The r.inund.fluv tool for flood-prone areas evaluation in GRASS GIS: application to the terminal reach of Magra River

Transcription

1 The r.inund.fluv tool for flood-prone areas evaluation in GRASS GIS: application to the terminal reach of Magra River Ilaria Ferrando1, Bianca Federici1, Domenico Sguerso1 & Roberto Marzocchi2 1 DICCA - Department of Civil, Chemical and Environmental Engineering University of Genoa, Via Montallegro 1, Genova, Italy 2 Gter srl Innovazione in Geomatica, GNSS e GIS Abstract The present work aims to illustrate the potentiality of the GRASS command r.inund.fluv, developed in the Laboratory of Geomatics of DICCA, in the evaluation of potentially flooded areas by means of its application to the terminal reach of Magra River in Italy. Such application underlined the need of a code improvement, in order to better manage the meander-shaped nature of the river through a rectification of river axis. The final results were compared with the official expected flooding map, achieving a satisfying correspondence, that proves the efficiency of the procedure. Keywords GRASS GIS, river flooding, flood-prone areas simulation, meanders, Magra River. 1 Introduction r.inund.fluv tool to flood-prone areas evaluation with the The first step of a risk assessment analysis is the evaluation of flood-prone areas. Its importance is considered for both managing and planning emergency activities. Nowadays, the use of GIS technology for risk assessment analysis is recommended (Peggion et al., 2008). However, it is not widely used for defining inundated areas. The proposed work uses a GIS module called r.inund.fluv, developed in the Laboratory of Geomatics of DICCA (Federici and Sguerso, 2007; Marzocchi et al., 2009), in order to compute perifluvial flood maps. r.inund.fluv runs under the version 6.x of GRASS GIS software and it is distributed as an add-on under GNU General Public License terms ( The tool starts from one-dimensional hydraulic models, and then it takes into account the two-dimensionality of both terrain and flooding event. The water surface profile along the river axis has to be calculated for a given water discharge through a generic one-dimensional hydraulic model (HEC-RAS, Basement, MIKE 11, etc.). The conformation of the river floodplain has to be described by a high-resolution Digital Terrain Model (DTM). The coupled integration between the 1D hydrodynamic model and the GIS environment takes the advantage of the simple and well developed 1D 501

2 hydrodynamic model, able to simulate the river flow in the channel, and of the GIS software to map the flooding extent. The r.inund.fluv tool was already successfully applied to the Tanaro River (Federici and Sguerso, 2007), an Italian river approximately 120 km long, and to Roggia Scairolo (Pozzoni et al., 2009), a stream in the Ticino canton, for the evaluation of flood hazard. The present study illustrates the application of such tool to the terminal reach of Magra River (Italy), a fluvial weakly meandering river, and a new improvement of the software code to better manage the meanders. 2 Overview to the r.inund.fluv procedure The procedure consists in 5 sequential steps that, starting from simple hypothesis and adding more strict conditions, allow to obtain the map of potentially flooded areas. The steps of procedure can be summarized as follows: 1. First evaluation of potentially flooded areas. The Thiessen polygons technique is used to extend the value of water surface elevation in the river axis to the whole analyzed region. In this way, the value of the water surface elevation in the nearest river point is assigned to every pixel of the DTM. Then this value is compared with the DTM height, so to realize a new map in which all the pixels that have a DTM height greater than water surface elevation are considered not potentially flooded, and conversely for the pixels in which the DTM height is smaller than water surface elevation. 2. Removal of lakes. The lakes, i.e. the areas considered flooded in the previous step but unreachable by the river flow because surrounded by not floodable terrain (DTM's height greater than water surface elevation in the nearest point of the river axis), have to be dried. 3. Hypothesis of orthogonal path from the river axis to the floodplain. The procedure checks which areas between the flooded ones at the end of the step 2 are surely reachable from the water by orthogonal path and which ones are protected, for example by levees, hence have to be dried. 4. Search for not orthogonal to river axis paths. The hypothesis introduced is that the water can flow along the maximum slope terrain direction, so it can reach areas previously considered not flooded. From a computational point of view, this step must follow the steps 2 and 3 to restrict the area that can be flooded by alternative paths. 5. Final map is obtained by the union of the maps computed in steps 3 and 4. The tool has innovative characteristics. In fact, even if it remains substantially one-dimensional, it takes into account the two-dimensionality of floodplain and flooding phenomena, introducing hypotheses that let to correct many typical errors of one-dimensional usually employed procedures. Hence, with respect to the use of one-dimensional model without GIS support, this procedure allows to obtain a more realistic perifluvial flooding map. With respect to a twodimensional model, it needs a lower computational effort, that allows to apply 502

