Περίληψη σε άλλη γλώσσα
Floods are among the most catastrophic natural hazards, creating significant human deaths and economic and environmental losses at the global scale. To cope with this the scientific community has invested a lot in developing appropriate tools. Hydrologists are using data and models to predict flood events in an uncertain environment with the purpose to design flood protection structures and water conservation works. At the catchment scale, environmental data, such as precipitation, elevation data (i.e., Digital Elevation Models or DEMs), and discharge timeseries are important as initial information for hydrological analyses. The generation of surface runoff is directly influenced by geomorphological characteristics of the studied catchment and requires the study of runoff generation mechanisms in a basin, and also the search for relationships connecting runoff and geomorphological characteristics. Obviously, this relationship is not linear due to the different time scales involved in h ...
Floods are among the most catastrophic natural hazards, creating significant human deaths and economic and environmental losses at the global scale. To cope with this the scientific community has invested a lot in developing appropriate tools. Hydrologists are using data and models to predict flood events in an uncertain environment with the purpose to design flood protection structures and water conservation works. At the catchment scale, environmental data, such as precipitation, elevation data (i.e., Digital Elevation Models or DEMs), and discharge timeseries are important as initial information for hydrological analyses. The generation of surface runoff is directly influenced by geomorphological characteristics of the studied catchment and requires the study of runoff generation mechanisms in a basin, and also the search for relationships connecting runoff and geomorphological characteristics. Obviously, this relationship is not linear due to the different time scales involved in hydrological processes, among others. On the other end of the spectrum, water is a valuable resource for human societies and ecosystems. It is therefore vital to predict water resources in view of the design of hydraulic works for the exploitation of such resources. As a result, the necessity of the accurate prediction of river runoff emerges again as in the above-mentioned case emphasizing floods. Despite the fact that our era is characterized by the development and application of remote and on-site sensing methods for measuring and collecting hydrological data (e.g., ground telemetry systems, radar, meteorological satellites), in terms of measuring the discharge in river cross sections, the problem remains. Most part of the drainage catchments worldwide remain ungauged. The Prediction in Ungauged Basins (PUB) initiative of the International Association of Hydrological Sciences (IAHS), launched in 2003 and concluded by the PUB Symposium 2012 held in Delft, aimed at improving scientific understanding of hydrological processes, as well as the associated uncertainties and the development of models with high predictive power. Furthermore, there is a need for better comprehension of the hydrological behavior of a catchment. In the presence of hydrological data scarcity worldwide, it would be useful to link catchment’s behavior with its physical properties, such us climate, topography, geology, soil type and land use, to address the challenge of ungauged basins (Wagener et al., 2007; Gupta et al., 2008). In addition to that, the changing environment (land-use changes, effects of a changing climate) plays a critical role in hydrological modeling. There was no clear understanding of the spatial and temporal scales at which these effects would emerge (Blöschl et al., 2007). On top of that, insufficient process understanding and the lack of concurrent data at multiple space–time scales enhance predictive uncertainty. Everyday engineering practices still follow simplistic approaches that are easy to implement in ungauged basins (Efstratiadis et al., 2014). In the last 20 years, there was a significant interest in basin classification (McDonnell and Woods, 2004; Wagener et al., 2007; Sawicz et al., 2011). The criteria for this classification consist mainly of physical catchment characteristics and geomorphological characteristic (Winter, 2001; Wolock, 2004; Gharari et al., 2011; Cheng, 2012; Sawicz et al., 2013; Papageorgaki and Nalbantis, 2016), and on streamow characteristics (Olden et al., 2011; Ley et al., 2011; Corduas, 2011). Classification of drainage basins into groups with similar response to meteorological forcing can be very helpful in cases of transfer of hydrological information in space such as in streamflow prediction in ungauged basins (Papageorgaki and Nalbantis, 2022). One of the focuses, in the present thesis, is testing the ability to classify