A Semi – Distributed Water Balance Model for Amaravathi River Basin using Remote Sensing and GIS
Jenifa Latha.C 1 , Saravanan.S 2 Palanichamy.K 3 1 Research Scholar is with National Institute of Technology, Tiruchirappalli. 2 Assistant Professor is with National Institute of Technology, Tiruchirappalli.
Sustainable water management in a river basin requires knowledge of the water availability and water requirements of the basin in the present and future for various purposes. The complexity of the water system in the region can be understood by calculating the regional water balance in a distributed scale considering the factors that affect it. Water balance is defined as the net change in water, taking into account all the inflows to and outflows from a hydrologic system. Spatial variation due to distributed landuse, soil texture, topography, groundwater level, and hydrometeorological conditions should be accounted for in the water balance estimation. However, conventional spatially and temporally lumped estimates of water balance do not help much in the planning and development of water resources in the water shed. This is an important and current issue in the socially and economically valuable Amaravathi River Basin of Tamil Nadu. In this study, a spatially semidistributed water balance model was developed to simulate mean monthly hydrological processes using landuse, soil texture, topography, and hydrometeorological data as input parameters in the Amaravathi River Basin, a semiarid region of Tamil Nadu in India. It is a physically based methodology for estimation of the average spatial distribution of water balance components. This model can be applicable in a public domain which can facilitate decision making. The water balance model is developed using SCS – CN (Soil Conservation Service – Curve Number) model to derive the runoff component and FAOPM (Food and Agriculture Organization – Penman Monteith) model to derive the evapotranspiration component spatially with the help of remote sensing and GIS techniques.
Keywords: Evapotranspiration, GIS, Runoff, Spatial Analysis, Water Balance.
1. Introduction
Water is the vital natural resource essential for the survival of mankind. Rainfall is the main source of water which is unevenly distributed spatially and temporally. Unprecedented increase in population, urbanisation, agricultural expansion and industrialisation leads to higher levels of human activities. As water demand increases, issues on water availability and demand become critical. This makes the management of water resources (assessing, managing and planning of water resources for sustainable use) a complex task. It has become more critical in places where rainfall is very low and erratic. Even though India is blessed with a higher average annual rainfall of 1,170 mm as compared to the global average of 800 mm, it faces the problem of water scarcity in most part of the year. (Jasrotia et al., 2009). A water balance, applied to a particular spatial unit is an application of the law of conservation of mass which states that mass can neither be created nor destroyed. Balancing the availability and the demand in a spatial scale will be the best way to cope up with the present trend. To achieve this balance, the rate of change of storage of water within the spatial unit must be equal to the difference between its rates of inflow and outflow across the unit. Calculation of water balance is a basic approach for determining stocks of water in different components (air, soil, water bodies) of the hydrologic cycle and fluxes between these components. The water balance model will be able to assess the water resources, in finding out the moisture deficit
and moisture surplus with different temporal and spatial resolutions. Knowledge of the quantity of water in different components of the water cycle serves in (a) Sustainable management of water resources and the protection against over exploitation and contamination (b) Cropping pattern – irrigation and drainage practices (c) Analysis of the effect of land use changes on water availability (d) Analysis of the effect of climate changes on water availability (e) Identification of recharge zones, (f) Rainwater harvesting, etc.
In the last few decades, changes in land use and land cover, changes in climatic conditions, population explosion, enhanced industrialization and urbanization has deteriorated the conditions of the Amaravathi watershed a semiarid region of Tamil Nadu, India. As a result, the effect of these changes on the water balance components is unknown. A serious problem recognized is that sufficient water is not available during the dry season. The water sector is very sensitive and is strongly influenced by the changes in climate and land use. Hence, it has the potential to impose additional pressures on water availability, water accessibility and water demand in the Amaravathi river basin. Even in the absence of climatic change, present population trends and patterns of water use indicate that the basin will exceed the limits of the economically usable, landbased water resources before 2025 (NWDA report, 1994 – T.S.No. 102). It is important to bridge the gap either by reducing the demand or by increasing the supply level to match the growing demand in future. Mechanisms must be developed for allocating the scarce water resources between the competing demand such as irrigation, rapidly expanding domestic and industrial needs, hydropower and environmental requirements.
A central support for decision making in sustainable water resources management arises from water balance modeling, which provides scientifically sound information of the current water resources and fluxes. Traditional approach of calculating the water balance using a spatially and temporally lumped scale does not give an accurate estimate of the water volume in a hydrologic component. Therefore there is a need to develop a methodology to model the distribution of the available and necessary water in the basin. A set of representative water balance models that have been used worldwide, their assumptions and limitations were considered. This eventually leads to the guidelines for the model selection. It is, however, not the intention of this paper to discuss all the models that have appeared in the literature.
