Download as pdf or txt
Download as pdf or txt
You are on page 1of 17

Nat Hazards (2008) 45:277–293

DOI 10.1007/s11069-007-9164-8

ORIGINAL PAPER

Flood hazard delineation combining geomorphological


and hydrological methods: an example in the Northern
Iberian Peninsula

J. Lastra Æ E. Fernández Æ A. Dı́ez-Herrero Æ J. Marquı́nez

Received: 16 December 2005 / Accepted: 18 September 2006 / Published online: 5 February 2008
Ó Springer Science+Business Media B.V. 2008

Abstract Flood mapping requires the combination and integration of geomorphological


and hydrological-hydraulic methods; however, despite this, there is very little scientific
literature that compares and validates both methods. Two types of analysis are addressed in
the present article. On the one hand, maps of flood plains have been elaborated using
geomorphological evidence and historical flood data in the mountainous area of north-
western Spain, covering an area of more then 232 km2 of floodplains. On the other hand, a
hydrometeorological model has been developed (Clark semidistributed unit hydrograph) in
the Sarria River basin (155 km2, NW Spain). This basin is not gauged, hence the model
was subjected to a goodness-of-fit test of its parameter (curve number) by means of Monte
Carlo simulation. The peak flows obtained by means of the hydrological model were used
for hydraulic modeling (one-phase, one-dimensional and steady flow) in a 4 km2 urban
stretch of the river bed. The delineation of surface areas affected by floods since 1918, as
well as those analyzed subsequent to the geomorphological study, reveals a high degree of
reliability in the delineation of the flooded areas with frequent recurrence intervals
(\50 years). If we compare these flooded surface areas with the estimate obtained by the
hydrological-hydraulic method we can see that the latter method overestimates the extent
of the surface water by 144% for very frequent recurrence intervals ([10 years) and
underestimates it as the recurrence interval increases, by up to 80% less floodplain for

Originally presented at the Sixth International Conference on Geomorphology.

J. Lastra (&)
Aguas de la Cuenca del Norte, S.A., La Regenta, 23, 33006 Oviedo, Spain
e-mail: jlastra@acunor.es

E. Fernández
INDUROT, University of Oviedo, Oviedo, Spain

A. Dı́ez-Herrero
Geological Hazards Unit, Spanish Geological Survey, Madrid, Spain

J. Marquı́nez
Environmental Government Department, Northern Spain Water Authority, Oviedo, Spain

123
278 Nat Hazards (2008) 45:277–293

exceptional events ([500 years). Finally, a management map is put forth combining the
most reliable results available by integrating both methods.

Keywords Hydrological model  Hydraulic model  Geomorphological map 


Return period  Flooding

1 Introduction

Flood hazard mapping can be performed using different methods that can be grouped into
two main categories: geological-geomorphological methods and hydrological-hydraulic
methods. The first category is based on the examination of aerial photographs and field
work of evidence of overflow (Baker et al. 1988), whereas the latter methods calculate the
peak flows for specific events or for return periods, thereby obtaining the extent of the
water surface (Wheater 2002). A third set of data sources, those taken from historical and
paleohydrological flood registries, can be used as a complement to calibrate and validate
the use of these methods (Thorndycraft et al. 2003).
Numerous investigators highlight the effectiveness of the geomorphological analysis
for flood studies (Wolman 1971; Baker and Pickup 1987; Blair and McPherson 1994;
Klimek et al. 2003). This method is indispensable and works better than one-dimensional
hydrological-hydraulic methods when studying rivers where channel and floodplain
morphology is highly variable and changeable over time, where the river has a high
erosive potential, and the load transported is significant, or simply when information is
scant, the river is not gauged or the rainfall registries and other information needed to
develop the hydrological model is relatively inaccurate. Geomorphological analysis has
been integrated into studies designed to draw up flood-risk prevention plans in the
Mediterranean regions of southern France (Ballais et al. 2005) and in Italy (Guzzetti and
Tonelli 2004; Guzzetti et al. 2005).
Recently, analyses of many events that have taken place in this kind of area in northern
Spain in the last 50 years have demonstrated that studies based solely on the hydrological-
hydraulic method tend to underestimate flood frequency (Marquı́nez et al. 2006).
For decades, numerous studies have stressed the need to combine and integrate
geomorphological and hydrological-hydraulic methods to elaborate flood hazard maps,
which has come to be known as the hydrogeomorphological approach or method (Sidle
and Onda 2004; Ballais et al. 2005). However, these studies have generated a great deal
of vague literature for local use with scant public dissemination, used in methodological
guides (Garry et al. 1996, 1999), doctoral dissertations (Chave 2003), odd chapters of
books (Chave 2002), but very little scientific literature has been published in interna-
tional peer-reviewed journals. The recent proposals put forth by Kenny (1990),
Thompson and Clayton (2002), Hudson and Colditz (2003), and Ballais et al. (2005)
serve to name just a few.
In this paper we examine the geomorphological characteristics of 232 km2 floodplains
within a 1,400-km reach located in Asturias, a province in northwestern Spain (Fig. 1). We
collected data from geomorphological field surveys and combined them with detailed
information about recent events since 1918 to analyze the relationship between flood fre-
quencies and geomorphic evidence preserved on these floodplains. Moreover, in the village
of Sarria, in a 4-km2 area located in the province of Lugo, the results of a geomorphological
study and historical data of recent floods were compared with hydrological-hydraulic return

123
Nat Hazards (2008) 45:277–293 279

Fig. 1 Study area and extent of the different methods used

periods. In this case, we compared the results to analyze the pros and cons of each method
for flood analysis in rivers where the channel and floodplain morphology is highly variable
and changeable over time.

