Study on initial evolution process of tidal creek network

Transcription

1 Study on initial evolution process of tidal creek network Toshiki IWASAKI, Yasuyuki SHIMIZU & Ichiro KIMURA Graduate school of engineering, Hokkaido University, Sapporo, Japan ABSTRACT: Prediction of morphological evolution of salt marsh, including development of tidal creek network, is important from the ecological and river engineering viewpoints. This paper presents field observation of salt marsh and numerical simulations of tidal creek network to clarify the formation mechanism of channel network. Field observation was conducted in Notsuke marsh which has the tidal creek network to understand hydrodynamic and morphodynamic phenomena on tidal marsh. Two kind of computations were performed under the different conditions, i.e, bed slope, vegetation and tidal condition, to clarify initiation process of tidal creek network and key phase for morphological development of tidal marshes. The results indicate that tidal channel network is mainly formed due to the drainage flow from tidal flat to main tidal channel with phase lag in ebb tide phase and the flow in ebb and flood phase have a different role for developing marsh morphology. 1 INTRODUCTION Tidal creek network is commonly salt marsh morphology in tidal environments. The natural environments associated with this complex marsh geography have contributed to numerous ecosystem and human activities as well as disaster prevention from hydrodynamic force such as tsunami in tidal area (Cochard et al. 28). However, morphodynamics behavior in intertidal area may be affected by a variation in hydrodynamic feature associated with human activities and sea level rise caused by recent green house effects. From the ecological and river-coastal engineering points of view, it is significant research topic to understand morphodynamics phenomena in tidal marsh and to predict flow field and morphological evolution of salt marsh. Many researchers have studied about hydrology, hydraulics and geomorphology in tidal environment (Allen 2). Various field observations have been carried out to understand hydrodynamic and morphodynamic phenomena of salt marsh. Geomorphic change of salt marsh and estuarine was investigated by using aerial photographs (Knighton et al. 1992). However, the dominant parameter for developing tidal creek is still not revealed because these complex network system is the result of long term morhodynamic process. Besides, there are no field information about tidal creek in Japan. Recently, numerical computations also have been performed to simulate morphological evolution of salt marsh, including development of tidal creek network, by using simplified model (D Alpaos et al. 25) and physicallybased model (Masuya et al. 27). In these studies, it is shown that physically-based model can reproduce channel network configuration qualitatively, but initiation process of tidal channel and development of network were not discussed. This work presents field observations of Notsuke marsh which is one of salt marsh in Japan to understand hydrodynamic and morphodynamic phenomena in this marsh. Numerical computations also conducted to discuss the initiation and developing process of tidal creek network by using physically-based morphodynamic model proposed by Masuya et al. (27) and Iwasaki et al. (28). Two computations were performed under the following different viewpoints, namely, (i)initial evolution process of channel network (Case1) and (ii) key phase for developing of salt marsh (Case2). The computations of Case1 are performed under the different bed slope and vegetation condition. Whereas, in Case2, different tidal condition is used for computations. 2 FIELD OBSERVATION 2. 1 Studied Area Figure 1 shows the aerial photograph of Notsuke marsh which has tidal creek network in Hokkaido, Japan. This marsh is located at the roof of Notsuke peninsula which was registered to Ramsar Convention and is exposed to various hydrodynamic phenomena, such as tidal current, wave current and inflow from Chashikotsu River that flows near this marsh.

