Investigation of Land Use Effects in Geomorphological Unit Hydrograph of Linear Reservoirs

The influence of urbanization, as one important form of land use, on runoff and floods within watersheds has been a major topic of research during the past few decades. Urbanization affects the hydrology processes of a watershed by replacing the vegetated land cover with impervious surfaces. This can have a substantial effect on the hydrological response of a watershed to rainfall, potentially resulting in faster response, greater magnitude of river flow, higher recurrence of small floods and reduced base-flow, and groundwater recharge. The direct runoff hydrograph generated by rain falling on a watershed reflects the characteristics of both the effective rain hyetograph and the relevant surface features that control the runoff generation and surface-water flow processes.
In this study, the effect of land use investigated using GUHCR model and adjusted GUHRLU model is presented. These models and Nash’s conceptual model used to investigate land use impacts for a small, well instrumented watershed consisting of two different land uses sub-watershed in the city of Sierra Vista, Cochise County, Southeastern Arizona. Geomorphological factors for the sub-watersheds extracted by GIS. In this study 13 storm events occurring on both sub-watersheds were selected to examine the proposed model’s performance. Nine events were selected for model calibration. The remaining four events were used to examine the simulated hydrographs for the outlet and the interior natural sub-watershed. The model parameter (k) was estimated for each event using the moment method and the average of the calibrated values was used for evaluation of the model. The model's performance demonstrated through four popular criteria (i.e. The Nash–Sutcliffe efficiency (NE), the Correlation Coefficient (R), the ratio of the absolute error of peak flow (EP) and the ratio of the absolute error of hydrograph’s volume (Ev)) using available hydro-geomorphological data.
The results show that although all studied models forecast the outlet hydrographs with acceptable accuracy, only the presented GUHRLU model shows appropriate results at sub-watershed outlet considering the effect of land use. Clearly, accounting for land use properties in the model formulation leads to improved efficiency at the internal sub-watershed. The Nash model as a lumped model, calculates the hydrography just at the watershed outlet without any information about the hydrological behavior of the interior watershed. Therefore, internal hydrography estimation is impossible via this model. In general, urban runoff tends to have a sharper rising limb and higher peak values while runoffs in natural watersheds have smaller peak values and the rising limb climbs more slowly. The hydrographs show that the overall shapes of the urban sub-watershed hydrographs are similar to each other, while those in the natural sub-watershed tend to be more different, as expected. Simultaneous consideration of geomorphological and land use parameters in the formulation of the proposed model (GUHRLU) provides this capability. As indicated by Ep and Ev, the error of peak flow and the volume of hydrographs show acceptable accuracy. It can be noted that some events show high values of error of peak flows (Ep), however, the model results in small values of Ev that is of great importance in water resource management. Note that, the performance values obtained for the watershed outlet were, for most events, higher than those of the internal sub-watershed outlet in both formulations, which may be due to the use of outlet hydrographs for calculating the model parameter (k)., This might also be due to less uncertainty in urban watersheds where runoff to rainfall ratios is much larger than in the natural sub-watershed. The GUHCR model has slightly better performance at watershed outlet, but it is unable to detect land use variability in its model formulation and so to estimate the internal watershed hydrographs appropriately. Overall, peak discharge and runoff volume for the natural sub-watershed was over-estimated via GUHCR model. The average values for Nash-Sutcliffe criteria at the internal watershed outlet for GUHCR and GUHRLU models are 0.47 and 0.78 respectively. Over 40% improvement is achieved in simulated peak discharge and runoff volume at interior watershed outlet using GUHRLU in compared with GUHCR model.
GUHRLU model considers not only the geomorphologic properties of the watershed, but also the land use variation of the sub-watershed in parameter formulation. This model can also reflect the hydrological conditions of the internal parts of the watershed with divergent land uses. The GUHRLU model is able to improve the efficiency of geomorphological rainfall-runoff simulations at the interior part of the study watershed, located in southeastern Arizona, by taking into account land use. Consideration of land use in the model leads to acceptable results at both watershed and interior sub-watershed outlets, particularly for watersheds like the studied watershed where different land uses sub-watersheds have. The overall efficiency of prediction was slightly poorer for the internal sub-watershed than for the outlet. Application of three models reveals that only the presented GUHRLU model shows appropriate results at sub-watershed outlet in which the land use variation is considered in the model formulation.