2 Study area

The study area is located in Asturias, in northwestern Spain. This area is in the Cantabrian
Range, a mountain chain that runs parallel to the northern coast of Spain from East to
West, with heights of up to 2,600 m and a maximum distance from the sea of 60 km,
giving rise to a hydrographic network that is highly sloped. Although the current
arrangement is the result of a tectonic rising movement during the Mesozoic and Tertiary
periods (Alvarez-Marrón et al. 1997; Boillot and Malod 1988), the bedrock is largely
Paleozoic. It is formed by a sequence of several sedimentary rocks in which formations
with different lithology (quartzite, sandstone, slate, and limestone) occur. The quaternary
geomorphological features recognized in the present landscape include landforms modeled
mainly by marine, glacial, fluvial, and slope mass movement processes, with fluvial
landforms being the most widespread. Rivers are incised onto both glacial landforms and
coastal plains, and have modeled an abrupt relief.
There are 85 independent catchments over the 23,177-km2 surface area (Fig. 1), the
largest of which measures 4,900 km2 and belongs to the Nalón River. Most of these
catchments do not have gauging station flood data available. Santos and Menéndez (2006)
suggest that 64% of the rivers in this area display behavior that is dominated by torrential
phenomena given that they are short, have high slopes, and great erosive power. The areas
with well-defined floodplains occupy a small extent and are located in the lower reaches

123
280 Nat Hazards (2008) 45:277–293

of large basins (Farias and Marquı́nez 1995). These reaches are characterized by high-
energy fluvial systems, where the rivers develop a meandering pattern and are dominated
by gravel sediment load (Fernández et al. 1997). Overbank erosive mechanisms and the
transfer of appreciable quantities of bed material onto the floodplain prevail in the
floodplain construction (Marquı́nez et al. 2006), offering the typical appearance of streams
with these features (Nanson 1986; Nanson and Croke 1992; Jarrett and Costa 1986). As
result of these processes, channel and floodplain morphology is highly variable and
changes greatly over time. Due to the abrupt relief, many urban and rural areas are located
on the floodplains and the societal and economic impact of flooding is high in these
regions.
Average rainfall and temperatures recorded over the last 20 years at 14 different
meteorological stations range from 1,000 to 2,000 mm per year and 5.5 to 11.5°C,
respectively (Menéndez and Marquı́nez 2002). The annual stream flow regime for the
rivers in northwestern Spain reflects the regional precipitation pattern. Stream flow is
uniformly low during the dry season in summer and both precipitation, and hence stream
flow, increase in winter and spring, with maximum discharge during the December to April
period. The Narcea River has a catchment size of 1,500 km2, which is average for the
catchments analyzed. The average daily discharge for this river is 53 m3 s-1, while the
average maximum discharge is 334 m3 s-1. The maximum bank height observed in the
channel is 6 m (60 m wide in the channel, and 700 m in the floodplain)
The hydrological study was conducted throughout the basin, over an area of 201.7 km2.
For the Sarria River (Lugo, Fig. 1) the geomorphological and hydraulic studies were
carried out over a smaller urban zone of about 4.1 km2. The orographic features are similar
to those found in the province of Asturias, in this case with altitudes ranging from 411 m to
1,471 m and with a mean slope of 12°. Annual rainfall is approximately 1,379 mm and
the average daily discharge for the Sarria River is about 3.6 m3 s-1, while the average
maximum discharge is 68 m3 s-1. The geological environment corresponds mainly to
silicic lithologies (granite and precambrian slates and sandstones).

3 Methodology

3.1 Historical flood information obtained from documentary sources and field
interviews with local residents

The three main sources of information used to compile the inventory of floods were
bibliographical data, newspapers, and field interviews with local residents.
In April 2001, the regional government requested an analysis of natural risks in the
province of Asturias, including floods. For this study, we compiled an inventory of sites
historically affected by floods during the period 1522–1983 using documentary sources and
prior national compilations (MOPU, 1985). The inventory was updated for the 1980–2001
period with data from 7,285 newspapers reports.
In June 2003, the regional department of civil protection requested a flood risk study at a
1:5,000 scale for Asturias. The inaccuracy of historical flood data obtained from the
previous compilation forced us to conduct 750 field interviews with local residents dis-
tributed over 232 km2 of the floodplain area. During these interviews we obtained
information regarding the characteristics of the different events and surface affected by
each, specifying the degree of certainty in the envelope line. Dates, damage (agriculture,
buildings, road connections, etc.), categories of water depth (\1 m, between 1 and 2 m,

123
Nat Hazards (2008) 45:277–293 281

[2 m), historic flood marks in urban areas, identification of sedimentation areas and
sediment grain size (lime, sand, gravel, and cobbles), secondary flows during flooding and
overflow zones, etc. were recorded. The information was drawn onto topographies at
1:5,000, providing a detailed description of the historical flood boundaries and charac-
teristics of each inundation. The information was stored in a database, including
photographs and videos of some floods and gauged flow and rainfall data for every event
when said information was available. This database is being updated to cover the period
since 2004 and we download reference information from newspapers on a daily basis. In
addition, we have conducted field surveys during recent floods in order to delineate the
surface affected by new inundations. We now have new data from 6 km2 (during floods
having return periods \10 years in October 2005, November 2005, and February 2006).
We applied the same method for the local study in the village of Sarria.