2 Bankfull A Water Depth(m) (m) Figure 1. Notsuke marsh in Hokkaido, Japan. Meandering channels extend from seaward to landward on this marsh. (From Hydrographic and Oceanographic Department, Japan Coast Guard) Many researchers have carried out the field observations to clarify the geological characteristics, hydrodynamic and morphodynamics feature as well as the developing process of tidal creek network (Allen 2). Time exchange of water depth, flow velocity and discharge in a tidal channel was measured and it is known that flow discharge in a tidal channel shows asymmetry feature in ebb/flood phase (Leopold et al. 1993; Boon 1975). Pye and French (Pye and French 1993) explored 12 British salt marsh which has channel network and they classified the properties of channel patterns in salt marsh into seven patterns from observational results. However, field investigations about tidal creek have never been conducted in Japan. Because of the lack of knowledge about tidal creek, it is difficult to discuss the development of tidal creek network. Thus, we carried out field observation of tidal creek at Notsuke marsh to examine (i) geological feature of tidal creek, (ii) hydrodynamic and morphodynamic phenomena on salt marsh and (iii) time exchange of water depth and flow velocity in a tidal channel Geological Characteristics Complicated and meandering channel network develops from seaward to landward on the near-horizontal muddy marsh platform in this site. The distinctive geological features of Notsuke marsh are as follows. Fine and cohesive material like mud composed the bed material load on whole marsh site, in particularly on the tidal platform which are flanked by tidal creek. Whereas, the sediment in a tidal channel consist of fine sand sediment and cohesive material like silty and distribution curve of this sediment shows wide range of which d 5 is approximately.3mm. In addition, marsh platform on channel bank is covered by Flood Phase Ebb Phase Velocity(cm/s) Figure 2. Time exchange of water depth and flow velocity measured in a tidal channel at point A (See Fig. 1). This figure shows the asymmetry feature of water depth and flow velocity between ebb and flood phase. dense vegetation in which diameter is about 2-3mm and density is 15-25/m 2. We also observed flow condition, shape of channel and sediment transport in this area. During observation, active sediment transport could not be seen and slope of bank showed right angle, partly overhanged. Vegetation and cohesive material makes bed material stronger. They reduced erosion rate from bed and protect bank of channel. This result indicates that development speed of tidal channel network is considerably slow in this marsh because of roles of cohesive material and vegetation Hydrodynamic Characteristics Marsh surface of Notsuke marsh is inundated at high water level associated with sea level fluctuations. Entire marsh platform is completely submerged when water level is high. While, when water level is low, marsh surface becomes dry condition except with a part of marsh surface and in the channel. Whereas, wave current can be neglected, because the water surface shows calm during observation. This results suggested that flow field of Notsuke marsh is dominated by tidal flow, therefore tidal creek network may be developed by mainly sea level fluctuations in this marsh. Measurements of water depth and flow velocity were performed at point A shown in Figure 1. Flow velocity was measured by electromagnetic current meter at 6% of water depth below water surface at centerline of a tidal channel. Figure 2 shows result of measurements. It is shown that flow velocity changed a range from to 4cm/s. However, this figure also implied that the water depth when the peak velocity appeared is different between ebb and flood phase. It is known that this asymmetry feature between ebb and flood phase is typical hydrodynamic phenomenon in a tidal channel. However, this feature depend on the discrepancy between sea level and bed elevation of marsh surface. Thus, more investigations will be required to clarify the reason of tidal asymmetry in this marsh.

3 2. 4 Discussion Although maximum flow velocity in a tidal channel reached to 4cm/s which can transport fine sediment, active sediment transport was not seen during a tidal cycle and flow velocity on tidal marsh showed slow. From this result, the process of developing network may require flow concentration that is transport sediment. It is suggested here that the flow in ebb phase is key factor for developing tidal channel network. In ebb phase, even if tide level in a channel becomes to undermarsh tide, inundated water in flood phase can not be drained due to mild slope of marsh platform. This residual water flows from marsh surface to a tidal creek with phase lag and high velocity associated with large water surface slope in edge of channel. This flow can occur sediment transport and bed changing. Thus, branching channel develops from a main channel and after then, this small incision gather the more residual water. As the result of this bed deformation processes, channel network may be formed in tidal marshes. 