Professional Documents
Culture Documents
Remotesensing 02 02607
Remotesensing 02 02607
3390/rs2112607
OPEN ACCESS
Remote Sensing
ISSN 2072-4292
www.mdpi.com/journal/remotesensing
Article
Received: 27 September 2010; in revised form: 3 November 2010 / Accepted: 10 November 2010 /
Published: 23 November 2010
Abstract: This paper shows the combined use of remotely sensed data and hydraulic
geometry methods as an alternative to rainfall-runoff models. Hydraulic geometric data and
boolean images of water sheets obtained from satellite images after storm events were
integrated in a Geographical Information System. Channel cross-sections were extracted
from a high resolution Digital Terrain Model (DTM) and superimposed on the image cover
to estimate the peak flow using HEC-RAS. The proposed methodology has been tested in
ephemeral channels (ramblas) on the coastal zone in south-eastern Spain. These fluvial
systems constitute an important natural hazard due to their high discharges and sediment
loads. In particular, different areas affected by floods during the period 1997 to 2009 were
delimited through HEC-GeoRAs from hydraulic geometry data and Landsat images of these
floods (Landsat-TM5 and Landsat-ETM+7). Such an approach has been validated against
rainfall-surface runoff models (SCS Dimensionless Unit Hydrograph, SCSD, Témez gamma
HU Tγ and the Modified Rational method, MRM) comparing their results with flood
hydrographs of the Automatic Hydrologic Information System (AHIS) in several ephemeral
channels in the Murcia Region. The results obtained from the method providing a better fit
were used to calculate different hydraulic geometry parameters, especially in residual
flood areas.
Remote Sens. 2010, 2 2608
1. Introduction
Hydraulic geometry methods commonly used for estimating flood peak discharges in ephemeral
channels usually incorporate the dominant features of the channel shape in its formulae (width, B;
depth, d; hydraulic radius, R; wet perimeter, P; slope, S) and its relations with the energy laws expressed
in terms of flow. Ever since Lacey‘s classic equations [1,2] until the hydraulic geometry relationships and
the Tinkler‘s critical flow regime [3], many proposals have been developed which, based on the regime
theories, are adequate approaches for estimating flood discharges. In recent decades many attempts have
been made to combine the remote sensing techniques with traditional channel geometry methods. In fact,
Dingman and Sharma‘s regime equation [4] makes it possible to estimate the channel using geometric
variables of the channel that are directly measurable in the field, and potentially measurable through
satellite images, without having to deduce the bed roughness parameters, normally required for
conventional hydraulic methods, such as the equations of Chezy or Manning. Another method for
calculating the flow is by using satellite data based on the consideration of channel geometry at a given
point together with the flow curves obtained by Leopold et al. [5] according to the depth and transversal
area of the flow. The precision of the satellite observations, together with the in situ calibration and the use
of high resolution DTMs in a GIS environment, have made it possible to construct new hydraulic models
in this respect [6-8].
For semiarid areas, the flow geometry method is especially useful and has already been used by many
authors as an indirect method for estimating flood discharges [9-12]. This method is based on the relations
between channel hydraulic geometry and high flow conditions (‗bankfull‘ and ―flood-prone area‖), and
therefore constitutes an excellent instrument for predicting extreme hydrological events and the planning
of areas at risk. The dry state of most of the channels in this area, and in south-eastern Spain in general,
makes it easy to measure the geometric and roughness parameters needed to calculate peak flows in
these systems which are barely instrumentalized. The satellite images provide valuable information
about the surface covered by flood waters. Therefore, when this information is added to precise field
data and high resolution terrain digital models, it could offer a more adequate methodological option
for this region than the standard hydro-meteorological approaches. The latter usually involves greater
uncertainty in the evaluation of peak flows, given the peculiarity of storms in this area, with varying
intensities, which are heterogeneous in space and very short in time.
In connection with this, the interest of the present study is two-fold: (i) to check the fitness level in
these types of methods supported by high resolution DTM and abundant field data as opposed to
meteorological methods; and (ii) to show its advantages for the evaluation of historic floods combined
with the use of remote sensing techniques. More specifically, the integrated use of remote sensing data,
in situ data, and digital elevation models, could be particularly useful for obtaining hydraulic geometry
data in flooded areas, and, therefore, for estimating discharges, water volumes stored in floodplains
and hydraulic regime parameters.
Remote Sens. 2010, 2 2609
2. Study Area
The present study has been developed in several drainage basins in the Region of Murcia
(south-eastern Spain), characterized by a semi-arid climate with heavy rain, the presence of ephemeral
channels (ramblas) and frequent flooding events. The basins analyzed consisted completely of a group
of independent ephemeral drainage areas with a direct outlet to the sea, a predominant dendritic
drainage pattern, steep slopes and a torrential regime. This coastal group within the Region of Murcia
consists of a total of 53 basins with different sizes, between 0.5 and 522 km2 (Figure 1). Almost half of
them have a very small watershed area of less than 10 km2, which could be totally covered by just one
storm. Another 11 basins are also small areas of less than 30 km2, which could be completely covered
by mesoscale precipitation events, and only six basins are equal to or greater than 70 km2; those
drained by the Albujón, Benipila, Cañarete, Moreras, Pastrana and Ramonete ramblas. The most
extensive and dangerous, due to their large floods, are the basins of the Albujón (522 km2), las
Moreras (246.5 km2) and Benipila (147 km2).
Figure 1. Study area and location of the sectors selected (A to F) for application of
hydraulic geometry method (HGM) and remote sensing techniques.
