Journal of Hydrology 385 (2010) 150164

Contents lists available at ScienceDirect

Journal of Hydrology
journal homepage: www.elsevier.com/locate/jhydrol

A method for estimating ash ood peak discharge in a poorly gauged basin: Case study for the 1314 January 1994 ood, Gioros basin, Crete, Greece
Aristeidis G. Koutroulis, Ioannis K. Tsanis *
Department of Environmental Engineering, Technical University of Crete, Chania, Greece

a r t i c l e

i n f o

s u m m a r y
A method for estimating ash ood peak discharge, hydrograph, and volume in poorly gauged basins, where the hydrological characteristics of the ood are partially known, due to stage gauge failure, is presented. An empirical index is used to generate missing hourly rainfall data and hydrologic and hydraulic models performs the basin delineation, ood simulation, and ood inundation. The peak discharge, hydrograph, and volume, derived from the analysis of measured hydrographs in a number of non-ood causing rainfall events with operating stage gauge, were used for calibration and verication of the simulated stage-discharge hydrographs. An empirical equation was developed in order to provide the peak discharge as a function of the total precipitation, its standard deviation, and storm duration. The peak discharge for a ash ood case based on the empirical equation was in close agreement with the results from a number of consolidated methods. These methods involved hydrological and hydraulic modeling and peak ow estimates based on Mannings equation and post ash ood measurements of the maximum water level observed at the control cross-section, for the 1314 January 1994 ash ood in the Gioros basin on the island of Crete, Greece. This method can be applied to other poorly gauged basins for oods with a stage higher than that dened by the rating curve. 2010 Elsevier B.V. All rights reserved.

Article history: Received 24 June 2009 Received in revised form 11 January 2010 Accepted 6 February 2010 This manuscript was handled by K. Georgakakos, Editor-in-Chief, with the assistance of Marco Borga, Associate Editor Keywords: Flash ood Poorly gauged basin Stage gauge failure Peak discharge estimation Hydrological simulation Flood volume

Introduction Flash oods are those oods that follow the causative storms in a short period of time, with water levels in the drainage network reaching a crest within minutes to a few hours resulting in a very limited opportunity for warnings to be prepared and issued (Borga et al., 2007; Collier, 2007). The severity of ash ood generating storms is poorly captured by using conventional rain-gauge networks (Norbiato et al., 2007). Poor knowledge of this phenomenon comes from the fact that, in most ash-ood cases, the rain accumulations and the discharges are unknown over most of the watersheds of concern (Creutin and Borga, 2003). The number of rainfall stations in poorly gauged basins is not sufcient to fully describe the spatiotemporal hydrometeorologic characteristics of the storm events. In addition during the ash ooding process, the stream gauges usually fail. This renders each methodology almost unique and coupled with some special catchment characteristics difcult to generalize and apply at a different location. Various methods have been developed in order to study extreme oods in ungauged or poorly gauged basins. An ungauged basin is one with inadequate records of hydrological observations to enable computation of hydrological variables of interest at the
* Corresponding author. E-mail address: tsanis@hydromech.gr (I.K. Tsanis). 0022-1694/$ - see front matter 2010 Elsevier B.V. All rights reserved. doi:10.1016/j.jhydrol.2010.02.012

appropriate spatial and temporal scales, and to the accuracy acceptable for practical applications. There are a number of methods that can be applied to study extreme oods on ungauged watersheds including the so-called indirect peak discharge estimates, rainfallrunoff modeling through hydrological models and empirical methods. Gaume (2006), refers to the peak discharge as a key issue of post ood studies and further hydrological analysis and makes an extensive reference to a number of indirect estimation methods, discussing the estimates accuracy. The methods presented, vary among those based on the ManningStrickler formula like slope area and slopeconveyance methods, non-parametric methods like critical depth and super-elevation in bends and in front of obstacles, water surface velocity evaluation on lms and rainfallrunoff checking through the rational-method or rainfallrunoff simulation. Gaume et al. (2009), enumerates the various methods of peak discharge estimation used for 550 documented ash ood events across seven European hydrometeorological regions. This categorization includes ManningStrickler formula estimation, extrapolation of calibrated stagedischarge relation, hydraulic 1D and 2D simulation, reconstruction from reservoir operation and direct current-meter measurement, with the rst two being more popular. Webb and Jarrett (2002), provide a particularly concise review of the mathematic and hydrologic assumptions that underlie the commonly used slopearea, step-backwater, and critical-depth

A.G. Koutroulis, I.K. Tsanis / Journal of Hydrology 385 (2010) 150164

151

Nomenclature index that correlates the daily precipitation of each station with the daily precipitation of the gauge with hourly data 0 total daily precipitation of gauge x0 on day j Px j total event runoff volume, m3 VT PT total event precipitation, mm D event duration, h rP standard deviation of precipitation over a rainfallrunoff event, mm peak discharge, m3/s Qpeak a, b, c, d coefcients related to the characteristics of the basin and the precipitation time series Q total ow rate, m3/s S bed slope, m/m area of cross-section i for water level j Aij A Rij ni fc pt pet Id tc R CR API L r D = R2 hydraulic radius of cross-section i for water level j, m roughness coefcient of area i (river channel or oodplain) maximum potential rate of precipitation loss, mm/h storm mean areal precipitation depth during a time interval t to t + Dt, mm excess during a time interval t to t + Dt, mm initial decit, mm basin time of concentration, h Clark coefcient for storage, h constant rate (percolation rate), mm/h average precipitation intensity over the basin, mm/h basins travel length in the Kirpich formula, ft correlation coefcient coefcient of determination

methods. Bathurst (1990), tested three discharge gauging techniques in mountain rivers and concluded that ood ow can be estimated, with limited accuracy, by the slopearea and criticaldepth methods. Costa (1987), combined the maximum rainfall runoff oods in basins from 0.39 to 370 km2 in the conterminous USA where the slopearea indirect method primarily was used to estimate the peak discharge. The slopearea method computes discharge based on an adaptation of a uniform-ow equation (the Bernoulli energy equation) using channel-geometry characteristics, water-surface proles, and Manning roughness coefcients. The uniform-ow assumption and Mannings equation were used by Gaume et al. (2004) for the estimation of the peak discharge of several cross-sections during the 12 and 13 November 1999 ash ood of the River Aude in France. Wohl (1995), used stepbackwater method in order to convert peak stage estimates from scour lines to discharge estimates in ungauged mountain channels. Hydrological models of varying complexity approaches are applied to provide detailed estimates of ow processes for ungauged sites (Sangati et al., 2009; Bonnifait et al., 2009; Bloschl et al., 2008; Reed et al., 2007; Bohorquez and Darby, 2008). However, empirical methods have also been developed and applied at ungauged sites. Empirical methods can be of great value in view of constrains of application effort when compared to alternative traditional methods. Wharton (1992) and Wharton et al. (1989) applied an alternative indirect peak discharge technique at ungauged sites, rst developed by the US Geological Survey. The channel-geometry method is based on the assumption that channel cross-section size and geometry at specic location reects the drainage process during a storm. The method can be applied when estimating relations have been dened for a region and channel width or cross-sectional area, are the only available information. Specic procedures have to be followed for accurate and consistent measurements during eld survey. Nasri et al. (2004), presented a geomorphological model methodology to predict shape and volume of hydrographs for oods in small hillside semiarid catchments. Their model was built around a production function that denes the net storm rainfall (portion of rainfall during a storm which reaches a stream channel as direct runoff) from the total rainfall (observed rainfall in the catchment) and a transfer function based on the most complete possible denition of the surface drainage system. Very high resolution (5 min) observations of rainfall were used in the model. The model runoff generation was based on surface drainage characteristics, which can be easily extracted from maps and observed runoff, was satisfactory reproduced. Detailed and well organized post ash ood surveys provide valuable information for the understanding of the event processes.

Guidance for ash oods surveys is provided by Marchi et al. (2009), describing the investigation process of a ash ood in Western Slovenia. Peak discharges and associated uncertainty were estimated for 22 cross-sections using the slopeconveyance method that proved to be practically useful and simple. Pruess et al. (1998), presented the methodology and implications of 15 eld reconnaissance surveys for estimation peak discharges of extreme ash oods along mountain channels of southwestern Colorado, using step-backwater ow modeling. Thirty selected major oods in the US were analyzed by Costa and Jarrett (2008) aiming to the most possible accurate estimate of the peak discharges using the best available practices. The dominant method used was the slopearea method for 21 of the events. For the rest of the events, direct current-meter measurements, culvert measurement, ratingcurve extension, and interpolation and rating-curve extension was used, as well as combination of the above. The study makes an extensive report to the errors and uncertainty embedded to the peak ow estimation. The estimation of peak discharges without direct current-meter measurements is, above all, a question of sound engineering judgment (Gaume et al., 2004; Gaume, 2006). Empirical relations must be used with caution and estimates should be made at a minimum of two or three cross-sections for the same river reach to reduce uncertainties (Gaume et al., 2004; Gaume, 2006). Errors in the estimation of peak discharges can be controlled by careful hydraulic and laboratory analysis (NRC, 1999). The estimation accuracy of peak discharge based on slopearea method in small steep watersheds in the US was evaluated by Jarrett (1987) by examining the effect of various parameters such as the roughness coefcient, the geometry of the cross-section, and the unsteady nature of large discharges, leading to overestimation of peak discharge by up to 100% or over. Dottori et al. (2009), presented a dynamic rating curve approach for indirect discharge estimation, based on simultaneous stage measurements at two adjacent cross-section. The authors compared various approaches described in the literature, examined the signicant uncertainty produced in the extrapolation beyond the range of measurements of the rating curve, and presented the improvement in the discharge estimation using their proposed approach. In order to avoid signicant errors and reduce the uncertainties, it is necessary to use various sources of information to enable cross-checking using different approaches (Gaume and Borga, 2008). The purpose of the paper was the comparison between a number of consolidated methods for estimating peak discharge under a stage gauge failure, or, in general, when limited information are available. The Gioros basin (158 km2) located in the central-north