3 it to very long river reaches (of the order of 100 km). However, the tool is applicable with reliable results for slowly rising fluvial floods, while it may lead to considerable errors in case of highly dynamic flooding phenomena (mountainous streams, dam or levee breaching, etc.) where a 2D approach is more appropriate (Marzocchi et al., 2014). 3 Application to the terminal reach of Magra River The Magra River flows at the border between Liguria and Toscana Region, in Italy (figure 1). The whole basin covers about 1700 km 2. The main tributary is Vara River, at 15.7 km from the Magra estuary. The Magra River estuary is taken as case study because of its remarkable critical situation, both historically and recently proven, as 2009 and 2011 flood events demonstrated. In the present application, the final 6 km of the Magra River, which are the most critical part of the river mainly due to dense population and extensive land use, were analyzed. Figure 1: The localization of the Magra River terminal area (2013 Liguria Region ortophoto). 3.1 Input data preparation The input data used for the application are: 1 m resolution DTM, supplied by Autorità di Bacino Interregionale del Fiume Magra. It comes from a LIDAR survey carried out for Ministero dell Ambiente, in WGS84 reference system and geographical coordinates; 55 cross sections, supplied by Autorità di Bacino Interregionale del Fiume Magra, in WGS84-UTM32 cartographic coordinates; the official expected flooding areas, supplied by Autorità di Bacino Interregionale del fiume Magra in WGS84-UTM32 cartographic coordinates, with respect to the 200 years return period flow rate. 503

4 The hydraulic profile was computed in HEC-RAS software knowing the flow rate and the river cross sections, assuming the flow as one-dimensional and stationary. A xyz file, describing the cartographic position of every cross section and the associated water surface elevation, was imported in GRASS. Such file was created by the script hec2grass.sh developed by Roberto Marzocchi ( combining the cartographic coordinates of the first upstream cross section, the geo-referenced river axis and the HEC-RAS profile file, in which, for every cross-section, the water surface elevation is associated to the progressive distance from the first upstream one (figure 2). Figure 2: hec2grass.sh working procedure. 3.2 The first results The procedure was applied to the 200 years return time flow rate, corresponding to a 6400 m 3/s water discharge. The results of the 5 sequential steps of the procedure, described in par. 2, are illustrated in figure 3. They showed a fault in the procedure dealing with curves in the river axis, in the 5 th step map. In fact, there are important and unrealistic differences in water surface elevation between neighboring points downstream the meanders (figure 4). Hence a modification of the code was performed. Figure 3: The results of the 5 steps of the r.inund.fluv procedure applied to the terminal reach of Magra River for the 200 years return time discharge. The Figures 3.4a and 3.4b show the identified non-orthogonal to river axis paths and the resulting flooded areas respectively. 504

5 Figure 4: A zoom of Figure 3.5 that highlights an example of unrealistic discontinuity of water surface elevation downstream the meander. 3.3 r.inund.fluv code modification The previously highlighted unrealistic discontinuity in the final map (Figure 4) is due to the Thiessen polygons approach of the first step of the procedure. The presence of a curve produces a singularity in the center of the curvature (point P in Figure 5a), so that many different values of water surface elevation (H1, H2, ) can be assigned to point P, through a central projection. The problem was simply solved by using a Matlab code to achieve a river axis rectification, connecting the first and the last point of the reach with a line and orthogonally projecting on it the water surface elevations computed by HECRAS along the river axis (as showed in Figure 5b). Figure 5: A scheme representing the singularity (5a) and the rectification principle (5b). The starting and final points for axis rectification have to be chosen by the user, depending on the analyzed river configuration. In the present case, the first and the last point of the river reach were used. Moreover, note that, in case of more marked meandering rivers, the procedure might need a meander rectification through a polyline instead of a simple line. After the modification, the results of the first step of the procedure is a 505