drainage basins using climate-based variables and geomorphometric characteristics as predictors. In such an effort, the stream network plays an important role, as it affects the accurate estimation of stream head positions and, hence, water transfer processes which are important for an accurate runoff prediction. The spatial distribution of geomorphological characteristics used for basin classification is also affected by the stream network, i.e., the drainage density, or the Height Above the Nearest Drain (HAND). HAND and ground slope, which can be easily obtained from a DEM, appear to be the dominant topographical controls for hydrological classification (Gharari et al., 2011; Papageorgaki and Nalbantis, 2016). The extraction of hydrographic networks and catchment delineation based on a DEM are performed in an automatic way in a Geographic Information System (GIS). The geomorphic and hydrologic importance of the extraction of hydrographic network is depicted in literature (Montgomery and Dietrich, 1989; Montgomery and Foufoula-Georgiou, 1993). Various methods have been proposed for automatically extracting channel networks from DEMs (O'callaghan and Mark, 1984; Tarboton et al., 1991; Band and Moore, 1995), the most common one being the adoption of a value of the minimum drainage area required for a channel to initiate (Band, 1986; Morris and Heerdegen, 1988; Tarboton et al., 1988). Tucker et al. (2001) proposed a combination of two criteria, i.e., a threshold upstream contributing area and a slope threshold. The upslope contributing area threshold, herein referred to as the Critical Support Area (CSA), is the critical quantity that allows for the extraction, from DEMs, of useful hydrological information. CSA is commonly selected as a constant quantity without consideration of its variation in time and space. In this thesis, an objective way for channel initiation is investigated, taking into consideration topographic information from classical maps at the scale 1:50000. So, these maps are used to identify stream heads and the corresponding values of CSA for two seasons of the hydrological year: the wet and dry season. For the extraction the Shuttle Radar Topography Mission (SRTM) DEM is used, while GIS and MatLab are employed for data processing. Landscape classification is considered as a promising tool for constructing rainfall runoff models that enable runoff predictions in ungauged basins (Papageorgaki and Nalbantis, 2022), Topography can be regarded as an integrated indicator to distinguish between landscape elements with different hydrological functions (Gao et al., 2014). More specifically, HAND calculates the elevation of each point in the catchment above the nearest stream it drains to, following the ow direction. HAND increases the hydrologic information content of elevation data (Nobre et al., 2011).Runoff prediction in ungauged basins requires the transfer of hydrological information in space, which is greatly facilitated by using physically observable quantities, such as geomorphological characteristics. These allow the landscape classification or, else, the identification of areas that are distinct with respect to the predominant runoff generating mechanisms. For example, Winter (2001) proposed a classification of the catchment inner space into hydrological landscape units (upland, valley slopes and lowland) through exploiting the combination of topographic, geological and climatic conditions. Topography has also a considerable inuence on the dominant hydrological processes in different parts of a catchment, which could be used to dene hydrologically different response units (Savenije, 2010). Topography is also linked to geology, soil characteristics, land cover and climate through co-evolution (Sivapalan, 2009; Savenije, 2010).Although topographic data of the ground surface are globally available, in hydrological models topography is sometimes explicitly used, whereas in other cases, this is used implicitly or even ignored. For example, the lumped topography-driven model known as TOP-MODEL (Beven and Kirkby, 1979) uses the topographic wetness index (TWI) (Beven and Kirkby, 1979), which is a proxy for the probability of saturation of each point in a catchment, which is directly linked to the occurrence of Saturated Overland Flow (SOF). The topography-driven conceptual modelling approach known as FLEX-Topo (Savenije, 2010) attempts to exploit topographic signatures in order to design conceptual model structures that represent the complexity and heterogeneity of hydrological processes. This model follows a middle way between parsimonious lumped and complex distributed models. It exploits topographic information as the main indicator of landscape classes and dominant hydrological processes. In this thesis, the FLEX-Topo model is adapted to the test basins used. The well-known procedure for the estimation of parameters of