152

A.G. Koutroulis, I.K. Tsanis / Journal of Hydrology 385 (2010) 150164

Fig. 1. Location and available hydrometeorological data for Gioros basin.

part of the island of Crete (Fig. 1) was used as case study and, in particular, the ash ood occurred on 13th January 1994 was analyzed. Methodology Reconstruction of precipitation realizations of variable spatiotemporal distribution The precipitation reconstruction approach combines the Thiessen polygons method, the extrapolation of the hourly rainfall intensity variations to the daily rain-gauges and the generation of gridded realizations using the assumption of multiplicative space-time separability as presented by Woods and Sivapalan (1999). In the case of the Gioros basin the hourly rainfall data was available from the station with the higher elevation that recorded at least four times higher rainfall than the other stations in the lower elevations in the valley with daily rainfall data. Therefore the hourly distribution of the rainfall in the stations with lower rainfall will not signicantly alter the hydrograph. The transformation is done with the introduction of index A for each of the daily recording stations. Index A correlates the daily precipitation of each station with the daily precipitation of the gauge with hourly data. Index A is introduced as:

the hourly precipitation of gauge x0. At a second step, the method by Woods and Sivapalan (1999) can be applied by splitting the spatial and temporal variability into two separate terms. Different realizations of stationary hourly rainfall of different intensity and timing reproduce the spatiotemporal variability. The average hourly or distributed precipitation, over the basin is then estimated by the Thiessen method (Thiessen, 1911) or by other interpolation methods such as inverse distance weighted IDW (Shepard, 1968), Kriging (Matheron, 1965). The proposed methodology takes into account the spatial distribution of precipitation due to topography (e.g., orographic precipitation) or locality of the meteorological phenomena. There are limitations of the size of the basin that can be applied since the spatial structure of convective rainstorms depends on the local meteorology and topography, but generally the distance of decorrelation between measured rainfall intensities lies between 10 and 20 km at an hourly time step (Lebel et al., 1987). Empirical peak discharge estimation In order to ll in missing data for runoff, an attempt was made to correlate the total event runoff volume with precipitation characteristics and peak discharges. The suggested method does not use directly soil properties and inltration processes during runoff but rather compares the total runoff volume VT (m3) occurring during a single event with the total precipitation PT (mm) that causes it, the standard deviation of the precipitation time series (rP in mm) and the rainfall event duration D (h). When the total runoff volume VT (m3) for a sufcient number of gauged oods are plotted against the product of the square of total precipitation PT and the standard deviation of precipitation rP their relation can be represented via a linear regression equation:

Aj k

Pj k
0 Px j

where P j k is the total daily precipitation at gauge xk (k = 1n for n 0 daily precipitation gauges) on day j and P x j is the total daily precipitation of gauge x0 on day j. Index A reects the correlation of the total amount of precipitation between gauges x0 and xk, within a period of 1 day. The hourly estimations for gauge xk is produced x by multiplying the index Aj k for each daily recording gauge xk with

V T aP2 T rP

A.G. Koutroulis, I.K. Tsanis / Journal of Hydrology 385 (2010) 150164

153

where a is a coefcient for the basin. Eq. (2) expresses that the average ow rate during the rainstorm is proportional to the product of the total precipitation, its intensity and its standard deviation while the coefcient of proportionality a is related to the characteristics of the basin. The nonlinearity of the peak discharge Qpeak with the total volume is expressed as:

Q peak D bV c T

where D is the event duration in seconds, and b and c are coefcients related to the characteristics of the basin. Due to limited data availability a number of non-ood causing rainfall events with an operating stage gauge meter are used to evaluate the coefcients in Eq. (3). Peak discharges of selected rainfallrunoff events, cover a wide range of discharge conditions for optimal method validation. Combining Eqs. (2) and (3) results in the elimination of VT:

tion storage is a consequence of absorption of precipitation by surface cover, including plants in the watershed. Depression storage is a consequence of depressions in the watershed topography; water is stored in these and eventually inltrates or evaporates. This loss occurs prior to the onset of runoff. Until the accumulated precipitation on the pervious area exceeds the recovered initial loss volume, no runoff occurs. Thus, if pt is the storm mean areal precipitation depth (mm/h) during a time interval t to t + Dt, the excess, pet (in mm/h), during the interval is given by:

8 > <0 pet pt fc > : 0

if if if

P P P

pi < Id pi > Id and pt > fc pi > Id and pt < fc 6

Q peak

1 dP2 rP c D

ModClark transformation and recession methods

4
The modied Clark (ModClark) model in HEC-HMS is a distributed parameter model in which spatial variability of characteristics and processes are considered explicitly (Kull and Feldman, 1998; Peters and Easton, 1996). This model accounts explicitly for variations in travel time to the watershed outlet from all locations of a watershed. The ModClark algorithm is a version of the Clark unit hydrograph transformation modied to accommodate spatially distributed precipitation (Clark, 1945). Runoff computations with the ModClark model explicitly account for translation and storage. Storage is accounted for within the same linear reservoir model incorporated in the Clark model. Translation is accounted for within a grid-based travel-time model t cell tc dcell =dmax (HEC, 2000), where tc is the time of concentration for the subwatershed and is a function of basins length and slope, dcell is the travel distance from the cell to the outlet, and dmax is the travel distance from the cell furthest from the outlet. The method requires an input coefcient for storage, R, where R accounts for both translation and attenuation of excess precipitation as it moves over the basin toward the outlet. Storage coefcient R is estimated as the discharge at the inection point on the recession limb of the hydrograph divided by the slope at the inection point. The translation hydrograph is routed using the equation:

where d bac . Given the coefcients related to the characteristics of the basin and the precipitation time series, the peak discharge of a ood can be determined via Eq. (4). Flood peak ow estimation from cross-sectional geometry and parameters In the absence of eld ow data, the evaluation of the results can be based on the estimation of the observed peak ow from a specic cross-section if the geometric and hydraulic characteristics of the cross-section along with the water level are available. The cross-section can be divided in two parts, the rst of which is considered as the main channel and the second is the oodplain. Using Eq. (5) (Streeter and Benjamin, 1988) the total ow between two cross-sections can be estimated as:

p Q K 1 K 2 S

=3 1 where Q is the total ow rate in m3/s, K i n Aij R2 ij , S is the energy i slope along ow, Aij is the area of cross-section i for water level j, Rij is the hydraulic radius, and ni is the roughness coefcient (i = 1 for the main channel and i = 2 for the oodplain).

Q t Dt=R 0:5DtIt 1 Dt =R 0:5Dt Q t 1

Hydrological and hydraulic modeling The US Army Corps of Engineers Models, developed at the Hydrologic Engineering Center (HEC) in Davis, California, were applied for ood simulation and mapping. The hydrological analysis was divided in four main steps according to the HEC modules: (a) the extraction of basins characteristics (HEC-GeoHMS), (b) the distributed rainfallrunoff simulation (HEC-HMS), (c) the simulation of the ood wave (HEC-RAS) and (d) the mapping and representation of ood extend (HEC-GeoRAS). In HEC-HMS model, the decit and constant was used as the loss method, the ModClark was selected as the transform method and the recession method for sub-basin baseow (HEC, 2001). For the precipitation representation, the generated hourly precipitation data were interpolated to hourly gridded precipitation datasets over the study area using the IDW method. Decit and constant loss method Decit and constant is a quasi-continuous model of precipitation losses (HEC, 2000). The concept of the decit and constant rate loss model is that the maximum potential rate of precipitation loss, fc (in mm/h), is constant throughout an event and the initial loss can recover after a prolonged period of no rainfall. An initial decit, Id (in mm), represents interception and depression storage. Intercep-

where Q(t) is the outow from storage at time t, Dt is the time increment, R is the storage coefcient, I(t) is the average inow to storage at time t and Q(t 1) is the outow from storage at previous time (t 1). The exponential recession baseow method, used in HEC-HMS (Chow et al., 1988), approximates the watershed drainage behavior when channel ow recedes exponentially after an event (Linsley et al., 1982). This recession model denes the relationship Qt, the baseow at anytime t, to an initial value as:

Q t Q 0k

where Q0 is the initial baseow and k is an exponential decay constant dened as the ratio of the baseow at time t to the baseow 1 day earlier. In HEC-HMS the parameters of this model include the initial ow, the recession ratio, and the threshold ow. The parameters can be easily estimated if gauged ow data are available. Calibration of the model Various indices of evaluation were used in order to calibrate the model. They are divided in two categories depending on time scale issues. The rst group of indices that evaluate the model is in the scale of entire rainfall events

