Abstract Of Dissertation On Environmental Water Resources

SUMMARY OF DISSERTATION (Ph.D)

Title: Assessment for Hydrological Changes due to Land-use Modification KEYWORDS: GIS, Remote Sensing, Distributed Hydrologic Modeling, Hydrological Similar Units (HSU), Land-use change model, Artificial Neural Network Abstract: There has been a growing need to quantify the impacts of land-use changes on hydrology from the standpoint of anticipating and minimizing potential environmental impacts. Thus, it is necessary that the hydrologic models need to account the effect of land-use modifications. In the development of hydrological models, fixed grid sizes in vector analysis and pixel sizes in raster analysis were considered in many studies as computational elements. In this study, the computational elements of ‘Hydrological Similar Units’ are considered in distributed hydrological modeling with accurate mapping of land-use at micro level using remote sensing and GIS. The widely accepted, the SCS-CN technique is adapted here to compute the runoff from the several hydrological similar units of the basin for the given rainfall. The integration of spatial data and application of a distributed model in GIS with remote-sensing data provides a powerful tool for assessment of effects due to land use change. The daily runoff values are calculated by three modeling approaches namely, Distributed Hydrological Modeling (DHM), Semi-Distributed Hydrological Modeling (SDHM) and Lumped Hydrological Modeling (LHM). It is found that DHM could predict runoff values very close when compared with that of the other models, and difference between the calculated storm runoff volume and observed runoff volume is less than 3 %.

From the sensitivity analysis on initial abstraction ratio λ, it is found that the average values of λ are 0.27 for premonsoon, 0.20 for monsoon and 0.17 for postmonsoon can be adopted irrespective of land-use and soil type in SCS-CN method for agricultural mountainous basin. On the other hand, an average value of λ could be taken as 0.25 for forestland, 0.20 for agricultural land, 0.22 for settlement and 0.12 for pastureland for estimation of runoff using SCS-CN method irrespective of season and soil type. The Land-use Change model (LuC Model) was developed to compute the change in peak runoff value at basin outlet due to change in land-use. The model was applied to analyze with future plan scenarios and land-use of different percentage of urbanization and deforestation. The variation of CN values, which represent the landuse changes, is found to have more effect on the rising limb and the peak runoff than on the falling limb. It is also found that as CN value increases the rising limb is shifted backwards. It is concluded that the percentage change in runoff due to the land-use change almost constant for different land-use irrespective of the rainfall pattern (different years) and irrespective of time of occurrence (premonsoon, monsoon and postmonsoon). It is found that the effect of land-use modification is more predominant in groundwater than on the surface water. This is due to fact that the extraction of groundwater to fulfill the demand of growth urbanization is increased and portion of water infiltrating for groundwater is reduced due to increase in imperviousness. The individual ANN models developed for each subbasin were integrated into a single ANN model to predict the change in runoff due to change in land-use for the entire basin. The performance of Feedforward Backpropagation Network (FFBPN) was

found to predict the runoff quite well when compared to the performance of Recurrent Neural Network (RNN). The total runoff volume for different percentage of urbanization predicted by DHM and FFBPN model are found to be closer than that of predicted values by RNN model. During low flow periods, RNN models underestimated the runoff and resulted in less R2 as well as large RMSE value. Artificial Neural Network (ANN) models were found to be capable of mapping the non-linearity between rainfall and runoff and able to predict the runoff efficiently. The determination of optimal network architecture is found to be critical for efficient mapping of rainfall-runoff relationship. It is concluded that the integrated distributed hydrologic model considering spatial variability through Hydrologic Similar Units (HSU) using HEC-1, GIS and Remote Sensing is a powerful tool for assessment of the hydrologic effect due to land-use changes. It is finally concluded that the coupling of GIS and hydrologic models seems to be a logical direction in the rainfall-runoff modeling.

Study area: Kathmandu Valley watershed