conceptual models relies on the availability of streamflow data for model calibration, which, however, are frequently unavailable. Alternative techniques, such as regionalization have been developed to bypass model calibration (Yadav et al., 2007; Zhang et al., 2008; Kling and Gupta, 2009; Samaniego et al., 2010; Kumar et al., 2010; Wagener and Montanari, 2011; Kapangaziwiri et al., 2012; Viglione et al., 2013). So, the identification of the relation between catchment characteristics and model parameters (Merz and Blöschl, 2004; Kling and Gupta, 2009; Nalbantis, 1995; Nalbantis et al., 2011; Nalbantis et al., 2002) has been and still is a challenging research topic. However, the lack of representation of processes heterogeneity in conceptual models limits the realism of these models. To handle this issue the concept of Hydrological Response Units (HRUs) was proposed. Landscape classification using topographical indices, from DEMs becomes a promising tool (McGlynn and McDonnell, 2003; Hrachowitz et al., 2009; Gourgoulios and Nalbantis, 2017). Consequently, within a exible modelling framework (Fenicia et al., 2008; Fenicia et al., 2011), different model structures have been developed to represent the different dominant hydrological processes in different landscape classes. FLEX-Topo is a modelling framework that makes exhaustive use of topographic information in hydrological models and it can be applied to any type of conceptual model. The design of a variety of hydraulic works is based either on design hydrographs, which are constructed through hydrological analysis of historical data, or through the use of geomorphological characteristics of the study basin. For ungauged basins, the use of a Synthetic Unit Hydrograph (SUH) seems the only way to handle the lack of data. Within the SUH (Singh et al., 2014) approach two forms of the Instantaneous Unit Hydrograph, or, else, the response hydrograph to an instantaneous pulse of rainfall excess, have been proposed: the Geomorphologic Instantaneous Unit Hydrograph (Rodriguez-Iturbe and Valdez 1979; Rinaldo and Rodriguez-Iturbe 1996), and the Width Function based Instantaneous Unit Hydrograph (WFIUH) or, else, the Rescaled Width Function (RWF) (Rinaldo et al. 1995) based on the idea of Width Function (Gupta et al. 1986) which reflects the probability density function of flow path length from each point within the studied basin to the basin outlet. The RWF method takes into consideration the heterogeneity of the ground relief. This has been further developed by Di Lazzaro et al. (2014) to make use of the drainage density and the resulting method is denoted as the drainage density weighted Rescaled Width Function, or ddRWF. Drainage density can be easily derived from a DEM, after the extraction of stream network. The ddRWF method is modified to make use of findings on the CSA (Papageorgaki and Nalbantis, 2018) for extracting the hydrographic network. The hydrographic network is important for the computation of drainage density, so that an objective method for its extraction is preferable to others. The CSA value is selected on the basis of heads of perennial and ephemeral streams shown on maps at the scale 1:50000. The extracted hydrographic network based on CSA is later compared with the one proposed by Tucker et al. (2001). As already said, Tucker et al. (2001) proposed a combination of two criteria, i.e., a threshold upstream contributing area and a slope threshold. The latter method has also been applied by Di Lazzaro et al. (2015) within the WFIUH framework. Both analytical forms of WFIUH include kinematic parameters which are effective flow velocities for hillslopes and channels. According to the proposed methodological framework, the estimation of kinematic parameters of WFIUH for ungauged basins is performed following a relationship described by Di Lazzaro (2009) after his regional analysis of 12 gauged basins in Central Italy. So, the channel flow velocity has been calculated using the average based on ground elevations flow path slope using ground elevations from a DEM. The latter framework is used to estimate the kinematic parameters of the WFIUH in a framework with lack of the necessary runoff and rainfall data. All flood simulations are known to suffer from significant uncertainties, since those are based on hydrological and hydraulic models that make use of land information. The impact of uncertainty in ground elevation on the extent of areas that are inundated due to flooding is critical. General guidelines on using Digital Surface Models (DSMs) in flood hazard assessments are lacking due to the difficulty in processing DSM-based topographical information within many existing models, and due to the lack of knowledge on the required DSM accuracy (or uncertainty). The typical procedure for flood hazard mapping includes the construction of design hyetographs; the transformation of these into