the response of braided planform configuration to flow variations, bed reworking and vegetation: the...

The response of braided planform configuration toflow variations, bed reworking and vegetation: thecase of the Tagliamento River, ItalyMatilde Welber,1* Walter Bertoldi1,2 and Marco Tubino11 University of Trento, Department of Civil and Environmental Engineering, Trento, Italy2 Queen Mary University of London, School of Geography, London, UK

Received 26 December 2010; Revised 9 December 2011; Accepted 12 December 2011

*Correspondence to: M. Welber, University of Trento, Department of Civil and Environmental Engineering, Trento, Italy. E-mail: matilde.welber@ing.unitn.it

ABSTRACT: Morphological features of braided rivers (bars, channels and pools) experience major changes in area, shape andspatial distribution as a response to (i) the pulsation of discharge during a flood and (ii) the bed evolution induced by floods. In thiswork, at-a-station relationships between water level and planform configuration were investigated on the Tagliamento River, a largegravel-bed braided river in northeast Italy, over a 2-year study period comprising three bankfull events and several small-to-mediumfloods. The analysis was performed on two 1-km-long reaches, characterized by different riparian vegetation cover. Ground-basedimages with an hourly temporal resolution were acquired using software-controlled, digital cameras. Bars, channels, pools andvegetated patches were manually digitized on more than 100 rectified images. Sequences of constant-level images spanning the studyperiod were used to quantify the impact of floods on the stability of at-a-station relationships and on the turnover rate of water bodies.The analysis shows that wetted area increased almost linearly with water level in both reaches. The average number of branches

per cross-section peaked at intermediate flow levels, increasing from 2 at low flow up to 6–7. The number of branches displayed thelargest fluctuations over time, with significant changes produced also by moderate floods. Turnover rates were high in both reaches,with more than 30% of wetted areas at low flow converting into bare gravel in less than 2months. Vegetation colonization was foundto limit the mobility of the low flow channels over time by concentrating the flow in fewer, deeper anabranches. The number ofchannels per cross-section was 30–40% less in the vegetated reach and the proportion of low flow water bodies in the same positionafter 12months increased from 3% to 14%. Copyright © 2011 John Wiley & Sons, Ltd.

KEYWORDS: braided rivers; inundation dynamics; ground-based imagery; planform configuration; Tagliamento River

Introduction

Braided rivers show distinct inundation dynamics, wherethe river landscape completely changes during floods. A multi-channel pattern reacts to water level fluctuations with the activa-tion and/or merging of branches, the submersion of exposed barsand the modification of groundwater/surface water interactions.As a consequence, major changes of both the morphologicaland the ecological features take place on the time span of singleflood events (Mosley, 1982, 1983; van der Nat et al., 2002, 2003;Lorang et al., 2005; Malard et al., 2006; Doering et al., 2007).Recent field and flume based investigations of braided river

morphology (Ashmore and Sauks, 2006; Bertoldi et al.,2009b) reported at-a-station relationships markedly differentfrom those typical of single-thread channels. In particular thesestudies highlight a much faster increase of the water surfacewidth with discharge. However, further field evidence of thisbehaviour is needed. Moreover, it is not yet clear how thenumber of branches in a multi-channel pattern changes (i) duringa single flood and (ii) as a consequence of the morphologicalevolution induced by floods. Different (sometime contrasting)relationships between discharge and braiding index, defined asthe average number of branches in each cross-section (following

Howard et al., 1970; Egozi and Ashmore, 2008), have beenproposed (Mosley, 1983; Robertson-Rintoul and Richards,1993; Chew and Ashmore, 2001; van der Nat et al., 2002; Luchiet al., 2007; Bertoldi et al., 2009b). These relationships are oftenbased on small databases, because of the difficulty in surveyingthe network configuration at different flow levels.

The observed extensive changes in the pattern of wet and dryenvironments are responsible for a highly dynamic fluvialsystem. The pulsing discharge is the dominant driving force thatgoverns the abundance and persistence of geomorphic units(such as bars, main and side channels, isolated pools, vegetatedislands) and therefore of specific meso-habitats (Tockner et al.,2000). Inundation cycles influence several aspects of the rivermorphology as sediment transport initiation (Surian et al.,2009); flow–sediment–vegetation interaction (Bertoldi et al.,2009a); fine sediment deposition and floodplain building(Haschenburger and Cowie, 2009). The inundation regime alsoaffects river habitat dynamics, e.g. mass/energy exchanges(Latterell et al., 2006); nutrients supply through biomassdecomposition (Langhans and Tockner, 2006); aquatic ecosys-tem fragmentation (Doering et al., 2007); disturbance drivenrejuvenation and habitat succession (Arscott et al., 2002;Steiger et al., 2005); thermal heterogeneity induced by

M. WELBER, W. BERTOLDI AND M. TUBINO

groundwater upwelling patterns (Tonolla et al., 2010). Under-standing the temporal scales of inundation dynamics and their ef-fect on the spatial distribution of different geomorphic units istherefore essential to improve the knowledge on large braided riv-ers dynamics.At present, the complexity of multi-channel patterns and the

different spatial and temporal scales on which sediments movelimit the possibility to quantify and predict bed activity. Recentanalysis of morphological processes in braided rivers showedthat bedload transport occurs only on a (small) fraction of theriver width, even during bankfull events (Ashmore, 2001; Egoziand Ashmore, 2008; Bertoldi et al., 2009b, 2010; Mao andSurian, 2010). Therefore, relevant morphological changes arelimited to a few main branches, possibly reducing theturnover of geomorphic units and habitats. However, further re-search is needed to assess the effect of network rearrangementinduced by floods on reach-averaged inundation patterns.In this work we use ground-based oblique imagery to monitor

planform river configuration at high temporal resolution.Ground-based surveying systems can be automated and theiracquisition frequency can be increasedwithout facing prohibitivecosts (Chandler et al., 2002; Ashmore and Sauks, 2006). Recentimprovements in remote sensing techniques (Marcus and Fonstad,2010) greatly increased the possibility tomap river features (Winter-bottom and Gilvear, 1997; Antonarakis et al., 2008; Feurer et al.,2008; Legleiter et al., 2009). LiDAR and multispectral surveys givethe possibility to produce a three-dimensional image of the river,computing bed elevation and/or water depth. Nevertheless thesetechniques are limited in terms of temporal accuracy, as before/afterflood surveys are the best possible scenario.For this study hourly pictures were acquired by digital cameras

overlooking two different reaches of the Tagliamento River, alarge gravel bed braided river in northeast Italy. The high acquisi-tion frequency of this system allows the observation and

Figure 1. Location of the study sites, ground-based cameras and gaugingdelimit the study area.

quantification of morphological parameters as a function of theflow level and at the temporal scale of single floods.