154

A.G. Koutroulis, I.K. Tsanis / Journal of Hydrology 385 (2010) 150164

peak flow error % PFE diff Q % phase errorPE diff sh sobs ssim

Q obsmax Q simmax 100 Q obsmax

9 10 11

jV obs V sim j volume error % VE diff V % 100 V obs

where Q obsmax Q simmax are the maximum observed and simulated ow values for each rainfallrunoff event, sobs, ssim the peak times for the observed and simulated process, and Vobs, V sim the total runoff volume for the observed and simulated process, respectively. The second group of indices is related to the time step of the rainfallrunoff simulation. Nash and Sutcliffe (1970):
i1 E1 P n

Pn

Q obsi Q simi 2 Q obs 2

i1 Q obsi

12

Coefcient of correlation:

30:5 i Q Q Q Q obs i obs i sim i sim 6 7 q 5 r 4q Pn Pn 2 2 Q Q Q Q obsi obs simi sim i1 i1 Pn h

i1

13

Explained variance:

EV 1

2 Pn Pn i1 Q simi Q obsi 1=N i1 Q simi Q obsi Pn 2 i1 Q obsi Q obs

14

Root mean square error:

s Pn 2 i1 Q simi Q obsi RMSE N

Average absolute percent errors:

15

AAPE

   1 Xn  Q simi Q obsi   i1  Q obsi N

16

where Qobs(i), Qsim(i) the observed and simulated ow values for time step i, Q obs , Q sim the mean observed and simulated ow values for each event (Marsigli et al., 2002) and N the number of observations. Case study of Gioros basin Description of the study area Gioros basin is located in the central-north part of the island of Crete, Greece (Fig. 1) and has a drainage area of 186 km2 at the outlet to the Aegean Sea, while the drainage area at the river gauge is 158 km2 (about 6 km upstream). The southern upstream part of the basin is rugged mountainous terrain, whereas, the northern part has milder rolling hillslopes. Geologically, the basin is composed of extensive strata of marly limestone, ysch and sandstone, gypsum and alluvial deposits. The watershed has a Mediterranean climate, thus, the Gioros basin has seasonal ow during the wet fall-winter period (typically from October to March), when the majority of the mean annual rainfall of 827 mm occurs (Ganoulis, 2003). Land use in Gioros basin is mainly vineyards, olive trees and farmlands, which are also present along the stream banks. Nevertheless, tourism has caused a rapid development increase in the northern part of the stream, along the riverbanks and the shorelines of the Mediterranean Sea. Due to the absence of planning, natural passages of water gradually give their place to construction and inltration rates are declining continuously (Ganoulis, 2003). This allows ash oods induced by heavy winter rainstorms that cause damages at the outlet areas.

On 13th January 1994 and around 15:00, the intensity of the light precipitation that had saturated Gioros soils during the previous days started to increase. Reaching a maximum 5 h accumulated precipitation of 123 mm at 21:00 this extreme rainfall eventually stopped at 24:00 lasting almost 9 h. By that time the resulting ash ood had catastrophic impacts on Gioros basin. Many houses located near the coast were ooded leaving 49 people homeless. In terms of precipitation severity, the 100-years recurrence interval of 5 h accumulated precipitation of a nearby meteorological station (Heraklio EMY in Fig. 1) is 98 mm according to Gumbel distribution (Ganoulis, 2003). The maximum 5 h accumulated storm rainfall of 123 mm that was observed at the south part of the basin (Ag. Varvara station in Fig. 1), far exceed the 100-years recurrence interval of 98 mm of the nearby, lowland, Heraklio EMY station. The single most adverse effect of the ood was the damage caused to the citys wastewater treatment plan which was still under construction at the time of the ood. Many of the concrete tanks were rendered unserviceable or were completely destroyed by the force of the oncoming ood wave (Ganoulis, 2003). The exact ood characteristics cannot be known since the only stage gauge was destroyed at 20:00 during the ash ood. During a recent ash ood data compilation effort in the frame of the EC funded research project HYDRATE (CN: 037024), an inventory rst step towards an atlas of extreme ash oods in Europe was developed containing over 550 documented events (Gaume et al., 2009). The 1994 Gioros ash ood event was one among the 22 Greek oods that enriched this ood inventory. In terms of ash ood magnitude and under a pan-European and international perspective (Gaume et al., 2009), the 1994 Gioros storm had a moderate unit peak discharge of 1.85 m3/s/km2 (300 m3/s/158 km2). Due to lack of high resolution precipitation data both spatially and temporally it is possible that the unit peak discharge could be much higher, i.e., (a) assuming that the majority of ood was produced in the upper 1/5th of the basin (186/ 5 = 37.2 km2) due to orographic effect the unit peak discharge could be 8.1 m3/s/km2, (b) simulations with the peak hourly precipitation occurring in 15 min instead of an hour reveal an increase in velocity from 3.62 m/s to 4.08 m/s, in water depth from 5.1 m to 5.4 m and in peak discharge from 300 m3/s to 375 m3/s, that brings the unit peak discharge as high as 9.67 m3/s/km2. Since 1994, several studies have been conducted on Gioros basin, related to that ash ood. Most of them focused on the estimation of the hydrological parameters of the basin, so that appropriate ood prevention measures could be taken, and the rest aimed at simulating the 1994 ood. Due to lack of ood recording data these studies could not verify the simulation results but rather empirically estimated the total ood volume. It should be noted that ood researchers have come up with peak discharge estimates that vary signicantly from 300 m3/s up to 600 m3/s (Ganoulis, 2003; Barboudakis, 1998). Field data Flash-ood monitoring requires rainfall estimates at small spatial scales (1 km or ner) and short-time scales (1530 min, and even less in rugged mountainous areas and urban areas) (Borga et al., 2008; McCain and Shroba, 1979). Floods research at hydrological basin scale requires high resolution data, particularly for using the most robust hydrologic models that have been recently developed. Existing data usually does not included all the data required for newer models. From the three rain-gauges of Gioros basin, one has hourly records and two stations providing daily precipitation measurements (Fig. 1). The stations that provided the daily precipitation records and the station with the hourly measurements are located along the southnorth direction, which is the main orientation axis of Gioros basin. Along this line,

A.G. Koutroulis, I.K. Tsanis / Journal of Hydrology 385 (2010) 150164

155

topography changes signicantly from sea level to 570 m. Ag. Varvara hourly precipitation station is located near the mountain headwaters at 570 m above sea level (a.s.l.) (see Fig. 1) and recorded 182.8 mm of total precipitation during the 14th of January 1994, while Protis Ilias (380 m a.s.l.) and Finikia (40 m a.s.l.) precipitation stations recorded 43.9 mm and 37.7 mm, respectively. Using the previously described methodology for generating distributed hourly precipitation data-sets, total average areal precipitation for the 1994 ood event was Ptotal = 75.7 mm. A water ow level gauge is located in the north part of the basin, but about 6 km upstream from the outlet on the Mediterranean Sea (Fig. 1). At the same location, the geometric characteristics of the certain cross-

section site of Gioros stream were available, as well as the ood extent of the 1994 ood event. The only water ow level gauge with hourly records was destroyed during the ood of 1994. Results and discussion Synoptic meteorological analysis The meteorological analysis is focused on the synoptic prevailing background conditions for heavy, localized rainfall over Crete. ERA-40 re-analysis data (ERA-40 is a re-analysis of meteorological observations from September 1957 to August 2002 produced by

Crete Island
Mean sea level analysis at 1200 UTC 13/1/1994 (isobars in hPa)

Libya

Egypt

Hih resolution Visible METEOSAT 4 - 12:00 UTC 13/1/94

Convective precipitation at 1200 UTC (units are mm) 13/1/1994

Mean sea level analysis at 1800 UTC 13/1/1994 (isobars in hPa)

Infrared METEOSAT 4 - 18:00 UTC 13/1/94

Convective precipitation at 1800 UTC (units are mm) 13/1/1994

Fig. 2. (ab) Mean sea level pressure analysis and convective precipitation based on the ECWMF 40 years re-analysis daily dataset and Meteosat satellite images for the 13 January 1994 ash ood event.