3.2 The geomorphological method

Geomorphological analysis is recognized as an essential part of our understanding of


floodplains, based on a simple principle: the floodplain boundaries of a stream correspond
to the envelope curve of its past floods (Lambert et al. 2001; Garry et al. 2002; Ballais
et al. 2005). Frequent analyses of the impact of flood events on the channel and the
floodplains have been performed, identifying morphological changes associated with
erosion and deposition processes using sedimentological studies and observations on the
floodplain flat (e.g., Magilligan et al. 1998; Benedetii 2003; Macklin et al. 2006).
Geomorphological maps delineate the morphologies created by these overbank
processes (Fig. 2), including crevasses and abandoned channels, recent and older overbank
deposits and erosive evidence, showing a landscape with heterogeneous topography (e.g.,
Baker and Costa 1987; Leopold et al. 1992; Brow 1996). Generally speaking, several
terrace levels of different heights from the main channel are also drawn, representing
active terraces with different flood frequencies in the floodplains (Leopold et al. 1992;
Smith 1997; Florsheim and Mount 2002). Steep banks or cliffs with different heights
(Fig. 2) demarcate the boundary between these terraces (Hudson and Colditz 2003; Ballais
et al. 2005).

Fig. 2 (a) Heterogeneous topography on the floodplain in relation to overbank deposits. (b) Typical scarp
or cliff between terraces

123
282 Nat Hazards (2008) 45:277–293

In the areas studied, the outer limit of the floodplains is easily drawn where the fluvial
systems are confined within high valley walls and are often defined by breaks in the slope.
Traditional aerial photo interpretation and field surveys have long been used to map
floodplains, identifying 232 km2 in the regional study in the province of Asturias (2.2% of
the territory) and 4.1 km2 in the local study of the village of Sarria. The geomorphological
analysis of the floodplain is based on fieldwork focused on delineating the overbank
evidence and qualitative observations of a 1,400-km-long channel section; channel bank
height was also estimated, given the importance it has in controlling flood processes
(Hudson and Colditz 2003). It was considered that these elements delineated four different
terraces on the floodplain: low, middle, high, and very high. The limits between terraces
are represented by steep banks or cliffs with a height ranging from a few decimeters to
1–2 m (Fig. 2). The maximum height observed in the highest terraces with respect to the
riverbed is 6 m in the biggest rivers (60 m wide in the channel and 700 m in the flood-
plain). Typically, however, we found very high terraces (up to 5 m), high terraces (3–4 m),
middle terraces (2–3 m), and low terraces (less than 2 m). The variability depends on the
river size and position of the floodplain in the basin.

3.3 The hydrological method

The aim of this method is to obtain peak flows and/or hydrographs for several return
periods (Fig. 3). Hydrological hazard analysis is usually carried out using two methods:
flood frequency analysis, which statistically analyzes the recorded peak flow series to
design the values associated with return periods, and the hydrometeorological one, which
obtains this data from rainfall series analysis. Normally, rainfall series are longer than peak

Fig. 3 Methodology scheme

123
Nat Hazards (2008) 45:277–293 283

flow series. This study has proved to be no different (only five-year peak flow data, lacking
statistical significance, were recorded); consequently, a hydrometeorological model was
developed for the hydrological analysis.
The HEC-HMS model was used for precipitation-runoff simulation (USACE 2000;
USACE 2001). This model requires information from the basin model, the meteorological
model, the infiltration losses and runoff transform methods, the reach routing method, and
the digital terrain models.

3.3.1 Rainfall analysis: the meteorological model

There are three main data sources for the meteorological analysis in this area: (1) a 58-year
24-h precipitation dataset, recorded in a rainfall gauge located in the village of Sarria; (2) a
national study (INM 2000) of the variable 24-hour precipitation conducted by the National
Meteorological Institute of Spain, which fits precipitation to a Gumbel function and plots
isovalue maps for the entire country, and (3) a national study (Ministry for Public Works
2001) of the same variable developed by the Hydrological Research Centre (CEDEX),
which fits the variable to a SQRT-ETmax function (Etoh 1986) using a regional scale factor.
These three data sources were compared in the analysis. The first data source was also
adapted to a SQRT-ETmax function because it is well known that this provides more-
realistic results than the Gumbel function for long return periods. Finally, the results
obtained by the CEDEX study were used, on the supposition that they provided a better
regional estimation. An intensity-duration-frequency (IDF) curve and a design storm
(synthetic hyetograph) were subsequently developed for each return period.

3.3.2 Terrain analysis: the basin model

Many studies support the importance of using a precise digital elevation model (DEM) for
characterizing the important relief patterns in hydrological studies (Krysanova et al. 1998;
Vanacker et al. 2003); therefore, a 10-m-resolution DEM was developed in this analysis.
The DEM was derived from the digitalization of the contour lines of the national topo-
graphical map 1:5,000 of the National Geographic Institute (IGN) and was created with the
ArcGIS software (ESRI Inc) using Hutchinson’s (1989) algorithm. Normal hydrological
revision of the DEM was brought about by filling artificial holes and ensuring continuity of
the hydrographic network, and a visual revision was performed in search of a realistic,
artifact-free pattern. Other digital terrain models required by the study were derived from
the DEM, such as: a flow-direction model, a flow-accumulation model, a hydrographic
network model using different threshold values in an attempt to achieve a realistic pattern
distribution and a subbasin model. Table 1 indicates the characteristics of the digital terrain
models.
Precipitation that does not infiltrate becomes direct runoff. The Clark unit hydrograph
was used to estimate runoff based on the estimation of two parameters: concentration time
(TC) and storage coefficient (R), which are usually expressed by the equation
R ¼ 0:6  TC :
The Muskingum method is based on the continuity equation. The method estimates two
parameters: the travel time constant for the reach (K) and a weighting factor (x) that varies

123
284 Nat Hazards (2008) 45:277–293

Table 1 Characteristics of the digital terrain models derived for the study area
DEM data source
The DEM was generated from a 1:5,000 topographic map with contour lines every 5 meters
DEM statistics
Minimum value (m) Maximum value (m) Mean value (m) Standard variation
(m)
411.10 1471.24 723.09 202.18
DEM error estimation
Absolute mean error (m) Quadratic mean error (m) Root quadratic mean error (m) Bias (m)
2.22 -0.16 0.97 4.94
DEM derived models
Digital model Notes
Flow direction Flow direction (ArcInfo 9.0)
Accumulated flow Flow accumulation (ArcInfo 9.0)
Stream definition Hec-GeoHMS (ArcView 3.2) with a 50,000 cell
threshold; 35 streams have been obtained
Watershed definition Hec-GeoHMS (ArcView 3.2); 24 basins have been
obtained

from 0 to 0.5 for a given reach, ranging from 0.2 to 0.3 for natural streams, where
2 Kx \ Dt B K for numerical accuracy.

