MODULE 8 LECTURE NOTES 2 REMOTE SENSING APPLICATIONS IN RAINFALL-RUNOFF MODELLING 1. Introduction The most common application of the remote sensing techniques in the rainfall-runoff studies is the estimation of the spatially distributed hydro-meteorological state variables that are required for the modeling, e.g., rainfall, temperature, ET, soil moisture, surface characteristics and land use land cover classes. Ability to achieve high spatial resolution and aerial coverage is the major advantage of the remote sensing techniques over the conventional methods. Hydrologic models that incorporate the remote sensing information include regression models, conceptual models, and distributed models. While selecting the hydrologic model for integration with the remote sensing data, spatial resolution of the hydrologic model structure and the input data must be comparable. Fine resolution data is relevant only if the hydrologic model uses spatially distributed information of all the relevant input parameters sufficient to capture the spatial heterogeneity, and also when the highly dynamic processes are monitored. This lecture gives the details of the remote sensing-aided rainfall-runoff modeling using the ArcGIS integrated Soil and Water Assessment Tool (ArcSWAT). Most of the figures and the results shown in this lecture are from Reshmidevi and Nagesh Kumar (2013). 2. SWAT and ArcSWAT Reshmidevi and Nagesh Kumar (2013) used the Soil and Water Assessment Toll (SWAT) for rainfall-runoff simulation. SWAT is a river basin scale hydrological model developed for the United States Department of Agriculture (USDA), Agricultural Research Service (Neitsch et al. 2005). Being a semi-distributed, continuous time model, it requires numerous spatial and attribute inputs that represent weather, hydrology, soil properties, plant growth, nutrients, pesticides, bacteria and pathogens, and land management. Integration of SWAT with a user interface in a Geographic Information System (GIS) environment provides the facility to input spatially referenced data and thereby enhances its D Nagesh Kumar, IISc, Bangalore 1 M8L2
capability to represent spatial heterogeneity (e.g., AVSWAT, ArcSWAT). The schematic flow of the SWAT integrated with a GIS framework (ArcView) is provided in Figure 1. Fig. 1. Schematic of GIS integrated SWAT (Di Luzio et al., 2002) In the study by Reshmidevi and Nagesh Kumar (2013), ArcSWAT (Winchell et al., 2007), a recent version of the GIS integrated SWAT was selected. ArcSWAT is the ArcGIS interface of SWAT. ArcSWAT uses various spatial and attribute data as input to the model and produces the output of hydrologic simulations in the form of tables showing various water budget components. 2.1 Study region and inputs to the ArcSWAT As a case study, the catchment of Malaprabha reservoir in the Karnataka state of India was taken up. It has an area of 2,564 km 2. Fig. 2 shows the location map of the study area. D Nagesh Kumar, IISc, Bangalore 2 M8L2
Fig. 2 Location map of the Malaprabha catchment Spatial data inputs Spatially referenced data used in the ArcSWAT include DEM, land use / land cover map and soil map. DEM used in the study was the Advanced Space-borne Thermal Emission and Reflection Radiometer (ASTER) Global DEM (GDEM) released by the Japan s Ministry of Economy, Trade and Industry (METI) and NASA, at a spatial resolution of 30m, generated using the satellite remote sensing techniques. Fig. 3 shows the DEM of the study area. In ArcSWAT, DEM was used to delineate the catchment boundary and to extract the topographic characteristics related to hydrology. Land use / land cover (LU/LC) map at 30m spatial resolution was generated from multiseason Landsat-7 ETM + imageries. Seven main LU/LC classes viz., water, agricultural land, barren / fallow land, rocky area, forest, settlement and grass land were extracted in the first step. Fig. 4 shows the LU / LC of the Malaprabha catchment. D Nagesh Kumar, IISc, Bangalore 3 M8L2
Fig. 3 ASTER GDEM of the Malaprabha catchment Fig. 4 Land use / land cover map of the Malaprabha catchment D Nagesh Kumar, IISc, Bangalore 4 M8L2
Based on the field information and the district statistical information about the crop production, the agricultural area was further classified into various crop classes. Each of the LU/LC classes was assigned to a corresponding SWAT class (Fig.5). Soil map of the area was procured from the National Bureau of Soil Survey and Land Use Planning (NBSS & LUP), Nagpur. Fig.6 shows the soil map of the study area. Fig.5 Assigning LU/LC information in ArcSWAT Fig. 6 Soil map of the Malaprabha catchment D Nagesh Kumar, IISc, Bangalore 5 M8L2
Attribute data Attribute data used as input to the model includes observed hydro-meteorological variables namely stream flow, precipitation, maximum and minimum temperatures, wind speed and relative humidity. Data at the streamflow gauging station recording the inflow to the Malaprabha reservoir on daily time scale was obtained from Water Resources Development Organization (WRDO), Karnataka, India, for the period 1978-2000. Observed daily data of temperature, wind speed, relative humidity, and cloud cover at one gauging stations namely Santhebasthewadi were obtained from the Directorate of Economics and Statistics, Bangalore, for the period 1992-2003. Spatial variation in the rainfall was accounted by using the rainfall observations at 9 stations in the catchment. Fig. 7 shows the locations of the raingauges, and the meteorological observatory in the catchment. Fig.7 Locations of the hydro-meteorological stations in the Malaprabha catchment D Nagesh Kumar, IISc, Bangalore 6 M8L2