156

A.G. Koutroulis, I.K. Tsanis / Journal of Hydrology 385 (2010) 150164

Mean sea level analysis at 0000 UTC 14/1/1994 (isobars in hPa)

Infrared METEOSAT 4 - 00:00 UTC 14/1/94

Convective precipitation at 0000 UTC (units are mm) 14/1/1994

Mean sea level analysis at 1200 UTC 14/1/1994 (isobars in hPa)

Hih resolution Visible METEOSAT 4 - 12:00 UTC 14/1/94

Convective precipitation at 1200 UTC (units are mm) 14/1/1994

Fig. 2 (continued)

the European Center for Medium-Range Weather Forecasts (ECMWF) in collaboration with many institutions Uppala et al., 2005) supported the synoptic analysis of the conditions prevailing to heavy local rainfall, leading to ash ooding. Moreover, high resolution visible channel METEOSAT 4 snapshots used during daylight and infrared channel images used during night for cloudy features (storm) generation and tracking (Ottenbacher et al., 1997). The satellite images of 30 min temporal resolution were provided from the European Organisation for the Exploitation of Meteorological Satellites (EUMETSAT). On January, 1994, a frontal depression in south-east Mediterranean moved eastward and north crossing the island of Crete. On the surface analysis, a low pressure area was already present at

12:00 UTC 13 January (Fig. 2a) all over the Eastern Mediterranean area. The lower pressure area of 1010 hPa appeared at 12:00 UTC 13 January over Egypt moving in a northerly direction. This low pressure area deepens to 1007 hPa centered just south of Crete at 00:00 UTC 14 January, producing high precipitation over central Crete and the ash ood of Gioros basin. High convective precipitation rates up to 45 mm per 6 h (00:00 UTC 14/1/1994) were observed during 13 and 14 January 1994 (Fig. 2b). The average areal daily precipitation over the island of Crete was calculated in order to examine the correspondence with the ERA40 convective precipitation dataset. The average areal daily precipitation estimated based on 53 daily temporal resolution rain-gauges through IDW interpolation method (Fig. 3) and then

A.G. Koutroulis, I.K. Tsanis / Journal of Hydrology 385 (2010) 150164

157

compared with the spatial aggregated convective precipitation over Crete. The average areal daily precipitation of the 14th of January 1994 based on gauged data was 61.5 mm while ERA40 estimated 12.3 mm resulting in an underestimation of total precipitation. For the 13th of January measured precipitation was 20 mm while ERA40 results to 1.5 mm and for the 15th of January measured and ERA40 precipitation was 9 mm and 3.3 mm, respectively. The comparison of ERA-40 re-analysis daily convective precipitation (Uppala et al., 2005), results in underestimation of measured precipitation due to lack of incorporating orographic effects on the storm. The studied ood produced by a heavy, orographic effected, thunderstorm based on the fact that the precipitation in Ag. Varvara station deep in the mountain at the furthest south point of the basin was almost ve times larger (182.8 mm in total with peak intensity of 37 mm/h) than the ones recorded in the other two sta-

tions (Fig. 4). In fact, a station close to the sea east of the Mediterranean Sea and east of the outlet of Gioros basin (Heraklio EMY station) recorded less than 28.4 mm of rain with the peak rainfall intensity of 7.8 mm/h (Fig. 5). The warm air masses with high dew points with a southwestnortheast ow (analysis at 500 hPa) and high positive relative vorticity were primary blocked by the Psiloritis Mountains steep topography (Fig. 4). The air mass was forced to higher elevations, thus temperatures cooled and produced high rates of precipitation of up to 37 mm/h at Ag. Varvara station. Hydrological analysis A digital terrain model DTM of the island of Crete in 1:10 000 scale (Fig. 1) was used with the HEC-GeoHMS extension of Arc-View GIS in order to delineate the Gioros watershed. The

Fig. 3. Average areal daily precipitation based on 53 rain-gauges through inverse distance weighting spatial interpolation method for the 13 January 1994 ash ood.

Fig. 4. Gioros basin, Psiloritis mountain range location and total daily precipitation during 1314 January 1994 ood.

158

A.G. Koutroulis, I.K. Tsanis / Journal of Hydrology 385 (2010) 150164

40 35 30 25 20 15 10 5 0 13/1/94 9:00 13/1/94 10:00 13/1/94 11:00 13/1/94 12:00 13/1/94 13:00 13/1/94 14:00 13/1/94 15:00 13/1/94 16:00 13/1/94 17:00 13/1/94 18:00 13/1/94 19:00 13/1/94 20:00 13/1/94 21:00 13/1/94 22:00 13/1/94 23:00 14/1/94 0:00 14/1/94 1:00 14/1/94 2:00 14/1/94 3:00 14/1/94 4:00 14/1/94 5:00 14/1/94 6:00 14/1/94 7:00 14/1/94 8:00

slope, and geological maps of the study area, provided by the Water Resources Department of the Prefecture of Crete. Hourly gridded precipitation of 1 km2 spatial resolution was generated according to the previously described methodology. Rainfallrunoff simulation of several events were conducted for the area (158 km2) upstream from the Finikia ow stage gauge (Fig. 1). The extent of the hydraulic simulation of the 1994 ood event were downstream of the ow stage gauge to the outlet of the basin (a distance of about 6 km), where the most signicant impacts of the ood were observed. Sensitivity analysis For a successful simulation of the 1994 ood, model calibration was based on the simulation of eight rainfallrunoff events of various intensities, prior and posterior to the 1994 ood. These events were derived from two time periods (from 1981 to 1995) with total areal rainfall up to 74.4 mm and were the only clearly gauged and consistent for simulation events. Before the calibration of the rainfall-produced oods and the simulation of the ood, one of these oods (1112 January 1995) was used to assess the sensitivity analysis of the model, which was based on all the parameters of the model. The results of sensitivity analysis are presented in Table 1a and b, based on the total runoff volume, the peak ow, the

Precipitation (mm)

Heraklio EMY raingauge

Ag. Varvara raingauge

Fig. 5. Hourly Precipitation intensity of Ag. Varvara rain-gauge (total precipitation of 182.8 mm at 570 m a.s.l.) and Heraklio rain-gauge (total precipitation of 28.4 mm at 40 m a.s.l.).

same DTM was also used for the delineation of the stream network and the sub-basins. Subsequently, HEC-GeoHMS output was imported in HEC-HMS. Regarding the rest of the input for HECHMS, basin parameters were estimated through the calibration of gauged rainfallrunoff events and based on hydrologic and physiographic characteristics of the basin from land use, soil type, basin

Table 1 Sensitivity analysis results on the 1112 January 1995 rainfallrunoff event with total areal rainfall of 16.0 mm. (a) Variable = initial decit, steady values: constant rate 5.7 mm/h, impervious 5%, Clark time of concentration 3.45 h, Clark storage coefcient 3.7 h, recession constant = 0.99, baseow threshold ratio = 0.14 Initial decit (mm) 5 6 7 8 9 10 11 12 13 Volume error (%) 40.1 36.7 32.9 28.3 23.8 19.0 14.4 10.3 6.5 Peak ow error (%) 28.5 24.5 19.4 13.6 7.7 1.8 4.9 11.3 17.4 Phase error (h) 0 0 0 0 0 0 0 0 0 Variable = constant rate, steady values: initial decit = 10.3 mm, impervious 5%, Clark time of concentration 3.45 h, Clark storage coefcient 3.7 h, recession constant = 0.99, baseow threshold ratio = 0.14 Constant rate (mm/h) 1.5 2 2.5 3 3.5 4 4.5 5 5.5 Volume error (%) 74.5 71.0 66.9 62.0 55.8 48.3 39.4 30.6 22.6 Peak ow error (%) 68.2 63.8 58.3 52.6 46.3 38.3 28.5 18.2 6.9 Phase error (h) 1 1 1 0 0 0 0 0 0

14 3.0 24.1 0

6 14.1 6.9 0

Variable = impervious, steady values: initial decit = 10.3 mm, constant rate 5.7 mm/h, Clark time of concentration 3.45 h, Clark storage coefcient 3.7 h, recession constant = 0.99, baseow threshold ratio = 0.14 Impervious (%) 1 2 3 4 5 6 7 8 9 10 Volume error (%) 36.9 16.6 1.5 10.0 19.0 26.8 32.8 38.1 42.5 46.4 Peak ow error (%) 54.3 35.0 20.0 8.0 1.8 10.0 16.9 22.3 27.5 32.1 Phase error (h) 0 0 0 0 0 0 0 0 0 0 (b) Variable = Clark time of concentration, steady values: initial decit = 10.3 mm, constant rate 5.7 mm/h, constant = 0.99, baseow threshold ratio = 0.14 Clark time of concentration (h) 0.5 1 1.5 2 2.5 Volume error (%) 6.6 12.2 15.9 17.9 19.2 Peak ow error (%) 1.8 6.1 0.9 2.7 2.7 Phase error (h) 2 2 2 1 1 Variable = Clark storage coefcient, steady values: constant = 0.99, baseow threshold ratio = 0.14 Clark storage coefcient (h) 2.5 Volume error (%) 24.6 Peak ow error (%) 20.6 Phase error (h) 0

impervious 5%, Clark storage coefcient 3.7 h, recession 3 19.4 0.9 0 3.5 19.1 1.8 0 4 18.4 4.9 1 4.5 17.4 0.9 1 5 16.3 4.9 1

initial decit = 10.3 mm, constant rate 5.7 mm/h, impervious 5%, Clark time of concentration 3.45 h, recession 3 22.7 12.9 0 3.5 20.3 5.3 0 4 17.5 2.9 0 4.5 14.6 11.3 0 5 11.7 18.7 0 5.5 8.8 25.6 0 6 5.6 33.3 0 6.5 2.1 42.1 0 7 0.7 47.9 0

Variable = recession constant, steady values: initial coefcient 3.7 h, baseow threshold ratio = 0.14 Recession constant 0.01 Volume error (%) 7.0 Peak ow error (%) 6.9 Phase error (h) 0 Variable = baseow threshold ratio, steady values: coefcient 3.7 h, recession constant = 0.99 Baseow threshold ratio 0.01 Volume error (%) 19.0 Peak ow error (%) 1.8 Phase error (h) 0

decit = 10.3 mm, constant rate 5.7 mm/h, impervious 5%, Clark time of concentration 3.45 h, Clark storage 0.1 9.6 4.9 0 0.2 11.7 3.8 0 0.3 12.9 2.9 0 0.4 14.3 1.9 0 0.5 15.0 0.9 0 0.6 16.1 0.0 0 0.7 17.0 0.0 0 0.8 17.9 0.9 0 0.99 19.0 1.8 0

initial decit = 10.3 mm, constant rate 5.7 mm/h, impervious 5%, Clark time of concentration 3.45 h, Clark storage 0.1 19.0 1.8 0 0.2 19.0 1.8 0 0.3 19.0 1.8 0 0.4 20.0 1.8 0 0.5 22.8 1.8 0 0.6 26.6 1.8 0 0.7 30.8 1.8 0 0.8 35.2 1.8 0 0.9 39.6 1.8 0