The first issuewe address in this paper is the definition of the at-a-station relationship between water level and morphologicalconfiguration, described in terms of water surface width (orproportion of wetted area), channel connection, braiding indexand shoreline length. The study site also provides the opportunityto isolate and quantify the role of vegetation in the inundationdynamics of a braided river. The two selected reaches are charac-terized by significantly different vegetation abundance, butsimilar values of discharge, slope and grain size (Bertoldi et al.,2011a). Previous work recognized vegetation as a crucial param-eter controlling fluvial forms and processes (Tal and Paola, 2007,2010; Braudrick et al., 2009; Gurnell et al., 2009; Eaton et al.,2010; Bertoldi et al., 2011b; Crosato and Saleh, 2011). Labora-tory and numerical modelling showed that bed stability inducedby root reinforcement is able to limit the development of braidednetworks, reducing the number of branches. However, very fewfield data are available on this topic. The second objective of thiswork is therefore to investigate between-reach differences of theplanform configuration of the network, in terms of width andbraiding index. Moreover, we assess the morphologicalchanges induced by floods, quantifying the variability of thereach-averaged planform parameters over time. Finally, weanalyse the temporal persistence of river features, comparingmaps of wet and dry areas (at the same water level) on a timeperiod of 2 years, thus providing data on the temporal scalesof geomorphic units turnover.

Study site

The Tagliamento River (Figure 1) is a large, gravel-bed, braidedriver that represents one of the main riverine European

stations within the Tagliamento River catchment. White dotted lines

Earth Surf. Process. Landforms (2011)

Figure 3. Water stage record at Venzone hydrometric station duringthe study period and timeline of selected pictures for the constant-levelanalysis. Filled dots represent picture availability for both sites, circlesrepresent images for single sites.

THE RESPONSE OF BRAIDED PLANFORM CONFIGURATION TO FLOW VARIATIONS

corridors. It is located in northeast Italy and connects the Alpsto the Adriatic Sea. It shows an essentially pristine, highlydynamic braided morphology for most of its 172-km-longcourse (Ward et al., 1999; Tockner et al., 2003). The catchmentcovers an area of about 2580 km2 and is characterized bytransitional, alpine to Mediterranean climatic conditions.Average annual precipitation can be as high as 3000mm inthe prealpine area of the catchment (Doering et al., 2007). Asa result, the river shows a bimodal, flashy pluvio-nival regime.Discharge peaks in spring and in autumn but floods of differentmagnitude can occur throughout the year (Gurnell et al., 2001;Tockner et al., 2003). Although water abstraction for hydro-power generation and irrigation occurs in the upper course, thisis limited compared with other catchments in the Alps. Theriver maintains a near-natural, dynamic flow regime (Bertoldiet al., 2009a), where a wide range of discharges is responsiblefor the river evolution (Surian et al., 2009).The active floodplain in the braided sections of the river is

up to 1.5 km wide and is bordered by nearly-continuousriparian vegetation. Patches of riparian shrubs and trees arefound throughout the reach. Dominant riparian tree generaare Populus and Salix, with P. nigra being the most commonspecies (Karrenberg et al., 2003). Vegetative regeneration,particularly from uprooted trees deposited on gravel bars, isthe main process leading to the formation of vegetated islands(Gurnell et al., 2001, 2005; Gurnell and Petts, 2006). A relevantportion of the braided riverbed is covered by large patches ofmature trees (approximately 8%) that show a relativelyfast turnover, on average 20 years (Zanoni et al., 2008). Theprealpine section of the river is characterized by specificpatterns of surface–groundwater interaction, induced by thepresence of a natural rocky constraint near Pinzano (Figure 1).The alternation between downwelling and upwelling reachescontrols water availability and hence growth performance ofthe riparian trees, determining large, local variations in thevegetation abundance (Bertoldi et al., 2011a).In the present study we focus on two 1-km-long reaches

located in the prealpine section. Both sites are well suited forground-based remote sensing of riverine landscape featuresdue to their proximity to steep cliffs. The Cornino reach islocated 9 km upstream of Pinzano at the foot of Monte Pratand has a maximum width of 900m (Figure 2(a)). This reachcan be defined as bar-braided, with a few vegetated patchesformed by sparse, shrubby vegetation. The Flagogna reach,located 3 km upstream of Pinzano, has a maximum width of650m and is dominated by Monte Ragogna. Several islandsare present in this reach (that can be defined island-braided),with dense woody vegetation located mainly on the right partof the braided riverbed (Figure 2(b)). Slope in both sites isapproximately 0.35% and mean grain size is about 40mm,

Figure 2. Ground-based camera on Monte Prat overlooking the bar-braidedMonte Ragogna (B). Flow is from left to right in (A), from right to left in (B).

with a considerable fraction of fine sediments trapped by thevegetated patches, particularly at the Flagogna reach.

Data Acquisition and Analysis

River stage data

A study period of two years, from March 2008 to February2010, was defined. During this period, 14 bed-moving floodevents occurred, three of which were approximately bankfull(Bertoldi et al., 2009a). Figure 3 shows the hydrometric recordwith 30-min interval data for the Venzone gauging station,located 15 km upstream of the Cornino reach. A water stageequal to 3m at Venzone corresponds roughly to a 2 years returnperiod. Data were made available by the Servizio Idrografico ofthe Friuli Venezia Giulia Region. Currently, no official stage–discharge relationship is available for this or any other gaugingstation on the Tagliamento and therefore water level is usedto describe flow conditions. A flow level duration curve wascomputed for the Venzone station, using a 6-year record coveringthe 2004–2009 period.

Instantaneous flow conditions occurring at Cornino andFlagogna reaches were associated to stage data acquired 1.5and 2 h earlier at Venzone, respectively. Flood propagationcelerity was estimated for a set of 20 events over the 2002–2009 period by comparing stage data recorded at Venzonewith records at the Villuzza gauging station, located 23.5kmdownstream. An average travel time of 2.65 h between the twohydrometric stations was computed, though no clear relationship

Cornino reach (A) and the island braided Flagogna reach as seen from

between peak water level and travel time was found. We notethat a reliable estimation of the water level at the study reachesis particularly relevant during the flashy summer floods, whenstage can vary by up to 2m in a few hours.