6 sequence of strips orthogonal to the rectified axis, to each of which corresponds the water surface elevation of the closest point of the rectified axis. The rectified profile was used only in the first phase of procedure, while the following ones refer to the real river axis. The new final flooding map is illustrated in Figure 6, where the elimination of discontinuities downstream the meanders can be appreciated. Figure 6: Final flooding map obtained using the rectified profile (in blue) instead of the original one (in red) in the first phase of the r.inund.fluv procedure. 3.4 Comparison with the official expected flooding map A comparison between the calculated final map and the official expected flooding map for the 200 years return time, supplied by Autorità di Bacino Interregionale del fiume Magra, was performed. The qualitative agreement is illustrated in Figure 7. The blue areas are the perfect correspondence areas, the red ones are areas flooded in the official map and not considered flooded by r.inund.fluv procedure, conversely the yellow ones. It is possible to notice a satisfying correspondence from a qualitative point of view, that confirms the good quality of the procedure. From a quantitative point of view, a performance index (Bates et al., 2000; Di Baldassarre et al., 2006) was computed as follows: PI= AreaF Area I AreaF Area I where AreaF is the flooded area in the official map and Area I is the flooded area computed with r.inund.fluv. The value assumed by the performance index is 72%, which means that the two maps have a correspondence for the 72% of their extent (the blue area in Figure 7). 506

7 Figure 7: Comparison between the calculated final map and the official expected flooding one. The main differences are due to the fact that the r.inund.fluv procedure does not consider the tributaries, which play a fundamental role in the official expected flooding map, as Figure 7 shows. 4 Conclusions The presented work consists in the realization of potentially flooded areas maps by means of an automatic GIS procedure using the r.inund.fluv tool, starting from a water surface elevation profile, computed through HEC-RAS along the river axis, and a high resolution DTM. Despite the procedure is based on a mono-dimensional hydraulic code to obtain the water surface profile, it can reproduce the bi-dimensionality of the terrain and of the flooding phenomena, keeping the simplicity and the computational speed of a 1D model. The whole procedure is automatic, but it is subdivided in 5 consecutive phases that, starting from very simple hypothesis, allow to refine the results and obtain the map of potentially flooded areas. The application of the r.inund.fluv tool on the terminal reach of Magra River led to a modification of the code to take into account the meanders of the river, introducing a rectified profile to be used only in the first step of the procedure. The resulting map for the 200 years return period flow rate was compared with the official expected flooding map, showing a good correspondence, both from qualitative and quantitative point of view. A deeper application of r.inund.fluv to the terminal reach of Magra River, analyzing different water discharges and constructive solutions for the reduction of flood-prone areas, will be object of a future publication. Moreover, an improvement of the code to take into account the contribute of tributaries in flooding event will be studied. Acknowledgment We wish to thank the Autorità di Bacino Interregionale del fiume Magra for the cartographic material provided. 507

8 References Bates, P.D., & De Roo, A.P.J. (2000). A simple raster-based model for floodplain inundation. Journal of Hydrology, 236, Di Baldassarre, G., Castellarin, A., Brath, A., Horritt, & M., Bates, P. (2006). Modellistica idraulica monodimensionale: alcune considerazioni applicative sul grado di dettaglio ottimale della descrizione topografica. XXX Convegno di Idraulica e Costruzioni Idrauliche. Federici, B., & Sguerso, D. (2007). Procedura automatica per la creazione di mappe di potenziale inondazione fluviale. Bollettino SIFET, 4, GRASS Development Team (2013). Geographic resources analysis support system (GRASS) software, version Marzocchi, R., Federici, B., & Sguerso, D. (2009). Procedura automatica per la creazione di mappe di potenziale inondazione fluviale in GRASS: il modulo r.inund.fluv. Atti del IX Meeting degli Utenti Italiani di GRASSGFOSS febbraio 2008, Marzocchi, R., Federici, B., Cannata, M., Cosso, T., & Syriou, A. (2014). Comparison of one-dimensional and two-dimensional GRASS GIS models for flood mapping, Applied Geomatics, 6(4), doi: /s , ISSN: , Publisher: Springer Berlin Heidelberg. Peggion, M., Bernardini, A., & Masera, M. (2008). Geographic information systems and risk assessment. Scientific and Technical Research series EUR EN. Pozzoni, M., Marzocchi, R., & Graf, A. (2009). Roggia Scairolo Zonazione della pericolosità per alluvionamento. Technical report of Institute of Earth Sciences. 508