FLOOD RISK RELATED TO A FLUVIAL SYSTEM 1 MODIFIED BY DAMS WITH EMPHASIS ON 2 MORPHODYNAMIC AND HYDROLOGICAL ASPECTS 3

Abstract. Villa Dolores and peripheral localities (Córdoba province, Argentina) flood risk was analyzed. They have expanded their urban area on the inactive channel of De Los Sauces River after the dam construction. Downstream receives water from 5 sub-basins whose upper basins, located in Mountains, generate the largest flash floods affecting these locations. The risk was analyzed considering the hazard in three different threat scenarios and vulnerability. The areas of greatest risk are reduced and they are confined to urbanized fluvial belt sectors in historical floodplain and low terraces. Geomorphological studies were effective for risk estimation, being irreplaceable in the hazard mapping.


Especially in the Córdoba province the major rivers whose upper basins are developed in the Pampeans Mountains (Suquía, Xanaes, Ctalamochita, Cruz del Eje, De Los Sauces, among others) have been intervened with dams since the first decades of the last century to generate electricity and, secondly, to store water for irrigation, drinking and recreational purposes.
Although there are some studies related to the problems associated with flooding from these rivers (Barbeito and Ambrosino, 2004, Barbeito et al., 2004, Echevarria et al., 2017), there are still few cartographic works on flood hazard and risk.
Regarding the methodologies used to predict flood risk, the studies based on the mathematical treatment of idealized parameters are the most used (Baker, 1994).They assume that they have similar real events, whose results are imposed on society through engineering designs, zoning of flood hazards, among others.In contrast to this perspective, the geomorphological studies make a more comprehensively approach to floods (Sayed and Haruyama, 2016), considering sedimentological, morphological, and paleohydrological records, among others, and larger spatial and temporal scales (Garzón Heydt, 1985;Schumm, 1977Schumm, , 2005;;Baker, 1986Baker, , 1994;;Baker et al., 1988;Benito y Thorndycraft, 2005).The complementation of conventional hydrological approaches and structural and non-structural measures (Dewan et al., 2007, Masood andTakeuchi, 2012) arising from geological-geomorphological studies, allow the development of more comprehensive hazard and risk flood maps, which would favor an adequate territorial planning.
In relation to environmental maps different meanings of the concept of risk are used (Hermelin, 1991;Panizza, 1992;Bosque Sendra et al., 2005;Kron, 2005 Fedeski andGwilliam, 2007;Merz, et al., 2007;Vilches and Reyes;2011, Field et al., 2012, Koks et al., 2015, among others) and therefore they may have different readings.In this work we use the definition of Panizza (1992) who considers risk as the product of the interaction of the hazard derived from a natural process with the vulnerability of the anthropogenic environment.Using this conceptual basis, the flood risk of the city of Villa Dolores and peripheral localities (Córdoba province, Argentina -Fig. 1) was analyzed.This objective was planned taking into account that the mentioned towns have expanded their urban areas and services on the inactive channel of the De Los Sauces river after the Medina Allende dam construction in 1942.

Sect 3. Regional setting
The study area is part of the Pampean Mountains geological province of Córdoba (Ramos, 1999) which exhibit a classic tectonic setting defined by regional faults that control the outstanding elements of the landforms.Towards the East the Grandes and Pocho Mountains stand out.These are composed by metamorphic and granitic rocks (Precambrian-Lower Paleozoic), where the highest altitudes (maximum height 2,900 masl) and slopes (maximum 30 %) were observed.Next to the mountainous front, two levels of alluvial fans of Pleistocene and Holocene age (Bonalumi et al., 1999) are developed, forming strongly ondulating reliefs (maximum slopes on the order of 10 %), which show neotectonic activity evidences.This piedmont environment is formed by conglomerates with clasts of very variable sizes, with a psefitic to sand-silt matrix, covered in the middle distal sectors by loess and/or fluvio-aeolian materials (Fig. 2).A noticeable drainage network has been developed in these alluvial fans, nowadays with very deepened courses.
The western zone of the study area is part of a major intermontane depression (Quines-Ulapes-Chancani) and is constituted, mainly by Cenozoic fluvial sequences of varied energy pertaining to the alluvial fan of De Los Sauces River and, secondarily, by Quaternary aeolian loessic and sandy sediments (Fig. 2).The climate of the zone is semiarid mesothermal (Thornthwaite, 1948).Average annual precipitation is 628.2 mm yr -1 for 1961-2014 period (Villa Dolores weather station-National Meteorological Service).However, there are variations in the precipitation values due to landform.In this way, the rainfall gradually decrease from the mountains towards the plain (Gorgas et al., 2003).Approximately 77 % of the precipitation is concentrated in the spring-summer period and is responsible for the largest flood events associated with the watercourses.

