Professional Documents
Culture Documents
3 Alpine watershed
6 *Corresponding author
8 1. Introduction
9 Rivers have been modified by human activities for centuries, with a remarkable
10 acceleration since the industrial revolution. One of the most frequent issues in river
13 a large number of variables and processes to consider, and also interactions and
14 feedbacks between processes. In fact, during the last fifty years, fluvial
16 the behavior of rivers. One of the most important drivers that motivated these
18 human induced disturbances (Wohl, 2014). There are many possible types of
23 relevant and impactful causes of fluvial system alteration (Grant, 2012). The
24 changes produced by dams and diversions on fluvial processes – mainly alteration
25 of both sediment and water fluxes – can be dramatic (Petts & Gurnell, 2013).
26 Despite the large amount of technical and scientific literature about the alterations
27 of human activities, in particular large dams, on river morphology (e.g., Graf, 2006;
28 Schmidt and Wilcock, 2008; Baker et al. 2011; Csiki and Rhoads, 2010; Draut et
29 al., 2011), this field of investigation is still of significant interest for both engineers
32 river reach. The response to multiple drivers and stressors in a watershed is still a
33 considerable challenge. In fact, there are only limited studies about the drivers of
37 classification of these tools was presented by Grant (2012). Based upon Grant’s
40 This class includes single case studies, empirical analysis and synthesis of
43 More detailed model outputs in turn require more detailed inputs for
44 construction and validation. In general these models are applied only for
45 single reaches and the input data are seldom fully obtained. Numerical
48 between the perturbation and the response of the system. However, while
50 of this approach is the need for highly accurate data inputs that are not
51 generally available at the watershed scale, particularly, when the spatial and
52 temporal scales are larges (watershed and couple of decades) and if more
56 weak predictive value. Conceptual models assume that the fluvial system is
60 in the responses.
61 For this paper we propose a predictive tool for geomorphological trajectories of the
62 gravel-bed rivers based on (i) conceptual models in which relations were derived
63 from the observation and understanding of study cases, using the experiences
64 from systemic analyses; (ii) that take into account the effect of disturbances of an
65 initial state of the river system and; (iii) can be applicable and developed at the
66 watershed scale with assessment of the connectivity from the upstream part of the
73
75 The following sections present a relevant background that guides the analysis and
77 disturbances. Because of the large spatial variability of processes and forms at the
78 scale of a complex watershed, and due to the paucity of input data, a conceptual-
79 analytical model approach was chosen to predict the river morphology response
80 using the Isère Alpine watershed as a test case and the complex hydropower
85 and sediment supply conditions (second order drivers are for example large woody
86 debris and animals); (ii) the ability of the system to adapt, in case of alteration on
87 hydrological flow regime and sediment supply; and (iii) the propagation of these
90
91 2.1. Fluvial morphological responses: driver variables and responses variables
94 sinuosity, sediment grain size and distribution) are the result of erosion, transport
98 responses (adjustment or fit) variables, which respond and adjust to the driver
100 corresponding to the internal characteristics of the river reach, such as riparian
101 vegetation and bank confinement, which can limit the lateral displacement of the
102 channel.
103
104 Driver variables for a river reach are considered from upstream watershed input
105 conditions: these are spatially distributed variables such as the climate and
106 geology, which determine the topography, soil, and the natural type of vegetation
107 cover. Another relevant control is human activities (e.g. land use, management
108 practices, hydraulics works). These variables define the first order of local drivers,
109 namely the sediment and water fluxes (Werritty, 1997; Church, 2002; Grant et al.,
110 2003; Buffington, 2012). However, some authors also give a relevant importance
111 to large wood and in-channel vegetation (e.g. Buffington et al., 2003 and 2012;
112 Piégay and Gurnell, 1997; and in this ESPL special volume Takebayashi, 2016
113 and; Bertoldi and Ruiz-Villanueva 2016), which are considered as second order
114 drivers if compared with sediments and hydrological flow regimes at the spatial
115 scale of this investigation. It is for this reason that hydrological flow regimens and
116 sediment supply are considered as the more important drivers variables in this
117 study.
120 These responses depend on the magnitude, intensity and duration of the
121 disturbance, river typology (i.e. their sensitivity to changes according Schumm,
122 1991), and on the geomorphic activity of its tributaries. A first type of response
124 (adaptation without changes) as a part of its inherent variability. A second type of
128 must first surpass a morphological threshold (see also Schumm, 1977; Champpell,
129 1983; Werritty, 1997; and Huggett, 2012). In this work, nine response variables
130 were selected to describe the reaction of the fluvial morphology due to a
131 perturbation on the driver variables, which are presented later (in section 3.2 and
133 Boundary conditions (internal conditions of the reach) express the degree of
134 freedom of mobility. For the development of this model, the local singularities and
135 some limitations of the variability of adjustment variables were taken into account.
136 The lateral mobility of the riverbed can be restricted if the riverbed it is confined or
137 semi-confined, for example if there are physical barriers such as levees or by
138 biological characteristic of the riverbank, in particular the role of the riparian
139 vegetation (e.g. Tal and Paola, 2007). Horizontal mobility is determined by the
141 transported and depth of this layer to bedrock (rock outcrops fix the bed level).