Only a few ramblas have discharge data provided by the AHIS. The low reaches of these courses
(Benipila Rambla, in Cartagena, and Nogalte Rambla, in Puerto Lumbreras), equipped with AHIS
gauging stations, have been used for estimating the flow in recent storms and for checking the
accuracy of hydraulic geometry methods (HGM) against hydrometeorological models. However, the
methodological proposal to combine HGM with remote sensing techniques has been trialed in coastal
areas flooded by the overflowing of torrential watercourses during the intense rains of 23 October
Remote Sens. 2010, 2 2610
2000. Six rambla reaches were studied in this case (letters A to F in Figure 1). These reaches are
located in the Cañarete-Charco (A), Ramonete (B), Moreras (C), Lorentes (D), Honda (E) and
Benipila-Dolores (F) ramblas. All of these have been selected because, in addition to being located in
the zones most affected by this hydrological process, enough field data are available to be able to apply
the proposed methods. All of these basins are characterized by torrential drainage systems established
on steep headwaters of metamorphic materials and low areas with a slight gradient formed by soft
sediments (marls, silts and clays).
All of them have been selected because, besides being placed in the sectors most affected by the
mentioned hydrological events, they have sufficient field data to apply the proposed methods. All these
basins are characterized by ephemeral drainage networks established on steep headwaters of
metamorphic terrains and low sectors of scanty slope formed by soft materials (marls, silts and clays).
These conditions, together with the existence of a scanty vegetation cover and the action of very
irregular and intense rainfall (>50 mm/h), favor the development of large and flash floods.
3. Methodology
Firstly, to determine the validity of the HGM applied to the fluvial systems combined with remote
sensing methods, its fitness level has been tested in comparison with the results obtained by
hydrometeorological methods (SCS Dimensionless Unit Hydrograph, SCSD, Témez gamma Unit
Hydrograph Tγ, and the Modified Rational Method, MRM). As comparative data for this validation,
discharge values of the AHIS in the gauging stations of Cartagena (Benipila Rambla) and Puerto
Lumbreras (Nogalte Rambla) have been used. The confrontation of such methods is carried out using
the comparative analysis of 10 important storm events that occurred between 1997 and 2009. Given
the best fit of the HGM with real gauging data, this method makes it possible to calculate laminated
flows in the flood zones, once water surface is delimited using satellite images according to the
methodology described in Bustamante et al. [13]. This model is based on obtaining ROIs (Regions of
Interest) in areas that show a characteristic water spectral behavior from the differential response
between the reflectivities of 4, 5 and 7 channels, and the posterior conversion of these ROIs in
vectorial archives. In this study the HGM has been trialed in some of the main areas of flooding
observed after the storm on 23 October 2000 (sectors denominated with the letters A to F in Figure 1).
3.1. The Hydraulic Geometry Method (HGM) Compared to Hydrometeorological Methods for
Calculating Flood Discharges
In order to provide flow information the environment data of each basin were introduced into the
hydrometeorological models, especially precipitation values. There are many methods in this respect:
pluviometer weights in each subbasin according to Thiessen‘s method of polygons [14], the rainfall
calculation from the distance inverse method, rainfall by cells (meteorological radar), and SCS storms
(24 h volume) associated to frequency. In this work daily rainfall data have been used from 16 stations
of the National Meteorological Institute (NMI) located at or near to the study area. The types of
Remote Sens. 2010, 2 2611
hydrograph methods adopted here are: dimensionless Unit Hydrograph (UH) of SCS adjusted to a
gamma function [15,16]; Témez‘s triangular UH; and Témez‘s modified rational method.
- SCS Dimensionless Hydrograph Method. The SCS Dimensionless Unit Hydrograph method has been
applied using a rainfall raster map obtained by the spatial interpolation of precipitation data. Using
GRASS (r.statistics command) the mean precipitation for subbasins has been calculated. A SCS type
temporal distribution has been applied to the rainfall quantities in 24 hours, choosing the most
adequate temporal interval [17]. The SCS Dimensionless is a synthetic unitary hydrograph in which
discharges are expressed in relation to peak flow (qp), and times in relation to peak time (Tp), once
estimated qp and Tp from the triangular UH of SCS. Studies carried out in the United States in rural
basins of various sizes [15], show that the lag time between peak rainfall and peak discharge is about
0.6 times the basin concentration time [18].
- Témez’s Unit Hydrograph Method (TUHM) and the Modified Rational Method (MRM). Témez‘s
formulas (TUHM) and the Modified Rational Method (MRM) use the channel slope to calculate the
concentration time which generates values which are lower than those of the SCS, given that the
channel slope is always less than the mean of the basin. Témez‘s triangular UH, in spite of having been
used very much in Spain, is usually more appropriate for extreme hydrological events of a large
magnitude. What is more, it is a modification of the SCS Triangular UH adapted to local conditions so
that the peak discharge, in this case, tends to be bigger and the base time lower [19-21]. In fact, there is
a restriction imposed by a unit rainfall time which is less than the fifth part of concentration time, a
fact that rules out less torrentiality events. Likewise the rational method applied in this work follows
the formula proposed by Témez [19-22] adapted for the climatic conditions in Spain. This method,
widely used in Spain, is known as the Modified Rational Method (MRM) and is able to estimate in a
simple way the peak discharges in small and medium basins with concentration times (Tc) between
0.25 and 24 hours [22,23].
Q p (1 / 3.6) C·
I ·A·
K (1)
where Qp (m3 s−1) is the peak discharge, I (mm h−1) the maximum mean intensity for a duration
interval which is equal to concentration time, C the runoff coefficient for the interval in which I is
produced, A the area of the drainage basin (km2) and K the uniformity coefficient. The precipitation
analysis has been carried out using the daily precipitation values of storms analyzed in
12 meteorological stations belonging to or near the basins under study. Due to the lack of precipitation
intensity-duration-frequency (IDF) curves in the study area, it has been necessary to apply the Norm
5,1-I.C., MOPU‘s Road Instruction [23] and Ferrer‘s work [20], in which it is accepted that the hourly
and daily intensity ratio is a value 11 times higher for this semi-arid area. With regard to the time of
concentration (Tc), this can relate to hourly rainfall intervals because it was estimated before and is
included in Equation 2 to obtain It (rainfall intensity in this time interval).