3 NUMERICAL APPROACH It is pointed out that drainage flow from marsh surface to tidal channel with phase lag is important factor for developing of the branched channel from a main tidal channel from field observations. However, it is difficult to prove this morphodynamic phenomenon by field observation because tidal creek network is the result of long-term morphological evolution of salt marsh. Therefore, we use a numerical model which was proposed by Masuya et al. (Masuya et al. 27) and Iwasaki et al. (Iwasaki et al. 28) to discuss the development processes of channel network. Two kind of computations were conducted to focus the different point of view as follows. (i)case1: to discuss the initial branching process of tidal channel network and the effect of slope of marsh platform and vegetation. (ii)case2: to clarify the role of each ebb/flood phase flow for development of marsh morphology Flow Model The simulation was performed utilizing twodimensional shallow water hydrodynamic model to calculate flow field on tidal marsh. It is assumed herein that flow field is governed by only sea level fluctuations. Continuity and momentum equations are expressed as follows. h t + (uh) x + (vh) = (1) u t + u u x + v u H = g x τ ( x 2 ) ρh + ν u x + 2 u Dx 2 2 ρh v t + u v x + v v = g H τ ( y 2 ) ρh + ν v x + 2 v Dy 2 2 ρh (2) (3) where, t:time, h:water depth, u,v:depth-averaged flow velocity in x,y directions, g:gravity acceleration, H:water level, ρ:water density, ν:kinematic viscosity, τ x,τ y : bed shear stress in x,y directions and D x,d y :drag force of vegetation in x,y directions. The component of drag force of vegetation is estimated under the assumption that the vegetation does not bend and act like circular poles as follows. D x = 1 2 ρc Da s min[h, l]u u 2 + v 2 (4) D y = 1 2 ρc Da s min[h, l]v u 2 + v 2 (5) where, C D :bulk drag coefficient of vegetation, l:height of vegetation, a s :projected plant area per unit volume Sediment Transport Model Continuity equation of bed is governed by bed and suspended load transport as following equation. z t + 1 ( qx 1 λ dx + q ) y dy + q su w f c b = (6) where, z:bed elevation, λ:void ratio of bed material, q x,q y :bed load transport in x,y directions, q su :upward flux of suspended sediment from bed, w f :settling velocity and c b :relative suspended sediment concentration. Bed load is calculated by using Ashida and Michiue s formula (Ashida and Michiue 1972) as below. ( q b = 17τ sgd τ ) ( c 1 τ ) τ c τ (7) where, q b :total bed load flux, s:specific weight of grain in fluid, d:sediment diameter, τ :nondimensional bed shear stress and τ c :non-dimensional critical bed shear stress which is estimated by Iwagaki s formula (Iwagaki 1956). Upward flux of suspended sediment from bed is estimated by using Kishi and Itakura s formula (Itakura and Kishi 198). q su = K ( α sgd u Ω w f ) (8)

4 Table 1. General computational condition Case1 Case2 Grain Size.1mm.3mm Tidal Amplitude.8m.6m Mean Sea Level Initial Bed Elevation of Downstream End Computational Domain 4m 12m (see Fig. 3) 4m 2m Dx, Dy 5m, 2m 2.5m, 2.5m Initial Bed Shape A Meandering Channel Random Perturbations Slope Different (see Tab. 2) Flat Vegetation Different (see Tab. 2) Not Consider Transport of suspended sediment is obtained by using following depth-averaged two-dimensional advection equation. (ch) t + (uch) x + (vch) = q su w f c b (9) where, c:depth-averaged suspended sediment concentration. Sediment transport model does not contained the effect of vegetation directly, i.e. trapping rate due to vegetation (D Alpaos et al. 25). This model considers the interaction between sediment transport rate and vegetation as indirectly, namely, increment of sedimentation and decrement of erosion associated with the effect of vegetation are estimated by the reduction of flow velocity due to drag force of vegetation Vegetation Growth Model It is necessity to consider several natural factor, such as sunlight, nutrient and bed condition, for modeling of vegetation growth. However, it is difficult to construct the vegetation growth model which is considered all parameters. D Alpaos et al. (D Alpaos et al. 25) expressed plant biomass of vegetation as a function of bed elevation. It is assumed here that whether vegetation can grows or not depend on only bed elevation, namely when bed elevation, z, exceeds critical bed elevation, z, vegetation start to grow. This assumption can be written as. z > z (1) Time and bed elevation when bed elevation satisfy vegetation growth condition here are defined as T and z, respectively. Besides, function of vegetation growth l and a s is set to be liner function of time until growth time, T, reaches as maximum time T max as following equations. l = T T T max l max (11) a s = T T T max a smax (12) Table 2. Computational condition of Case1 Slope Vegetation Case1-1 Flat Not Consider Case1-2 1/25 Not Consider Case1-3 Flat Consider 12 Tidal Flat Channel 4 Upstream end Downstream end Figure 3. Plane view of initial bed in Case1. A meandering channel is set in center region of tidal flat. where, l max and a smax :maximum value of l and a s, respectively. After the growth time becomes T max, l and a s are set to be maximum value. While, disappearance process of vegetation can be modeled under the hypothesis that this phenomena is caused by only bed erosion as follows. z < z (13) 3. 4 Computational Condition We carried out two kind of computations under the different viewpoint and condition. General computational condition were given in Table 1. The details of computational conditions were described as follows Case1 The developing process of tidal channel network as well as the effect of slope and vegetation was examined by using proposed morphodynamics model in this case. Three short computations were performed under different bed slope and vegetation condition as given in Table 2. Initial bed configuration is set as simple rectangular domain containing a meandering channel of width of 6m and depth of.5m in center region of computational domain as shown in Figure 3. Boundary conditions are given as follows. Water level of downstream end is set to be equal to sinusoidal semi-diurnal tide which amplitude is.8m and return period is 12 hours. In other three boundaries, flow and sediment transport were completely restricted.