3.3.3 Loss rate parameters: SCS curve number method

The SCS curve number (SCS 1972) is a well-known method used to calculate losses from
precipitation. This method was chosen for the analysis because: (1) it is commonly used in
different environments and provides good results; (2) its calculation is made easier by the
fact that only a few variables need to be estimated (hydrological soil groups, land use, and
slope); and (3) despite its simplicity, it yields results that are as good as more-complex
models.
The hydrological soil groups (NCRS 1999) was the most difficult variable to estimate
given the absence of a homogenous soil cartography of the study area. The SCS classifies
soils into four groups ranging from A (high infiltration rate) to D (low infiltration rate) and
has established a quantitative relationship between the minimum infiltration rates, the
texture characteristics of the soil, and hydrological groups (SCS 1986). They were esti-
mated from the only cartography available, the 1:50,000 scale geological map, relating
lithological textures to the hydrological soil groups (Témez 1987; Martı́nez de Azagra and
Navarro Hevia 1995) given that soil texture depends on lithology, topography, and
hydroclimatic characteristics.
The land use groups (SCS 1986) were developed for use in the United States and require
the use of conversion tables (Ferrer et al. 1995; CEDEX 1997) to adapt to the cartography
of the area, which is obtained from the year 2000 CORINE land-cover project (Joint
Research Centre, European Commission).
Slope was derived from the DEM, classifying the study area as more or less than 3%.
These three layers were combined in a GIS obtaining a curve number digital model,
which was aggregated for each watershed in the hydrological model.

123
Nat Hazards (2008) 45:277–293 285

3.3.4 Sensitivity analysis: Monte Carlo simulations

The hydrological model presented in this work has been developed in an ungauged basin
and where very little catchment information is available; it is therefore senseless to con-
sider calibrating parameters, since this is simply not possible. In cases like this,
a sensitivity analysis of the parameter with the greatest uncertainty is generally performed,
that is, the curve number (Singh and Quiroga 1996; Schneider and McCuen 2005). Hence,
a sensitivity analysis of simulated flows to uncertainties in the curve number parameter was
conducted by means of 250 Monte Carlo simulations for each return period rainfall input.

3.4 The hydraulic method

The peak flows obtained by the hydrological method were used as input data in the
hydraulic analysis using the HEC-RAS software (USACE 2002) (Fig. 3). This model is
based on the St. Venant equations when flow conditions are unidirectional, one phase
(it does not take sediment load into account), and gradually varying, and makes it possible
to estimate the water surface extent (from water height), depth, and velocity.
Hydraulic studies of flood areas require the use of a highly detailed topographic
dataset, which is decisive for the accuracy of the results. In fact, several studies (Aronica
et al. 1998; Wilson 2004) have shown that small errors in topography substantially impact
the model results. With this aim in mind, a topographic campaign was conducted using
GPS and survey techniques, obtaining 65,347 topographical points. Other features like
buildings, roads, and any other type of obstacle were also integrated into the topography
when developing the TIN model. Geometric data were derived from the TIN model,
channel roughness characteristics were estimated using Manning’s n coefficient in the
visual comparison approach (Barnes 1967; Arcement and Schneider 1989; Chow et al.
1988), and finally, the geometric data on eight bridges were incorporated into the model.
The HEC-RAS model was applied using a subcritical flow and, since there are no critical
flow sections upstream or downstream nor is there any possibility of using a rating curve,
normal depth boundary conditions were simulated given that the channel is more or less
longitudinally uniform.

4 Results

In the regional study conducted in the province of Asturias, the analysis of the docu-
mentary sources revealed that 109 sites had been affected by floods in the period 1522–
1983, but rarely more than once. One hundred and fifty-two new localities affected by
floods were identified by our revision of 7,285 newspaper reports published between 1980
and 2001. With this information, we found that 59% had been affected once or more, 31%
had been affected twice or more, 9% five times or more, while only 1% had been affected
10 times or more during this period. The information displays a lack of homogeneity
insofar as geographical distribution is concerned and displays a clear inclination towards
urbanized floodplains.
In order to carry out the flood study over a surface area of 232 km2 at 1:5,000 scale, 750
field interviews of local residents were needed, yielding a total of 70 different dates of
events identified during the period from 1918 to 2004. We also delineated areas affected by
recent floods in 71% of the active floodplain. We obtained water depth data in 43% of the