280 ,1 t 0 ,1
I 1 280 ,1 1
I t I d (2)
Id
where Id = Mean daily rainfall intensity (daily P/24 hr); I1 = Rainfall intensity in the most rainy hour;
t = Time interval (h) to evaluate the rainfall intensity.
Remote Sens. 2010, 2 2612
To correct the fact that the spatial rainfall distribution is not uniform, the area reduction coefficient
has been used (ARC) proposed by Témez [22], through the expression 1 − ((log A)/15), where A is the
area of the basin (km2). The possible errors generated by the lack of uniformity of net rainfall over
time are corrected using a uniformity coefficient that varies according to concentration times. The
runoff coefficient is a result of the relationships between daily precipitation and runoff threshold (P 0),
the latter obtained from the land uses and soil hydrological characteristics.
HGMs have been used frequently in the reconstruction of palaeofloods [24-26]. They are generally
based on the identification in the field of indicators of water height during flooding which together
with geometric characteristics of the channel section are used in standard hydraulic calculations for
estimating peak flows. In this study, two calculation methods have been applied, according to the
hydraulic conditions of each reach: (1) The critical depth method in those sections that experienced an
abrupt change in the slope and it is assumed therefore that there are critical flow conditions during
flooding; and (2) the Manning equation commonly used in well-defined open channels, adopting
Jarrett‘s approach [27-29] for the calculation of the roughness coefficient in reaches with a steep slope.
The reaches to which these equations have been applied in the analyzed channels have very varied
slope ranges, between 0.007 and 0.039 (m m−1). The critical depth used here for channels with steep
slopes (>0.02 m m-1) is the one proposed by Jarrett [29] according to the expression:
Q (m3 s−1) = Ac Vc (3)
where Ac is the area of the section and Vc the critical velocity. Using Jarrett‘s formula [28,29]
proposed for the calculation of the bed roughness in steep reaches (n = 0,32R0,38 S−0,16), in such cases
Manning‘s equation has been reformulated estimating the flow velocity and the flood discharge from
the expressions:
V = 3.17 R 0.83 S 0.12 (4)
Q = 3.17 AR 0.83 S 0.12 (5)
where V is the mean flow velocity (m s−1), Q the peak discharge (m3 s−1), A the area of the wet
cross-section (m2), R the hydraulic radius (m) and S the slope of the water surface (or the bed slope in
relatively uniform channels) [30]. In the validation of HGM several cross-sections have been used
which are located upstream from the AHIS control stations (lower reaches of the Benipila and Nogalte
ramblas). However, for the estimation of laminated discharges through the studied flood (23 October
2000) a total of 63 topographic profiles have been extracted and grouped by series within the six
flooding sectors defined using Landsat images. The first two series of cross-sections correspond to
flooded sectors in the lower reaches of the Cañarete and Charco ramblas (Águilas) (sector A), the third
series was located in the mid-low reach of the Ramonete rambla (sector B), the fourth and fifth are
extracted from sectors C and D flooded by the Moreras and Lorentes ramblas respectively.
The E sector is made up of two series of profiles, located to the right and left of the final reach of
the Valdelentisco Rambla, while sector F is made up of another two series running along both sides of
the Dolores-Benipila Rambla.
Remote Sens. 2010, 2 2613
3.2. Combination of Hydraulic Geometry Methods (HGM) with Remote Sensing Methods for Recent
Floods
The measurement of the direct discharge is difficult and very costly, which is why indirect
procedures are often used based on measuring the water level and hydraulic geometry data with cost
curves known a priori. This is the case of the AHIS, which through various sensors and devices (float
and counterweight, piezoresistives, compact tire, ultrasound and radar) makes it possible to measure
the height of water and to obtain relatively precise hydraulic data from periodic record campaigns. In
ephemeral channels, it is quite common to calculate peak flood discharges by applying hydraulic
models based on channel geometry data, area of the wet cross-section, longitudinal slope and bed
roughness (methods based on hydraulic geometry), but the level of precision is not so high. Even so,
they could generate good results if the used topographic and roughness data have high quality. The
limits of the flooded areas in the flood in October 2000 have been established using the analysis of
Landsat images. These areas are defined by polygons from a planimetric information layer without
heights, through vectorial files (.shp and .evf formats). To obtain the water heights in these sectors, the
masks extracted through remote sensing techniques have been superimposed on a digital terrain model.
The DTM used is one of the products of a digital photogrammetric flight financed in 2008 by the
Agriculture and Water Ministry of the Region of Murcia. The basic reference system of the whole
project is ETRS89, supported by the REGENTE network through the permanent Meristemum GNSS
stations. In this flight a LIDAR rising was carried out (Figure 2), of which DTM is derived (4 × 4 m)
and used here, with more precision than the traditional automatic correlation photograms. For the
specific case of flat zones that have been flooded, it is very useful to have LIDAR pulses available to
complete the information provided by the DTM, given that precision is greater in a small space.
The Hec-RAS program and its Hec-GeoRAS extension for ArcView3.x have been used for the
hydraulic modeling. Specifically from this extension, a total of 63 cross-section profiles have been
extracted which have served to determine the geometric features of the analyzed flood areas. On the
DTM (transformed into a model of elevations obtained by Triangular Irregular Network—TIN) the
channel and the cross-sections are drawn up, generating a 3D file containing elevations and distances
with regard to the adjacent profiles. By superimposing the flooded areas defined using remote sensing
onto the model of elevations (TIN) a raster map was obtained, with the same cell size as the one
chosen in the file exportation which provides data about the flow width and depth. The roughness data
have also been included in this file, with the addition of a polygon layer on land uses that represent
different Manning ―n‖ roughness values (Table 1). Already in Hec-RAS, discharges—which cover the
wet perimeter sections—are simulated through steady flow analysis, considering that being
floodplains, the flow regime tends to be subcritical and the roughness somewhat higher than in the
channel, as reflected in its predominant uses (scrubs, crops under plastic, and trees). The use of
multiple profiles in the flow analysis allows to select the best profiles that adjust at the flooded area.