Pictures acquisition and selection

Planform configuration of the two study reaches was surveyedthrough a ground-based remote sensing system (Figure 2(a)),consisting of solar-powered digital reflex cameras (NikonD40, with an 18mm lens and a 6 Mpixel CCD sensor). Thecameras are located on Monte Prat (installed in June 2006)and Monte Ragogna (set in March 2008), approximately350m above the floodplain. Horizontal distance between thecamera and the floodplain ranges from 0.5 to 1.4 km for bothreaches. The set-up is similar to that used by Chandler et al.(2002) and by Ashmore and Sauks (2006), but in this casepicture acquisition and storage is software controlled, withthe possibility to set up acquisition time depending on seasonalconditions. A time resolution of 1 h during daylight was chosenfor this survey, in order to gather a sufficient number of imagesduring the rapid floods. The system provided a high-frequencyphotographic record for the whole study period. A few picturesobtained by this survey system were used by Bertoldi et al.(2010) to study bank erosion and channel shift as afunction of flood intensity. Here we exploit the full capacityof the acquisition system analysing a much larger set of picturesand investigating during-flood processes.A set of 68 images of the two sites was used to characterize

the different inundation conditions, with water level at theVenzone hydrometric station spanning a 3m range. Availablepictures cover five different flood events and span a wide rangeof level duration, from around 330days per year to few hoursper year. Highest points are close to bankfull, though peakflood often occurred at night or under low visibility conditions(fog or heavy rain).A second set of pictures was used to investigate channel

reworking induced by floods. For both sites, 12 imagescorresponding to low flow conditions and 12 for high flowconditions were selected (Figure 3). The low flow level(approximately 0.05m at Venzone) corresponds to a waterstage that is exceeded approximately 180 days per year. Thehigh flow level (approximately 0.8m) represents the conditionin which about 50% of the riverbed is submerged and isexceeded 6 days/year. Pictures for the high flow level coverall but two of the 14 bed-moving events that occurred in thestudy period (the threshold for sediment movement was evalu-ated following Surian et al., 2009). For the two remainingevents, power shortage prevented image acquisition.Overall, 114 pictures of the two reaches were used to

completely characterize the inundation dynamics over the2 years study period.

Image processing and relevant landform typesidentification

Oblique pictures were rectified using about 15 GPS-surveyedground control points, through the Leica PhotogrammetricSuite of ERDAS ImagineTM. The final image resolution is 1mfor the Flagogna reach and 2m at Cornino. Planform configura-tion (gravel bars and water bodies) was manually digitized oneach rectified image using the open source GRASS GISsoftware. Manual classification was preferred due to the presenceof disturbances such as shadows produced by trees, reflections

onwater surfaces, variability of water and gravel colour that werelikely to cause misinterpretations.

In the present study we use four classes to map the planformconfiguration and describe the inundation dynamics: gravelbars (class 1), upstream connected water bodies (class 2),partially or totally disconnected water bodies (class 3) andvegetated patches (class 4). In the last case we consider onlyriparian vegetation (trees and shrubs), excluding from the anal-ysis aquatic plants and herbs, as the stream power is too largefor these plants to affect the bed morphology on a reach scale.Class 3 comprises ponds and all water bodies whose upstreamend was clearly identifiable inside the study reach area, such asbackwater channels and groundwater-fed channels. Thesewater bodies differ substantially from full connected channelsin terms of water depth, velocity and temperature and theirpresence significantly enhances aquatic habitat diversity(Arscott et al., 2002; Gray et al., 2006; Sukhodolov et al.,2009). Figure 4 reports on four examples of digitized maps ateach site, showing variations in inundation patterns underincreasing water levels (H in Figure 4 is the water levelmeasured at Venzone hydrometric station, in cm). Vegetatedpatches are reported only at the lowest flow level, for clarityof the pictures (their area does not change with water level).

A set of four geometrical parameters was automaticallycomputed for each map, in order to characterize the planformconfiguration. Namely: (i) the wetted area proportion,evaluated as the ratio between the total wetted area (sum ofclasses 2 and 3) and the total river corridor area; (ii) the proportionof disconnected water bodies, computed as the ratio betweenclass 2 and the sum of classes 2 and 3; (iii) the braiding index(BI), evaluated as the reach-averaged value on a set of 16 cross-sections with a spacing interval of about 60m (Egozi andAshmore, 2008); (iv) the shoreline length, computed as km ofshoreline per km of river reach. The latter two parametersdescribe the network complexity, with the shoreline lengthmeasuring the extent of aquatic/terrestrial habitat interfaces.

For each cross-section, the braiding index was assigned byvisually counting all downstream-connected water bodiesintersecting the cross-section line, regardless of their conditionat the upstream end (in this way excluding all the ponds, butconsidering the groundwater-fed channels). Gravel bars weretaken into account as emerging landforms separating twochannels according to their size compared to the width of theadjacent channels. Values of BI obtained by visual inter-pretation were compared with values obtained through anautomatic procedure, which takes into account only gravelbars larger than a threshold area. We found that the choice ofthe threshold does not influence the shape of the braidingindex–water stage relationship and that a minimum area of 10000m2 minimizes the difference between the values of BIcomputed by the two procedures.

Results

Planform configuration at variable water level

Morphological parameters extracted from the digitized mapsare shown in Figure 5 as a function of the water level atVenzone. Inundation dynamics are described through the wettedarea extension (Figure 5(A)), as a proportion over the total activecorridor area, i.e. excluding floodplainwoodland (the area delim-ited by the white dotted line in Figure 1). The 68 images analysedfor the two sites cover a range from 10 up to 90% of wetted area.Both sites show an almost linear increase of thewetted proportionwith water level. The island braided Flagogna site is characterisedby a lower wetted area, particularly at medium to high levels.

Figure 4. Examples of inundation maps for the Cornino (A) and Flagogna (B) reaches covering a wide range of flow conditions, with the four classes(gravel, upstream connected channels, upstream disconnected channels, and vegetation). For figure clarity vegetated areas are reported only at thelowest flow level. In figure legend H represents water level measured at Venzone hydrometric station (in cm).

