GIS BASED STUDY OF GEOMORPHOLOGY, LANDUSE AND WATER RESOURCE IN SMALL WATERSHEDS

Abstract

A proper understanding of morphological parameters, land use and land cover, underlying geology and hydrological behavior of a watershed can be significantly useful in watershed planning particularly in the absence of observed time series data. Literature review has brought out the following observations which motivated the present research work. (i) Comparative study of different watersheds in terms of several morphological parameters is tedious and lacks clarity. There is need to evolve suitable watershed indices which represent combined effect of several parameters on permeability of geological formation and intensity of erosion. (ii) Land use and land cover is undergoing significant changes at the level of small watersheds. Such changes can not be ignored in developmental planning. (iii) Error in flood estimation and hydrologic design of structures may occur if watershed is assumed to be linear (unit hydrograph theory) while in fact its response may be nonlinear. (iv) Observed hydrological data is usually not available for small watersheds. In such situations regional approach is followed. There is a need to evolve methods for establishing homogeneity among the watersheds. (v) Potential of Remote Sensing and GIS techniques is considerably greater than the research work has addressed so far. (vi) Watershed development and management, to be sustainable, has to be based on satisfying the basic needs of the local population. This aspect needs to be integrated in watershed planning process. In this context, a GIS based study of an area covering geomorphology, geology, dynamic changes in land use caused by human interference and hydrologic behavior has been carried out. Study Area The study area covers watersheds of three adjacent rivers namely Barureva (488 km2), Sher (1635 km2) and Umar (699 km2) which conjoin together to form an important southern sub-basin of Narmada basin in its upper reaches in Madhya Pradesh State of India. Umar and Barureva rivers are, in fact, tributaries of Sher River. From the south of the Satpura highlands down to the Narmada in the north, drainage system of the three rivers represents an accretional plain of alluvium deposits. The study area has been divided into 89 sub watersheds (68 are of the 4th order). Size of these is in the range of 1.77 sq km to 219 sq km. Morphological Analysis of the Study Area Morphological parameters of the three watersheds (Barureva, Sher and Umar) and their corresponding fourth order sub watersheds have been calculated with help of data attributes generated from the GIS analysis. A major part of Barureva (77.4%) and Umar (89%) watersheds are within 0 to 3% slope range. Sher watershed is comparatively hillier exhibiting considerable range of slope (nearly flat to very steep slope zone). Q-Q plots and frequency histograms of 1-4 orders length suggest the normality of the data hence all the 89 sub watersheds have been retained for further analysis. The fractal dimensions have been computed from power relationship of drainage parameters with area using data of 68 sub watersheds. Along with these fractal dimensions, degree of randomness (Cheng et al., 2001) is determined by combining several fractal dimensions into a single factor with help of principal component analysis. The spatial distributions of chosen fractal dimensions and degree of randomness are depicted for 68 sub watersheds to explain the pattern of drainage evolution and geological control in relation to various geological formations. The evolution of drainage pattern and shape of sub watersheds formed on the alluvium is highly controlled by alluvium formations, where as the evolution of drainage pattern and shape of sub watersheds formed on Deccan trap is found to be least controlled by its formation. The extent of geological control on drainage pattern goes on decreasing as the share of the alluvium formation decreases. Geomorphological Permeability Index A Geomorphological Permeability Index (GPI) considering length ratio (Rt), drainage density (Dd), drainage frequency (D f) and relief ratio (Rh) has been proposed to assess the nature of permeability and ground water recharge potential in eighty nine sub watersheds. GPI values are in the range of 0.05 to 119. Sub watersheds on alluvium formation have GPI higher than 20 whereas sub watersheds with GPI in the range of 0.05-2.55 are formed on the Deccan trap which is massive compact and impermeable. Field visits have shown that in those watersheds having GPI values less than 1, ground water structures are either very less or nonexistent. Settlements in these watersheds are very scanty. ii Sub watersheds with GPI values in the range of 4.8-6.3 have alluvium formation (37-51% of area) in lower part while upper part