C. W. Thornthwaite (1955) worked with the water balance approach to assess water needs for irrigation and other waterrelated issues. Waterbalance methodology has been used in a lumped water shed scale in order to develop climate classifications. Savanije (1997) developed a monthly time scale water balance model for the Mali sub catchment of the Niger River Basin, to determine the total evaporation from the earth’s surface using the relation between transpiration and soil moisture storage. A semidistributed conceptual model that simulates selected components of the water balance on a subcatchment scale, has been developed at a much finer spatial resolution and has mainly focused in the accurate estimation of the semidistributed model parameters. (Eder et. al., 2005). The BROOK90 model, a lumped hydrologic simulation model, has been calibrated and used for the water balance analysis to determine the model performance and the fractions of precipitation that become stream flow, evapotranspiration, and ground water flow by Combalicer et. al.,(2008). A semidistributed groundwater recharge model based on the concept of the tank model has been proposed to quantify the temporal changes of water table and water balance variables by He et. al.,(2008). Tilahun and Merkel (2009) has used a spatially distributed water balance model WetSpass to simulate longterm average recharge using land use, soil texture, topography, and hydrometeorological parameters in a semiarid region of Ethiopia. Jasrotia et. al., (2009) has performed a water balance study using the Thornthwaite and Mather model with the help of Remote Sensing and GIS for finding out the moisture deficit and moisture surplus of a watershed.
INTERNATIONAL JOURNAL OF GEOMATICS AND GEOSCIENCES Volume 1, No 2, 2010
Gridbased water balance modelling at high spatial resolution involves high complexity, and is data intensive. Semidistributed water balance model involves a simple modeling procedure and minimum data requirement. Quantifying water balance is thus a prerequisite for efficient and sustainable water resource management. Unfortunately, there has been no such study conducted in a distributed scale for Amaravathi River Basin, a semiarid region in Tamil Nadu, India. In this study, a spatially semi distributed water balance model is developed on a sub watershed scale using mean monthly hydro meteorological data. It is a GISbased hydrologic model to estimate surface runoff and evapotranspiration using landuse, soil texture, topography, and hydrometeorological parameters in the economically and socially important Amaravathi River Basin.
2. Study Area
The River Amaravathi (Fig.1) is one among the main tributaries of the river Cauvery, in its mid reach. It rises from Naimakad at an elevation of 2300 m above MSL in the Western Ghats (Anaimalai) in Idukki District of Kerala State. It flows towards northeast till the confluence with the river Cauvery on its right bank. Throughout its course of 256 km, the Amaravathi receives a number of small streams. The Amaravathi basin lies between the latitudes 10°06’51” N and 11°02’10” N and longitudes 77°03’24”E and 78°13’06” E. It has a catchment area of 8280 km2 constituting of four districts namely Coimbatore, Erode, Dindigul and Karur in Tamil Nadu.
Figure 1: The Study Area: Amaravathi River Basin, Tamil Nadu, India
The basin has four distinct seasons, SouthWest monsoon from June to September, NorthEast monsoon from October to December, the winter season from January to February and summer from March to May. The highest monthly mean of daily maximum temperature is around 40.6°C in April and the lowest monthly mean of daily minimum temperature is 22.4°C during January at Coimbatore (Pilamedu). The maximum average wind speed at Coimbatore is 32.6 kmph in the month of June. The minimum average wind speed is 9.5 kmph in November at Coimbatore (Pilamedu). The mean relative humidity is low in dry weather and high in the monsoon season. The sky is very cloudy during the monsoon season and is lightly clouded during nonmonsoon season. The monthly potential evapotranspiration value varies from as low as 66 mm in November to as high as 130.9 mm in May. The Amaravathi basin has two distinct topographical features. The hilly terrain occupies the southern, southwestern and western part of the basin between the elevation 2300 m and 500 m and the undulating plains with stray hillocks lie between 500 m and 40 m. The basin is more or less fan shaped
INTERNATIONAL JOURNAL OF GEOMATICS AND GEOSCIENCES Volume 1, No 2, 2010
and has a length of about 188 km from west to east and an average width of 81 km from south to north. Generally, surface water drains from the south western hilly area towards the plain area in the north east. The basin area is underlained by the crystalline complex of Archaen age. The oldest of the rock types in the basin comprises of Granite, Gneiss, Khondalite, Charnockite, Ultramafic and associated intrusive rocks. The hard rock aquifers are fairly heterogeneous as indicated by the variations in lithology, structure and texture within short distances. The occurrence and movement of ground water mainly depends on the degree of weathering, topography, extent of fracture zones, etc. The important hard rock aquifers noticed are the weathered gneiss of Coimbatore, weathered and jointed charnockite and khondalite of Coimbatore. The precipitation in the study area is moderate to high intensity, long duration rainfall resulting in floods that have disastrous effects on property and human life. The rainfall for the months from January to April is not significant. The average annual rainfall of the basin for the period from 190102 to 198384 is 1029 mm (NWDA report, 1994).