123
286 Nat Hazards (2008) 45:277–293

floodplains and these results revealed that the water depth on the floodplains is \1 m in
43%, C1 m in 50%, and C2 m in only 7%. Moreover, in the database we obtained
information about overbank deposits in 65 km2 of the floodplains. These sources yielded
information of different quality and in different amounts, but they provided data as to flood
frequency, the importance of overbank deposits during the events, and their potential
magnitude in the region.
Most of the information has to do with the big floods that have taken place during the
last seven decades, with the events of 1938, 1959, and 2003 being the most important and
covering the largest area. In approximately 40% of the floodplains, we have information of
two or more inundations. With this information we can estimate the return time or interval
of at least 50 years between two consecutive flood events (Forte et al. 2005).
The cartography of geomorphological evidence in 1,400 km of streams and 232 km2 of
channel floodplain enabled us to identify 44 km2 of channel, 70 km2 of low terrace,
40 km2 of middle terrace, 29 km2 of high terrace, and 33 km2 of very high terrace. The
remaining surface area corresponds to water in reservoirs and non-flooded artificial
landfills. The lower terraces conserved abundant overbank flood evidence. The maximum
height of overbank sediment observed on the floodplain was 2 m in the lower terrace, but
was typically between 1 m and 0.5 m (Fig. 2), visible in fieldwork and aerial photographs.
The low terraces commonly show large wooded portions of the floodplain adjacent to the
river channel. In the middle terraces, generally used for agricultural purposes, we observed
that the height of the overbank deposits was typically\0.5 m and that the microtopography
we identified in our fieldwork only was heterogeneous. In this case, the deposits show
limits gradually sloped to the floodplain surfaces. The higher terraces showed flat surfaces
without morphological evidence.
The boundaries of the terrace levels were compared with the surfaces historically
affected by some kind of flooding. We observed that 70% of the low and middle terraces,
where the morphologies and microtopography associated with erosive and depositional
flood evidence are located, have been affected by recent floods, in many cases by two or
more events. Ninety percent of the inundated areas identified during frequent floods after
this study coincided with low terraces. With the data regarding recent floods and geo-
morphic evidence, we assume a minimal flood return time of 10 years for the lowest
terraces and 50 years for the middle terraces, whereas for the highest terraces lacking
geomorphological evidence and flood data, the frequency is assumed to be lower.
The same analysis was applied to the 4 km2 of floodplains in the village of Sarria
(Fig. 1). Figure 6a shows the map of the hazard categories for flooding, delimited using the
geomorphological method. The differences in floodplain geomorphology made it possible
to determine that the low terrace occupies 23% (88 ha), the middle terrace covers 35%
(132 ha), the high terrace 25% (95 ha), while only 13% (52 ha) corresponds to very high
terrace. We found that 10 noteworthy floods had occurred in the study area over the last
century, the most extensively documented flood being that of 2000.
We compared the geomorphological mapping with areas that had been flooded two or
more times in the last 50 years. In 84% of the low and middle terrace, we have information
of recent floods assuming a return time of at least 50 years, although for the remaining area
we are uncertain of the return time.
Table 2 presents rainfall data obtained by meteorological analysis and some of the
main physical and hydrological parameters obtained using the previously described
methodology. The basin model generated is broken down into small subbasins (\23 km2)
with concentration times of less than 5 h. The estimated curve number in the area varies
from 60 to 89 owing to: the slope (98% of the study area has a slope greater than 3%), the

123
Nat Hazards (2008) 45:277–293 287

Table 2 Estimated data obtained from rainfall and basin models


Meteorological model

Return periods (years)


500 100 50 10
Estimated rainfall (mm) 198 155 138 100
Basin model

Physiographic parameters Minimum Maximum


Subbasin area (km2) 0.1 23
Mean subbasin elevation (m) 415 870
Mean slope of the river reach (m/m) 0.002 0.06
Loss rate
Curve number 60 89
Clark transform
Concentration time (hours) 0.3 4.6
Storage coefficient (hours) 0.2 2.8
Muskingum routing
Muskingum K (hours) 0.2 3.4
Muskingum X 0.3 0.3

predominance of agricultural land use and forests, and the predominance of hydrological
groups types B and C.
The peak flows estimated by the hydrometeorological model are shown in Table 3 and
graphically in Fig. 4. These peak flows were estimated using the value of the curve number
parameter that was deemed most appropriate based on the data available and knowledge
about the area. The influence of curve-number estimation error on the uncertainty of the
hydrological response was further examined with Monte Carlo simulation. The results are
presented in Table 4 with the corresponding hydrographs shown in Fig. 5. It is clear that
the differences between the simulated mean peak flow value and the mean peak value
estimated by the hydrometeorological method are small, ranging from -2.9% for the 500-
year return period to -6.6% for the 10-year return interval, at any rate, well within the 95
percentile in all cases.
These peak flows were used as inputs for the hydraulic model that delineates flooding
for different return periods as shown in Fig. 6b. In addition to that of estimating peak
flows, the uncertainties inherent to this kind of study must be added (Turner-Gillespie et al.
2003), such as inaccuracies in representing the valley-bottom topography or in estimating

Table 3 Estimated peak flows for 10, 50, 100 and 500 years return periods
Return periods (years)

500 100 50 10

Estimated peak flows (m3/s) 506 360 261 127

123
288 Nat Hazards (2008) 45:277–293

Fig. 4 Hydrograms estimated by means of the HEC-HMS model for return periods of 2, 10, 50, 100, and
500 years in the confluence

Manning’s roughness. This hazard map shows that 59% (215 ha) of the floodplain has a
10-year return period, 31% (115 ha) has a 50-year period, 7% (25 ha) corresponds to the
100-year return period, and 3% (10 ha) to the 500-year return period.
A flood hazard management map (developed from the combination of both the
geomorphological and the hydrological maps) (Fig. 6c) was elaborated based on our
knowledge of the geomorphology together with the limited rainfall runoff observations
used in developing the hydrological-hydraulic model. This map shows that 30%
(112 ha) of the study area is very often subject to flooding, 42% (160 ha) undergoes
frequent flooding, 15% (58 ha) suffers occasional flooding, while 9% (33 ha) is seldom
flooded.

