Simulation and Characterization of Streamflow Time Series

Abstract

Streamflow can be partitioned into distinct time scales to better understand the underlying governing processes: fast flow, representing surface runoff controlled mainly by meteorological processes that operate over relatively shorter time scales and slow flow, representing the baseflow through groundwater flow and subsurface flow that occur at a slower rate. Because of the significant differences between fast and slow process controls, each may be used to explain streamflow variability independently. The thesis considers this concept by including seasonal (timing) streamflow variability in a regional context. In order to isolate the effects of these drivers on the observed Flow Duration Curves (FDCs), a modeling framework is developed that comprises of partitioning streamflow in multiple ways: seasons/months in the time domain, east-west/north-south directions in the space domain, fast/slow flows in the process domain, and small to large catchments to account for the size of the catchment in a basin. This modeling framework is demonstrated through the Peninsular Indian river system, in which the hydrology is primarily governed by the monsoon rains. The framework of stratifying streamflow variability consists of three independent elements: (i) process partitioning: partitioning of the total streamflow into the fast flow and slow flow, each of which is impacted differently by climate and landscape properties, (ii) time scale partitioning: stratifying the temporal streamflow variability into distinct seasons namely-South-West monsoon, North-East monsoon, and Non-monsoon, and (iii) investigating the directional aspects of west-east and north-south gradients in the space domain. Findings of the study showed that South-West monsoon and fast flows are the major contributors to mean annual flow and total flow respectively, in these river systems. The spatially increasing mean rainfall towards the northern part and favourable geology compounded by seasonal rainfall pattern in the southern parts of the Peninsular region control the directional variation of fast and slow flow contributions to total flow respectively. The rainfall variability due to monsoons and mountain ranges, and regional geology, are dominant climatic and landscape drivers of streamflow variability across the region. Moreover, the seasonal rainfall patterns and subsurface flows regulate the combined influence of time scale and process controls on total flow. Synthetic daily streamflow generation requires a critical understanding of the underlying dynamics resulting from the inherent time irreversibility in the rising and falling limbs of the hydrograph. Most models in literature considering the time irreversibility deal with single site streamflow simulation. Addressing intersite dependencies is, however, crucial for interconnected stream networks. The thesis presents a multisite streamflow generating framework to simulate concurrent streamflow sequences. This framework explicitly takes into account the spatial correlation and time irreversible dynamics of streamflow. A few streamflow gauging stations in the Godavari River Basin, located in Southern India, are considered to demonstrate the applicability of the framework. Through this work, the thesis contributes to the methodological development of streamflow simulation by capturing temporal asymmetry hydrograph and spatio-temporal dependency structure among multiple stations. The modeling procedure consists of the following steps: (i) fitting statistical models to the ascension and recession limbs of the hydrographs in the historical streamflow time series, (ii) constructing a set of correlated streamflow states through a nonparametric approach to address the dependence characteristics of streamflow across multiple stations, and (iii) simulating the multisite streamflow sequences that are consistent with the asymmetry of the ascension and recession limbs of the hydrograph. The proposed framework shows the ability to adequately generate multisite simulations capturing at-site statistics as well as intersite correlations for the case study. Furthermore, the approach ensures that the simulated flow values are not merely resampled from the historical data but uses the physical features of the hydrograph and shows variability beyond that observed in the historical sequence. Such a rich variety of streamflow sequences can help water managers to investigate how existing water resources systems on interconnected stream networks will operate in scenarios which have not been observed in the historical record. The thesis further deals with streamflow indices for large sample hydrology that take into consideration the time asymmetry. The study aims to provide flow descriptors on larger catchment scales and use these metrics to examine the driving forces of catchment attributes governing rising and falling limbs. The indices related to hydrograph limbs are primarily associated with distinct catchment attributes, forming a relationship between indices and catchment attributes to demarcate the governing drivers in corresponding hydrograph segments. The study presents a collection of streamflow indices with temporal asymmetry for 671 catchments in the United States. The regional variations across catchments are discussed using the spatial maps of streamflow indices. The flow metrics considering temporal asymmetry of hydrographs offer an alternative technique for determining the driving forces of streamflow hydrographs and opening up new possibilities to explore how the interaction of topography, soil, climate, vegetation, and geology defines the hydrological behavior of catchment. In summary, the main research contribution of this thesis is on developing novel statistical approaches towards providing an understanding of streamflow dynamics and demonstrate their applicability in different hydrological settings.