Figure 5(B) reports on the ratio between the area of upstreamdisconnected water bodies (e.g. pools, backwaters andgroundwater-fed channels) and total wetted area. At low flow,up to 16% of water bodies area is not connected and thereforecharacterized by quite different values of flow velocity andwater temperature. As water level increases, this ratiodecreases rapidly at Cornino, where almost no disconnectedwater bodies are present at water stage higher than 0.5m.On the contrary, at Flagogna the ratio remains higher up to1m, with nearly 10% of wetted area still partially or totallydisconnected from the network at a water level equal to0.5m. At Flagogna, more intense upwelling occurs due tothe downstream narrowing in the Pinzano gorge, originatingseveral groundwater-fed branches at low flow. In this case,some scatter of the data may be due to the difficulty inmapping relatively small ponds that are likely to be hidden

by vegetation. Nevertheless, the distinct decreasing trendmatches those found by Arscott et al. (2002) and van der Natet al. (2002) for the same study reaches.

Network configuration is also described by the braidingindex (Figure 5(C)). In both reaches the number of branchesincreases from two or- three at low flow up to a maximum forwater levels of approximately 0.5–1m. At higher flows thebranches start to coalesce, determining a drop of the BI tovalues close to 1 for a water level higher than 2m. The tworeaches are quite different in terms of maximum braided index:at Cornino we observed up to seven branches per cross-section(average on the reach), whereas at Flagogna the maximum isabout five, with most of the data falling below four. Finally, inFigure 5(D) the shoreline length is plotted. On the y-axis, theminimum value of 2 km per river km corresponds to the shore-line length of a straight channel. In both study reaches we

Figure 5. At-a-station variability of morphological parameters with water level: (A) wetted area proportion; (B) disconnected water bodies ratio; (C)braiding index; (D) shoreline length.

observed values higher than 20 km per river km, with theshoreline length exceeding 6 km per river km even at verylow or high flow. This parameter shows a trend similar to thatof the braiding index, with a maximum at a water level ofapproximately 0.5–1.0m and generally lower values at Flagogna.To add a temporal dimension to the inundation analysis the

wetted area and the disconnection ratio are plotted in Figure 6as duration curves, i.e. substituting the water level with thenumber of days per year in which the given level is exceeded.Figure 6(A) shows the wetted area proportion for the tworeaches. Water occupies approximately 20% of the activecorridor area for most of the year, exceeding 40% for less than10days per year. This suggests that the Tagliamento braidedcorridor is a mainly terrestrial habitat, subject to strong flooddisturbances. Figure 6(B) highlights the role of disconnectedwater bodies in the island-braided Flagogna reach. Thesefeatures represent a significant proportion of the total wetted area

Figure 6. At-a-station variability of morphological parameters with water letion; (B) disconnected water bodies ratio.

for most of the year, dropping below 8% for only 10days per year.In contrast, upstream disconnected water bodies do not exceed8% of the total for approximately 100days at the Cornino reach.

The effect of morphological changes

The two sets of images taken at given flow levels (roughlycorresponding to 0.05m and 0.80m at Venzone station) wereused to quantify the role of floods in modifying planformmorphology and the related parameters (namely wetted areaproportion and braiding index). Results are reported in Figure 7for both sites and water level ranges. Figure 7(A) and 7(B) showthe changes in the wetted area proportion for the lower andhigher water level, respectively. In both cases we observedvariations up to 10-15%. Differences are smaller at Flagognareach, particularly at low water level, where the wetted

vel, shown as a function of water level duration: (A) wetted area propor-

Figure 7. Variation of morphological parameters over time at constant water level for the two chosen level ranges. The wetted area proportion andthe braiding index for the lower (A, C) and higher water level (B, D) are reported. Grey lines represent the water level recorded at the Venzonehydrometric station.

proportion fluctuates between 13 and 22%. On average,greater variations are observed at the higher water level (at Cor-nino the wetted area ranges from 39% to 55%).The braiding index shows large fluctuations at both low and

high water level (Figure 7(C), (D)). Changes in the cross sectiongeometry induced by floods can increase (or decrease) theaverage BI by two or three channels for a given flow stage. Atthe higher water level the variability is slightly lower, withchanges limited to one or two channels. At low flow, the BIshows a similar fluctuating behaviour at both sites (Figure 7(C)), with two minima separated by approximately 16months.The BI increases from 2.5 to 4.5 at Flagogna and from 3.5 to7.0 at Cornino between August 2008 and March 2009; subse-quently, a decreasing phase starts and the number of branchesreduces close to the minimum observed value. These behav-iour can be interpreted in terms of the stochastic variability ofbed topography and/or of sediment supply. We have noevidence of major changes in the sediment input in the tworeaches, therefore a possible explanation may be related tothe aggrading and degrading phases that have been alsoobserved in flume experiments with constant water discharge(Hoey and Sutherland, 1991). Here in particular the simultaneousincrease in braiding index and wetted area at low flow thatoccurred between August 2008 and March 2009 might havebeen produced by aggradation leading to a larger number ofwide, shallow channels. Such a configuration in turn increasesthe chances of merging of channels as water level rises, whichmay explain the decrease of the braiding index at intermediateflow over the same period observed in Figure 7(D).In principle, one could expect a relationship between flood

magnitude and the extent of change in morphological para-meters; however, no consistent trend is found in our dataset.Generally, larger floods induce more substantial changes inthe planform configuration, though this is not always the case.

Figure 7 also reports the water level recorded at the Venzonehydrometric station, to facilitate the comparison between floodmagnitude and changes in the planform configuration. Forexample, the relatively small July 2008 flood (peak level lowerthan 1.5m at Venzone) caused the braiding index to drop byapproximately 1.5 channels at Cornino at the lower water level;in contrast, after the December 2009 bankfull flood the braidingindex decreased by less than one channel. We also observed thatsingle avulsion/deposition phenomena are able to significantlymodify reach-averaged morphological parameters at low flow.For instance, the opening of a bifurcation in the main channelat Flagogna after the November 2008 flood was almost entirelyresponsible for the increase in braiding index observed in Figure 7(C) between October and December 2008. Hence, singlechannel/bar-scale phenomena may have a disproportionatelylarge influence on reach-averaged parameters, which may alsobe a consequence of the limited length of the study reaches.