Table 4 Peak flows estimated by a Monte Carlo sensitivity analysis of the curve number parameter
Simulated peak flows (m3/s) Return periods (years)

500 100 50 10

Mean 521 376 278 136


Maximum 576 412 330 174
Minimum 471 330 225 102
Stand. dev. 19 18 22 14
95% probability (max/min) 554/481 409/341 320/235 165/107
90% probability (max/min) 550/487 403/344 313/240 161/111
80% probability (max/min) 545/492 399/349 306/248 154/118
Difference between the mean simulated peak -2.9 -4.3 -6.1 -6.6
flow and the estimated peak flow using
the hydrometeorological model (%)

123
Nat Hazards (2008) 45:277–293 289

Fig. 5 Hydrograms estimated by a Monte-Carlo sensitivity analysis of the ‘‘curve number’’ parameter

5 Discussion and conclusions

Floodplain delineation and recurrence periods obtained by the geomorphological method


are tremendously reliable when there is abundant geomorphological evidence and registers
of recent events are available. In this paper, we have found a relation between the
preservation of overbank evidence on the floodplain flat (low and middle terraces) with

123
290 Nat Hazards (2008) 45:277–293

Fig. 6 Flood hazard maps created by the: (a) geomorphological-historical method, (b) hydrological-
hydraulic method, and (c) combination of both methods to elaborate a management map

high flood frequency (return time of 10 and 50 years) using a high-quality database of
recent flood data since 1900. This close relation could have to do with the fact that
sediment is readily available in the studied rivers (Fernández et al. 1997; Marquı́nez et al.
2006) and the low bank height of the channel-floodplain boundary to the riverbed in an
area with young fluvial relief (Farias and Marquı́nez 1995).
There is scarce geomorphological evidence for the highest terraces, either because it has
not been conserved or because the extreme events that caused them were low impact, and
hence caused minimal erosion to deposition (Dury 1973; Costa 1974; Nanson 1986; Nash
1994; Hooke and Mant 2000). We have assumed a lower frequency for the surfaces of
highest terraces that do not present geomorphological evidence or flood data; however, we
lack the information needed to consider an exact return time.
We applied both geomorphological and hydrological-hydraulic methods in a local study
of the village of Sarria. The application of the latter methods was complicated by two
issues: calibration of the hydrological model and the special geomorphological charac-
teristics of the area. On the one hand, it is an ungauged basin, making it impossible to
calibrate the hydrological model. Instead, as generally done in these circumstances (Singh
and Quiroga 1996; Schneider and McCuen 2005), we performed a sensitivity analysis of
the parameter with the greatest uncertainty involved in its estimation (the curve number) as
a result of the scarcity of available data. This sensitivity analysis enabled us to verify how
the flow rates obtained by means of the hydrometeorological model were very slightly
underestimated with respect to the random mean obtained by using Monte Carlo simula-
tions; we therefore considered its use appropriate for this study.
On the other hand, as previously commented, the study area has a young hydrographic
network, with steep slopes that endow it with a great deal of energy and transport abundant
bed load and where the channel frequently changes course. Hence, the magnitudes of rare
flood events have been underestimated in this study area when based solely on the
hydrological-hydraulic method described by several researchers (Black and Burns 2001;
Thomson and Clayton 2002; Ballais et al. 2005). The hydrological-hydraulic method
overestimates the extension of the water surface compared to the estimates provided by the
geomorphological method and the recent flood data by 144% for very short recurrence

123
Nat Hazards (2008) 45:277–293 291

periods ([10 years) and underestimates them as the recurrence period increases, by up to
80% less floodplain area for exceptional events ([500 years) compared to the surface area
delineated as exceptional by the geomorphological method.
Some authors (Black and Burns 2001; Thompson and Clayton 2002; Ballais et al. 2005)
recommend the use of recent flood extent and geomorphological analyses to complement
hydraulic modeling and improve the identification of flood hazards. This study combines
both methods to delineate the extent of 10-, 50-, 100- and 500-year flooding on the basis of
the associations established between geomorphological evidence and estimated return
times using high-quality data based on recent flood scenario reconstructions and hydraulic
modeling. The latter method has been taken into account particularly in delineating
floodplains in places where less geomorphological evidence and historical data was
available.
Methodology and findings from this study are significant because most hydrogeomor-
phological approaches (Hudson and Colditz 2003; Arnaud-Fassetta et al. 2005) are based
mainly on flood event analyses and not on delineating the extent of floods for different
intervals. This kind of analysis is fundamental for planners and river engineers to answer
management-related questions (Sidle and Onda 2004).
In accordance with other authors (Garry 1994; Garry et al. 2002; Ballais et al. 2005)
and to conclude, we would like to point out that, in addition to being economical and easily
applied to large areas, the geomorphological method has proven to be more consistent than
using only hydrological-hydraulic methods, especially in rivers with paths that vary in
terms of both space and time or that carry abundant load, since it is based on physical
criteria that reflect the present evidence of fluvial activity.