142
144 Riverbeds are constantly adjusting and adapting in response to the sequences of
145 flood events associated with the regional climate and local meteorological
146 conditions.
147 In modeling and predicting channel changes, the concept of equilibrium condition
148 is often used (Buffington, 2012; Church and Ferguson, 2015). In this work, we
149 consider a river reach to be in equilibrium when there are no disturbances on driver
150 variables. This implies that the sediment supplies (from upstream reaches and
151 from hillslopes and banks within the same reach) are substantially equivalent to
152 the transport capacity over a time period of several years. This assumption allows
153 for geometric (bedforms, depth, width and local slope) and grain size changes at
154 a shorter time scale, but considers these variations negligible over the long-term.
155 The initial condition of the river morphology in this work corresponds to the current
157 A fluvial system reacts to a change of water and sediment regimes at a wide range
158 of spatial and temporal scales. According to Graf (1977) and Brandt (2000b),
161 evolution of bed level in 12 rivers after dam construction was fast immediately after
163 Buffington (2012), based on the Knigton’s work (1984) presented responses
164 variables in alluvial rivers as function of the spatial and temporal scales ranging in
165 orders of magnitude. These changes can last from minutes (for local changes in
166 grain size and textural patches) to many decades or centuries (for stream gradient
167 and channel sinuosity). Taking into account these physical processes in rivers, the
168 order of the temporal scales considered is from many years to some decades.
169 These conceptual models, including the model presented in this work, do not
170 consider the temporal evolution, which is only possible to take in account in
172 Gregory (2006) and Gregory and Downs (2008) used the Schumm (1979) and Graf
173 (1977) approaches to describe the kinetics of river response after a disturbance. If
175 a transient period will start. The transient period is called the response time,
176 composed of a reaction time and relaxation time. During the transient period the
177 system evolves to achieve a new equilibrium state, marking the end of this
178 transient. According to many previous studies (Phillips, 1996; Grant et al., 2003;
179 Petts and Gurnell, 2005; Grant, 2012, Ziliani and Surian, 2012), for an
181 as a trajectory (from an initial equilibrium state to a new equilibrium state). Each
182 response variable can have a different response time and the response time of the
183 whole river system can last from decades to centuries or millennia.
184
185 2.3. From upstream to downstream: river reaches segmentation and connectivity
187 The river network of a watershed is naturally or artificially segmented into different
189 characteristics and the type of disturbance and its localization (e.g. changes in
191 reaches are based mainly on bed slope, lateral mobility (river bank conditions:
192 confined and whether), vertical mobility (a fixed bed level) and river confluences.
193 Modifications on the continuity of water and sediment fluxes are two important
194 disturbances on the more relevant fluvial geomorphic drivers (water and sediment).
195 The continuity of these fluxes can be affected if the magnitudes, frequencies and
197 For the tested case (application of the model to the Isère river watershed) the
198 sources of disturbance taken into account were the hydropower structures. The
199 structures considered to partition a reach in the hydrographic network are, intake
200 water points, release water points, and large dams, considering that these kinds of
201 structures can affects the processes of sediments transport and hydrological flow
202 regimes. Taking in account the timescale of the river response, structures like
203 barrage dams (diversion dam which large gates) capable to generate flushing;
205 for the continuity of the sediment transport). The river confluences for the more
206 important streams in terms of sediments yield and water discharge are also
208 From upstream to downstream in the river network, hydrological and sedimentary
209 disturbances are transmitted and attenuated by the compensation of the water and
210 sediment supplies from tributary streams. In the case of the sediments surplus or
211 deficit, the effects can also be balanced by sediments delivered from the bed and
212 banks or sedimented in the bed and banks, translated as a channel aggradation
214
216 This section describes how changes on the main drivers are computed. A proposal
217 is made relatively to initial equilibrium conditions in the present state of the model.
218 Proposed formulations are simplified and can be discussed, improved and adapted
219 to other cases and particular situations. Their use in the framework of the
221
226 discharges. Large and infrequent floods can completely change the morphology of
227 a system (Tamminga et al., 2015), but low and constant discharge can also
228 contribute to evolve the river morphology (Asahi et al., 2013). From a
229 morphological point of view, two characteristic discharges are the most important,
230 namely i) the critical discharge associated to the incipient sediment motion (e.g.
231 Recking, 2009 and 2013) and ii) the channel-forming discharge. The critical
232 discharge depends on the size of bed sediments (e.g. d50 or d84) and the local
234 The channel-forming or dominant discharge (e.g. Wolman and Miller, 1960;
235 Leopold et al., 1964; Andrews, 1980) can be considered the discharge that (as in
236 steady state) produces the same morphological result as the combined effects of
237 the entire hydrological flow regime. Even if difficult to assess, channel-forming
238 discharge is often used for river restoration and natural channel design (Doyle et
239 al., 2007; Lave 2009), and it is usually considered as: a) being equal to the bankfull
240 discharge (Navratil et al., 2006), b) the effective discharge (calculated according
241 to the Wolman and Miller, 1960 procedure), and c) the discharge associated with
242 a certain return period (see also, FISRWG, 1998), which in most cases roughly
244 In this paper describing the first version of the model, the channel-forming
245 discharge is identified as the discharge with a return period of two years. For each
246 reach, flow duration curves are used to determine the change in the number of the
247 days that the channel-forming discharge was equal or was exceeded during the
248 period before and after the perturbation on the hydrological flow.