Morphological changes induced by floods are likely to be themain cause of the scatter of the data in Figure 5, which is particu-larly evident in braiding index (Figure 5(C)). This parameter fluctu-ates at a given level on a similar range of values as found with thevariable stage analysis, thus confirming field observations byMosley (1983). Thewetted area proportion shows greater stability,as the linear increase with the water level (Figure 5(A)) is notundermined by the fluctuations at constant level (Figure 7(A), (B)).

Water bodies turnover and their interaction withvegetation

The turnover of water bodies (and hence aquatic habitats) wasinvestigated by comparing the planform pattern of channelsbefore and after bed-moving events. The analysis was

igure 9. Distribution of water bodies over persistence classes.

performed by combining the 12 maps at low flow level(approximately 0.05m at Venzone station, see Figure 3), reclas-sified as binary wet/dry maps. For each area unit (pixel) wecomputed the maximum number of consecutive maps in whichthe area was classified as wet. Values in the resulting persis-tence map range from 0 (areas that were never occupied bywater on the 12maps at lowwater level) to 12 (areas permanentlyoccupied by water on the 12 maps). Class 1 corresponds tochannels that are wet in a single picture, but not in the previousand the following one; these areas were assigned a duration of2months, which corresponds to the average time intervalbetween subsequent images.Persistencemaps are shown in Figure 8(A) and 8(B) for Cornino

and Flagogna, respectively. Here, persistence is expressed interms ofwater bodies’ lifespan, assuming an average time intervalof 2months between maps. For clarity, wetted areas lasting formore than 12months have been aggregated to a single class.The intense reworking of the braidplain occurring in the bar-

braided Cornino reach is evident, with 62% of the activeriverbed occupied by low flow channels at least for 2months.In contrast, the island-braided Flagogna reach exhibits a morestable channel pattern, with low flow patterns confined toapproximately 46% of the active tract area. Here, the mainbranch on the left side of the corridor and a secondary channelflowing along the right bank are quite stable and thereforeclearly recognizable in Figure 8(B) (darker areas). In particular,the channel on the right experienced only minor variationsduring the whole study period, due to the confinement ofvegetation, especially flourishing in this area. Conversely,water bodies persisting for more than 1 year correspond to 3%of the wetted area at Cornino, as opposed to 13% at Flagogna(Figure 9). Short-lived water bodies, i.e. areas classified as weton at least one picture, but for a period shorter than 2months,sum up to more than 30% of the wetted area in both reaches.Since only small changes occurred in the vegetation distribu-

tion during the study period, in Figure 8 we report themaximum extension of vegetated patches. It is worth pointingout that typical time scales of vegetation evolution are gener-ally longer than 2 years, though we observed the erosion offew patches after the bankfull floods. Statistics on vegetationand channel mobility are summarized in Table I. In theisland-braided Flagogna reach vegetation occupies on average21% of the whole active corridor area (see Table I), whereastrees and shrubs cover only 4% of the active area in theCornino reach. On average, at both sites 87% of trees coverfalls in persistence class 0 (i.e. areas that were never occupiedby water at low flow), showing that riparian vegetation isgenerally at a higher elevation and is able to reduce bankerosion.

Figure 8. Water bodies persistence maps at Cornino (A) and Flagogna (B) re

Analysing in detail the composition of class 0 in the tworeaches we observe that vegetation covers 9% of this class atCornino, as opposed to 34% at Flagogna. Moreover, if weexclude vegetated areas from class 0, we find that a similarproportion (approximately 35%) of the two study sites is occu-pied by areas of bare gravel throughout the 2 year period.

Discussion

At-a-station relationships

Data on water area extension at increasing flow stage can beexpressed as at-a-station relationships between discharge (Q)and reach-averaged free surface width (W). The latter wascomputed as the ratio between wetted area and reach length,consistently with Ashmore and Sauks (2006). Figure 10 showsthe measured points, with the fitting curves in the form ofpower law W~Qn. A similar trend is found in both reaches,with the exponent n falling in the range 0.52–0.53. A widerange of values of n has been documented for braided rivers.For example Mosley (1983) and Smith et al. (1996) reportedlower values (n~0.3) falling in the typical range of singlethread channels (Ferguson, 1986). More recently, Ashmoreand Sauks (2006) showed that in the proglacial Sunwapta Riverthe width/discharge relationship can be described by a linearfunction (or by a power law with a coefficient nearly equal to1). Such a large exponent means that cross-sectional averagedflow velocity and depth do not increase with discharge as fastas in single thread channels, therefore affecting sedimenttransport processes. This behaviour has been also confirmedby the laboratory and numerical findings of Bertoldi et al.(2009b) who computed an exponent equal to 0.7.

aches. Vegetated areas within the active tract are outlined on the maps.

Table I. Statistics of vegetation abundance and turnover rate for thetwo study reaches

Corninoreach

Flagognareach

Total vegetated area / total reach area 4% 21%Area of persistence class 0 / total reach area 38% 54%Vegetated area in persistence class 0 / area ofpersistence class 0

9% 34%

Vegetated area in persistence class 0 / totalvegetated area

86% 88%

Bare gravel area in persistence class 0 / totalreach area

35% 36%

Figure 10. Wetted width (W)–discharge (Q) relationships for the twosites and power-law fitting curves.

The cause of these discrepancies in the at-a-station width/discharge relationship is not yet known. Different explanationshave been proposed that involve the effect of vegetation (Smithet al., 1996) and the flow regime (Ashmore and Sauks, 2006).Our data on the Tagliamento River give support to the latter,as no significant difference can be observed in the two reacheswith contrasting vegetation abundance. On the other hand, theTagliamento river experiences large discharge fluctuations,with floods of different magnitude being responsible formorphological changes, in contrast to the limited flow variabilityobserved in the Sunwapta river and in the flume experiments ofBertoldi et al. (2009b). An extensive field study by Booker(2010) on 326 gauging stations across New Zealand showed thatat-a-station width prediction can be improved by considering thecatchment area, therefore underpinning the relevance of the flowregime. More theoretical and empirical research is needed inorder to understand how fluvial pattern and flow regime affectthe at-a-station relationship.Different relationships have been proposed in the literature

also for the braiding index. For example, Mosley (1983)surveyed four braided rivers in New Zealand and found thatthe number of channels at a cross-section remains constantover a wide range of discharges, from low flow up to the meanannual flood. For the Tagliamento, we observed a unimodalrelationship between water level and BI that apparently doesnot match the almost linear increase proposed by van der Natet al. (2002) for the same reaches. However, all but one of thebraiding index values reported by van der Nat et al. (2002)correspond to water levels lower than 2m at Villuzza, which setsthe threshold between a predominantly terrestrial environmentand a fully connected surface aquatic system (Bertoldi et al.,2009a). Therefore, the lack of a falling limb in the plot by vander Nat et al. (2002) may be attributed to an under-representationof medium-to-high flow conditions.