3. Data Processing The base map of the study area was prepared from topographic data. LANDSATETM data of 143/52, 143/53, 142/52 and 142/53 path/row was first rectified, corrected and georeferenced to the UTM projection system. The desired information was brought out by applying digital imageprocessing techniques on the image. The SOI toposheets were used for digitizing the contour and drainage network of the study area. The 30 x 30 meter DEM is used as input for the model. 8D flow direction algorithm was used to derive the flow direction map. Routing an accumulation of water as it flows down slope is computed using the GISintrinsic Flow Accumulation functions. A suitable threshold value was assumed to derive the drainage network and it is to be used in delineation of the sub watersheds.
Six different landuse/landcover features were classified such as Agricultural land, Waste land, Forest, Water Bodies, Settlement and Water logged area as shown in Fig.2. A field visit was made in order to verify the preparation of landuse/ landcover map of the study area.
Figure 2: Land Use/Land Cover map of the Amaravathi River Basin
A hydrological soil type map as shown in the Fig.3 was prepared from the soil texture map of the study area obtained from the Bureau of Soil Survey, Nagpur. The contour map was used to prepare the slope map of the study area. The model developed was used to calculate the water balance of the watershed from the rainfall and temperature data. A runoff potential map is prepared by using the
INTERNATIONAL JOURNAL OF GEOMATICS AND GEOSCIENCES Volume 1, No 2, 2010
runoff calculated from the model. The input data were prepared in the form of maps of selected meteorological, hydrological, and geographical elements in the basin.
Figure 3: Hydrological soil group map of the Amaravathi River Basin
The data for the purpose of the study for four stations in the Amaravathi catchment was obtained from the Central Water Commission and the Indian Meteorological Department. The spatial variation of the monthly average temperature as shown in Fig.4 and precipitation as shown in Fig.5, were interpolated using the Inverse Distance Weighted (IDW) kriging method on 30 × 30 m cell size. The hydrological soil group grid and the land use grid were overlaid and used for determining the Curve Number grid, depending on the Antecedent Moisture Condition. Based on the land use classes the crop coefficient value (Kc) grid was generated for various months. All these input grids were used for calculating the water balance components.
Figure 4: Spatial variation of the monthly average temperature (Aug. 2005) in the Amaravathi River Basin
INTERNATIONAL JOURNAL OF GEOMATICS AND GEOSCIENCES Volume 1, No 2, 2010
Figure 5: Spatial variation of the monthly average precipitation (June 2005) in the Amaravathi River Basin
4. Methodology
The proposed methodology for water balance estimation using remote sensing and GIS approach is shown in the flow chart (Fig. 6). This is developed for finding out the water deficit and water surplus in a subwatershed scale for the entire river basin. The base map of the study area was prepared from the Survey of India toposheets of scale 1:50,000. The topo sheets were used for digitizing the contour network. 30m x 30m Digital Elevation Model was developed from the contour map.
The sinks and imperfections of the DEM were identified and filled because they will cause the surface flow to disappear and invalidate the water balance. The slope of the area was calculated in degrees. The DEM was used to derive the flow direction map using the D8 ‘eight direction pourpoint algorithm’. Routing an accumulation of water as it flows down slope is computed using the GIS intrinsic Flow Accumulation functions. The Flow Accumulation map is derived, which indicate the size of the contributing area in each grid cell. The Stream Network is delineated by applying a threshold of 50 to the Flow Accumulation. The threshold is defined by specifying a minimum contributing area in square km, which is further used in the delineation of the watershed. The desired information on land use classification was brought out by applying digital image processing techniques from the LANDSATETM image. A field visit at specific time intervals is necessary in order to verify the preparation of landuse/ landcover map of the study area. Six different landuse/landcover features has been classified such as Agricultural land, Waste land, Forest, Water Bodies, Settlement and Water logged area as shown in Fig. 2. A hydrological soil type map of the study area, as shown in the Fig. 3 was prepared from the soil texture map obtained from the Bureau of Soil Survey, Nagpur. Temperature and rainfall data of five weather stations were used to develop a temperature map as shown in Fig. 4 and rainfall map of the catchment as shown in Fig. 5, by interpolation using inverse distance weighted kriging method.