References

Álvarez-Marrón J, Rubio E, Torne M (1997) Alpine age subduction structures in the North Iberian Margin.
J Geophys Res 102:22495–22511
Arcement G, Schneider V (1989) Guide for selecting Manning’s roughness coefficient for natural channels
and flood plains. US Geological Survey Water-Supply Paper 2339
Arnaud-Fassetta G, Cossart E, Fort M (2005) Hydro-geomorphic hazards and impact of man-made structures
during the catastrophic flood of June 2000 in the Upper Guil catchment (Queyras, Southern French
Alps). Geomorphology 66:1–4
Aronica G, Hankin B, Beven KJ (1998) Uncertainty and equifinality in calibrating distributed roughness
coefficients in a flood propagation model with limited data. Adv Water Res 22(4):349–365
Baker VR, Pickup G (1987) Flood geomorphology of the Katerine Gorge, Northern Territoriy, Australia.
Geol Soc Am Bull 98:635–646
Baker VR, Costa JE (1987) Flood power. In: Mayer L, Nash D (eds) Catastrophic Flooding. Unwin, Boston
and London, pp 1–21
Baker VR, Kochel RC, Patton PC (1988) Flood geomorphology. Wiley Interscience, Toronto (Canada), 503
pp
Ballais JL, Garry G, Masson M (2005) Contribution of hydrogeomorphological method to flood hazard
assessment: the case of French Mediterranean region. CR Geosci 337(13):1120–1130
Blair TC, Mcpherson JC (1994) Alluvial fans and their natural distinction from rivers based on morphology,
hydraulic processes, sedimentary processes, and facies assemblages. J Sediment Res 64(3):450–489
Barnes HH (1967) Roughness Characteristics of Natural Channels. USGS Water Supply Paper 1849
Black AR, Burns JC (2001) Re-assessing the flood risk in Scotland. Sci Total Environ 294:169–184
Benedetii MM (2003) Controls on overbank deposition in the Upper Mississippi River. Geomorphology
56(3–4):271–290
Boillot G, Malod J (1988) The north and north-west Spanish continental margin: a review. Rev Soc Geol
España 1:295–316
Brown AG (1996) Floodplain and palaeoenvironments. In: Anderson MG, Walling DE, Bates P (eds)
Floodplain processes. J. Wiley & Sons, Chichester, pp 95–138

123
292 Nat Hazards (2008) 45:277–293

CEDEX: 1997, Utilización de la Teledetección para la estimación del parámetro hidrológico del número de
curva. Non published report of the Hydrological Research Centre (CEDEX), Ministerios de Fomento y
Medio Ambiente
Chave S (2002) Pertinence de la cartographie hydrogéomorphologique dans l’approche des inondations
rares à exceptionnelles: exemples de sept bassins fluviaux dans les Corbières et le Minervois. Géo-
morphologie: relief, processus, environnement 4:297–306
Chave S (2003) Elaboration d’une méthode intégrée de diagnostic du risque hydrologique. Mémoire de
Thèse de géomorphologie, Université de Provence (Aix-Marseille I), 284 pp (unpublished)
Chow VT, Maidment DR, Mays CW (1988) Applied hydrology. McGraw-Hill, New York
Costa JE (1974) Response and recovery of a Piedmont watershed from tropical storm Agnes. Water Resour
Res 10:106–112
Dury GH (1973) Environmental geomorphology and landscape conservation. In: Coates (eds) Dowden,
Hutchinson and Ross, Stroudsburg, Pa., and Wiley, Chichester, Earth Sci Rev 9(3):286–287
Etoh T, Murota A, Nakanishi M (1986) SQRT—Exponential type distribution of maximum. In: Proceedings
of international symposium on flood frequency and risk analysis. Louisiana, pp 235–265
Farias P, Marquı́nez J (1995) El relieve. In: Aramburu C, Bastida F (eds) Geologı́a de Asturias. Ed. Trea,
Gijón, pp 163–173
Fernández FJ, Menéndez Duarte R, Marquı́nez J (1997) Aplicación de un Sistema de Información Geog-
ráfica en la cartografia temática y clasificación Geomorfológica de los sistemas fluviales en Astúrias.
Rev Soc Geol España 10(1–2):117–130
Ferrer M, Rodrı́guez J, Estrela T (1995) Generación automática del número de curva con Sistemas de
Información Geográfica. Ingenierı́a del agua 2(4):43–58
Florsheim JL, Mount JF (2002) Restoration of floodplain topography by sand-splay complex formation in
response to intentional levee breaches, Lower Cosumnes River, California. Geomorphology 44:67–94
Forte F, Pennetta L, Strobl R (2005) Historic records and GIS applications for flood risk analysis in the
Salento peninsula (southern Italy). Nat Hazard Earth Syst Sci 5:833–844
Garry G (1994) Évolution et role de la cartographie dans la gestion des zones inondables en France, Mappe
Monde, 4:10–16
Garry G, Masson M, Ballais JL (1996) Cartographie des zones inondables: approche hydrogéomorpho-
logique. Minist. de l’Equipement, Minist. de l’Environnement, Les Editions Ville et Territoires, Paris
La Défense, 100 pp
Garry G, Graszk E (1999) Plans de prevention des risques naturels (PPR). Risques d’inondation. Guide
méthodologique. Minist. de l’Environnement, Minist. de l’Equipement, La Documentation Française,
Paris, 123 pp
Garry G, Ballais JL, Masson M (2002) Cartographie des zones inondables. Approche hydrogéomorpho-
logigue, Éditions Villes et Territoires, Paris La Défense
Guzzetti F, Stark CP, Salvati P (2005) Evaluation of flood and landslide risk to the population in Italy.
Environmental Mangement 36(1):15–36
Guzzetti F, Tonelli G (2004) Information system on hydrological and geomorphological catastrophes in
Italy (SICI): a tool for managing landslide and Flood hazards. Nat Hazard Earth Syst Sci 4:213–232
Hooke JM, Mant JM (2000) Geomorphological impacts of a flood event on ephemeral channels in SE Spain.
Geomorphology 34:163–180
Hudson PF, Colditz RR (2003) Flood delineation in a large and complex alluvial valley, lower Panuco basin,
Mexico. J Hydrol 280(1–4):229–245
INM (2000) Las precipitaciones máximas en 24 horas y sus perı́odos de retorno en España, vol 1: Galicia,
Ministerio de Medio Ambiente
Jarrett RD, Costa JE (1986) Hydrology, geomorphology, and dam break modelling of the July 15, 1982,
Lawn Lake Dam and Cascade Lake Dam failures, Larimer County, Colorado. U.S. Geological Survey
Professional Paper 1369, 79 pp
Kenny R (1990) Hydrogeomorphic flood hazard evaluation for semi-arid environments, Q J Eng Geol
23(4):333–336
Klimek K, Malik I, Owczarek P, Zygmunt E (2003) Historical flood evidence using geomorphological and
dendrochronological records, Sudetes mountains, Central Europe. In: Thorndycraft VR et al (eds)
Palaeofloods, historical data & climatic variability. Applications in flood risk assessment, pp 61–72
Krysanova V, Müller-Wohlfeil DI, Becker A (1998) Development and test of a spatially distributed
hydrological/water quality model for mesoscale watersheds. Ecological Modelling 106:261–289
Lambert R, Gazelle F, Gholami M, Prunet C (2001) La cartographie informative des zones inundables.
L’exemple de Midi-Pyrénées, in: Actes du colloque Au chevet d’une catastrophe, Presses universitaires
de Perpignan, Perpignan, pp 147–164
Leopold LB, Wolman MG, Miller JP (1992) Fluvial processes in geomorphology. Dover, 522 pp