Sect. 4. Methodology
In this work the risk concept of Panizza (1992) was used which is synthesized in the following equation:

RISK= HAZARD x VULNERABILITY
The Hazard represents the Susceptibility or natural fragility of a region exposed to a certain Threat.The susceptibility includes the geological, geomorphological, lithological, hydrological, geotechnical aspects, among others, that together determine the behavior of an area in front of a natural process (Panizza, 1992), whereas the Threat, according to Hermelin (1991), is the probability of occurrence of a potentially destructive phenomenon within a specific time period for a specific area.Finally, the Vulnerability includes the population aspects, social organization, economy, programming, cultural, historical and natural values of interest for the preservation (Panizza, 1992, Cendrero, 1987).
Currently there is a wide range of procedures adopted for the realization of flood maps with the use of GIS tools (Domínguez Chávez et al., 2015).
In this work, layers overlay tools in vector format for the elaboration of cartography at a detailed scale were used.
The Villa Dolores topographic maps, scale 1: 50,000 (National Geographic Institute), aerial photographs from1970 (approximate scale 1: 20,000), satellite images of the Google Earth software were used.However, and taking into consideration the scarce accuracy that the satellite images have for this detailed work, the field survey data was the main input for this study.Thus, water-level marks (considering vegetation, sediment distribution, erosion features and witness For the analysis of the flood susceptibility, the geomorphological characteristics of the area were taken into account, considering the topographic aspects (elevation, slope).Particularly, the morphological and morphometric changes of the main river channel as a result of the Medina Allende dam operation were evaluated.The previous scenario was reconstructed from aerial photographs from 1970 and the field survey evidences.
To assess the threat, three flood scenarios of different magnitude and recurrence were considered, resulting from the combination of discharges controlled by the partial or total sluice gates opening in the Medina Allende dam and the flow coming from the not intervened sub basins.For the flood transport assessment, land use types in the channel and in the flood plain were taken into account, especially the percentage vegetation cover and the activities and structures that affect the flow distribution and the roughness coefficient.
Due to the lack of records in the courses without interventions the flood flows were estimated for the extraordinary event on February 4, 2014 using the Manning equation (Chow, 1994) Where: R is the hydraulic radius (A/Wp), A is the cross section area, Wp is the wet perimeter, S is the channel slope (calculated for a stream reach from contour lines from topographic maps) and n is the roughness coefficient.Based on channel and floodplain characteristics (dominant particle sizes, type and percentage of cover vegetation, etc.) a "weighted n" was established (Chow, 1959).
The flood flows derived from the opening of sluice gates were informed by official sources (Epec-Provincial Energy Company of Córdoba).
To estimate the Vulnerability, the density of houses in urban and rural areas and the vial infrastructure (roads, bridges and accesses) were considered.

Subsect 5.1. Geomorphological and Topographic Characterization
The study area is located in the proximal sector of the alluvial paleofan (Neogene-Quaternary) of the De Los Sauces River, where the current course presents different incision degree and varied development of the fluvial belt.
The oldest terrace level (T1) has a slope on the order of 3-4 m and the lower level (T2) of 2-3 m.
The channel of De Los Sauces River shows variability, not only linked to geological controls but as a result of the operation of the Medina Allende dam.In general, the bedrock segment do not exhibit changes, while the alluvial channel lost its braided behavior, although it maintained its sinuosity, prevailing a semiconfined single channel with and erosive behavior.The channel width was reduced up to 85 %, generating a historical floodplain.The channel was segmented in three parts considering the most relevant morphological and morphometric characteristics in pre and post dam conditions (Table 1).

. Hydrology and Hydrometry
In the hydrographic map the medium and low reach of the De Los Sauces River, downstream of the Medina Allende dam, and also Las Tapias and Chuchiras streams are shown (Fig. 3).These courses drain the western scarp of the Grandes Mountains and have a torrential regime controlled by lithology, high slopes and summer rainfall intensity.In the piedmont sector, although the main collectors are incised in inactive alluvial fans, avulsion processes and transfers to neighboring basins can be registered in extraordinary floods, attenuating then the flood peaks.These sub basins are not instrumented therefore there is no systematic discharge records.