is dominated by hilly and hard rock formations. This type of situation is suitable for ground water recharge in lower part and rain water harvesting in the upper hilly part. Sub watersheds comprising of 16-28% alluvium formation show low values of GPI (1.2-1.85). These sub watersheds are in runoff production zone and suitable for surface water harvesting. These watersheds require erosion control measures. Therefore on the basis of proposed GPI, the sub watersheds may be identified for suitable treatment measures in terms ground water recharge, surface water harvesting and erosion control structures. Morphological Index of Erodibility Part of Sher, Barureva and Umar watersheds near the confluence with Narmada river and entire area of small tributaries (Dhamani and Saras rivers) were affected by badland formation in the year 1972. Over the years, these badlands have been mostly reclaimed for agriculture use as discussed in Chapter 7. However an index of erodibility has been proposed and used to identify and compare severity of erosion as existed in the year 1972 in different watersheds. The MIE index uses morphological parameters such as drainage density (pd), drainage frequency (Df), texture ratio (T) and relief ratio (Rh) which have direct relationship with soil erosion while shape parameters(Re, Re and Rf) have inverse relation with soil erosion. The isopach map shows alluvium deposits underneath, in the range of 30 m to more than 150 m in depth. Intensity of badland network is found to be maximum within 1 km distance to major river course. It is also observed that encroachment of badland formation is more intense on alluvium deposits which have the depth 120 m or more. Morphological index of erodibility (MIE) have been estimated for eight watersheds which are under the badland formation. MIE index values for these watersheds vary from 811 to 9208. A watershed which has alluvial formation and under agricultural use (not affected by badland formation) has the morphological index value of 200. It is recommended that a watershed in this region can be characterized as badland if its MIE is more than or equal to 4 times the MIE of normal watershed under agricultural use having same geological formation (alluvium). MIE index can be used as simple tool to quantify the degradation of watersheds. iii Analysis of Land Use and Land Cover Changes Land use and land cover of the Barureva, Sher and Umar watersheds have been determined for three different years i.e. 1972, 1989 and 2000 using satellite imageries. Processing of satellite imageries: Band layers of the satellite imagery for the years 1972, 1989, 2000 are based on different sensors. Classification of satellite imageries has been done using visual interpretation technique. The recent satellite imagery pertaining to year 2000 was selected initially. Recent photographs and spatial data base information are used to understand and recognize color, texture and tone of intended land use and land cover in the study area. Classified superimposed polygon layer of year 2000 is used as guide layer to identify the changes in size, colour, texture and tone of patches of land classes in a satellite imagery layer of year 1989. According to changes observed in the size of the land classes, these have been modified in superimposed polygon layer and saved as land use and land cover of 1989. Similar procedure is applied for land use and land cover classification of year 1972. Recognizable changes have taken place in land classes in the three watersheds during period from 1972 to 1989 (17 years) and during period from 1989 to 2000 (11 years). The land use changes are analyzed in terms of magnitude of area, percent change and dynamic rate of change per year for the intended periods. Moreover dynamic transition matrixes for three watersheds have been used to explain the conversion of land classes. Agricultural area has now become dominant in Barureva (72%) and Umar (77%) watersheds. The expansion in agricultural areas in these watersheds has occurred through reclamation of badland areas. Rate of deforestation in recent time period (1989-2000) has been comparatively higher than for the previous period (1972-1989). The barren land in Barureva and Umar watersheds has decreased in recent period (1989-2000) due to conversion into agriculture land. On the other hand, barren land in Sher watershed has increased by 13.71% due to deforestation in recent period (1989-2000). The expansion of urban settlement has mostly occurred by replacing agricultural area. The upper part of Sher watershed shows higher amount of water body area in comparison to Barureva and Umar watersheds. The appearance of water bodies in the upper part of watersheds areas suggests that surface water storage is necessary for expansion of agricultural area. Relation of GPI and LULC: Land use changes have also been studied at sub watershed level and correlated with their GPI values. Following inferences are drawn based on study. iv 1) Increase in surface water bodies has occurred in those sub watersheds whose GPI values are less than 15. Without further increase in water bodies these sub watersheds (remotely located and scattered settlements) will undergo more deforestation to increase the rainfed agriculture area for meeting food demand. 