INTERNATIONAL JOURNAL OF GEOMATICS AND GEOSCIENCES Volume 1, No 2, 2010
P is precipitation (mm), Q is the surface runoff (mm), ET is the actual evapotranspiration (mm), R is the ground water recharge (mm).
Runoff is that part of precipitation which does not evaporate or remain in soil storage. Soil Conservation Service (USDASCS, 1972) developed an equation to estimate surface runoff from storm rainfall, which is used in this study:
( ) ( )
2 0.20.8
P S Q
P S −
= + (2)
25400 254 S CN
= − (3)
where Q is the surface runoff (mm); P is precipitation (mm); S is the watershed storage parameter (mm);
CN is a dimensionless curve number which depends on landuse, hydrological soil group and antecedent moisture conditions II (i.e., the soil is near field capacity).
Figure 6: Proposed methodology for the water balance estimation
Evapotranspiration is the major source of uncertainty in evaluating the Water Balance Equation. Reference crop evapotranspiration according to FAO56PM model (Allen et al., 1998) is:
WATER BALANCE IN EACH CELL
RAINFALL
SOIL
SATELITTE IMAGE
TEMPERATURE
DEM
LAND USE DERIVATION
EXCESS RAINFALL
PET (FAO PENMAN)
ACTUAL E T
RUNOFF SCS CN
STREAM NETWORK DELINEATION
FLOW ACCUMULATION
FLOW DIRECTION SLOPE FILLED
DEM
WATER SHED DELINEATION
CROP COEFFICIENT
INTERNATIONAL JOURNAL OF GEOMATICS AND GEOSCIENCES Volume 1, No 2, 2010
where ET is the reference crop evapotranspiration (mm/d), ∆ is the slope of saturation vapor pressure versus air temperature curve (kPa/°C), Rn is the net solar radiation at the crop surface (MJ/m2/d), G is the soil heat flux (MJ/m2/d), γ is the psychrometric constant (kPa/°C), Ta is the mean air temperature at 2m height (°C), u2 is the wind speed at 2m height (m/s), es is the saturation vapor pressure (kPa), and ea is the actual vapor pressure (kPa).
The derived runoff and actual evapotranspiration on a monthly scale in each cell is to be used for calculating the water balance in each cell. The percolation and deep percolation components are not considered separately in the model development to avoid complexity. The water balance can be found out from equation (1) as S P Q ET ∆ = − −
5. Results and Discussion
For better understanding of the spatial water balance distribution, it is essential to first analyze the distribution of surface runoff and evapotranspiration which occur in the river basin. The amount of mean monthly rainfall is varying from 5 mm to 210 mm. The catchment is receiving rainfall both in the southwest and northeast monsoon. Mean Runoff simulated on the monthly time step for the year 2005 is presented in Fig. 7. The mean monthly surface runoff is found to be maximum in June, July and October, November months. A high spatial variation in runoff rates was observed, which can be due to the complex topographic and landuse characteristics.
The mean monthly evapotranspiration is found to be varying from 60 mm to 130 mm. Analysis of the simulation results from 2002 to 2007 shows that the mean monthly evapotranspiration values are higher during May through October and lower values are recorded at September to February. The simulated potential evapotranspiration for the year 2005 is presented in Fig. 8. The highest ET was from the Agricultural land (paddy field and banana plantation) followed by open forest and the least ET was from urban and builtup areas. The urban and builtup areas are mainly impervious surfaces with lower evaporation and transpiration.
INTERNATIONAL JOURNAL OF GEOMATICS AND GEOSCIENCES Volume 1, No 2, 2010
Figure 7: Spatial Distribution of Runoff simulated in a monthly time step (July 2005)
Figure 8: Spatial Distribution of actual evapotranspiration simulated in a monthly time step (July 2005)
Monthly simulation of the hydrological components of the Amaravathi river basin for the year, 2002 is shown in Fig 9. It shows that losses exceed precipitation in most of the months in the year 2002, which makes it a deficit year. Similar analysis was done for each year from 2002 to 2007. The simulated annual water balance for the five years period is given in the Table 1. The average annual water balance components were determined for the period of 2002 to 2007. Based on the results year 2005 is found to be a water surplus year and the remaining periods are found to be deficit years.