graphs of river discharge, water velocity, and water depth; the calculation of the peak (i.e., maximum) water elevations; and the construction of maps with the maximum inundated area. The latter is known to contain uncertainty, which was studied only in the last decade (e.g., Merwade et al., 2008). This is mainly caused by uncertainty that is inherent in hydrological processes (Carpenter and Georgakakos, 2004; Chaubey et al., 2005; Crosetto et al., 2000; Huang and Liang, 2005; Wilby, 2005; Anderson et al., 2009). Also, DSMs contain uncertainty whose effect has been investigated only in the last 15 years by focusing either on peak discharge (Brasington and Richards, 1998; Chaubey et al., 2005; Hancock, 2005; Valeo and Moin, 2000), or on water depth and inundation extent (Bales and Wagner, 2009; Bates et al., 2003; Colby et al., 2000; Marks and Bates, 2000; Omer et al., 2003; Tate et al., 2002; Vazquez et al., 2002; Wang and Zheng, 2005; Werner, 2001; Werner, 2004). The effect of DSM uncertainty on the extent of inundated areas is investigated using the Monte Carlo method to quantify the uncertainty. A typical photogrammetric procedure and conventional maps are used to obtain a reference DSM, later altered to provide DSMs of lower accuracy. Also, data from the Shuttle Radar Topography Mission are used. It is worth noting that, in the past, the effect of DSM uncertainty was studied within the frame of international research initiatives such as the Flood Risk Management Research Consortium Initiative (Pappenberger et al., 2005) and the Prediction in Ungauged Basins Initiative (Hrachowitz et al., 2013; Sivapalan et al., 2003). To characterize the inundated area, the 90% quantile of the inundation extent and inundation topwidth for peak water level at specific river cross sections are employed. For topwidths, apart from point estimates, also interval estimates are acquired. The aim of the doctoral thesis is to contribute to the classification of ungauged basins in terms of the mechanisms of runoff generation and the improvement of hydrological models in cases of catchments with poor or complete lack of hydrological information. Specifically, the thesis is aimed at contributing to responding to the following questions: Question 1: How is it possible to transfer hydrological information from a gauged basin to an ungauged one, when the approach is lumped in space and the time scale is coarse (e.g., daily, monthly)?Separate questions are rising, concerning: Is it possible to recognize the hydrological class of an ungauged basin, for which only information that is readily available, can be used? Is it possible to identify geomorphological features of a basin, that can be used as criteria to allow the recognition of the hydrological class of an ungauged basin among classes obtained via hydrological signatures, and, also, allow the transfer of hydrological information in space? Is it possible to set an appropriate methodological framework for the classification of ungauged basin? Is it possible to determine an objective way of assessing the geomorphological characteristics to be used as classification criteria? Question 2: How is it possible to extract the hydrographic network of a drainage basin based on objective information derived from a DEM, and, moreover, what is its spatiotemporal variation? Separate questions are rising, concerning: Which is the suitable value of Critical Support Area (CSA) for the extraction of hydrographic networks from a DEM?Is the dynamics of the hydrographic network depicted in the value of CSA, as the latter is known to be related to the size and scale of a stream network and the related basin area? In which way the variation in time and space of CSA can be investigated and modeled? Question 3: How is it possible to transfer hydrological information from a gauged basin to an ungauged one, when the approach is lumped in space and the time scale is fine (e.g., hourly)?Separate questions are rising, concerning: How does the drainage density weighted Rescaled Width Function can be used for the derivation of the Synthetic Unit Hydrograph of an ungauged drainage basin? How to estimate the kinematic parameters of the Width Function based Instantaneous Unit Hydrograph (WFIUH) in a framework with lack of the necessary runoff and rainfall data? How does the uncertainty in DSM affect the inundation extent of rainfall-induced floods? Question 4: How is it possible to transfer hydrological information from a gauged basin to an ungauged one, when the approach is semi-distributed in space and the time scale is coarse? A separate question is rising, concerning: It is possible to accurately predict the runoff of an ungauged basin, adopting the approach of semi-distributed rainfall-runoff modelling using landscape classification into hydrological landscape classes, and, also allow the transfer of hydrological information in space?
περισσότερα