The occurrence of a maximum in BI is consistent with theobvious observation that at very high flow its value must dropto 1 as the entire active corridor is likely to be submerged. Ananalogous behaviour was also documented by Egozi andAshmore (2008) both on the Sunwapta river and in flumeexperiments. The above observations, along with the resultsof 1D numerical modelling (Bertoldi et al., 2009b) and the datareported here show that the peak of the BI occurs at a levellower than the mean annual flood. This implies thatanabranches tend to merge at this stage and therefore most ofthe previously exposed bars are submerged. Significantly, ourdata show that the maximum of the braiding index correspondsto a flow level where the proportion of exposed area falls under50%.

Data presented here show that the relationship betweenbraiding index and discharge is complex, difficult to measure,highly variable over time due to network reworking and quitesensitive to the occurrence of vegetated patches.

The role of vegetation

The two study reaches are characterized by largely differentvegetation abundance (4% and 21% of the total area, atCornino and Flagogna, respectively), whereas discharge, longitu-dinal slope and grain size keep similar. Vegetation colonizationalong the Tagliamento is controlled by groundwater dynamicsand hence moisture availability (Bertoldi et al., 2011a). Annualtree growth rate can be two to three times faster in the Flagognareach, due to groundwater upwelling forced by the Pinzanonarrow gorge (Doering et al., 2007 and Tom Gonser, personalcommunication).

Recent analysis of the Tagliamento river morphologyhighlighted the effect of vegetation inmodifying the bed elevationdistribution (Bertoldi et al., 2011b). Vegetated islands present amuch higher bed elevation (due to fine sediment deposition)and are rarely inundated by water, thus reducing the flow areaat Flagogna reach. This is clearly visible in Figure 5(A), wherethe island-braided reach shows lower values of wetted areaproportion especially at medium to high stages. The occurrenceof stable islands determines also the confinement of the flow(particularly at low to medium stages) in smaller and deeperchannels. This effect is evident when considering the braidingindex (Figure 5(C)). The average number of branches at Flagognais roughly half of that at Cornino for a water stage rangingbetween 0 and 0.7m at Venzone. On the contrary, BI is onaverage larger at Flagogna at higher flow stage (1.5–2m) as theoccurrence of vegetated patches (that are not submerged at thisstage) are more likely to split the flow in a larger number ofbranches.

Furthermore, vegetation contributes to the higher proportionof upstream disconnected channels at Flagogna, particularly atflow levels ranging between 0 and 0.5m. The presence ofvegetation enhances the formation and persistence ofelongated ponds in large, deep scour holes located along thesides of established islands (Gurnell et al., 2005).

A strong link between vegetation distribution and channelpattern persistence is evident at both sites, with tree coverconcentrating in areas that were never occupied by low flowchannels during the study period. Turnover data show thatthe proportion of class 0 to the active tract area is much largerin the island-braided Flagogna reach, where it is 54%,compared with the Cornino reach where it is 38%. Therefore,vegetation colonization affects the bed morphology mainly byconfining and stabilizing the channels, whereas the bare gravelarea which is never part of a low flow channel is not reduced(approximately 35% in both reaches).

We note that these differences were not recognized in thestudy by van der Nat et al. (2002), who showed a largely similarbehaviour of both reaches in terms of wetted area and braidingindex. However, vegetation developed massively in theFlagogna reach in the last 10 years after two large floods in1996 and 2000 (Bertoldi et al., 2009a). Different sequences offloods (and in particular prolonged absence of intense floods)may cause fluctuations of vegetation diffusion, changing themorphological response of the river as well (Bertoldi et al.,2011a). The persistence maps obtained in the present studydemonstrate that the presence of vegetated patches contributesto the higher bed stability of the Flagogna reach, reduces theturnover of bars and islands and hence produces a morefavourable environment for vegetation growth. This sequenceof processes can establish a positive feedback (described as a‘fluvial biogeomorphic succession’, Corenblit et al., 2007)and lead to the transition from a fully braided system (like theCornino reach) to a more stable, vegetated fluvial form.

Conclusions

In this paper we present a detailed dataset on planform config-uration of two 1-km-long reaches of the gravel-bed braidedTagliamento River (northeast Italy), acquired at different waterlevels through ground-based imagery. The high temporalresolution of the survey (1 h, during daylight) allows the inves-tigation of the planform dynamics during single floods. The tworeaches are characterized by different vegetation colonizationand can be described as a bar-braided and an island-braidedreach. This difference offers the possibility to investigate therole of vegetation on the inundation patterns (in particular onthe wetted width and the braiding index) and on the turnoverrate of the low flow anabranches.We investigated inundation dynamics over a wide range of

flow conditions, with water occupying from 10% up to 90%of the active corridor. Our data show that the increase of waterlevel determines large variations in the planform configuration,transforming a mainly terrestrial system into a predominantlyaquatic environment. Wetted area proportion increases almostlinearly with water level, with small differences in the tworeaches. In terms of width/discharge at-a-station relationship,we found exponents slightly larger than 0.5 in both reaches.This analysis therefore confirms that braided rivers are morelikely to show larger exponents compared with single-threadchannels, where values generally range between 0.3 and0.35. A much faster increase of the width with flow stage hasa direct effect on the variability of depth and velocity, hence af-fecting the sediment transport rate.The number of branches varies considerably with the water

level, increasing from two up to six or seven at Cornino andfour to five at Flagogna and dropping again to one or two forvery high values of water level. At low flow conditions discon-nected water bodies, i.e. ponds, groundwater-fed branches orbackwaters, represent a significant proportion of the wettedarea, thus considerably increasing the morphological diversityof the anabranches in terms of water velocity, depth andtemperature.The effect of morphological changes induced by floods on

the above at-a-station relationships was evaluated using twosets of images acquired at the same water level. The analysishighlights that the braiding index is subject to large fluctuationsthat are comparable with the at-a-station variability. The num-ber of branches is strongly affected even by moderate floods,making this parameter quite difficult to monitor. In contrast,the wetted area proportion is more stable, although large floodscan modify it by up to 10%.