249 The indicator used to subsequently assess the hydrological alteration downstream
250 in each river reach was calculated as (Alcayaga et al., 2012; Alcayaga, 2013):
NQ post
251 FQ 1 (1)
NQ pre
252 where FQ is the Indicator of alteration for hydrological regime based in the
253 frequency change of the channel-forming discharge, and NQpre and NQpost:
256 Evidently, this choice is a simple approach and other approaches can be adapted
257 to compute FQ, according on data availability and the specific characteristics of
258 the site. In every case, FQ is a proxy to quantify the alteration of the “energy”
259 available for morphological transformations of the channel. FQ=-1 means that in
260 the final state, the transport capacity is null; FQ=0 in the case of unchanged
261 conditions, and FQ positive means an increase of the energy and transport
262 capacity.
263
265 Although bedload can represent a relatively small percentage of the total sediment
266 load (Collins and Dunne, 1990; Turowski et al., 2010), bedload determines the
267 morphology and habitat of gravel-bed rivers (Kondolf, 1997). For this reason only
269 Particularly, volumes of sediment available on internal sources (banks and channel
270 bed), external sources (hillslopes), and the degree of connectivity between
271 hillslopes and streams are considered. To evaluate the intensity of sediment
272 supplied to each reach we applied a simplified method, which only takes in account
273 the external supply from both the hillslope and from the upstream reach, as a
276 in-channel processes relative to sediment supply (internal sources) were not taken
277 into account, due to the complexity of the temporal and spatial scales.
278 Following the same logic to assess the alteration on the hydrological flow regime,
279 the indicator we used was (Alcayaga et al., 2012; Alcayaga 2013):
SS post
280 AS 1 (2)
SS pre
281 where AS is an indicator of alteration for sediments supply (bed load sediment),
282 based on the changes of the intensity of sediment supply from upstream reaches,
283 and lateral sediments supply in the reach under analysis, and SSpre and SSpost are
284 potential intensity of the sediments supply estimated with a simplified method,
285 before and after a disturbance in the continuity of the supply, respectively.
286 AS=-1 means that in the final stage, there is no sediment supply from the upstream
287 reach; AS=0 in the case of unchanged conditions, and AS=+1 means a doubling
289 In the present state of our model, the intensity of sediment supply SS of a sub-
290 basin that drains to a river reach is the combination of hillslope capacity and of
291 outlet properties for transferring it downstream. Hillslope capacity of the sub-basin
293 variables: lithology (surface geology), land cover and relief (hillslope gradients). All
295 classified according to their capacity to deliver sediments (see Tables 2, 3 and 4).
296 More details are given in section 4 of this paper. The result is a new GIS layer
297 representing an overlapping of the previously classified layers (Table 5). This layer
298 represents the potential supply of sediments from the hillslopes and is called SS.
299 This combination and classification assumes that all three factors (geology, terrain
300 gradient and land cover) have an equivalent impact on the sediment supply. SS
301 hillslope is computed from pre and post values as the sum of individual pixel
302 indexes within the sub-basin, considering the path of the sediment in the watershed
304 sediment path such as a lake, a reservoir, check dams, or a large dam at the outlet
305 of the sub-basin, there is no continuity of bedload downstream (Brune, 1953; Lewis
306 et al., 2013; Kondolf et al., 2014) and the above-computed hillslope capacity is
308
310 The values of FQ and AS for each river reach form a vector named the disturbance
311 vector. The direction of these vectors is associated with a trend or trajectory of
312 channel changes and its magnitude represents the intensity of the perturbation.
314 The indicators FQ and AS are represented as two principal axes forming a
315 Cartesian coordinate system (the explanations hereafter refer to Figure 1), defining
316 four directions associated with the angles (0°, 90°, 180° and 270°). The FQ axis
318 river reach “energy”. In the same way the AS axe (positive or negative) is a core
319 of the increment or decrease in sediment supply from upstream. Two secondary
320 axes are also relevant and they define others four directions (45°, 135°, 225° and
321 315°). For the direction 45°, the fluvial processes increased in intensity (more
323 For the direction 225°, the processes decreased in intensity (less sediment supply
324 and a reduction of the channel forming discharge frequency). These two directions
325 (45° and 225°) bisect the Cartesian plane, with zones associated with two basic
326 morphological responses of the channel: sedimentation and erosion. The first zone
329 morphologic work). The second zone –erosion- occurs just the opposite, when AS
330 decreased and FQ increased. In the border of these two directions (45° and 225°)
331 the morphological response of the channel depends on other more complex and
332 specific factors, such as the geological channel history (Grant et al., 2003).