Remote Sens. 2010, 2 2614
Figure 2. LIDAR rising. Flight financed in 2008 by the Agriculture and Water Ministry of
the Region of Murcia.
Table 1. Manning roughness values used in the analysis. U. S. Army Corps of Engineers,
Hydrologic Engineering Center [32].
Type of Channel and Description Minimum Normal Maximum
A. Natural Streams
1. Main Channels
a. Clean, straight, full, no rifts or deep pools 0.025 0.030 0.033
b. Same as above, but more stones and weeds 0.030 0.035 0.040
c. Clean, winding, some pools and shoals 0.033 0.040 0.045
d. Same as above, but some weeds and stones 0.035 0.045 0.050
e. Same as above, lower stages, more ineffective slopes and sections 0.040 0.048 0.055
f. Same as ―d‖ but more stones 0.045 0.050 0.060
g. Sluggish reaches, weedy, deep pools 0.050 0.070 0.080
h. Very weedy reaches, deep pools, or floodways with heavy stands of 0.070 0.100 0.150
timber and brush
2. Flood Plains
a. Pasture no brush
1. Short grass 0.025 0.030 0.025
2. High grass 0.030 0.035 0.050
b. Cultivated areas
1.No crop 0.020 0.030 0.040
2. Mature row crops 0.025 0.035 0.045
3. Mature field crops 0.030 0.040 0.050
c. Brush
1. Scattered brush, heavy weeds 0.035 0.050 0.070
2. Light brush and trees, in winter 0.035 0.050 0.060
Remote Sens. 2010, 2 2615
Table 1. Cont.
Type of Channel and Description Minimum Normal Maximum
3. Light brush and trees, in summer 0.040 0.060 0.080
4. Medium brush and trees, in winter 0.045 0.070 0.110
5. Medium brush and trees, in summer 0.070 0.100 0.160
d. Trees
1. Cleared land with tree stumps, no sprouts 0.030 0.040 0.050
2. Same as above, but heavy sprouts 0.050 0.060 0.080
3. Heavy stand of timber, few down trees, little undergrowth, flow 0.080 0.100 0.120
below branches
4. Same as above, but with flow into branches 0.100 0.120 0.160
5. Dense willows, summer, straight 0.110 0.150 0.200
3. Mountain Streams, no vegetation in channel, banks steep, with trees
and brush on banks submerged
a. Bottom: gravels, cobbles and few boulders
b. Bottom: cobbles with large boulders
B. Lined or Built-Up Channels
1. Concrete
a. Trowel finish 0.011 0.013 0.015
b. Float Finish 0.013 0.015 0.016
c. Finished, with gravel bottom 0.015 0.017 0.020
d. Unfinished 0.014 0.017 0.020
e. Gunite, good section 0.016 0.019 0.023
f. Gunite, wavy section 0.018 0.022 0.025
g. On good excavated rock 0.017 0.020
h. On irregular excavated rock 0.022 0.027
2. Concrete bottom float finished with sides of:
a. Dressed stone in mortar 0.015 0.017 0.020
b. Random stone in mortar 0.017 0.020 0.024
c. Cement rubble masonry, plastered 0.016 0.020 0.024
d. Cement rubble masonry 0.020 0.025 0.030
e. Dry rubble on riprap 0.020 0.030 0.035
3. Gravel bottom with sides of:
a. Formed concrete 0.017 0.020 0.025
b. Random stone in mortar 0.020 0.023 0.026
c. Dry rubble or riprap 0.023 0.033 0.036
4. Brick
a. Glazed 0.011 0.013 0.015
b. Incement mortar 0.012 0.015 0.018
5. Vegetal lining 0.030 0.050
C. Excavated or Dredged Channels
1. Earth, straight and uniform
a. Clean, recently completed 0.016 0.018 0.020
Remote Sens. 2010, 2 2616
Table 1. Cont.
Type of Channel and Description Minimum Normal Maximum
b. Clean, after weathering 0.018 0.022 0.025
c. Gravel, uniform section, clean 0.022 0.025 0.030
d. With short grass, few weeds 0.022 0.027 0.033
2. Earth, winding and sluggish
a. No vegetation 0.023 0.025 0.030
b. Grass, some weeds 0.025 0.030 0.033
c. Dense weeds or aquatic plants 0.030 0.035 0.040
d. Earth bottom and rubble side 0.028 0.030 0.035
e. Stony bottom and weedy banks 0.025 0.035 0.040
f. Cobble bottom and clean sides 0.030 0.040 0.050
3. Dragline-excavated or dredged
a. No vegetation 0.025 0.028 0.033
b. Light brush on banks 0.035 0.050 0.060
4. Rock cuts
a. Smooth and uniform 0.025 0.035 0.040
b. Jagged and irregular 0.035 0.040 0.050
5. Channels not maintained, weeds and brush
a. Clean bottom, brush on sides 0.040 0.050 0.080
b. Same as above, highest flow stage 0.045 0.070 0.110
c. Dense weeds, high as flow depth 0.050 0.080 0.120
d. Dense brush, high stage 0.080 0.100 0.140
The delimitation of flooded areas defined from Landsat images immediately after the flood of 23
October, 2000 have been contrasted with the marked limits, according to hydromorphological criteria,
through the bankfull discharges and flood prone areas. This has made it possible to catalogue these
extreme hydrological events depending on the reached morphological thresholds (main channel
associated with dominant discharges and active floodplains) and therefore to determine the
hazard levels.
3.3. Using Bankfull Stage and Flood-Prone Area as Thresholds of Flood Hazard
For the analysis of bankfull and flood prone areas, 150 cross-sections along the middle and lower
reaches in the studied coastal ramblas have been obtained, between Águilas and Cartagena. The
measurement of these cross-sections was carried out using TOPCON Hiper+ GPS, the TOPCON
FC100 notebook and the INLAND group Triton program. Cross-sections were located in straight or
non-sinuous reaches with uniform depth, avoiding proximity of gully junctions and turbulent sectors.