5 DAY= 1 day Both Phase f e n d Flood Phase Ebb Phase o v e l a m le r n stre a te w o W d Time Figure 4. Water depth of downstream end in Case2. Tidal fluctuation was divided into ebb and flood phase. It takes long time computation to evaluate morphodynamic behavior by using physically-based model. Smaller sediment diameter and larger tidal amplitude compared with field data measured in this study given in Table 1 were therefore used to evaluate developing process of channel network efficiently. As the result, however, computational morphodynamics evolution may occur with different time scale in comparison with real scale morphodynamic development. Thus, we set the parameter of vegetation growth to adjust time scale between morphodynamic behavior and vegetation growth from preliminary computations Case2 The remarkable point of this case is how the ebb and flood phase flow affect for morphodynamics of tidal marshes. Three computations were performed under different tidal condition by separating tidal fluctuation, i.e, only flood phase (Case2-1), only ebb phase (Case2-2) and both phase (Case2-3) shown in Figure 4. Return period of a cycle in Case2-1 and 2-2 was defined as six hours. Because of dividing tidal fluctuation, a function of tidal fluctuation becomes discontinuous at slack tide in Case2-1 and 2-2. Thus, it is assumed that flow stage can change to next cycle immediately and after shifting, flow velocity is set to be zero and water level on whole computational domain is given as same water level at downstream end. The computation domain given in Case2 is set as a simple flat rectangular domain of 4m 2m in which the initial surface has a random perturbations of order of grain size. Grain size and tidal amplitude used in this case were determined based on field data of Notsuke marsh. Boundary conditions were given as same condition of Case1. Resolution of computational domain significant affects to the computational results and time. From preliminary computation and previous study of Masuya et al. (27), it is found that development of tidal creek network in narrow area considering in this study can be reproduced with less than 5m grid spacing in DAY= 1 day Figure 5. Calculated channel network in Case1-1. both directions. Therefore, we use the grid size shown in Table 1. However, relation between grid spacing and size of computational domain may strongly affect computational result. Thus, detailed sensitivity analysis about grid spacing will be required for quantitatively evaluation of computations. 4 RESULT AND DISCUSSION 4. 1 Case Development of tidal channel network Figure 5 shows the developing process of channel network calculated during 1days in Case1-1. It is shown that complex channel network which has dendritic feature is generated from a meandering channel given initial condition on computational domain. Besides, initial meandering channel migrates toward landward due to flow associated with flood phase. From comparison between this result and network configuration in Figure 1, it is found that a model can reproduce channel network which has like actual field characteristics of complex and meandering. Figure 6 shows calculated depth-averaged flow vector and suspended sediment concentration on a computational marsh when tidal stage is at mean sea level in ebb and flood phase. It is shown that active sediment transport of suspended load occurs in edge of channel network associated with shallow flow which concentrated from marsh surface to a main channel in ebb tide phase with phase lag. This results suggest that ebb flow with phase lag is key factor for development of tidal channel network. Whereas, in flood phase, water flows from channel to marsh surface with sediment. This flow migrates the channel and transports sediment to upstream side. It indicates that the flood flow also affects network extension. The result implied that the ebb and flood flow have a different role for geometric change of salt marsh, including formation of channel network.