A.G. Koutroulis, I.K. Tsanis / Journal of Hydrology 385 (2010) 150164

159

16 14

0 4 8 12 16 20

10 8 6 4 2 0

0:00

2:00

4:00

6:00

8:00

10:00

12:00

14:00

16:00

18:00

20:00

18:00

20:00

11-Jan-95

22:00

12-Jan-95

60 50

22:00

0 5 10

40 30 15 20 10 0 20 25

12:00 14:00 16:00 18:00 20:00 22:00 0:00 2:00 4:00 6:00 8:00 10:00 12:00 14:00 16:00 18:00 20:00 22:00 0:00 2:00 4:00 6:00 8:00 10:00

15-Feb-1981

16-Feb-1981

17-Feb-1981

80 70 60 50 40 30 20 10 0

0 5 10 15 20 25 30

Precipitation (mm)

Outflow (m3/s)

Precipitation (mm)

12

hydrograph, and the time phase of the peak discharge. Constant rate (CR) parameter, Clark storage coefcient (R), impervious percent of the basin, and initial decit were found to be the parameters that could mostly affect the simulation results. The initial decit varies depending to the antecedent soil condition of each event. A range from 5 mm to 14 mm has been tested for this rainfall event of 16 mm of total average precipitation. Sensitivity analysis of this parameter resulted to a peak ow error from 28.5% to 24.1% and a volume error from 40.1% to 3.0%. Timing of the peak ow is not affected by initial decit (Table 1a, rst rows). The CR parameter depends on the physical properties of the watershed soils, the land use and the storm precipitation characteristics (e.g., intensity) and represents the inltration rate. Different values of CR were tested during sensitivity analysis varying from 1.5 to 6 mm/h, taking into account Gioros basin soil properties and land use. Results show a peak ow error from 68.2% to 6.9%, a volume error from 74.5% to 14.1% and a phase error of 1 h for CR values between 1.0 and 2.5 mm/h (Table 1a, middle rows). Impervious parameter represents the percent impervious area of the basin and is dened based on the land use and soil properties of the basin. Gioros basin can be characterized by 5% impervious based on watershed properties. Moreover, values ranging from 1% to 10% tested in order to examine the effect of the parameter, resulting to a peak ow error from 54.3% to 32.1% and a volume error from 36.9% to 46.4%, with no effect on phase error (Table 1a, lower rows). The R parameter accounts for both translation and attenuation of excess precipitation as it moves over the basin toward the outlet and depends on basin attributes and precipitation characteristics. Values ranging from 2.5 to 7 h tested resulting to a peak ow error from 20.6% to 47.9% and a volume error from 24.6% to 0.7% (Table 1b, middle rows). Calibration procedure and parameter estimation Following the sensitivity analysis, rainfallrunoff simulation was carried out for the eight precipitation events calculating the optimum hydrological model parameters using the indices given in Eqs. (9) to (16). The estimated and observed hydrographs with the corresponding hourly precipitation data of three most representative rainstorms (in terms of peak discharge amplitude) are presented in Fig. 6. Model calibration was based on the minimization of the error of the water volume (VE) from Eq. (11), the peak ow error (PFE) from Eq. (9), the phase error (PE) from Eq. (10) and the best performance of the time step based evaluation criteria from Eqs. (12) to (16). The calibration procedure resulted to the estimation of the characteristic parameters for the Gioros basin representing the physical properties of the watershed soils and land use and the antecedent-moisture condition. The initial decit of the storms varied from 3.9 to 37 mm (Table 2) and is used as a

Outflow (m3/s)

Fig. 6. Flow hydrographs for three of the rainstorms, used for the hydrological analysis.

Table 2 Model calibration results for the simulation of the eight raistorms. Event 1 2* 3 4 5 6 7* 8* Date 3031 January 1993 1112 January 1995 1314 January 1995 2931 January 1994 1415 January 1995 78 February 1994 1517 February 1981 2224 February 1981 Volume (m3) 33,812 180,579 151,577 482,020 408,760 834,502 1,394,986 3,333,347 Average rainfall intensity (mm/h) 0.83 2.67 0.32 1.26 2.91 1.06 2.21 1.58 P** (mm) 2.5 16.0 10.4 26.4 32.0 21.8 53.1 74.4 Id (mm) 3.9 10.3 11.0 9.1 8.0 5.1 5.0 37.0 Peak ow (m3/s) 2.3 10.8 10.0 9.5 20.1 20.3 40.5 61.0 Peak ow error (%) 0 1.8 4.2 0.0 2.0 0.0 1.7 4.1 0.2 Phase error (h) 0 0 0 0 0 0 0 0 0 Nash Sutcliff 0.90 0.90 0.69 0.88 0.88 0.53 0.74 0.70 0.78 Volume error (%) 1.9 13.4 18.5 13.1 13.0 3.2 5.4 9.2 7.9 RMSE 0.19 0.96 1.10 0.90 1.60 3.07 3.51 3.93 1.91 AAPE (%) 22.0 16.5 17.4 28.4 16.6 26.9 27.0 15.3 21.3 r 0.99 0.99 0.99 1.00 1.00 0.98 1.00 0.98 0.99 EV (%) 90.5 91.7 82.2 90.5 93.0 53.2 88.5 92.9 85.3

Average values
* **

Storms corresponding to ow hydrographs of Fig. 6. P is average rainfall over the basin.

1:00 3:00 5:00 7:00 9:00 11:00 13:00 15:00 17:00 19:00 21:00 23:00 1:00 3:00 5:00 7:00 9:00 11:00 13:00 15:00 17:00 19:00 21:00 23:00 1:00

22-Feb-1981

23-Feb-1981

24

Precipitation

Q simulated

Q observed

Precipitation (mm)

Outflow (m3/s)

160

A.G. Koutroulis, I.K. Tsanis / Journal of Hydrology 385 (2010) 150164

tting parameter (during the present study Id was poorly linked to its physical meaning). The maximum decit estimated at 91 mm and the impervious area to 5% of the basin based on land-use, geology, and soil maps. The parameter CR varied from 2.5 to 5.5 mm/h according to the average precipitation intensity (API) (high API results to high CR for all the storms). Fig. 7a contains the values of CR for the calibration events plotted against the API. The time of concentration for Gioros basin was estimated from the Kirpich formula (Kirpich, 1940) t c 0:0078L0:77 =S0:385 3:45 h, where L is the basins travel length and S is the slope along main channel. R varied from 2.5 to 5.5 h according to API (high precipitation rates results to faster basin response and eventually to lower R values).

Fig. 7b contains the values of R plotted against API, for the calibration events. Base ow for each event was estimated and recession ration and threshold ow calibrated at 0.99 and 0.14, respectively. The calibration for the eight events resulted in (Table 2) an average R2 of 0.99 and a zero phase error (PE) for all the events. The average volume error was 7.9% and the average value of the peak ow error was approximately 0.18%, while the average Nash and Sutcliffe (1970) criterion was 0.78. Overall, the calibration of the model on the eight events was successful, showing good efciency.

Flash ood peak ow estimation through hydrologic and hydraulic simulation Time series of 1 km2 spatial resolution hourly precipitation elds were generated from the three rainfall stations of Gioros basin according to the method that was previously described. The effect of initial decit on the ood peak discharge estimate was tested and proved to slightly affect the result. The Id was set at 8 mm, used as a tting parameter to the measured hydrograph until the stage gauge failure (Fig. 8).The CR based on Fig. 7a was estimated at 7.1 mm/h and R at 1.5 h based on Fig. 7b. The simulated ow hydrograph with the data from the damaged river gauge is presented in Fig. 8. The estimated total water volume was 5.20 Mm3. The peak ow value was 296 m3/s. The maximum precipitation intensity as well as the center of mass of precipitation time series was observed at 20:00 of 13 January 1994, while the peak discharge based on the simulation was observed at 24:00 (Fig. 8). According to the rain-gauge measurements and the meteorological analysis, high precipitation rates were observed over the

9 8

Constant Rate-CR (mm/h)

7 6 5 4 3 2 1 0 1

CR= 1.73API + 0.99 R = 0.8801

(a)

StormEvents
2 3 4

Flood
5

API (mm/h)
7 6 5 4 3 2

Clark Storage Coefficient-R (h)

(b)
R= -1.31API + 6.71 R = 0.8696

StormEvents
1

Flood
2 3 4 5

API (mm/h)
Fig. 7. Correlation of event scale API (average precipitation intensity mm/h) with CR (a) and R (b) values for the calibration events and estimation of parameters values for the ash ood event.

500 450 400 350 Outflow (m3/s) 300 250 200 150 100 50 0
Uncertainty boundaries Precipitation Q observed Q simulated Stage gauge failure

0 5 10 Precipitation (mm) 15 20 25 30 35 40 45 50

Fig. 8. Simulation of 1994 Gioros ood hydrograph at Finikia location by HECHMS model and uncertainty bars. The river gauge washed out at a very low discharge, probably from the suspending large particles of the bouldery front (Pierson, 1986).

