RAINFALL - RUNOFF MODELLING USING DIGITAL ELEVATION MODEL AND REMOTE SENSING DATA

Abstract

Water is an important natural resource for life, agriculture, forestry, navigation, irrigation, power generation etc. Over the years, its demand is rising due to rapid growth in population, industry and urban areas. Therefore, it is necessary to know the availability of water in an area for its proper management and utilization. The modern technique of remote sensing, integrated with a Digital Elevation Model (DEM), can be used for the estimation of runoff from a catchment, accurately and efficiently. The present research work, thus, focusses on the estimation of runoff, and various hydrologic processes using Watershed Module of Strathclyde River Basin Model (SRBM). The main objectives of the research work are (i) to identify and assess those hydrologic parameters which can be derived from satellite data and DEM, for runoff generation (ii) integration of remote sensing, DEM, and hydrometeorological data into a hydrologic model in order to compute daily runoff, and (iii) to study the sensitivity of input parameters of the model in relation to runoff and identify the important parameters. The study area covers a part of Giri river catchment upto Yashwant Nagar, lying between latitudes 30° 45'N to 31°30'N and longitudes 77° 00'E to 77° 45'E, ranging in elevation from 900m to 3300m above m.s.l. Topographical maps no. 53E and 53F , IRS LISS I digital data of year 1989 and available meteorological data such as rainfall, evaporation and runoff of the catchment are the various data products used to carry out this study. The SRBM model used here is based on Stanford Watershed Model with modifications to watershed segmentation on the basis of elevation range. The primary input data required by the model are hourly or daily precipitation, daily evaporation and daily runoff data. The entire work has been divided into three components i.e. (i) Development of DEM and its application, (ii) Analysis of remote sensing data and (iii) Runoff simulation using SRBM model Contours from topographic maps at 200m interval have been digitized to generate a DEM at 500m grid size, using interpolation module of ILWIS (ii) (Integrated Land and Water Information System). The depressions in the DEM have been filled by the lowest elevation present at the rim of the pit. The depressionless DEM is then used for catchment segmentation and computation of slope, flow direction, flow accumulation and overland flow length data. Flow directions have been computed using flow line approach where flow direction of the previous cell identifies the next cell to be processed. Channel network has been obtained using criteria of minimum contributing area to form a channel. An area equal to 0.75 km . has been found as the threshold value to get a channel network, which is comparable to the channel network shown on topographic maps. This channel network is used to compute flow path slope and overland flow length for each segment as well as for the entire catchment, as required by the model. The land use information has been extracted from IRS LISS I data using image processing module of ILWIS system, implying Unsupervised clustering technique. Four classes are spectrally separated out, viz. thick forest, thin forest, cultivation and grass land. The overall classification accuracy has been achieved 89%. Interception and Potential Evapotranspiration from Lower Zone parameters have been obtained from landuse data, using area weighted technique for each segment as well as for the entire catchment. The study has been carried out for two different situations. In first situation, the catchment has been divided into three segments according to their elevation range, i.e., segment 1 is above 2300m, segment 2 is between 1600m to 2300m and segment 3 is below 1600m. In another situation the entire catchment has been considered homogeneous.as one segment. The values of uniformly distributed rainfall for each segment have been computed using Thiessen Polygon Method. Thus, raingauge adjustment factor in the SRBM model may now be considered as redundant. The parameters of SRBM model have been calibrated for runoff volume for year 1989 using measured runoff values. The calibrated values have been used to compute runoff for years 1991 and 1992 and compared with the measured values to validate the model. The measured and simulated daily flow volume at outlet have been (iii) found to have correlation coefficient varying between 0.991 to 0.999, variance of residual between 4 to 42, and explained variance between 92% to 99%. Sensitivity analysis of the model parameters has been carried out segmentwise on 1989 data, by changing the value of parameters upto ±30%. It is found that the runoff is more sensitive for lower zone soil moisture and ground water recession parameters as compared to infiltration and interflow indices. It is concluded that the model can be effectively utilized to estimate runoff by integrating it with DEM and remote sensing data. Further, it is found that the segmentation of the catchment improves the computed runoff values when compared with entire catchment as a single homogeneous unit. The proposed model is found to be very sensitive for lower zone soil moisture, hence the value of this parameter requires to be estimated with great care. As the results are found to be promising, the model can be applied to ungauged catchments, with similar environmental and land conditions.