Channel width, maximum and mean depth, wetted perimeter, entrenchment ratio, and bankfull and
Remote Sens. 2010, 2 2617
flood-prone areas were measured in three to four locations within the survey reach and averaged to
produce representative reach values.
4.1. Validation of the Hydraulic Geometry Method compared to the Rainfall-Runoff Models
Table 2 shows the main morphometric data of the basins and channels used in the validation of
described methods, in addition to its respective runoff thresholds (P0) and curve numbers (CN). The
influence of the lithological control in runoff indexes is notable because the metamorphic basin of
Nogalte has a greater curve number and a lower P0 than the limy basin of Benipila, in spite of it having
a higher mean slope. However, the characteristic times of the unit hydrograph vary according to the
method applied. Témez‘s UH methods and MRM offer very similar values, which are always greater
in the Nogalte Rambla (Table 3). On the other hand, the time of concentration, lag time and time to
peak estimated for the SCS UH method are shorter in this rambla and slightly longer in the Benipila
Rambla, which denotes a greater influence of the soil infiltration capacity compared to the other
hydrological abstractions.
Table 2. Morphometric parameters of basins and channels and curve numbers (CN) used
for elaborating unit hydrographs. Watercourses included in the study area.
Watershed area Basin slope Channel length Channel slope P0
Basin CN
(km2) (m/m) (km) (m/m) (mm)
Benipila 97.10 0.1370 22.84 0.016 34.3 81.4
Benipila+Dolores 144.64 0.1145 24.35 0.013 26.9 84.8
Nogalte 128.65 0.1943 32.40 0.022 23.2 86.6
P0 = runoff threshold; CN = curve number.
Table 3. Characteristic times of the unit hydrograph obtained for each watershed area from
the SCS Dimensionless UH method (SCSD), Témez‘s Gamma HU (Tγ) and the Modified
Rational Method (MRM).
The flood peak discharges obtained using hydrometeorological methods are in many cases different
from those facilitated by the AHIS (Table 4). The main cause is supported by the condition of
homogenous rainfall on which rainfall-runoff models are based, especially the unit hydrograph method
(UHM). The locally strong storms on the middle-lower reach, where the observatories with
Remote Sens. 2010, 2 2618
pluviographs are located, lead to high values which could generate small flows because they are not
representative episodes of the basin.
Table 4. Flood peak discharges estimated using the SCS Dimensionless UH method
(SCSD); Témez‘s Gamma HU (Tγ); Modified Rational Method (MRM); and the Hydraulic
Geometry Method (HGM); compared with the peak flows measured by AHIS.
The interpolation of the rainfall intensity data of this reach with others with less detail in the rest of
the basin, introduces an error factor that is often appreciable. In fact, the location of the meteorological
stations in the Benipila basin, in an area of less than 10 km2 near the city of Cartagena (lower reach)
which hardly accounts for 10% of the whole watershed area, together with the lack of pluviometric
information in the headwater area, means that the unit hydrograph method (UHM) is scantily
satisfactory when storms are slightly extended. In these cases, the basin surface of the study is too wide
to be completely affected by isolated storms which are mainly of a convective origin.
Only when the storms are sufficiently generalized, affecting most of the basin with low intensities,
do these methods provide acceptable approximations. The following are examples of this: the storms
of 30 September, 1997, 23 October, 2000 and 28 September, 2009 for the Benipila Rambla basin and
29 September, 1997 for the Nogalte Rambla basin (Table 4). However, these methods overestimate
peak flow even more when the measured value is higher. The Dimensionless UH with SCS gamma
function (SCSD) usually provides values between 6 and 11% higher than those registered by the
AHIS, while Témez‘s Gamma HU (Tγ) gives values which are between 2 and 9% higher. The MRM
shows values that are less fitted despite correction factors applied.
Remote Sens. 2010, 2 2619
The best-fit with the AHIS data is provided by the HGM, with a determination coefficient
2
R = 0.98. This explains why the AHIS determines the discharges using cost curves associated to water
levels, having a close correlation with geometric parameters of the channel cross-sections in which
height sensors are installed. There is a linear relationship between both types of values (Equation 6). In
most cases the flood peak discharges estimated using the HGM are lower than those measured using
AHIS, which is due to the reduced influence of the bed roughness in the AHIS control sections, which
are normally channeled and have a flat bottom. The cross-sections used in the calculation of the peak
flows with HGM belong to naturally open channels, located immediately upstream from the reaches
channeled with AHIS gauging stations.
These natural sections generally have more irregular beds, provided with vegetation and alluvial
forms which give it a greater degree of roughness.
In Figure 3 a comparison has been made of the hydrograph calculated using UH methods, the
hydrograph construct from AHIS data and the peak discharge estimated by HGM, corresponding to the
flood of 23 October, 2000.
h
10
11
12
13
14
15
16
17
18
19
20
21
22
23
10
11
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0 600
5
10 500
15
20
QHGM = 406 m3/s 400 m 3/s
mm
25
30 300
35
40 200
45
50 100
55
60 0
By overestimating the real flood discharges, SCSD and Tγ UH could be used to establish greater
safety levels when there are extreme hydrological events. Table 5 shows the hydraulic geometric
values extracted using Hec-RAS from the maximum flows estimated using these methods in the three
floods of the Benipila Rambla with the best theoretical fits. In these cases, the following hydraulic
geometry variables have been obtained: the water sheet width, the mean flow depth, the area of the
flow section and the wetted perimeter. The bed slope is determined using the DEM and the
cross-section elevations measured by GPS, while the channel roughness (n) is a field data.
Remote Sens. 2010, 2 2620
Given that we are dealing with the same reach, the slope and the bed roughness are kept constant.
The remaining values closely reflect flow variations and are somewhat higher when the flood
simulation is run with SCSD UH. However, the difference between the hydraulic geometry values
obtained using SCSD UH and Tγ UH discharges is minimal in large floods and practically inexistent in
small floods.