2.2 Runoff simulation in ArcSWAT The modified SCS curve number method (USDA-NRCS, 2004) included in the ArcSWAT interface was used for the runoff simulation. Delineating the sub-watershed boundaries, defining the Hydrologic response units (HRUs), generating SWAT input files, creating agricultural management scenarios, executing SWAT simulations, and reading and charting of results were all carried out by various tools available in the interface. Data contained within HRU can include topographic characteristic, information about water flow, land cover, erosion, depressional storage areas etc. In ArcSWAT, hydrologic processes are simulated in two phases: land phase and the channel phase. The land phase was divided into various sub-basins, which were further disaggregated into spatially homogeneous HRUs. Each HRU was vertically divided into the surface layer, root zone, shallow aquifer and the deep aquifer layers as shown in Fig. 8. Hydrologic processes considered at each layer are also shown in the figure. Fig. 8 Schematic representation of the land phase hydrologic process simulated in SWAT (Reshmi et al., 2008) D Nagesh Kumar, IISc, Bangalore 7 M8L2
The SWAT model estimates the water yield from a HRU for a time step, using eqn. 1. The water leaving a HRU contributes to streamflow in the reach. WYLD = SURQ + LATQ +GWQ TLOSS Pond abstractions (1) where SURQ, LATQ and GWQ represent contribution to streamflow in the reach from surface runoff, lateral flow and groundwater, respectively, during the time step. TLOSS refers to the amount of water lost from tributary channels during transmission. The groundwater is primarily contributed by shallow aquifers. Fig. 9 shows the sub-basins delineated in the Malaprabha catchment. Fig.9 Sub-basins delineated in the Malaprabha Catchment Soil, slope and LU / LC information were integrated with the sub-basin data and 94 HRUs were delineated. ArcSWAT simulation estimated the hydrologic processes at each HRU and for each sub-basin and the output was given in a tabular format as shown in Fig. 10 and 11, respectively. D Nagesh Kumar, IISc, Bangalore 8 M8L2
Fig.10 HRU output in tabular form Fig.11 Sub-basin output in tabular form D Nagesh Kumar, IISc, Bangalore 9 M8L2
Water yields from the sub-basins were routed through the channel and the streamflow at the basin outlet (in this case outflow from sub-basin 1) was given in tabular format as shown in Fig. 12. Fig. 12. Channel output in tabular form generated by ArcSWAT Setting up of the SWAT for any catchment involves calibration and validation phases. Here, period from 1992 to 1999 was used for model calibration and the remaining period 2000-2003 was used for validation. Fig. 13 shows the monthly streamflow hydrograph in the calibration and validation periods. Table 3 shows the model performance indices for the monthly stream flow simulation. D Nagesh Kumar, IISc, Bangalore 10 M8L2
Fig.13 Monthly streamflow simulation for the calibration and validation periods Table 3. Model performance indices for the Malaprabha catchment Statistical index Correlation RMSE NMSE NSE coefficient (M.cu.m) Calibration period 0.963 41.34 0.074 0.925 Validation period 0.961 18.10 0.075 0.923 RMSE: Root mean square error NMSE: Normalized mean square error NSE: Nash-Sutcliffe efficiency 3. Assessment of the impact of land use changes on the streamflow With the help of satellite remote sensing, the dynamics of the land use pattern can be effectively captured and assimilated in the model. The calibrated model can be run using different LU/LC maps by altering the land use definitions at the HRU level. This lecture demonstrates the use of satellite remote sensing data in rainfall-runoff modeling. D Nagesh Kumar, IISc, Bangalore 11 M8L2
Bibliography / Further Reading 1. DiLuzio, M., Srinivasan, R., Arnold, J. G., and Neitsch, S. L. (2002). ArcView interface for SWAT2000, user s guide. TWRI report TR-193, Texas Water Resources Institute, Collage Station, Texas. 2. Nagesh Kumar D and Reshmidevi TV (2013). Remote sensing applications in water resources J. Indian Institute of Sci., 93(2), 163-188. 3. Neitsch S L, Arnold J G, Kiniry J R, Williams J R. 2005. Soil and Water Assessment Tool theoretical documentation. SWAT technical manual, Texas. 4. Reshmi T V, Christiansen A B, Badiger S, Barton D N. 2008. Hydrology and water allocation: comprehensive database and integrated hydro-economic model for selected water services in the Malaprabha River Basin. Report SNO 5695-2008, Norwegian Institute for Water Research: Oslo, Norway. 5. Reshmidevi TV and Nagesh Kumar D. (2013) Modelling the impact of extensive irrigation on the groundwater resources. Hydrological Processes. Doi:10.1002/hyp.9615. 6. USDA-NRCS, (2004), Part 630: Hydrology. Chapter 10: Estimation of direct runoff from storm rainfall: Hydraulics and hydrology: Technical references, In NRCS National Engineering Handbook.Washington, D.C.: USDA National Resources Conservation Service. Available at: www.wcc.nrcs.usda.gov/hydro/ 7. Winchell M, Srinivasan R, Di Luzio M, Arnold J. 2007. ArcSWAT Interface for SWAT2005: User s Guide. Blackland Research Center, Texas Agricultural Experiment Station, Texas and Grassland, Soil and Water Research Laboratory, USDA Agricultural Research Service: Texas. D Nagesh Kumar, IISc, Bangalore 12 M8L2