The persistence maps highlight the extraordinary dynamicityof this fluvial system. We observed a rapid and widespreadturnover of wetted areas, with almost none of the low flowwater bodies in the same position after 12months. In thebar-braided reach 62% of the active corridor was occupiedby low flow channels in the 2 year period. The colonizationof riparian vegetation plays amajor role in stabilizing the channelpattern. The island-braided reach shows a reduced lateralmovement of the anabranches and a lower network complexity,in terms of braiding index and shoreline length.

Acknowledgements—The authors gratefully acknowledge funding fromCARIPARO, MIUR and the Leverhulme Trust, which supported theresearch reported in this paper. We also thank Lorenzo Forti, MartinoSalvaro and Luca Zanoni (University of Trento) for their help withthe fieldwork. The paper has benefitted greatly from comments andsuggestions by two anonymous referees and the Special Issue Editorwho helped us to clarify parts of the paper.

ReferencesAntonarakis AS, Richards KS, Brasington J. 2008. Objects-based landcover classification using airborne lidar. Remote Sensing of Environ-ment 112(6): 2988–2998.

Arscott DB, Tockner K, van der Nat D, Ward JV. 2002. Aquatic habitatdynamics along a braided alpine river ecosystem (Tagliamento River,Northeast Italy). Ecosystems 5: 802–814.

Ashmore PE. 2001. Braiding phenomena: statics and kinetics. InGravel-bed Rivers V, Mosley MP (ed). New Zealand HydrologicalSociety: Wellington; 95–121.

Ashmore PE, Sauks E. 2006. Prediction of discharge from water surfacewidth in a braided river with implications for at-a-station hydraulicgeometry. Water Resources Research 42(W03406): 1–11.

Bertoldi W, Gurnell AM, Surian N, Tockner K, Zanoni L, Ziliani L,Zolezzi G. 2009a. Understanding reference processes: linkagesbetween river flows, sediment dynamics and vegetated landformsalong the Tagliamento River, Italy. River Research and Applications25: 501–516.

Bertoldi W, Zanoni L, Tubino M. 2009b. Planform dynamics of braidedstreams. Earth Surface Processes and Landforms 34(4): 547–557.

Bertoldi W, Zanoni L, Tubino M. 2010. Assessment of morphologicalchanges induced by flow and flood pulses in a gravel bed braidedriver: the Tagliamento River (Italy). Geomorphology 114: 348–360.

Bertoldi W, Drake NA, Gurnell AM. 2011a. Interactions between riverflows and colonising vegetation on a braided river: exploring spatialand temporal dynamics in riparian vegetation cover using satellite data.Earth Surface Processes and Landforms doi: 10.1002/esp.2166.

BertoldiW,Gurnell AM,DrakeNA. 2011b. The topographic signature of veg-etation development along a braided river: results of a combined analysisof airborne lidar, colour air photographs and ground measurements.Water Resources Research 47: W06525 doi:10.1029/2010WR010319.

Booker D. 2010. Predicting wetted width in any river at any discharge.Earth Surface Processes and Landforms 35: 828–841.

Braudrick CA, Dietrich WE, Leverich GT, Sklar LS. 2009. Experimentalevidence for the conditions necessary to sustain meandering incoarse-bedded river. Proceedings of the National Academy ofSciences USA 106: 16936–16941.

Chandler J, Ashmore P, Paola C, GoochM, Varkaris F. 2002. Monitoringriver channel change using terrestrial oblique digital imagery andautomated digital photogrammetry. Annals of the Association ofAmerican Geographers 92(4): 631–644.

Chew LC, Ashmore PE. 2001. Channel adjustment and a test of rationalregime theory in a proglacial braided stream. Geomorphology 37:43–63.

Corenblit D, Tabacchi E, Steiger J, Gurnell AM. 2007. Reciprocalinteractions and adjustments between fluvial landforms and vegetationdynamics in river corridors: A review of complementary approaches.Earth-Science Reviews 84(1–2): 56–86.

Crosato A, Saleh MS. 2011. Numerical study on the effects of floodplainvegetation on river planform style. Earth Surface Processes andLandforms 36(6): 711–720.

Doering M, Uehlinger U, Rotach A, Schlaepfer DR, Tockner K. 2007.Ecosystem expansion and contraction dynamics along a large Alpinealluvial corridor (Tagliamento River, Northeast Italy). Earth SurfaceProcesses and Landforms 32(11): 1693–1704.

Eaton BC, Millar RG, Davidson S. 2010. Channel patterns: Braided,anabranching, and single-thread. Geomorphology 120: 353–364.

Egozi R, Ashmore PE. 2008. Defining and measuring braiding intensity.Earth Surface Processes and Landforms 33: 2121–2138.

Ferguson RI. 1986. Hydraulics and hydraulic geometry. Progress inPhysical Geography 10: 1–31.

Feurer D, Bailly JS, Puech C, Le Coarer Y, Viau A. 2008. Very highresolution mapping of river immersed topography by remote sensing.Progress in Physical Geography 32(4): 1–17.

Gray D, Scarsbrook MR, Harding JS. 2006. Spatial biodiversity patternsin a large New Zealand braided river.New Zealand Journal of Marineand Freshwater Research 40(4): 631–642.

Gurnell AM, Petts GE. 2006. Trees as riparian engineers: the TagliamentoRiver, Italy. Earth Surface Processes and Landforms 31: 1558–1574.

Gurnell AM, Petts GE, Hannah DM, Smith BPG, Edwards PJ, KollmannJ, Ward JV, Tockner K. 2001. Riparian vegetation and islandformation along the gravel-bed Fiume Tagliamento, Italy. EarthSurface Processes and Landforms 26: 31–62.

Gurnell AM, Surian N, Zanoni L. 2009. Multi-thread river channels: aperspective on changing European alpine river systems. AquaticSciences 71: 253–265.

Gurnell AM, Tockner K, Edwards PJ, Petts GE. 2005. Effects of depositedwood on biocomplexity of river corridors. Frontiers in Ecology andthe Environment 3: 377–382.

Haschenburger JK, Cowie M. 2009. Floodplain stages in the braidedNgaruroro River, New Zealand. Geomorphology 103(3): 466–475.

Hoey T, Sutherland AJ. 1991. Channel morphology and bedload pulsesin braided rivers: a laboratory study. Earth Surface Processes andLandforms 16: 447–462.