Table 5. Hydraulic geometry data of the peak flow on 23 October, 2000 estimated using
the SCS Dimensionless UH method (SCSD) and Témez‘s Gamma HU (Tγ). AHIS station
of the Benipila Rambla through Cartagena.
Q h W A wP V S n
Rambla Date
(m) (m) (m2) (m) (m/s) (m/m)
Benipila 30/09/1997 SCSD 0.28 20.5 5.71 20.6 1.66 0.0025 0.04
Tγ 0.28 20.4 5.64 20.5 1.65 0.0025 0.04
Benipila 23/10/2000 SCSD 2.18 47.8 103.24 48.0 4.61 0.0025 0.04
Tγ 2.16 47.2 101.97 47.9 4.58 0.0025 0.04
Benipila - 28/09/2009 SCSD 1.59 46.0 73.10 46.5 3.93 0.0025 0.04
Dolores Tγ 1.55 45.8 70.90 46.3 3.88 0.0025 0.04
Q = discharge; h = flow depth; W = flow width; A = wet cross-section area; wP = wet
channel perimeter; V = flow velocity; S = channel slope; n = bed roughness.
With the help of satellite images, it has been possible to reconstruct flooded areas in the study
ramblas. Comparison of the bankfull limits and flood-prone area established using field work and
DEM could prove whether the magnitude of a given flood is greater than the hazard thresholds defined
according to hydromorphological criteria.
In the flood used as an example for the application of the proposed methodology, of 23 October
2000, the extension of the water sheet and the energy developed by this flood exceed the flood-prone
area thresholds at several points. In the case of Lorentes Rambla (Figure 4), it is clear that there is a
greater surface covered by this flood compared to the limits marked by the flood-prone area, especially
in the final reach. Even the discharge calculated for the same event and reach using HGM, applied to
the delimitation of the flooded area from Landsat images, it is twice larger than the discharge in the
flood-prone area (Table 6).
As Table 6 shows, the QHGM*/Q(fp) ratio of all the analyzed cases is between 1.6 and 2, which
clearly shows the high rate of hydrological and morphological hazard of the studied flood, especially
due to the great width and depth reached by its waters. The width and depth of the channel in the
bankfull and flood-prone area stages vary longitudinally according to the hydraulic regime variations,
the geological constrictions and the grain-size distribution. This could explain why the area affected by
the flood on 23 October, 2000 is downstream considerably different to the flood-prone area. In the end
reach, the Lorentes Rambla has a channel that is constrained by protective walls and buildings which
invade the floodplain, in such a way that the width of the flood-prone area is only slightly different
from the bankfull width and is widely exceeded by the width of the flooded area in this event.
Remote Sens. 2010, 2 2621
Figure 4. Delimitation of the flood area on 23 October, 2000 compared to the flood prone
area. Lower reach of the Lorentes Rambla.
Remote Sens. 2010, 2 2622
Table 6. Comparison of HGM flow estimated for the flood of October 2000 with predicted
flood-prone area discharge in the study cross-sections.
In bankfull conditions more than 58% of the culvert systems located on the road-stream crossings
is incapable of evacuating flood waters, a percentage which increases to 88.2% when the flow reaches
the flood-prone stage. However, the drainage capacity of these works is often greatly reduced due to
the obstruction caused by the bedload transport, especially in ramblas with strong slopes close to
sediment sources (e.g., the Ramonete and Pastrana ramblas). The greatest hazard is found in the stream
ford reaches crossed by roads, where the hydraulic and morphological processes of the flood maintain
their continuity.
In extreme floods, where the flood-prone area is completely covered, the bed of these reaches is
very unstable, registering Relative Bed Stability (RBS) values and Log Relative Bed Stability (LRBS)
values which are often below 0.25 and −0.60 respectively [32]. This bed instability means that these
reaches also have greater potential bed transport (6.5·10−3−0.013 m2/s) [34].
The morphological adjustments produced by the flood on 23 October, 2000 (scour holes, bank
erosion, spill-over lobes and lateral accretion deposits in the floodplain) overpass the flood-prone area
limits and affect the overall fluvial system. In the main channel the flow caused minor changes, from
undercutting of bank base and toppling to morphological bed variations associated with the mobility of
alluvial bars and other bedforms. Most of these morphological adjustments are superficial changes
which are easily observed in situ. What is more difficult to detect is channel bottom erosion (transitory
bed erosion), the cause of direct and indirect effects which could damage or endanger the stability of
exposed hydraulic structures. The highest transitory erosion values were registered in the lower reach
of the Moreras Rambla near Mazarrón, where it was above 1 m. Under the hydraulic conditions that
characterized this flood, in nearly 47% of the analyzed reaches, transitory erosion values were above
0.6 m, making this the minimum critical threshold for the design of existing drainage works and those
planned in these cases.
The comparison of the anterior and posterior Landsat images in flood events has made it possible to
define their spread. Landsat 7-ETM+ false color images showing the meridional coastal ramblas of the
Murcia Region on 8 August and 27 October, 2000, that is to say 75 days before and four days after the
October flood, have been compared. Figures 5 and 6 are an example of this comparison of images
before and after the flood on October 2000 in the Cañarete and Charco rambla systems.
Remote Sens. 2010, 2 2623
Figure 5. False color composition (4, 5, 7) of a Landat 7-ETM+ image on 8 August, 2000
(left) and on 27 October, 2000, 4 days after the flood (right). Lower reach of the Cañarete
Rambla (Águilas area).
Figure 6. False color composition (4, 5, 7) of a Landat 7-ETM+ image on 8 August, 2000
(left) and on 27 October, 2000, 4 days after the flood (right). Middle-lower reach of the
Charco Rambla (Águilas area).
Remote Sens. 2010, 2 2624
Figure 7. Delimiting of flooded areas after the event on 23 October, 2000 from Landat 7-
ETM+ images and location of cross-sections. South coastal zone of the Murcia Region.
Using these images it has been possible to obtain vector files with information about the limits of
the flooded sectors. Figure 7 shows the flooded areas to which the described methodology has been
applied and Table 7 has geometric data of different cross-sections representing the water layer in
these areas.