333 All these eight directions (defined by the four axes) correspond to morphological
334 trends or trajectories of the expected changes in the channel (see Table 1) based
335 on an extensive review of the literature, empirical relations, case studies, and
336 conceptual models, which have been developed during the last sixty years. The
337 main references we used to associate the eight directions with to build this model
338 are: Lane (1955); Schumm (1969; 1977); Petts (1980); Williams and Wolman
339 (1984); Kellerhals and Church (1989); Brandt (2000a); Brandt (2000b); Huang and
340 Nanson (2002); Grant et al., (2003); Grant (2012); Petts and Gurnell (2005);
341 Schmidt and Wilcock (2008); and Dust and Wohl (2012). Modified from Alcayaga
342 et al. (2012) and Alcayaga (2013), Table 1 and Figure 1 show the nine
343 morphological variables (bed elevation, bed slope, width, depth, wetted area, d50,
344 terrace formation, and colonization of vegetation) their responses, associate with
346
348 The disturbance vector for each river reach has an origin from an initial non-
349 perturbed condition (the origin of the Cartesian coordinate system) and an end
350 point given by each FQ and AS; this distance represents the magnitude of the
351 disturbance vector. The magnitude corresponds to the intensity of the disturbance
352 classified in three classes as low (< 0.15), medium (0.15 – 0.30) and strong (>
353 0.30).
354 Additionally, there is continuity in the trends for each response variable, as
355 depicted in the graph of Figure 1. For example, the bed slope (S) increases in the
356 graph from 90 until 180; afterwards, there is a zone in which an inflexion in the
357 trend occurs at 225; then S decreases from 270 until 0 and the next inflexion
358 zone occurs at 45. If the direction of the disturbance vector does not correspond
359 exactly to one of these eight principal directions (the 8 axes in the Figure 1), the
360 final response is determined using a weighted function of the distance between the
361 two closed axes with respect to the calculated disturbance vector. Regarding the
364
365 Figure 1. Graphical representation of the disturbance vector (shown in red) and
368 The methodological approach proposed here could be applied to a wide range of
369 fluvial systems which suffer alterations in hydrology and sediment supply due to
371
373 The application of our model has been exercise in order to assess it in a complex
374 watershed with numerous and important perturbation sources. The model was
375 tested on the Isère watershed upstream of the city of Grenoble (5700 km2). The
376 Isère mountain watershed is a sub-basin of the Rhône River (upper part of the
377 Rhone basin), located in southeastern France (Figure 2). Its main tributary is the
378 Arc River, and the second one is the Arly River. This watershed is a strongly
380 sources of perturbations, the hydrological flow regime and the sediment supply are
381 altered by several hydropower systems (Vivian, 1994; Marnezy, 1999). According
382 to Peiry et al. (1999), the hydropower systems of the Isère watershed (Figure 3)
383 are among the oldest, densest, and likely most sophisticated in the word. In the
384 watershed there are three different types of hydropower plants: impoundment
385 (large dams), diversions (heavy transbasin diversions with barrages) and pumped-
386 storage. The applied model considers the effect of the hydropower systems built
387 during the 1950 and 1970, but with the following limitations:
388 - The whole system is considered to be built at the same period, from an initial
392 Other perturbations sources were not taken in account. Older ones include climatic
393 perturbation (Little Ice Age), the stabilization of mountain hillslopes (forestation,
394 engineering torrent control) (Provansal et al, 2014) and levees in the downstream
395 reaches. Recent perturbations are mainly intensive gravel mining from the end of
396 the Second Word War; these were completely stopped at the beginning of 1980.
397
398 Figure 2. Location of the Isère watershed in the geographical French context.
399
400 According to the criteria mentioned in section 2.3, river network segmentation was
401 built using its morphological characteristics (i.e. bed slope, confluences) and the
402 hydropower structure locations (i.e. water intakes, releases and large dams). In
403 total, 23 reaches (see figure 3) were identified. Codes were assigned to each reach
404 starting with the main stream from the headwater reach of Isère (code 100) to the
405 lower downstream reach (900); then the main affluent of the Isère (Arc River)
407
408 Figure 3. Location of the Isère watershed, sub-basin delineation, river reaches
409 with their codes and main hydropower canalizations and dams.
410
411 Overall, 16 discharge stations were used to calculate the FQ indicator. These
412 discharge time series were gathered from the Banque hydro data base
413 (www.hydro.eaufrance.fr). When the discharge stations were not located at the
414 beginning of the river reach, it was necessary to weight discharges using the
415 surface drainage area to estimate the discharge values associated with a return
417
418 The determination of AS was more complex, as it involved the use of information
419 from different sources, and a reclassification of this information according the
420 methodology explained above. For the Isère watershed, this classification,
421 organized by geology, relief and land cover, is described below (from Table 2 to
422 Table 4). The geological surface formations, and their mechanical characteristics,
426
427 The topographic gradient of hillslopes is taken as the local gradient of the terrain.
428 The gradient is calculated using a Digital Elevation Model (DEM). The results (in
429 degrees) are then reclassified according to the groups shown in Table 3.
430
431 The land cover data used comes from the European CORINE project (2000). The
432 land cover for the Isère watershed was grouped and then classified into three
433 distinct classes according their potential for delivery of sediment (Table 4).
434
435 The combination of geology, gradient and land cover (codes from the Tables 2, 3
436 and 4) resulted in 15 class types for the potential intensity of sediment supply.