Howard AD, Keetch ME, Vincent CL. 1970. Topological and geometricalproperties of braided streams.Water Resources Research 6: 1674–1688.

Karrenberg S, Kollmann J, Edwards PJ, Gurnell AM, Petts GE. 2003.Patterns in woody vegetation along the active zone of a near-naturalAlpine river. Basic and Applied Ecology 4: 157–166.

Langhans DS, Tockner K. 2006. The role of timing, duration, andfrequency of inundation in controlling leaf litter decomposition in ariver–floodplain ecosystem (Tagliamento, northeastern Italy).Oecologia 147: 501–509.

Latterell JJ, Bechtold JS, O’keefe TC, Van Pelt R, Naiman RJ. 2006.Dynamic patchmosaics and channel movement in an unconfined rivervalley of the Olympic Mountains. Freshwater Biology 51: 523–544.

Legleiter CJ, Roberts DA, Lawrence RL. 2009. Spectrally-based remotesensing of river bathymetry. Earth Surface Processes and Landforms34(8): 1039–1059.

LorangMS,WhitedDC, Hauer FR, Kimball JS, Stanford JA. 2005. Using air-borne multispectral imagery to evaluate geomorphic work across flood-plains of gravel-bed rivers. Ecological Applications 15: 1209–1222.

Luchi R, Bertoldi W, Zolezzi G, Tubino M. 2007. Monitoring andpredicting channel change in a free-evolving, small alpine river:Ridanna Creek (North East Italy). Earth Surface Processes andLandforms 32(14): 2104–2119.

Malard F, Uehlinger U, Zah R, Tockner, K. 2006. Flood-pulse andriverscape dynamics in a braided glacial river. Ecology 87: 704–716.

Mao L, Surian N. 2010. Observations on sediment mobility in a largegravel-bed river. Geomorphology 114(3): 326–337.

Marcus WA, Fonstad MA. 2010. Remote sensing of rivers: theemergence of a subdiscipline in the river science. Earth SurfaceProcesses and Landforms 35(15): 1867–1872.

Mosley M. 1982. Analysis of the effect of changing discharge on chan-nel morphology and instream uses in a braided river, Ohau River,New Zealand. Water Resources Research 18(4): 800–812.

Mosley M. 1983. Response of braided rivers to changing discharge.Journal of Hydrology New Zealand 22: 18–67.

Robertson-Rintoul MSE, Richards KS. 1993. Braided-channel patternand paleohydrology using an index of total sinuosity. In BraidedRivers, Best JL, Bristow CR (eds). Geological Society Special Publica-tion vol. 75, Geological Society: London; 113–118.

Smith LC, Isacks B, Bloom A. 1996. Estimation of discharge from threebraided rivers using synthetic aperture radar satellite imagery:Potential application to ungaged basins. Water Resources Research32: 2021–2034.

Steiger J, Tabacchi E, Dufour S, Corenblit D, Peiry JL. 2005. Hydrogeo-morphic processes affecting riparian habitat within alluvial channel-floodplain river systems: A review for the temperate zone. RiverResearch and Applications 21(7): 719–737.

Sukhodolov A, Bertoldi W, Wolter C, Surian N, Tubino M. 2009.Implications of channel processes for juvenile fish habitats in alpinerivers. Aquatic Sciences 71(3): 338–349.

Surian N, Mao L, Giacomin M, Ziliani L. 2009. Morphological effects ofdifferent channel-forming discharges in a gravel-bed river. EarthSurface Processes and Landforms 34(8): 1093–1107.

Tal M, Paola C. 2007. Dynamic single-thread channels maintained bythe interaction of flow and vegetation. Geology 35(4): 347–350.

Tal M, Paola C. 2010. Effect of vegetation on channel morphody-namics: results and insights from laboratory experiments. EarthSurface Processes and Landforms 35(9): 1014–1028.

Tockner K, Malard F, Ward JV. 2000. An extension of the flood pulseconcept. Hydrological Processes 14: 2861–2883.

Tockner K, Ward J, Arscott D, Edwards P, Kollmann J, Gurnell A, PettsGE, Maiolini B. 2003. The Tagliamento River: a model ecosystemof European importance. Aquatic Sciences 65: 239–253.

Tonolla D, Acuna V, Uehlinger U, Frank T, Tockner K. 2010. Thermalheterogeneity in river floodplains. Ecosystems 13(5): 727–740.

van der Nat D, Schmidt A, Tockner K, Edwards P, Ward J. 2002.Inundation dynamics in braided floodplains: Tagliamento River,Northeast Italy. Ecosystems 5: 636–647.

van der Nat D, Tockner K, Edwards PJ, Ward JV, Gurnell AM. 2003.Habitat change in braided rivers (Tagliamento, NE-Italy). FreshwaterBiology 48: 1799–1812.

Ward JV, Tockner K, Edwards PJ, Kollmann J, Bretschko G, Gurnell AM,Petts GE, Rossaro B. 1999. A reference river system for the Alps: the’Fiume Tagliamento’. Regulated Rivers: Research & Management 15(1–3): 63–75.

Winterbottom SJ, Gilvear DJ. 1997. Quantification of channel bed mor-phology in gravel-bed rivers using airborne multispectral imageryand aerial photography. Regulated Rivers: Research & Management1: 489–499.

Zanoni L, Gurnell AM, Drake N, Surian N. 2008. Island dynamics in abraided river from analysis of historical maps and air photographs.River Research and Applications 24(8): 1141–1159.

the response of braided planform configuration to flow variations, bed reworking and vegetation: the...

Documents

experimental evaluation of braided ekf for sensorless...

multiscale simulation of fracture of braided composites via...

effect of wing planform and canard location

reworking of the tarim craton by underplating of mantle...

regional variability of aquatic pattern in braided reaches...

braided textiles to improve the competitiveness of

the braided weave of mnemosyne: aby warburg, carl gustav...

braided reinforced composite rods for the internal...

inquiry into js bach's method of reworking in his...

endocardial pacing, cardioversion and defibrillation using a...

whole shaft visibility and mechanical performance for active...

development of hybrid braided composite rods for...

braid representations from unitary braided vector spaces

towards the hydrologic and bed load monitoring from...

culture against race: reworking the basis for racial...

immediate profile and planform evolution of a beach...

application of the equilibrium planform concept to natural...

simulation of braided river elevation model time series with...

“in scirpo nodum: symphosius’ reworking of the riddle...

monitoring and modelling morphological change in a braided...