The sectors analyzed are residually flooded areas included within the area reached by the flood
waters on 23 October, 2000. The residual water sectors delimited using the false color
Remote Sens. 2010, 2 2625
Landsat 7-ETM+ images on 27 October, 2000, four days after the flood, corresponded to the most
depressed topographic areas and most impermeable terrains. The water remains a long time in these
areas, increasing the danger for activities and land uses developed there. Water therefore remains on
the floodplain adjacent to the study ramblas.
Table 7. Geometric and hydraulic parameters of several flooded sectors after the event of
23 October, 2000. South coastal ramblas in the Murcia Region.
h W Rh Vi Re A Volume
Flooded sector W/h Fr
(m) (m) (m) (m/s) 104)
(1· (Ha) (m3)
A.1Cañarete 0.30 133.1 447 0.30 1.5 33,51 0.86 21.96 65331
A.2 Charco 0.27 138.1 512 0.27 0.6 13,43 0.39 13.32 35964
B. Moreras 0.37 97.1 260 0.37 1.7 51,75 0.90 8.17 30556
C. Ramonete 0.74 162.3 220 0.74 1.2 69,12 0.45 7.36 54219
D. Lorentes 1.09 75.7 70 0.96 0.9 66,81 0.28 10.26 111720
E.1. Honda (iz) 0.95 192.5 203 0.81 1.9 118,29 0.68 11.16 105797
E.2. Honda (der) 1.34 246.5 183 1.20 0.9 159,48 0.18 4.5 60450
F.1. Dolores (iz) 0.06 98.4 1575 0.06 0.6 2,81 0.73 3.15 1969
F.2. Dolores (der) 0.54 80.1 147 0.54 2.6 104,86 1.12 5.4 29391
Rh = hydraulic radius; Vi = flow initial velocity; Re = Reynolds number; Fr = Froude number.
By applying HGM to these flooded sectors it was possible to reconstruct not only the geometric
features of the stored water mass but also the initial conditions of its hydraulic regime. The hydraulic
geometry in these residual water bodies varies from case to case. In some areas, water accumulation
reaches a mean depth of 1 m and a width greater than 240 m (e.g., right floodplain of the Honda
Rambla, sector E2), and in others, such as sectors C and E1, the water ponds have between 0.75 and
1 m depth and between 150 and 200 m mean width. Sector D, located around the Lorentes Rambla,
adopts the shape and surface of a group of greenhouses installed on the same floodplain and the same
occurs, although at a lower water height (around 0.3 m), in sectors A1 and A2, affected by the
overflow of the Cañarete and Charco ramblas respectively. Finally, there are areas where the flow
lamination diffuses to a wide area of very shallow waters. This is the case for sector E1 located on the
left of the Dolores Rambla, to the north of the city of Cartagena. In spite of it being a floodplain
depression with a high slope (S = 0.02), the flood did not have the same magnitude here as in the south
west of the Region.
It is not surprising that some of these sectors (B and F1), which are sloping, register geomorphic
processes that are especially active during the rising flood stage. Very high W/h ratios
(250 < W/h < 1,600) are added to mean significant slope values (0.02 < S < 0.027) in these cases,
which suggests there is a progressive widening of the erosion grooves generated in the lateral
floodplains. These grooves tend to become wider due to the instability of their banks and the crossover
of many of these in extraordinarily high waters which leads to higher W/h ratios. This relationship
between flow width and depth is greater, although slows as the discharge increases, and it is clearly
affected by the lateral erosion of grooves and other active channels within the floodplain. This
Remote Sens. 2010, 2 2626
widening involves a reduction in the unit flow in the initial sheeting stage, at the same time as more
solid material comes into play, causing sedimentary accretion and, therefore, an even greater
W/h ratio.
The hydraulic regime that dominated the water entry in the floodplains was subcritically turbulent
in nearly all cases as shown by Reynolds‘ (Re > 30·104) and Froude‘s numbers (Fr < 0.9) (Table 7).
The celerity of the surface water waves did not exceed, except in sector F2 (right plain of the Dolores
Rambla), the flow velocity during the overflow stage and the high waters were quite calm on these
plains.
5. Conclusions
Based on high quality topographic and bed roughness data, the integrated use of hydraulic geometry
methods and remote sensing has proven to be a good option for finding out about the characteristics of
the water masses in flooded areas close to ephemeral fluvial systems. Its main advantage compared to
hydrometeorological ones, for the calculation of flood discharges, is the absence of uncertainties
inherent in irregular temporal and spatial distribution of precipitation in the basins that drain these
systems. However, the HGM described is only applicable for recent extreme hydrological events, as it
requires timely field work to obtain more reliable entry data related to the height of the water observed
on the markings in the channels, the bed roughness, and the geometry of cross-sections reconstructed
using topographic data which improves the detail of the DTM. The delimitation of the residual flood
zones by obtaining the ROIs from the differential reflectivity response between channels 4, 5 and 7,
with the help of DTM and the transversal profiles, has served to establish the hydraulic geometry of
these areas. The HGM-remote sensing data integration model (Landsat-TM5 and Landsat-ETM+7)
also makes it possible to estimate: (a) the hydraulic regime of the flow once it overflows and (b) the
volumes of water stored in the floodplains.
Acknowledgments
This paper was carried out within the research project PROMETEO/2009/086, funded by
Generalitat Valenciana, and the project RIFLUTME (Natural hazard processes associated with
torrential fluvial systems in Mediterranean Environment, application to South Coastal Streams in the
Murcia Region), financed by Fundación SENECA, Agencia de Ciencia y Tecnología de la Región de
Murcia, Reference 02955/PI/05.
1. Lacey, G. Stable channels in alluvium. ICE Minutes Proc. 1930, 229, 259-284.
2. Lacey, G. Uniform flow in alluvial rivers and canals. ICE Minutes Proc. 1934, 237, 421-453.
3. Tinkler, K.J. A hydraulic geometry of rockbed channels based on critical flow. In Proceedings of
Geological Society of America Annual Meeitng, Toronto, ON, Canada, October 25–29, 1998.