2014
) and flood discharge estimates with a recurrence of approximately 25-30 years (according to journalistic information and witnesses.In the De Los Sauces River, the only equipped station (La Viña Dam), located immediately downstream of Medina Allende dam, belongs to the National Water Resources Secretary and has been in operation from 1928 to 1980.The average discharge for this period is 5.6 m 3 s -1 , with a maximum of 900 m 3 s -1 .The Medina Allende dam has a storage capacity of 230 hm 3 and, through 8 sluice gates, it can evacuate an extreme discharge of 1,200 m 3 s -1 (EPEC, 2009).Lower discharge values, on the order of 30-40 m 3 s -1 , were registered in 2015 and 2016 by the partial opening of 4 sluice gates.
Subsect 5.3.Land Use: the defined units are presented in Fig. 5 and the main features are described below: 1) Urban Areas: The unit includes the Villa Dolores, Villa Sarmiento and San Pedro towns located on both margins of the De Los Sauces River.They show an important population growth (130-180 %) in the last 50 years (NU.CEPAL.CELADE, 2001).Although the expansion of these localities has been carried out in several directions, a moving towards the historic floodplain can be observed (Fig. 4), as occurred in the Villa Dolores western sector where the "Paso de la Virgen" densely populated neighborhood was placed.This leads to human impact increases such as waste disposal sites, soil infiltration capacity reduction, naturalness and functionality decrease (especially by loss of vegetation cover), water contamination, among others.
2) Agricultural -Horticultural: It extends mainly in the fluvio-aeolian plain (Fig. 4) covering most of the study area.This rural area has a very low population density.In general, the crops occupy small extensions and are irrigated through canals and/or ditches.Secondly, extensive livestock farming was observed.
3) Recreational: As it is observed in Fig. 5 all the recreational uses are located in the channel and the floodplain of the De Los Sauces River.Due to high pressure from land use in the summer period and the moderate to low vegetation cover degree, the Piedra Pintada bathing site highlights (Fig. 6).In this river reach, there are different facilities across the channel (a bridge, a ford, a small dike and a duct).A second bathing site is located immediately upstream of the bridge that connects Villa Dolores and Villa Sarmiento.This sector has scarce to null coverage of shrubs and trees, is heavily modified with channels, a swimming pool and roads, exhibiting significant environmental degradation.A motocross circuit, which generates local changes in the floodplain relief, interferes with the water flow distribution and also green spaces planned for hiking complements the recreational activities.These are the areas that show scarce human modifications and that preserves the highest percentage of tree species.

4) Mining:
the handheld extraction of sediments in the river channel and the floodplain of the De Los Sauces River is the main mining activity.It is carried out from the Villa Dolores towards downstream being the San Pedro area the more intensive exploited (Fig. 6).This activity generates a change in the channel morphology, causing changes in fluvial dynamics and a strong impact on the landscape quality.

5) Disposal of waste:
the dumping of illegal waste along the fluvial belt is a common practice, although three sites of greater relevance were detected.This activity is associated with the expansion of urbanized areas and the loss and null conservation of natural spaces (Fig. 6).Five susceptibility classes were defined (Table 3) which were evaluated in each geomorphological unit (Fig. 7).As can be observed in the map, the susceptible zones are those located in the most modern fluvial belt.Taking into account that it is incised in the paleo alluvial fan and then deepened, these zones have very low susceptibility.-Threat Analysis Three threat scenarios were defined according to the hydrological analysis, including flows of different magnitude and recurrence.
Scenario 1: Discharge values between 30 and 80 m 3 s -1 are considered, which include floods of low magnitude and recurrence periods less than 10 years.These are related to the streams not intervened and to the opening of the dam sluice gates.
The threat was divided into low and very low classes according to the characteristics of the channel and intervention degree and type, which condition the flow behavior (distribution, water stage).The first class corresponds to reach 1 (R1-Table 1) which is narrower, straight, on bedrock and with the highest slope.There the flow is conducted at high velocity and show the highest stages.Towards the end of this segment, with alluvial bed and vegetation, the roughness increases and the velocity decreases, increasing the water stage.On the other hand, the very low threat was defined for the alluvial channel reach, which is wider, sinuous, multichannel and highly impacted by sediment mining (Reaches 2 and 3 -Table 1).In this case, for the estimated flows, the water stage and flow velocity are lower.
In March 2015 a scenario of these characteristics occurred.The dam was at the limit of its storage capacity, so 4 sluice gates were opened evacuating a flow close to 30 m 3 s -1 .s -1 , respectively, while for De Los Sauces river a value of 130 m 3 s -1 was estimated.This scenario also considers discharges associated with the partial opening of sluice gates dam.
Scenario 3: Discharges of great magnitude and with recurrences greater than 50 years were estimated.This scenario would be associated to an extraordinary event added to an inadequate management of the dam.The reservoir would reach its maximum storage capacity evacuating a flow of approx.1,200 m 3 s -1 through the total opening of the 8 sluice gates.