2) There is no definite relation between increase in settlement size and increase in water bodies suggesting that domestic water supply is not dependant on surface water. On the other hand increase in settlements has occurred in sub watersheds having GPI greater than 15 suggesting groundwater as main source of water supply to the settlements. 3) Barren land existed in sub watersheds having GPI less than 10 but has now been converted into agricultural land. 4) Sub watershed having GPI greater than 15 do not depend on surface water bodies for increase in agriculture and water supply to settlements. Driving Factors for Change in Land Use and Land Cover The study area consists of rural watersheds. Driving factors for change in land use and land cover are related to basic human needs (food, fodder and fuel) and economic dependence on agriculture in the study area. Demand of food, fodder has to be met locally in absence of adequate infrastructure facilities and low purchasing power of population in the remotely located sub watersheds. Analysis of land use and land cover shows that the rate of deforestation has accelerated in recent period to expand agricultural area so as to meet demand of food and fodder and to improve economic status. A sample analysis of Umar watershed illustrates the following. Whereas population has increased by 79.42% during thirty years period of analysis, agriculture area increased by 42.97% only. Umar is an agricultural watershed with 67.02% percent area under alluvium. Pressure of food demand on available agriculture land has tremendously increased necessitating improvement in crop production through use of ground water for irrigation. Falling trends in ground water level are observed in alluvial sub watersheds. On the other hand rising trend is observed in wells located in upper part of study area over Deccan trap formation (19S, 53S, and 55S). Agriculture area in these sub watersheds has remained nearly static. Rise in water table is probably due to creation of water bodies in these sub watersheds. Pressure of fodder demand on forest and barren land has increased by 107.36% over 30 years period. However it is reasonable to believe that part of this pressure might have been eased by crop residue which is also used as fodder. Runoff Potential under Varying Land Use and Land Cover Curve number (CN) in the SCS-CN method represents runoff potential which is an important consideration in surface water utilization and for design of hydraulic structures and erosion control measures. The CN computed from observed rainfall (P) and runoff events (Q) is termed as CN (PQ). The CN computed using land use and land cover is termed as CN (LU). The analysis has been carried out to (1) use observed data sets of rainfall (P) and runoff (Q) events of period greater than 1-day and develop year wise series of Curve Number (CN(PQ)), (2) estimate yearly series of Curve Number using land use and hydrological soil cover data (CN(LU)) and compare with observed CN(PQ), (3) forecast runoff potential i.e. CN(LU) on the basis of change in land use,(4) test the performance efficiency of SCS-CN method on gauged Sher watershed and its application to nearby ungauged Barureva and Umar watersheds and (5) compare the CN values of popular SCS-CN method and slope adjusted SCS-CN method at watershed level and at sub watershed level for assessing effect of slope on runoff potential. The CN (PQ) values have been computed in Sher watershed for the selected 187 rainfall-runoff events spread over 26 years (1977-2002). Most of the selected events have duration of 4-7 days. Observed events mostly occur in month of July, August and September. The annual CN (PQ) is defined as average of CN values for rainfall-runoff events in a year. It varies in the range of 69 to 87 over 26 years. The median value of CN (PQ) for observed data period is about 74 and average value is about 75. Values in the range of 70-79 are most significant values and these truly represent the AMC II condition of the Sher watershed. Estimation of CN from land use and land cover: CN values estimated on the basis of land use and land cover are termed as CN (LU). The classified land use maps of different years are crossed with hydrological soil group map by GIS operation to generate the collective layer. Thereafter the collective layers have been assigned the CN values appropriate