437 Then, for each sub-basin that corresponded to the drainage area of each river
438 reach, a single unique intensity value was obtained. Finally, these unique values
439 for reach sub-basin are added (as a continuum) from upstream to downstream (as
441
443 FQ and AS were calculated for each reach of the river network. As an example,
444 we present the calculations FQ and AS for reach 300, which is influenced by the
445 hydropower system in the upper basin (mainly by Tignes dam and Malgovert power
447 For reach 300, the channel-forming discharge (2-yr return period) is 102 (m3s-1),
448 and the flow duration curves for the conditions pre and post hydropower are shown
449 in Figure 4. For this reach, the flow regime is regulated by dams in the upper part
450 of the watershed, and by diversions (transbasin diversion). The effects of these
452 of frequency. Then, the frequency of the channel-forming discharge was reduced
454 Figure 4. Flow duration curves for the Isère river at Moûtiers downstream of the
455 Tignes large dam and Montrigon regulating reservoir. The curves correspond to
456 both periods before and after the initial operation of these dams during 1954.
457 The calculation of AS for the reach 300 implies consideration of the contributions
458 from upstream of the sub-basins associated with reaches 200 and 100. The first
459 calculation estimated the hillslope capacity of each pixel and sub-basin (SS)
460 through the overlapping of geology (table 2), terrain gradient (table 3) and land
461 cover (table 4), resulting in 45 possible values (see column 1 in Table 5). The result
462 of this combination is reclassified in 9 classes (see column 2 in Table 5), assigning
463 values from 1 (lower productivity) to 9 (higher productivity). The hillslope capacity
465
466 Figure 5. Hillslope capacity reclassified for the sub-basins 100, 200 and 300
468 The second calculation consisted in obtaining a single value of sediment supply
469 (SS) from the hillslopes of each sub-basin. The SS value for each sub-basin is
471 pixels with this respective value) and normalized by the total number of pixels in
472 each sub-basin. The SS values represent the potential of sediment delivery from
474 The SS values for each sub-basin relates to reaches 100, 200 and 300 are 5.4,
475 5.7 and 4.7, respectively. To obtain the SSpre and SSpost required adding the SS
478 described in section 3.1.b. Figure 6 shows the AS calculation for the reach 300,
479 where SSpre is the sum of the SS contribution from the sub-basins 100, 200 and
480 300 (SSpre=15.8). SSpost was calculated using the same sum of SS values
481 considering the blockage effect in the sediments fluxes due to Tignes dam
482 (SSpost=10.4), and subsequently AS was calculated for reach 300 result -0.34.
484 fluxes, because it has Tainter gates located in the base of the dam.
485
486 Figure 6. SS values for the sub-basins, SSpre and SSpost values for the reaches
488
489 For the same reach (code 300), FQ and AS resulted to be -0.31 and -0.34,
490 respectively. Using these two values, the magnitude of the disturbance vector was
491 calculated as 0.46, with a direction of 227.4. This vector is located in an area
492 where the morphological processes decrease in intensity as both FQ and AS are
493 reduced proportionally, in accordance with Figure 3. In a case such as this when
494 the disturbance vector is near to 45 or 225, the trajectory of the changes are
495 difficult to predict, and according Grant et al. (2003) the response of the channel
496 depends on specific factors. Thus, for this reach, only the response variables
499 The results for all the reaches of the Isère watershed are shown in Figure 7 and
500 Table 6.
501
504 Figure 7 shows that all the disturbance vectors are located in the lower
505 hemisphere, suggesting that all reaches affected by the hydropower systems
506 experience incision. It is possible to appreciate also that a group of vectors (for
507 reaches 610 to 650) have nearly the same direction. These reaches correspond to
508 the main tributary of the Isère River, which is the Arc River. The Arc River is
510 diversion structures with water gates capable to flush (these flushing operations
511 are carried out at least once a year); therefore, the flow regime is modified without
512 modifying the continuity of sediment supply (considering a period over many
513 years). Figure 7 also shows that a group of vectors have trends with a direction
514 around 270° (reaches 200, 500, 600, 531 and 540). These trends are the result of,
515 in one case, the presence of a few large dams (e.g. the reach 200), and in another
516 case, the product of both a deficit in sediment supply remaining from upstream and
517 the water release points that “compensate” the discharge intake upstream.