-Hazard Analysis
In the Table 4 and Figs. 8, 9 and 10 the hazard maps for the three threat scenarios are showed.As it is observed only for the lower discharges, geomorphological differences and human interventions in the pre-dam channel, had incidence in the distribution of the threat and, therefore of the hazard, being between moderate and moderately low, in the straight and sinuous reaches, respectively.On the other hand, flows of this magnitude are conducted in the channel, without affecting the floodplain (Fig. 8).For intermediate discharges (Scenario 2 -Fig.9), a moderately high hazard in the pre-dam channel and moderate hazard for the floodplain were estimated.It was assumed for these flow values that differences between reaches are not relevant.On the other hand, considering the geomorphological aspects together with the occurrence of events of this magnitude, it is expected that the evaluated flows will not affect the terraces levels due to the degree of incision of the De Los Sauces River.Finally, the scenario 3 involves all geomorphological environments, resulting in very high and high hazard in the pre-dam and floodplain, respectively, until moderately low in the fluvio-aeolian plain, assuming the possible occurrence of overflows associated with paleochannels (Fig. 10).In general, the areas of greatest risk are reduced (10 %) and limited to the fluvial belt, in sectors where the population forms urban centres in the historical floodplain and in the low terraces of De Los Sauces river (Fig. 15 and 16).The fluvio aeolian plain presents no risk or it is low in extraordinary flood events.When scenarios of increasing threat are analyzed, the areas with risk increase from 5 % to 10 %, with risk classes from moderately low to very high and high.In turn, it is worth emphasizing that due to the advance of urbanization on the fluvial plain there are sectors with risk, even for small floods.

Sect. 5. Conclusions
The important incision degree of De Los Sauces River in its inactive alluvial fan explains the low percentage of area that has a moderately high to high flood risk.Only due to anthropogenic causes (channels, diversions to ditches, among others) could be considered cases where, in the event of extraordinary floods, overflows occur upstream of the urbanized areas and a risk scenario can be presented.
The morphological and morphodynamic changes of De Los Sauces River lower-middle reach, after the Medina Allende dam operation in 1942, caused an increase in the risk of flooding in Villa Dolores city and peripheral localities, for low to moderate magnitude floods.Although the dam controls the greater floods of De Los Sauces River, for minor events associated with not intervened tributaries and/or sluice gates opening, the advance of the urbanization in the pre-dam channel and floodplain causes changes in the hazard and vulnerability.In the first case, the spatial distribution of the threat is modified by obstructions, changes in the channel cross section, while in the second case, increases the exposure of population and associated infrastructure.

Figure 1 .
Figure 1.Location of the study area.Digital elevation model from SRTM (Shuttle Radar Topography Mission -data available from the U.S. Geological Survey).

Figure 6 .
Figure 6.Photos of different land uses in the fluvial belt of the De Los Sauces River.

Scenario 2 :
Discharge values considered are between 100 and 300 m 3 s -1 .In this case, moderate magnitude flood events are included, with a recurrence of 20-30 years associated to the tributaries that drain the scarp of the Grandes Mountains and come together downstream of Boca del Río dam.The events recorded in 1981 and most recently on February 4, 2014 represent this situation.In that event, Las Tapias and Chuchiras streams evacuated an estimated discharge of 129 and 200 m 3 Nat.Hazards Earth Syst.Sci.Discuss., https://doi.org/10.5194/nhess-2018-162Manuscript under review for journal Nat.Hazards Earth Syst.Sci. Discussion started: 28 June 2018 c Author(s) 2018.CC BY 4.0 License.

Figure 8 .
Figure 8. Flood hazard map associated with the De Los Sauces River for the first threat scenario.

Figure 9 .
Figure 9. Flood hazard map associated with De Los Sauces River for the second threat scenario.

Figure 10 .Figure 12 .
Figure 10.Flood hazard map associated with De Los Sauces River for the third threat scenario.

Figure 13 .Figure 14 .
Figure 13.Flood risk map associated with De Los Sauces River for the second threat scenario.

Figure 15 .
Figure 15.Photos of flooded sectors in 1981 with a discharge greater than 120 m 3 s -1 (scenario 2).The neighborhood seen in the photo was relocated after the flood.

Figure 16 .
Figure 16.Views of zone with moderate and high flood risk in overflows occurred on February 2014 (scenario 2).

Table 1 .
Most relevant morphological and morphometric characteristics of the channel in pre and post dam conditions.

Table 2 .
Hydrological and hydrometric characteristics of the main sub-basins
Figure 7. Flood susceptibility map associated with De Los Sauces River.

Table 4 .
Flood hazard classes considering three threat scenarios