INTERNATIONAL JOURNAL OF GEOMATICS AND GEOSCIENCES Volume 1, No 2, 2010
The simulation results obtained in the present study show that the spatial variation of the availability of water in the basin, on a monthly time scale, is useful for decision making and the effective utilization of water. The results obtained here compare favourably with other similar studies, done by Voudouris. K, (2007) and Asefa .T (1999) particularly considering the amount and quality of the input data.
6. Conclusion
Based on the observed rainfall and other hydrogeological data, the water balance of the water shed was modeled. The water balance components in a monthly time scale utilizing the spatial variability of the catchment characteristics were evaluated using the water balance model. Loss due to evapotranspiration is maximum in most part of the year, like in 2002 it is around 110 mm to 120 mm from March to October (Fig.9). It is higher in the central and north western part of the basin (Fig. 8). This may be due to the heavy wind speed, less amount of rainfall and more agricultural land use in those areas. The monthly mean runoff for July 2005 varies spatially from 79.77 to 104.81 mm and is available from June to December for the Rabi Crops and is found deficient during March to May. The
INTERNATIONAL JOURNAL OF GEOMATICS AND GEOSCIENCES Volume 1, No 2, 2010
water balance calculations show that, July to October are found to be surplus months in most of the years. The model was evaluated from 2002 to 2007, from which 2005 was found to be a surplus year and the remaining years are deficit. The results are useful in agricultural planning, recharge by rain water harvesting, ground water modeling, vulnerability studies, etc. in the area. It can be found that using input data at a higher spatial resolution, will further improve the performance of the model. Further studies on the impacts of climatic changes on the water balance components will be carried out.
7. References
1. Allen RG, Pereira LS, Raes D, Martin. S, “Crop evapotranspiration Guidelines for computing crop water requirements” FAO Irrigation and drainage paper 56, (1998) Rome.
2. Asefa T, Wang Z., Batelaan O. and Smedt F. De, “Open Integration of a Spatial Water Balance Model and GIS to Study the Response of a Catchment”, Proceedings of the Nineteenth Annual American Geophysical Union Hydrology Days, August 16 20, Fort Collins, Colorado, USA, (1999) pp. 11 – 22.
3. Combalicer. E.A., Sang Ho Lee, Sujung Ahn, Dong Yeob Kim, and Sangjun Im, “Modeling Water Balance for the SmallForested Watershed in Korea”, KSCE Journal of Civil Engineering, Vol.12, (2008) pp.339348.
4. Eder. G, Martin Fuchs, HansPeter Nachtnebel and Wolfgang Loibl, “Semidistributed modelling of the monthly water balance in an alpine catchment”, Hydrol. Process., Vol.19, (2005) pp.2339–2360.
5. He Bin and Keiji Takase and Yi Wang, “A semidistributed groundwater recharge model for estimating watertable and waterbalance variables”, Hydrogeology Journal, Vol.16, (2008) pp.1215–1228.
6. Jasrotia A. S., Abinash Majhi and Sunil Singh, “Water Balance Approach for Rainwater Harvestingusing Remote Sensing and GIS Techniques, Jammu Himalaya, India” Water Resource Manage., Vol.23, (2009) pp.30353055.
7. Report of National Water Development Agency (NWDA), Technical Study No. 102, (1994).
8. Savenije. H. G, “Determination of Evaporation from a Catchment Water Balance at a Monthly Time Scale”, Hydrology and Earth System Sciences, Vol.1, (1997) pp.93100.
9. Thornthwaite C.W and Mather J. R, “Instructions and tables for computing Potential evapotranspiration and Water Balance”, Laboratory of Climatology, Publication no. 10, Cenetron, New Jersey, USA (1957).
10. Tilahun. K & Broder J. Merkel, “Estimation of groundwater recharge using a GISbased distributed water balance model in Dire Dawa, Ethiopia”, Hydrogeology Journal, Vol.17, (2009) pp.14431457.
INTERNATIONAL JOURNAL OF GEOMATICS AND GEOSCIENCES Volume 1, No 2, 2010
11. USDA Soil Conservation Service, ‘Hydrology’, Section 4, Soil Conservation Service National Engineering Handbook, U.S. Government Printing Office, Washington, DC (1972).
12. Voudouris. K., Mavrommatis. Th. and Antonakos. A., “Hydrologic balance estimation using GIS in Korinthia prefecture, Greece” Adv. Sci. Res., Vol.1, (2007) pp.1–8.