518 The magnitude of the vectors is related to the intensity of the morphological
519 changes. The intensity of the changes is related to the time response and the
521 (e.g. the activity of the affluent downstream, which attenuates the disturbance
522 effects). We characterized the magnitude of the vector using three intensity
523 classes (presented in section 3.2) namely: > 0.15 low; 0.15 – 0.30 medium and;
525
526 The validation of conceptual models in fluvial geomorphology is limited, in
527 particular when applied to large spatial scales. While an evaluation of the results
528 was possible only for the elevation of the riverbed, the aim of this study was to
529 apply the model in the Isère watershed in order to test the model on a very complex
530 system, and not seek in detail all the possible sources of morphological
531 perturbations. We checked the model results against the Peiry et al.’s (1994) field
532 observations (see Figure 8) and with the evolution of longitudinal profiles, available
533 for some portions of the reaches (see Figures 9a and 9b). The data for the
534 longitudinal profiles were gathered from the Service de Nivellement Général de la
535 France (1908) and SOGREAH (1994). Of the 15 reaches in this study that have
536 field observations, six agreed with the model predictions by Peiry et al. (1994) (500,
537 531, 540, 600, 620 and 630). For three other reaches, the model results did not
538 agree (610, 640 and 650). The model predicts aggradation for these three cases,
539 but the field observation from the Peiry et al. (1994) shows a local incision. In
540 particular, reach 610 (a head water reach) is influenced by water derivation from
541 hydropower intakes, without changes in sediments supply due to HPs. Thus, the
542 morphological channel response expected due to the reduction of river discharge
543 and consequently the transport capacity was aggradation. However, gravel mining
544 took place in this reach (Fig. 8). This point source (gravel mining) of the
546 greater than the discharge perturbation introduced by water diversion. We recall
547 that for the case test in the Isère watershed, gravel mining was not considered. For
548 the remaining 6 reaches (300, 400, 670, 700, 800 and 900) the model was unable
549 to definitively predict the evolution in riverbed elevation, as the disturbance vectors
550 were too close to 225°. Overall, for 6 of 9 reaches with viable prediction, were in
551 agreement with Peiry et al. (1994). Specifically, for the longitudinal profiles of
552 reaches 300 and 500, we verify that the model was capable to predict the changes
553 in the bed elevation. In the case of reach 300 (mentioned in the first part of the
554 results section) the response was unchanged, and for reach 500 the result was
555 bed degradation (due to a heavy water release, displayed in Figures 3 and 6 as
557
558 Figure 8. Trends and intensities predicted by the model for bed elevation,
559 observations from Piery et al., (1994), and location of gravel mining on the active
561
562 Figure 9. Comparison of the longitudinal profiles from the years 1908 and 1988.
563 Figure 9a is reach 300 between Isère at Montrigon and confluence Isère - Doron
564 de Bozel, showing relative stability. Figure 9b is reach 500 between Isère at La
565 Bâthie and confluence Isère – Arly, showing aggradation experienced by the
566 longitudinal.
567
568 While the model satisfactorily compared with field observations, it is important to
569 consider that there are other sources of alteration that can disturb the control
570 variables in alpine mountain rivers, like the Isère River. These include land use
573 and gravel mining. Climate patterns could have also affected the Isère river
574 morphology. According to Peiry et al. (1994), the climatic period known as the Little
575 Ice Age affected the morphology of the Isère River, notable until the 1980. For
576 these reasons, the morphological changes of the Isère River are not solely a
578 is greatly restricted. Additionally, Comiti (2012) adds that a limited number of study
579 cases exist, which thereby hinders any attempt to infer the relative contribution of
581 rivers. One of these cases is presented by Provansal et al. (2014) for the Lower
582 Rhône.
583
585 The need for developing a simplified conceptual model predicting channel changes
586 due to anthropogenic disturbances is justified by the paucity of field data that
587 hamper the change of using more sophisticated approaches at the scale of an
588 entire watershed. From this point of view, the Isère watershed was no exception.
589 In this sense, the aim of the model application was to test the robustness of the
591 The particular difficulties in working at watershed scale with strong anthropogenic
592 disturbance, is that there are often more than one source of disturbance, which act
593 at different spatial and temporal scales. As a result, model predictions are difficult
594 to validate. Additionally, there are other assumptions and limitations within the
595 model that we must note. First, our modeling approach considers only the initial
596 and final states of the channel in dynamic equilibrium. Thus, the duration of the
597 transition period is not evaluated, as well as the overlapping effects of non-
598 simultaneous perturbations, but it is possible to do the model, if there are data
599 available. Second, the perturbations introduced during the transition period at a
600 specific time, or distributed over time (e.g. climate change) may potentially stop,
601 accelerate or shift the processes, and divert the original trajectory of morphological
602 changes (it can include the changes in the land cover dynamics). This also is valid
604 Despite the simplicity of this model, and inherent assumptions and associated
606 morphological trends, when the spatial domain is large and when the perturbations
607 are spatially distributed. Thus, this research presents a first attempt at obtaining a
608 “big picture” view regarding the behavior of morphological fluvial systems. The
609 simplicity of the model makes it flexible for a wide range of different applications
610 and scenarios similar to that tested here (e.g. sand and gravel extraction, sediment
611 replenishment, etc.). Possible future improvements of this model would allow the
612 evaluation of both the indicators related to the alteration for the hydrological regime
613 (FQ) and the alteration of sediment supply (AS). This model uses a channel-
615 and transport capacity have a non-linear relationship (e.g. using the transport
617 shear stress (Shields number) similar to the proposition by Schmidt and Wilcock
618 (2008). However, it is only possible at the watershed scale when the detailed data
620 The simple method to evaluate sediment supply intensity can be improved by
621 considering the stock of sediments in the active channel (available for transport),
622 for example, through time series of aerial photography or time series of bed slope.
623 However, this invariably brings us to the same challenge related to data availability.
624 An important issue that is not addressed by this type of model is the transient
625 duration period (reaction time), and -as mentioned before- possible trajectory
626 changes due to other changes introduced during this period (for example,
627 deforestation, climate change, etc.). In both cases this type of model is limited, and
629 challenge is not only to establish the trajectory of change, but also to know how
631
633 In general terms, anthropogenic disturbance to the river network are evaluated
634 individually at the scale of reach, and not as a system. In this study, we attempted
635 to evaluate the trend of morphological changes of a gravel-bed river at the scale
636 of river network, due to disturbance to two main morphological drivers: hydrological
637 flow regime and sediment supply, which are spatially distributed in a watershed.
