this study the statistical indicators used for evaluating the model performance were coefficient of determination (R 2 ), and Nash-Sutcliffe efficiency (NSE).The values for these indicators were checked during calibration ( ) and validation ( ).The closer the values of R 2 and NS to unity the better the model performance. The value for NS was 0.79 and 0.76 after calibration and validation respectively. The value of R 2 was 0.83 and 0.8 after calibration and validation respectively. The model showed a good agreement with the observed data R² = Observed daily discharge (cumecs) Figure :2 Scatteredplot of observed and simulated discharge after calibration R² = Observed daily discharge (cumecs) Figure :3 Scattered plot of observed and simulated discharge after validation 569

5 6. CONCLUSION The present study examined the applicability of SWAT model in simulating stream flow in Harangibasin. Sensitivity analysis showed that CN2, GW_DELAY,ESCO,EPCO and SOL_K were the most sensitive parameters.calibration and validation results showed that there is a good agreement between observed and simulated discharge.according to the model performance indicator the model was good in simulating the stream flow of Harangi river basin taken at Kudige outlet. REFERENCES Manish K. et. al., Simulation of the Streamflow for the Rio Nuevo Watershed of Jamaica for Use in Agriculture Water Scarcity Planning, Journal of Irrigation and Drainage Engineering,ASCE,August Nagaraj S. et.al., Runoff Modeling for Bhima River using Swat Hydrological Model, International Journal of Engineering Research & Technology, Volume 3,Issue 7,July 2014 Hadi M. et al., SWAT-based hydrological modeling of tropical land-use scenarios,hydrological Sciences Journal, September 2014 Sandra George and SathianK..K, Simulation of water yield of a humid tropic watershed using swat model, International Journal of Remote Sensing & Geoscience, Volume 4, Issue 6, Nov ZeyuanQiu and Lizhong Wang HydrologicalandWaterQualityAssessmentinaSuburban Watershed with Mixed Land Uses Using the SWAT Model, Journal of Hydrologic Engineering, Vol. 19, No. 4, April 1, 2014.,ASCE Alejandra S., Hydrological modeling with SWAT under conditions of limited data availability: evaluation of results from a Chilean case study, Hydrological Sciences Journal, June 2008 Etienne Levesque et al., Evaluation of streamflow simulation by SWAT model for two small watersheds under snowmelt and rainfall, Hydrological Sciences Journal, October 2008 RokhsareRostamian et al., Application of a SWAT model for estimating runoff and sediment in two mountainous basins in central Iran, Hydrological Sciences Journal,October 2008 Fait G., Estimating the impacts of land-cover change on runoff using the soil and water assessment tool (SWAT): case study of Nzoia catchment, Hydrological Sciences Journal,October 2009 SangjunIm et al., Comparison of HSPF and SWAT models performance for runoff and sediment yield prediction, Journal of Environmental Science and Health Part A,October 2007 Applied Hydrology, A Textbook by VenTe Chow et. al. S.L.Neitsch et al., Soil and Water Assessment Theoretical Documentation Version 2009,September 2011 Engineering Hydrology, K.Subramanya 570