123
Nat Hazards (2008) 45:277–293 293

Macklin M, Benito G, Gregory K (2006) Past hydrological events reflected in the Holocene fluvial record of
Europe. Catena 66:145–154
Magilligan FJ, Phillips JD, James LA, Gomez B (1998) Geomorphic and sedimentological controls on the
effectiveness of an extreme flood. J Geology 106:87–95
Marquı́nez J, Lastra J, Fernández E (2006) Metodologia utilizada para cartografiar la peligrosidad de
inundaciones en las cuencas del Norte. In: Dı́ez-Herrero A, Laı́n Huerta L, Llorente Isidro M (eds)
Mapas de peligrosidad de avenidas e inundaciones: Métodos, experiencias y aplicación. Instituto
Geológico y Minero de España, pp 125–141
Martı́nez de Azagra A, Navarro Hevia J (1995) Hidrologı́a forestal. El ciclo hidrológico. Universidad de
Valladolid, Valladolid
Menéndez Duarte R, Marquı́nez Garcı́a J (2002) The influence of environmental and lithologic factors on
rockfall at a regional scale: an evaluation using GIS. Geomorphology 43:117–136
Ministry for Public Works (2001) Máximas lluvias diarias en la España peninsular. Dirección General de
Carreteras
MOPU (Ministerio de Obras Públicas y Urbanismo): (1985) Cuenca del Norte de España: inundaciones
históricas y mapa de riesgos potenciales. Comisión Nacional de Protección Civil, 190 pp
Nanson GC (1986) Episodes of vertical accretion and catastrophic stripping: a model of disequilibrium flood
plain development. Geol Soc Am Bull 97:1467–1475
Nanson GC, Croke JC (1992) A genetic classification of floodplains. Geomorphology 4(6):459–486
Nash D (1994) Alluvial sedimentation. Sediment Geol 92(3–4):296–297
NCRS (1999) National Soil Survey Handbook. NCRS
Santos R, Menéndez R (2006) Topographic signature of debris-flow dominated channels: implications for
hazard assessment. In: Lorenzini G, Brebbia CA, Emmanouloudis DE (eds) Monitoring, simulation,
prevention and revealation of dense and debris flows, Wessex Institute of Technology, UK, pp 311–321
Schneider L, McCuen R (2005) Statistical guidelines for curve number generation. J Irrig Drain Eng –
ASCE 131(3):532–533
SCS (1986) Urban hydrology for small watersheds, USDA
SCS (1972) National Engineering Handbook, Section 4. Hydrology, Soil Conservation Service, USDA, U.S.
Dept. of Agriculture, Washington
Sidle RC, Onda Y (2004) Hydrogeomorphology: overview of an emerging science. Hydrol Process 18:
597–602
Singh VP, Quiroga CA (1996) Effect of highway alignment on flooding—A case study. J Environ Hydrol
4:1–17
Smith LC (1997) Satellite remote sensing of river inundation area, stage and discharge: a review. Hydrol
Process 11:1427–1439
Témez JR (1987) Cálculo hidrometeorológico de caudales máximos en pequeñas cuencas naturales. MOPU,
Dirección General de Carreteras
Thompson A, Clayton J (2002) The role of geomorphology in flood risk assessment. Proc Inst Civil Eng-
Civil Eng 150:25–29
Thorndycraft VR, Benito G, Barriendos M, Llasat MC (eds) (2003) Palaeofloods, historical data and
climatic variability: applications in flood risk assessment. CSIC, Madrid, 378 pp
Turner-Gillespie DF, Smith JA, Bates PD (2003) Attenuating reaches and the regional flood response of an
urbanising drainage basin. Adv Water Resour 26:673–684
USACE (2000) Hydrologic Modelling System HEC-HMS. Technical Reference Manual, US Army Corps of
Engineers
USACE (2001) Hydrologic Modelling System HEC-HMS. User’s Manual, US Army Corps of Engineers
USACE (2002) HEC-RAS River Analysis System. User’s Manual U.S. Army Corps of Engineers
Vanacker V, Vanderschaeghe M, Govers G, Willems E, Poesen J, Deckers J, De Bievre B (2003) Linking
hydrological, infinite slope stability and land-use change models though GIS for assessing the impact
of deforestation on slope stability in high Andean watersheds. Geomorphology 52:299–315
Wheater HS (2002) Progress and prospects for fluvial flood modeling. Phil Trans R Soc Lond A 360:
1409–1431
Wilson MD (2004) Evaluating the effect of data and data uncertainty on predictions of flood inundation.
University of Southampton, Southampton
Wolman MG (1971) Evaluating alternative techniques of floodplain mapping. Water Resour Res 7:
1383–1392

123
All in-text references underlined in blue are linked to publications on ResearchGate, letting you access and rea

You might also like