638 The outcome of this study was the creation of a tool in the form of a simple model
643 cumulative and synergistic effects at larger scales than just the river reach. Its
644 application is especially relevant for developing countries, for example, in Chilean
645 Patagonia or in Sub-Saharan Africa. In remote zones (where data are not
646 available), there are hydropower projects that consist of more than one
647 hydropower plant, and which could influence a large part of the watershed. In
648 cases such as these, conceptual models similar to the we present could be better
651
652 Acknowledgements
653 The authors thank the suggestions, comments and advices provided by the
654 reviewers and associate editors, and for the excellent support of the SI editors to
656 This work was supported by the FONDECYT project n. 11140764, funded by the
657 Chilean National Agency for the Science and Technology (CONICYT).
658
659 References
667 Andrews E. 1980. Effective and bankfull discharges of streams in the Yampa
668 River Basin, Colorado and Wyoming. Journal of Hydrology 46: 311-330.
670 meandering with self-evolving banks, J. Geophys. Res. Earth Surf., 118, 2208–
673 dams on sediment and hydraulic condition of Rocky Mountains streams. River.
677 Biedenharn DS, Watson CC, and Thorne CR. 2008. Fundamentals of fluvial
681 doi.org/10.1061/9780784408148.ch06#sthash.8mMb237m.dpuf
687 Brune GM. 1953. Trap efficiency of reservoir. Transaction, American Geophysical
689 Buffington JM, Woodsmith RD, Booth BD, Montgomery DR. 2003. Fluvial
690 processes in Puget Sound Rivers and the Pacific Northwest. Pages 46-76 in
691 Restoration of Puget Sound Rivers. University of Washington Press, Seattle, WA.
692 Buffington JM. 2012. Changes in channel morphology over human time scale.
696 Bellière J, Von Raumer J, Carte géologique de la France à 1/250 000 30, Annecy
708 Italian Alps. Earth Surface Processes and Landforms. Volume 37, Issue 7, pages
710 CORINE EEA. 2000. Land cover technical guide: Technical guidelines. Technical
714 10.1177/0309133310369435.
718 Draut A, Logan J, Mastin M. 2011. Channel evolution on the dammed Elwha
720 Dust D, Wohl E. 2012. Conceptual model for complex river responses using an
722 10.1016/j.geomorph.2011.10008.
725 (FISRWG)(15 Federal agencies of the US gov't). GPO Item No. 0120-A; SuDocs
727 Gao JH, et al. 2015. Reservoir-induced changes to fluvial fluxes and their
730 http://dx.doi.org/10.1016/j.quaint.2015.03.015
731 Graf W, 1977. The rate law in fluvial morphology. Amer. Jour. Sci. 272(2), pp178-
732 191.
733 Graf W, 2006. Downstream hydrologic and geomorphic effects of large dams on
735 Grant GE, Schmidt JC, Lewis SL. 2003. A geological framework for interpreting
736 downstream effects of dams on rivers: In a peculiar river, water science and
737 application 7. O’Connor JE, Grant GE. (eds.) Washington, D.C., American
740 Perspectives and prospects. In: 7th Gravel-bed rivers conference, Todoussac,
742 Gregory K. 2006. The human role in changing river channels, Geomorphology,
744 Gregory K, Downs P. 2008. The sustainability of restored rivers : Cathment scale
747 Huang H, Nanson G. 2002. A stability criterion inherent in laws governing alluvial
748 channel flow. Earth and surface Processes and Landform 27, 929-944.
750 environement, Springer-Verlag Berlin and Heidelberg GmbH & Co. KG. 270
751 pages.
752 Kellerhals R, Church M. 1989. The morphology of large rivers: characterization
754 (LARS). Can. Spec. Publ. Fish. Aquat. Sci. 106, pages 31-48.
755 Knigthon D. 1984. Fluvial forms and processes. Edward A. (ed). London, 218 p.
756 Kondolf GM. 1997. Hungry water: effects of dams and gravel mining on river
758 Kondolf GM, et al. 2014, Sustainable sediment management in reservoirs and
760 doi:10.1002/2013EF000184.
762 Proceedings of the American Society of Civil Engineers. Journal of the hydraulics
764 Lave R. 2009. The Controversy Over Natural Channel Design: Substantive
765 Explanations and Potential Avenues for Resolution. Journal of the American
767 1688.2009.00385.x
770 Lewis S.E, Bainbridge Z.T, Kuhnert P.M, Sherman B.S, Henderson B, Dougall C,
771 Cooper M, Brodie. 2013. Calculating sediment trapping efficiencies for reservoirs
772 in tropical settings: A case study from the Burdekin Falls Dam, NE Australia,
775 morphodynamic. The exemple of the upper Arc (French Alps-Savoie). In:
776 Quaternaire, vol. 10, n°1, 1999. pp. 49-59. DOI : 10.3406/quate.1999.1628
777 Navratil O, Albert MB, Herouin E, Grésillon JM. 2006. Estimation of bankfull
778 discharge of river reaches: comparison of existing methods with a steady flow
779 hydraulic model. Earth Surface Processes and Landforms, 31 (11), 1345–1363
783 Peiry JL, Salvador P, Nouguier F. 1994. L’incision des rivières dans les Alpes
786 Peiry JL, Girel J, Pautou G, Guyen N. 1999. Planning of the river Isère in France:
791 Petts G, Gurnell A. 2005. Dams and geomorphology: Research progress and
795 Geomorphology. Academic Press, San Diego CA. Vol 13, Fluvial
798 pattern: examples from S.E. France and S. England. Geomorphology. 19:99-116.