0:00 2:00 4:00 6:00 8:00 10:00 12:00 14:00 16:00 18:00 20:00 22:00 0:00 2:00 4:00 6:00 8:00 10:00 12:00 14:00 16:00 18:00 20:00 22:00
13-Jan-94 14-Jan-94
Fig. 9. A three dimension representation of the 1994 ood wave simulation 3 h time step facing upstream. The upper cross-section is the Finikia control crosssection presented in Fig. 10. The lower cross-section is near the outlet of Gioros basin on the Mediterranean Sea.

A.G. Koutroulis, I.K. Tsanis / Journal of Hydrology 385 (2010) 150164

161

Fig. 10. Channel-geometry characteristics of the Finikia River gauge control cross-section of the 1994 ood and measured ood extend.

south part of the basin far from the outlet. Moderate precipitation was recorded at the center and north part of the watershed. Post event eld campaigns conrm the timing of the ood peak through witness interviewing. An attempt to examine the propagation of uncertainty of the CR and R parameters on the simulated peak ow made by adding a percent of variation to these parameters, from 10% up to 10%. In addition, different precipitation realizations in terms of temporal (1 h offset) and intensity distribution (40% peak precipitation intensity) between the stations with daily and the one with hourly precipitation as described in Section Reconstruction of precipitation realizations of variable spatiotemporal distribution, constructed to examine the uncertainty of precipitation records. A set of 343 different parameter combinations and precipitation realizations, used in the hydrological model, resulting to an ensemble of output hydrographs. The effect of the CR and R parameters and precipitation component variation on the hydrograph and peak ow is presented in Fig. 8 within the uncertainty area (boundaries). The uncertainty boundaries are dened by the ensemble area formulated by the 343 output hydrographs. Peak ow shows a linear rate of decrease as Clark storage coefcient (R) increases (shorter response time of the basin). The rate of decrease is expressed by the equation peak flow 43:79R 361:91 taking into account a CR = 7.1 mm/h and values of R varying from 1.35 to 1.65. Moreover, peak ow shows a linear rate of decrease as CR increases (higher inltration rates). The rate of decrease given by the equation peak flow 26:28CR 504:03 for R = 1.5 and values of CR varying from 7.1 to 8.9 mm/h. HEC-RAS was used for hydraulic modeling in order to validate the ood extent at a control cross-section (cross-section with measured ood high-water marks) and to simulate the ood wave. The area of simulation is presented in Fig. 1 as the outow sub-basin. The ood hydrograph resulted from rainfallrunoff modeling used as input for the hydraulic model and the geometric setup of the model was based on information retrieved by cross-sections and the DTM of the study area. A three dimensional ood wave simulation is presented in Fig. 9 at a 3-h time step. The simulated ood extent corresponds to the measured ood extent in the control cross-section as presented in Fig. 10. The observed value of the peak ood ow level was 5.1 m and the simulated value was 5.08 m at 24:00 January 13. The maximum value of average ow velocity at the control cross-section and the corresponding Froude number were estimated at 3.62 m/s and 0.62, respectively for January 13th at 24:00. Moreover, the propagation of uncertainty through the hydraulic model was examined. The previously presented ensemble of 343 hydrographs imported to the hydraulic model providing 343 realizations of the maximum ood extend for the control cross-section. The maximum ow level varied between 4.96 m and 5.19 m, the maximum average ow velocity varied from 3.47 m/s to 3.77 m/s and the corresponding Froude number varied from 0.61 to 0.64.

Flash ood peak ow estimation from post ood measurements Because there were no observed eld ow data, the evaluation of the results was based on an indirect method for the estimation of the real peak ow, from a specic cross-section called Finikia (Fig. 10). This cross-section was at the location where the water ow level gauge was located, during the ood. Eq. (5) was applied using S (energy slope) equal to 0.003 m/m as calculated from the streams topography. Gioros is a natural stream channel with fairly regular sections with some grass weeds and little brush in it. Flood plains, during winter time, are covered by short grass and medium to dense bushes. The literature (Arcement and Schneider, 1989; Chow et al., 1988) suggests n1 value of 0.03 for the main channel (the main channel is a weedy stony earth channel) and n2 values for oodplains from 0.04 (short grass with minor obstructions and moderate rises and dips) to 0.08 (medium to dense brush with appreciable obstructions and very irregular shape of oodplain). Flow rates were available from the existing rating curves for up to 2.8 m water level. Furthermore the total ow was estimated for selected values of water level, starting from 3.0 m up to 5.1 m (Fig. 11). Roughness coefcient is the most important parameter of Eq. (5). A sensitivity analysis of ow output for different values of n1, ranging from 0.03 to 0.04 and for n2, ranging from 0.04 to 0.08 was performed. Table 3 contains the parameters of Eq. (5) for various ow levels up to 5.1 m that was the maximum ood level of the control cross-section. The results of this analysis showed that Q varied from 220 m3/s for n1 = 0.04 and n2 = 0.08 to 350 m3/s for n1 = 0.03 and n2 = 0.04 (Table 3). Peak ow for 1994 ood was estimated using Eq. (5) at

H = 1.16Q0.26 D=R2 = 0.99

H (m)

3
field measurements Estimated values for n1=0.03 and n2=0.06

2 H = 0.71Q0.43 D=R2 = 0.99 1

Estimated values for n1=0.03 and n2=0.08 Estimated values for n1=0.03 and n2=0.04 Estimated values for n1=0.04 and n2=0.06 Estimated values for n1=0.04 and n2=0.08 Estimated values for n1=0.04 and n2=0.04

0 0 100 200 300 400

Q (m3/s)
Fig. 11. Water level stage-ow curve at the cross-section of Finikia (grey triangles corresponds to the direct eld measurements).

162

A.G. Koutroulis, I.K. Tsanis / Journal of Hydrology 385 (2010) 150164

0.04 0.08 Qtotal (m3/s)

292 m3/s for n1 = 0.03 and n2 = 0.06, compares well with the simulation result based on hydrological model.
30.9 36.4 42.7 59.7 74.8 92.0 111.1 132.0 154.7 179.0 203.1 218.9

Flash ood peak ow estimation based on an empirical method A number of non-ood causing rainfall events when the waterstage gauge was operating were used for the estimation of the parameter d of Eq. (4). The hydrologic characteristics of the rainfallrunoff events are included in Table 4. When the total runoff volume VT (m3) for a sufcient number of gauged events are plotted against the product of the square of total precipitation P and the standard deviation of precipitation rP (Fig. 12), a linear relation is evident described by Eq. (2) with a = 215 as a coefcient for Gioros basin. The extrapolation of the regression line meets the values of the 1994 ash ood event. Moreover, when the total runoff volume VT (m3) for the same events, are plotted against the product of the peak discharge Qpeak and duration of precipitation time series D of each event (Fig. 13) provides the power regression Eq. (3) with b = 0.0118 and c = 1.3702 as coefcients for Gioros basin. The elimination of VT from Eqs. (2) and (3) results in d = 18.5 and consequently to Eq. (17), so given the coefcients and the precipitation time series allows estimating the peak discharge of an event for Gioros watershed.

0.03 0.08 Qtotal (m3/s) 0.04 0.06 Qtotal (m3/s) 0.03 0.06 Qtotal (m3/s) 0.04 0.04 Qtotal (m3/s)

41.2 48.6 56.9 79.6 99.7 122.6 148.1 176.0 206.3 238.7 270.8 291.8

30.9 36.4 42.7 62.8 79.9 99.7 121.8 146.2 172.6 201.1 229.4 247.8

41.2 48.6 56.9 76.5 94.5 114.9 137.4 161.9 188.3 216.7 244.5 262.9

Q peak

18:5 2 PT rP 1:3702 D

30.9 36.4 42.7 69.0 90.3 115.2 143.4 174.5 208.5 245.2 282.0 305.6

17

0.03 0.04 Qtotal (m3/s)

Based on Eq. (17) the peak ow of the gauged and the ash ood events was estimated. Table 4 contains the estimated peak ow. Peak ow for the 1994 ash ood event based on the empirical
41.2 48.6 56.9 85.8 110.0 138.2 169.6 204.4 242.1 282.8 323.4 349.6

7.0E+06
n1 n2 S (m/m)

6.0E+06

Total Event Volume (m3)

0.003 0.003 0.003 0.003 0.003 0.003 0.003 0.003 0.003 0.003 0.003 0.003

5.0E+06 4.0E+06 3.0E+06 2.0E+06 1.0E+06

V= 215P2P R = 0.9093

R2 (m /m)

0 0 0 0.53 0.71 0.9 1.08 1.26 1.44 1.62 1.77 1.88

R1 (m /m)

Events
0.0E+00 0
1.23 1.31 1.38 1.5 1.63 1.76 1.88 2.01 2.14 2.26 2.34 2.44

1994 FF Event
25000 30000

5000

10000

15000

20000

P2P (mm3)
Fig. 12. Total runoff volume VT (m3) as a function of precipitation squared times its standard deviation.

P2 (m)

Table 3 Variation of ow for different ow level and n1 and n2 values.