for Indian condition. The collective layers with their assigned CN values have been used to generate distributed CN map of years 1972, 1989 and 2000. The annual CN (LU) values show rising trend with the time. The increase in CN (LU) with time period is attributed to increase in agriculture area in all watersheds. The vi equations of trend of CN (LU) with time (year) for three watersheds can be used to predict runoff potential with change in land use and land cover in future. Comparison of CN (LU) and CN (PQ) values shows close agreement. Moreover, derived land use land cover data from satellite imageries from years 1972, 1989 and 2000 also gets validated. The SCS—CN method along with annual CN (LU) values has been used for computation of daily runoff over period of 26 years. The agreement between computed and observed event runoff has been judged on the basis of the NS efficiency and RMSE values. The NS efficiency for entire data set (for all events spread over 26 years) is about 75 % which is quite satisfactory. Model performance is again verified by plotting computed and observed runoff with the line of perfect fit. It is concluded that the SCS method with dynamic annual CN (LU) is capable to predict direct runoff satisfactorily in gauged Sher watershed. Therefore the dynamic CN (LU) estimated for ungauged Barureva and Umar watersheds can be used for runoff prediction being under same hydrometerological zone. Although the effect of the slope on runoff volume has been established by research studies, few attempts have been made to study effect of topography in the SCS-CN method. The present study shows that slope adjusted CN is less than conventional CN over areas with slope less than 5% and more than conventional CN for areas with slope more than 5%. Higher the deviation from 5% slope more is the difference. Significant difference in CN is observed in the forest lands which are usually located on slopes. For micro watershed planning, SCS-CN method should be modified to incorporate effect of change in land use also in addition to effect of slope. Further research is needed to study effect of morphological parameters on the curve number. Hydrologic Nonlinearity of Watersheds Hydrologic linearity is related to the mutual proportionality of hydrograph peaks and runoff depths for storms of same duration. The peak discharge volume relationship (logQp= b+mlogV) proposed by Rogers (1980) without consideration of storm duration is empirical in nature. In spite of its criticism, the relation between peak discharge-runoff volume has been subject of research around the world due to its simplicity and potential applications. Analysis of 1 hour unit hydrographs (V=1 cm) of 18 watersheds in Narmada basin shows strong correlation between peak discharge and catchment area (in log space) as the duration of rainfall excess is same (1 hour). However, in general, basin vii area alone can not be used to explain variance of b (b=logQp for unit hydrographs) if duration of storm is not same. Slope of PDVR in log-log space (m) can be used as a measure of non linearity and to identify family of hydrologically similar watersheds. Analysis of 30 flood hydrographs of four watersheds (Umar, Kolar, Teriya and Temur) in upper Narmada basin shows that these watersheds exhibit nonlinear hydrologic character. Regression analysis shows strong correlation between peak discharge and runoff volume (0.872 to 0.983) for these four watersheds. Analysis of relation in logarithm space between V and qp/V2 suggests hydrologic similarity between all the four watersheds. Error in hydrologic design can occur by over estimating or underestimating flood discharge when a watershed is assumed to be linear while in fact it may be nonlinear in terms of catchment's response to rainfall. Case study of Umar watershed shows that UH model is not applicable in this nonlinear watershed and PDVR can be reliably used for prediction of peak discharge. Therefore the popular usage of UH theory necessitates validation of linearity concept in the rainfall-runoff process. Peak discharge per unit excess rainfall in the 89 sub watersheds have been estimated using relation between b and the geomorphological parameters such as A, S and L. A large part of watershed is found to have flood potential in the range of 0.2 to 5 m3/s/km2 of the watershed. In a more realistic study, flood potential of different sub watersheds should be compared for unit rainfall and not for unit excess rainfall. However the value of m (degree of non linearity) is required for these ungaged sub watersheds. Keywords : Morphological analysis, Fractal analysis, Geomorphological Index of Permeability, Morophological Index of Erodibility, Badland formations, Land use land cover change, Driving factors, Runoff potential, SCS-CN method, Hydrologic nonlinearity and similarity.