800 Kondolf M, Piégay H. (eds.), Tools in Fluvial Geomorphology, J. Wiley and Sons,
803 2014. The geomorphic evolution and sediment balance of the lower Rhône River
804 (southern France) over the last 130 years: Hydropower dams versus other
806 http://dx.doi.org/10.1016/j.geomorph.2014.04.033
808 hydraulics on incipient bedload motion. Water Resources Research, vol. 45, n°
810 Recking A. 2013. Bedload transport in rivers : from the flumes to the field.
813 Schmidt JC. and Wilcock PR. 2008. Metrics for assessing the downstream
814 effects of dams. Water Resour. Res., 44, W04404, doi: 10.1029/2006WR005092.
815 Schumm SA. 1969. River metamorphosis, Journal of the Hydraulics Division.
817 Schumm SA. 1973. Geomorphic thresholds and complex response of drainage
823 Schumm SA. 1991. To interpret the Earth : Ten ways to be wrong. Cambridge
825 Service de Nivellement Général de la France. 1908. Etude des grandes forces
827 http://geodesie.ign.fr/fiches/index.php?module=e&action=e_profils.
829 Aigueblanche et la confluence avec l’Arly. Donnes regroupés des levés de 1993,
831 Takebayashi H. 2016. Modelling bars and braided channels under unsteady flow
832 and the effect of spatiotemporal change of vegetation on bed and channel
837 change following an extreme flood event. Earth Surf. Process. Landform 40,
839 Thorne C. 1997. Channel types and morphological classification. In Thorne C. R.,
840 Hey RD, Newson MD. (eds), Applied Fluvial Geomorphology for river
841 Engineering and Management, John Wiley & Sons, pp.175 – 222.
842 Turowski J, Rickenmann D, Dadson S. 2010. The partitioning of the total
843 sediment load of a river into suspended load and bedload: a review of empirical
846 Vivian H., 1994. L'hydrologie artificialisée de l'Isère en amont de Grenoble. Essai
847 de quantification des impacts des aménagements. Rev. Géogr. Alpine, 2, 97-112.
849 Werritty A. 1997. Short-term changes in channel stability. In: Thorne CR, Hey
850 RD, Newson MD, (eds.), Applied Fluvial Geomorphology for Engineering and
856 Wohl E. 2014. Time and the rivers flowing: Fluvial geomorphology since 1960.
858 doi:10.1016/j.geomorph.2014.04.012
861 doi:10.1016/j.geomorph.2012.06.001
862
863
864
865
866
867
868
869 Table 1
870 Directions of the most probable trends of river responses as a function of FQ and
871 AS
Angles in degree
Responses Variables
0° 45° 90° 135° 180° 225° 270° 315°
Bed elevation E D ? A A A ? D D
Bed slope S
Channel Width W
Channel Depth DP
Wetted area WA
Width to depth radio* WDR ? ? ? ?
Mean particle size d50 ? ?
Terrace formation T Y ? N N N Y/N Y Y
Colonization of vegetation V N Y/N Y Y Y ? N N
872 D: degradation; A: aggradation; : increased; : decreased; : increased or
873 decreased;
874 ?: Not referenced in the literature; Y: occurrence of phenomena; N:
875 nonoccurrence of phenomena; Y/N: occurrence or nonoccurrence of phenomena
876 *according to Schumm (1969) and Huang and Nanson (2002)
877
878
879 Table 2
880 Identification code used for the assessment of sediments supply based on the
881 classification of the geological map (BRGM, 1980).
Geological zone Mechanic
Characteristics Code
class resistance
Piedmont - Oceanic meta-sediments (ancient High 100
Bündner schist ocean detrital clay metamorphosed)
whose metamorphism is slight.
Pre-alps Sedimentary rocks (easily soluble in Medium 200
limestone water)
External Massif Plutonic magmatic rocks (e.g. High 300
crystalline granite/complex plutonic-volcanic of
the Belledonne mountain range)
Sedimentary Sedimentary rocks Medium 200
cover of the
central massifs
Alpine corridor Alluvial Platform Low 300
“Sillon alpin”
Intra-alpine zone Magmatic or metamorphic rocks High 100
(crystalline and metamorphic rocks)
882
883
884
885 Table 3
886 Identification code used in the sediment supply assessment and classification of
887 terrain gradient
Terrain
Description of potential
gradient Code
sediment supply delivery
(in degree)
1 – 15 Very low 010
15 – 30 Low 020
30 – 45 Medium 030
45 – 60 High 040
60 – 90 Very high 050
888
889
890 Table 4
891 Identification of code used for land cover factors in the assessment of sediment
892 supply, based on the database of CORINE project Level III.
Classification according to the
Land cover class Code
potential input of sediments
Urban, and forest areas Low 001