0 0 0 38.57 40.23 40.85 41.49 42.1 42.73 43.36 44.58 44.3

2.5E+07
P1 (m) 15.9 17.03 18.17 18.84 19.23 19.63 20.03 20.43 20.82 21.23 21.97 21.82

2.0E+07

QpeakD (m3)

QpeakD= 0.0118V1.3702
R = 0.9709

1.5E+07 1.0E+07 5.0E+06

A2 (m ) A1 (m )

0 0 0 20.61 28.5 36.59 44.8 53.13 61.57 70.13 78.81 83.2

19.63 22.28 25.14 28.18 31.3 34.48 37.74 41.07 44.46 47.92 51.46 53.25

0.0E+00

Events

1994 FF Event

h (m)

Total Event Volume (m3)

3 3.2 3.4 3.6 3.8 4 4.2 4.4 4.6 4.8 5 5.1 Fig. 13. Total runoff volume VT (m3) as a function of the product of the peak discharge and the duration of precipitation for each event.

A.G. Koutroulis, I.K. Tsanis / Journal of Hydrology 385 (2010) 150164

163

Table 4 Hydrologic characteristics of rainfallrunoff events for peak discharge estimation based on the empirical equation. Peak ow for ash ood event (case 9) is estimated through hydrological modeling. Case 1 2 3 4 5 6 7 8 9 Date 3031 January 1993 1112 January 1995 1314 January 1995 2931 January 1994 1415 January 1995 78 February 1994 1517 February 1981 2224 February 1981 (FF) 1314 January 1994 Total event runoff volume (m3) 33,812 180,579 151,577 482,020 408,760 834,502 1,394,986 3,333,347 5,195,000 Total precipitation (mm) 2.5 16.0 10.4 26.4 32.0 21.8 53.1 74.4 75.7

rp DEV of precipitation mm)

0.59 2.24 4.67 0.87 2.74 0.76 2.51 2.74 4.36

Duration of precipitation (h) 3 6 2 17 11 21 24 47 19

Runoff coefcient 0.09 0.07 0.09 0.12 0.08 0.24 0.17 0.28 0.43

Peak ow (m3/s) 2.3 10.8 10.0 9.5 20.1 20.1 40.5 61.0 296

Estimated peak ow (m3/s) 0.0 5.2 13.0 2.0 24.8 0.8 40.4 58.6 287.4

method estimated to 287 m3/s that compares very well with the rainfallrunoff simulation output. The empirical equation has smaller error in higher peak ows which is the operational range of interest. Storm events with low standard deviation of precipitation are not well represented by the empirical approach as shown in Table 4 and in Fig. 12. Empirical equation rationale Based on hydrological characteristics of the 1994 ash ood event (PT, rP, D, VT) we examine the rationale of the empirical Eq. (4). The duration of the event was D = 19 h, the total areal precipitation over the basin was PT = 75.5 mm, the standard deviation of the precipitation time series was rP = 4.36 mm and the total

500 450 400

(a)
Qpeak = 38.213P
1.3702

volume based on rainfallrunoff simulation was estimated VT = 5.20 Mm3. Assuming a set of the same total precipitation and duration but of different standard deviation varying from 1 mm up to 6 mm increasing by 0.5 mm, the expected peak ow was estimated to be 287 m3/s. Fig. 14a illustrates the expected peak ow versus a precipitation event with the same total precipitation and duration of the 1994 ash ood event but different temporal distribution of rainfall expressed by the standard deviation of the precipitation time series. Precipitation distributions with higher standard deviations results in higher peak discharges. By keeping the total precipitation the same and increasing the standard deviation by 20%, which corresponds four times higher precipitation intensity at the 2-h peak of the storm, the peak ow was estimated via Fig. 14a at 365 m3/s which agrees with the value obtained with the other two methods. With the duration varying from 13 h up to 46 h in increments of 3 h, but keeping the total precipitation and standard deviation the same, the peak ow was estimated. Fig. 14b shows that shorter duration events correspond to higher peak discharges.

Peak Flow (m3/sec)

350 300 250 200 150 100 50 0 0 2 4 6 8

Conclusions During many ash ood events there may be a water-stage gauge failure or the ood stage far exceeds the maximum value of the gauges rating curve that makes the estimation of the peak discharge difcult. A method for estimating ash ood peak discharge, hydrograph, and volume is presented that uses the measured rainfall data and river hydrographs in a number of nonood causing rainstorms with an operating water-stage gauge for calibration and verication of the simulated stage-discharge hydrographs. An empirical equation also was developed in order to provide the peak discharge as a function of the total precipitation, its standard deviation and storm duration while the value of the coefcients depend on the characteristics of the basin. The peak discharge for a ash ood case based on the empirical equation, is in agreement with the one calculated via the veried hydrological model and the one derived based on Mannings equation and post ood measurements of the control cross-section (e.g., all values were close to 290 m3/s for the 1994 ood on the Gioros River). These methods can be applied to other poorly gauged basins facing common stage gauge failures or poorly dened rating curves during ash oods caused by severe convective rainstorms. The coefcients that represent the characteristic of the basin have to be developed for the individual basin. A severe storm that produces a ash ood can be approached with a variety of methods. Among others, meteorological analysis, hydrological modeling, hydraulic modeling and analysis, post event campaigns for data retrieving (ood marks, peak ow timing through interviews) can be used to provide additional information for reliable peak discharge estimations. This multilateral approach,

Generated events

FF Event

P (standard deviation of precipitation in mm for a 19h rs duration event of 75.7 mm total P

450 400 350

(b)
Qpeak = 5459.9D-1

Peak Flow (m3/sec)

300 250 200 150 100 50 0 0 10 20 30 40 50

Generated events

FF Event

D (duration in hrs for an event of 75.7mm total P and P of 4.36 mm

Fig. 14. Flash ood peak ow variation with storm duration and standard deviation of the total precipitation (75.7 mm) (a) storm duration = 19 h and (b) standard deviation = 4.5 mm.

164

A.G. Koutroulis, I.K. Tsanis / Journal of Hydrology 385 (2010) 150164 Marchi, L., Borga, M., Preciso, M., Sangati, M., Gaume, E., Bain, V., Delrieu, G., nik, N., 2009. Comprehensive post-event survey of a ash Bonnifait, L., Pogac ood in Western Slovenia: observation strategy and lessons learned. Hydrological Processes 23 (26), 37613770. doi:10.1002/hyp.7542. Marsigli, M., Todini, F., Diomede, T., Liu, Z., Vignoli, R., 2002. Calibration of Rainfall Runoff Models, Project Report of Multi-Sensor Precipitation Measurements Integration, Calibration and Flood Forecasting (MUSIC-EC Project Contract No EVK1-CT-2000-00058), Presented at European Workshop on New Tools for Flood Forecasting and Warning, Helsinki, Finland, 2223 June 2004. <http:// www.geomin.unibo.it/hydro/music/index2.htm>. Matheron, G., 1965. Les variables rgionalises et leur estimation. Une application de la thorie des fonctions alatoires aux sciences de la nature. Masson, Paris (in French). McCain, J.F., Shroba, R.R., 1979. Storm and Flood of July 31August 1, 1976, in the Big Thompson River and Cache la Poudre River basins, Larimer and Weld Counties, Colorado, USGS Series Professional Paper, Report Number 1115-A,B. <http://pubs.er.usgs.gov/usgspubs/pp/pp1115AB>. Nash, J.E., Sutcliffe, J.V., 1970. River ow forecasting through conceptual models part I a discussion of principles. Journal of Hydrology 10 (3), 282290. doi:10.1016/0022-1694(70)90255-6. ISSN 0022-1694. Nasri, S., Cudennecb, C., Albergelc, J., Berndtssond, R., 2004. Use of a geomorphological transfer function to model design oods in small hillside catchments in semiarid Tunisia. Journal of Hydrology 287 (14), 197213. doi:10.1016/j.jhydrol.2003.10.001. National Research Council (NRC), 1999. Improving American river ood frequency analyses. In: Committee on American River Flood Frequency: Water Science and Technology Board, Committee on Geoscience, Environment and Resources. National Academy Press, Washington DC, pp. 1120. Norbiato, D., Borga, M., Sangati, M., Zanon, F., 2007. Regional frequency analysis of extreme precipitation in the eastern Italian Alps and the August 29, 2003 ash ood. Journal of Hydrology 345 (34), 149166. doi:10.1016/j.jhydrol.2007. 07.009. Ottenbacher, A., Tomassini, M., Holmlund, K., Schmetz, J., 1997. Low-level cloud motion winds from meteosat high-resolution visible imagery. Weather Forecasting 12 (1), 175184. doi:10.1175/1520-0434(1997)012<0175: LLCMWF>2.0.CO;2. Pierson, T.C., 1986. Flow behavior of channelized debris ows. Mount St. Helens, Washington. In: Abrahms, A.D. (Ed.), Hillslope Processes. Allen and Unwin, Boston, pp. 269296. Peters, J.C., Easton, D.J., 1996. Runoff simulation using radar rainfall data. Journal of the American Water Resources Association 32 (4), 753760. doi:10.1111/ j.1752-1688.1996.tb03472.x. Pruess, J., Wohl, E., Jarrett, R., 1998. Methodology and implications of maximum paleodischarge estimates for mountain channels, Upper Animas River Basin, Colorado. USA Artic and Alpine Research 30 (1), 4050. Reed, S., Schaake, J., Zhang, Z., 2007. A distributed hydrologic model and threshold frequency-based method for ash ood forecasting at ungauged locations. Journal of Hydrology 337 (34), 402420. doi:10.1016/j.jhydrol.2007.02.015. ISSN 0022-1694. Sangati, M., Borga, M., Rabuffetti, D., Bechini, R., 2009. Inuence of rainfall and soil properties spatial aggregation on extreme ash ood response modelling: an evaluation based on the Sesia river basin, North Western Italy. Advances in Water Resources 32 (7), 10901106. doi:10.1016/j.advwatres.2008.12.007. ISSN 0309-1708. Shepard, D., 1968. A two-dimensional interpolation function for irregularly-spaced data. In: Proceedings of the 1968 ACM National Conference, pp. 517524. doi:10.1145/800186.810616. Streeter, V., Benjamin, W.E., 1988. Fluid Mechanics. Translated by Tsimikalis G., 638, Fountas, Thessaloniki, Greece. ISBN 960-330-103-x. Thiessen, A.H., 1911. Precipitation averages for large areas. Monthly Weather Review 39 (7), 10821084. Uppala, S.M., Kllberg, P.W., Simmons, A.J., Andrae, U., Da Costa Bechtold, V., Fiorino, M., Gibson, J.K., Haseler, J., Hernandez, A., Kelly, G.A., Li, X., Onogi, K., Saarinen, S., Sokka, N., Allan, R.P., Andersson, E., Arpe, K., Balmaseda, M.A., Beljaars, A.C.M., Van De Berg, L., Bidlot, J., Bormann, N., Caires, S., Chevallier, F., Dethof, A., Dragosavac, M., Fisher, M., Fuentes, M., Hagemann, S., Hlm, E., Hoskins, B.J., Isaksen, L., Janssen, P.A.E.M., Jenne, R., Mcnally, A.P., Mahfouf, J.-F., Morcrette, J.-J., Rayner, N.A., Saunders, R.W., Simon, P., Sterl, A., Trenberth, K.E., Untch, A., Vasiljevic, D., Viterbo, P., Woollen, J., 2005. The ERA-40 re-analysis. Quarterly Journal of the Royal Meteorological Society 131 (612), 29613012. doi:10.1256/ qj.04.176. Webb, R.H., Jarrett, R.D., 2002. Ancient oods modern hazards, principles and application of paleoood hydrology, chapter one-dimensional estimation techniques for discharges of paleoood and historical oods. Water Science and Application, American Geophysical Union 5, 295310. Wharton, G., 1992. Flood estimation from channel size: guidelines for using the channel-geometry method. Applied Geography 12 (4), 339359. doi:10.1016/ 0143-6228(92)90013-D. ISSN 0143-6228. Wharton, G., Arnell, N.W., Gregory, K.J., Gurnell, A.M., 1989. River discharge estimated from channel dimensions. Journal of Hydrology 106 (34), 365376. doi:10.1016/0022-1694(89)90080-2. Wohl, E.E., 1995. Estimating ood magnitude in ungauged mountain channels, Nepal. Mountain Research and Development 15, 6976. Woods, R., Sivapalan, M., 1999. A synthesis of space-time variability in storm response: rainfall, runoff generation and routing. Water Resource Research 35, 24692485.