4. Dingman, S.L.; Sharma, K.P. Statistical development and validation of discharge equations for
natural channels. J. Hydrol. 1997, 199, 13-35.
Remote Sens. 2010, 2 2627
5. Leopold, L.B.; Wolman, M.G.; Miller, J.P. Fluvial Processes in Geomorphology; W.H. Freeman:
New York, NY, USA, 1964.
6. Koblinsky, C.J.; Clarke, R.T.; Brenner, A.C.; Frey, H. Measurement of river level with satellite
altimetry. Water Resour. Res. 1993, 29, 1839-1848.
7. Smith, L.C.; Isacks, B.L.; Bloom, A.L.; Murray, A.B. Estimation of discharge from three braided
rivers using synthetic aperture radar satellite imagery. Water Resour. Res. 1996, 32, 2021-2034.
8. Massonet, D. Satellite radar interferometry. Sci. Amer. Mag. 1997, February, 46-53.
9. Riggs, H.C. Flash flood potential from channel measurements. In Proceedings of Flash Floods
Symposium, Paris, France, September 1974; International Association of Hydrological Sciences
Publication No. 112, 1974; pp. 52-56.
10 Osterkamp, W.R.; Hedman, E.R. Perennial-Streamflow Characteristics Related to Channel
Geometry in Missouri River Basin; Professional Paper 1242; United States Geological Survey:
Washington, DC, USA, 1982.
11 Wharton, G. Flood estimation from channel size: Guidelines for using the channel-geometry
method. Appl. Geogr. 1992, 12, 339-359.
12 Wharton, G. The channel-geometry method: Guidelines and applications. Earth Surf. Process.
Landf. 1995, 20, 649-660.
13 Bustamante, J.; Díaz-Delgado, R.; Aragonés, D. Determinación de las características de masas de
agua someras en las marismas de Doñana mediante teledetección. Revista de Teledetección 2005,
24, 107-111.
14 Chow, V.T.; Maidment, D.R.; Mays, L.W. Hidrología aplicada; McGraw-Hill: Madrid, Spain,
1994.
15 SCS (Soil Conservation Service). National Engineering Handbook; Section 4, U.S. Department of
Agriculture: Washington, DC, USA, 1972.
16 SCS (Soil Conservation Service). Hydrology. In National Engineering Handbook; Section 4; U.S.
Department of Agriculture: Washington, DC, USA, 1985.
17 Solís, L. Hidrología, 2005. Available online: http://212.81.159.185:8080/hidrologia/index.html
(accessed on 10 September 2009).
18 Nanía, L.S.; Gómez Valentín, M. Ingeniería Hidrológica, Segunda Edición; Grupo Editorial
Universitario, Granada, Spain, 2006; p. 280.
19 Témez, J.R. Cálculo hidrometeorológico de caudales máximos en pequeñas cuencas naturales.
Dirección General de Carreteras, MOPU, Madrid, Spain, 1987.
20 Ferrer, J. Recomendaciones para el cálculo hidrometeorológico de avenidas; CEDEX M-37;
CEDEX: Madrid, Spain, 1993.
21 Singh, S.K. Transmuting synthetic unit hydrograph into gamma distribution. J. Hydrol. Eng.
2000, 5, 380-385.
22 Témez, J.R. Extended and Improved Rational Method: Version of the Highways Administration of
Spain. In Proceedings of XXIV Congress of IAHR, Madrid, Spain, September 1991; Volume A,
pp. 33-40.
23 MOPU. Norma 5,1-I.C. Instrucción de carreteras. Dirección General de Carreteras, MOPU:
Madrid, Spain, 1990.
Remote Sens. 2010, 2 2628
24 Benito, G.; Grodek, T.; Enzel, Y. The geomorphic and hydrologic impacts of the catastrophic
failure of flood-control-dams during the 1996-Biescas flood (Central Pyrenees, Spain). Zeitschrift
für Geomorphologie 1998, 42, 417-437.
25 Rico, M.; Benito, G. Estimación de caudales de crecida en pequeñas cuencas de montaña:
Revisión metodológica y aplicación a la cuenca de Montardit (Pirineos Centrales, España).
Cuaternario & Geomorfología 2002, 16, 127-138.
26 Benito, G.; Thorndycraft, V.R. Palaeoflood hydrology and its role in applied hydrological
sciences. J. Hydrol. 2005, 313, 3-15.
27 Jarrett, R.D. Hydraulics of high-gradient streams. J. Hydraul. Div. Am. Soc. Civ. Eng. 1984, 110,
1519-1539.
28 Jarrett, R.D. Determination of Roughness Coefficient for Streams in Colorado; USGS Water
Resources Investigations Report 85-4004; 1985.
29 Jarrett, R.D. Errors in slope-area computations of peak discharges in mountain streams. J. Hydrol.
1987, 96, 53-67.
30 Jarrett, R.D. Hydrologic and hydraulic research in mountain rivers. Water Resour. Bull. 1990, 26,
419-429.
31 Manning‘s n Information Table. In HEC-RAS, Version 3.1.1.; U.S. Army Corps of Engineers,
Hydrologic Engineering Center: Davis, CA, USA, 2003.
32 Conesa-Garcí a, C.; Garcí a-Lorenzo, R.; López Bermúdez, F. Bed stability variations after check
dam construction in torrential channels (South-East Spain). Earth Process. Landf. 2007, 32,
2165-2184.
33 Conesa-García, C. Les ‗ramblas‘ du Sud-est Espagnol: systèmes hydromorphologiques en milieu
méditerranéen sec. Zeitschrift für Geomorphologie 2005, 49, 205-224.
© 2010 by the authors; licensee MDPI, Basel, Switzerland. This article is an open access article
distributed under the terms and conditions of the Creative Commons Attribution license
(http://creativecommons.org/licenses/by/3.0/).