as described in the present study, reduces uncertainty in the event interpretations. References
Arcement, G.J., Schneider, V.R., 1989. Guide for Selecting Mannings Roughness Coefcients for Natural Channels and Flood Plains. Water-Supply Paper 2339, US Geological Survey. <http://www.fhwa.dot.gov/BRIDGE/wsp2339.pdf>. Barboudakis, M., 1998. Environmental Impacts Research of the Upstream Flood Protection Constructions of Gioros Hydrological Basin. Technical Report, Organization of Development for Eastern Crete, Athens (in Greek). Bathurst, J.C., 1990. Tests of three discharge gauging techniques in Mountain Rivers. In: Molnar, L. (Ed.), Hydrology of Mountainous Areas: Wallingford, UK, IAHS Publ. No. 190. Wallingford, UK, pp. 93100. Bloschl, G., Reszler, C., Komma, J., 2008. A spatially distributed ash ood forecasting model. Environmental Modelling & Software 23, 464478. doi:10.1016/j.envsoft.2007.06.010. ISSN 1364-8152. Bohorquez, P., Darby, S.E., 2008. The use of one- and two-dimensional hydraulic modelling to reconstruct a glacial outburst ood in a steep Alpine valley. Journal of Hydrology 361 (34), 240261. doi:10.1016/j.jhydrol.2008.07.043. ISSN 0022-1694. Bonnifait, L., Delrieu, G., Le Lay, M., Boudevillain, B., Masson, A., Belleudy, P., Gaume, E., Saulnier, G.M., 2009. Distributed hydrologic and hydraulic modelling with radar rainfall input: reconstruction of the 89 September 2002 catastrophic ood event in the Gard region, France. Advances in Water Resources 32 (7), 10771089. doi:10.1016/j.advwatres.2009.03.007. ISSN 0309-1708. Borga, M., Gaume, E., Creutin, J.D., Marchi, L., 2008. Surveying ash oods: gauging the ungauged extremes. Hydrological Processes 22 (18), 38833885. doi:10.1002/hyp.7111. Borga, M., Boscolo, P., Zanon, F., Sangati, M., 2007. Hydrometeorological analysis of the August 29, 2003 ash ood in the eastern Italian Alps. Journal of Hydrometeorology 8 (5), 10491067. doi:10.1175/JHM593.1. Chow, V.T., Maidment, D.R., Mays, L.W., 1988. Applied Hydrology. McGraw-Hill, New York, NY. Clark, C.O., 1945. Storage and the unit hydrograph. Transactions: American Society of Civil Engineers 110 (1945), 14191488. Collier, C.G., 2007. Flash ood forecasting: what are the limits of predictability? Quarterly Journal of the Royal Meteorological Society 133 (622), 323. doi:10.1002/qj.29. Costa, J.E., 1987. Hydraulics and basin morphometry of the largest ash oods in the conterminous United States. Journal of Hydrology 93 (34), 313338. doi:10.1016/0022-1694(87)90102-8. Costa, J.E., Jarrett, R.D., 2008. An Evaluation of Selected Extraordinary Floods in the United States Reported by The US Geological Survey and Implications for Future Advancement of Flood Science. United States Geological Survey Scientic Investigations Report 2008-5164, 232 pp. Creutin, J.D., Borga, M., 2003. Radar hydrology modies the monitoring of ashood hazard. Hydrological Processes 17 (7), 14531456. doi:10.1002/hyp.5122. Dottori, F., Martina, M.L.V., Todini, E., 2009. A dynamic rating curve approach to indirect discharge measurement. Hydrology and Earth System Sciences 13, 847863. <http://www.hydrol-earth-syst-sci.net/13/847/2009/>. Ganoulis, J., 2003. Risk-based oodplain management: a case from Greece. International Journal of River Basin Management 1 (1), 4147. Gaume, E., Livet, M., Desbordes, M., Villeneuve, J.P., 2004. Hydrological analysis of the river Aude, France, ash ood on 12 and 13 November 1999. Journal of Hydrology 286 (14), 135154. doi:10.1016/j.jhydrol.2003.09.015. Gaume, E., 2006. Post ash-Flood Investigation Methodological Note. Floodsite European Research Project, Report D23.2., 62 pp. <http://www.oodsite.net/>. Gaume, E., Borga, M., 2008. Post-ood eld investigations in upland catchments after major ash oods: proposal of a methodology and illustrations. Journal of Flood Risk Management 1 (4), 175189. doi:10.1111/j.1753-318X.2008. 00023.x. <http://dx.doi.org/10.1111/j.1753-318X.2008.00023.x>. Gaume, E., Bain, V., Bernardara, P., Newinger, O., Barbuc, M., Bateman, A., ov, L., Blschl, G., Borga, M., Dumitrescu, A., Daliakopoulos, I., Blakovic Garcia, J., Irimescu, A., Kohnova, S., Koutroulis, A., Marchi, L., Matreata, S., Medina, V., Preciso, E., Sempere-Torres, D., Stancalie, G., Szolgay, J., Tsanis, I., Velasco, D., Viglione, A., 2009. A compilation of data on European ash oods. Journal of Hydrology. doi:10.1016/j.jhydrol.2008.12.028. HEC, 2000. HEC, Hydrologic Modeling System: Technical Reference Manual. US Army Corps of Engineers Hydrologic Engineering Center, Davis, CA. HEC, 2001. HEC-HMS Users Manual. US Army Corps of Engineers Hydrologic Engineering Center, Davis, CA. Jarrett, R.D., 1987. Errors in slopearea computations of peak discharges in mountain streams. Journal of Hydrology 96, 5367. Kirpich, P.Z., 1940. Time of concentration of small agricultural watershed. ASCE Civil Engineers 10 (6), 362. Kull, D., Feldman, A., 1998. Evolution of Clarks unit graph method to spatially distributed runoff. Journal of Hydrologic Engineering 3 (1), 919. doi:10.1061/ (ASCE)1084-0699(1998)3:1(9). Lebel, T., Bastin, G., Obled, C., Creutin, J., 1987. On the accuracy of areal rainfall estimation: a case study. Water Resource Research 23 (11), 21232134. doi:10.1029/WR023i011p02123. Linsley, R.K., Kohler, M.A., Paulhus, J.L.H., 1982. Hydrology for Engineers. McGrawHill, New York, NY.