6 Time/T=.26 =.4(m/s) DAY= 1 day (a) Ebb tide phase (unit:%) Time/T=.76 =.4(m/s) (unit:%) (b) Flood tide phase Figure 6. Computational result of velocity vector and suspended sediment concentration in middle tide level in ebb and flood phase The effect of slope and vegetation Figure 7 shows calculated bed topography in Case1-2 in which slope of marsh platform is considered. From calculated result, branched channel is also generated from initial channel in this case, however, network feature of Case1-2 is different from that in flat condition, namely, result of Case1-1 shows more complicated network compared with Case1-2. This results suggest that the slope of marsh surface is important for channeling process in tidal environments. In sloping bed condition, flooded water on tidal marsh can be drained with decreasing of water level, because the platform slopes from landward to seaward. This effect decreases shallow flow with phase lag which contributes to developing channel network. Moreover, flow velocity associated with flood phase is also reduced by bed gradient which oppose flow direction. The result indicates that the marsh slope restricts the development of channel network. Figure 8 shows computational result of Case1-3 in which vegetation is considered. In Figure 8a, white dots mean vegetated area. Network geometry between each case has slightly difference in bifurcated part of channel network shown in Figure 8b. From Figure 8a, marsh platform around channel network is covered by vegetation. In vegetated area, erosion rate is reduced and deposition rate is increased by reduction of flow velocity associated with vegetation. This result indicates that vegetation acts to restrict channel migration to upstream Case2 This section discusses the role of ebb and flood phase flow for developing channel network by comparing network geometry calculated after 56 cycles in Figure 7. Calculated channel network after 1days in Case1-2 in which initial tidal flat have a slope of 1/ ) ( m y 1 5 DAY= 1 day Case1-1 Case1-3 (a) Upstream end Downstream end (b) Figure 8. Computational result of Case1-3. (a) Calculated channel network in Case1-3 that considered vegetation and white dots means vegetated area and (b) Comparison between the channel network configuration simulated Case1-1 and Case1-3. Channel network of Case1-1 migrated toward to upstream side with bifurcation of channel network in comparison with case of considering vegetation. Case2. The computations show considerably different bed topography between in case of considering only ebb phase (Case2-1) and only flood phase (Case2-2) as shown in Figure 9. In Case2-1, two main channels develop from downstream end to upstream side with a combination of main channel meandering and headward development. As the result, drainage channel system like gully erosion appears on the computational domain. Whereas, flow in only flood phase can not form clear channel network, but the complex geometry develops in upstream region of computational area. Figure 1 shows calculated network configuration after 28 ebb and flood cycles. This network property shows the dendritic patterns. Clear channel network is generated only in case of considering ebb phase. Hence, tidal creek network is mainly attributed to the flow in ebb tide phase. It is known that network properties developed in terrestrial and tidal region have similar feature (Knighton et al. 1992). However, network configurations simulated in Case2-1 and Case2-3 show the slight difference. This difference may be caused by

7 Cycle= 562 Cycle= Cycle= 566 (a) Case2-1(only ebb tide phase) Figure 1. Channel network simulated in Case2-3 that consider both tidal phase after 28 ebb and flood tide cycle. Three main meandering and branching channels form from downstream end (b) Case2-2(only flood tide phase) Figure 9. Comparison of computational channel network properties in Case2-1 and Case2-2 after 56 tide cycle. the deposition by the flow in flood phase. This result indicates that the ebb and flood flow have a different role for development of channel network. Ebb tide flow mainly works for channel initiation and meandering. While, flood tide flow gives the some perturbation to upstream region that affects formation of channel network. 5 CONCLUSIONS We conducted field observation and numerical simulations about tidal creek network formed in tidal environments to examine hydrodynamic and morphodynamic phenomena on salt marsh and to clarify developing process of tidal creek network. Conclusions of this study can be summarized as follows. (i) Marsh platform of Notsuke marsh is composed with high cohesive material and a part of marsh surface is covered by dense vegetation. Hydrodynamic feature in a tidal channel shows velocity asymmetry. (ii) The key factor of developing channel network is ebb tide flow which concentrates from marsh platform to main channel with phase lag. (iii) Mild slope of marsh surface makes channel network more complex and vegetation protects migration of channel and bank erosion. (iv) The flood/ebb tide flows have a different roles for evolving tidal channel network. Ebb flow mainly acts developing channel network, while flood tide flow gives perturbation to upstream side by deposition. This study shows developing process of tidal creek network qualitatively by the result of field observation and numerical simulations. As the next step, it is necessary to performed detailed field investigations and experiments for understanding morphodynamic phenomena in tidal environments. Moreover, validation of numerical model and computation which reflects field observation and experimtn are also needed for prediction of developing of tidal creek network. REFERENCES Allen, J. R. L. (2). Morphodynamics of holocene salt marshes: a review sketch from the atlantic and southern north sea coasts of europe. Quaternary Science Reviews 19, Ashida, K. and M. Michiue (1972). Hydraulic resistance and bed transport rate in alluvial streams. Proceedings of JSCE 21, 59 69,(in Japanese). Boon, J. D. (1975). Tidal discharge asymmetry in a salt marsh drainage system. Limn. & Ocean. 2(1), Cochard, R., S. L. Ranamukhaarachchi, G. P. Shivakoti, O. V. Shipin, P. J. Edwards, and K. T. Seeland (28). The 24 tsunami in aceh and southern thailand: A review on coastal ecosystem, wave hazards and vulnerability. Perspectives in Plant Ecology, Evolution and Systematics 1, 3 4. D Alpaos, A., S. Lanzoni, M. Marani, S. Fagherazzi, and A. Rinald (25). Tidal network ontogeny: Channel initiation and early development. Jornal of Geophysical Research 11, F21, doi:1.129/24jf182. Itakura, T. and T. Kishi (198). Open channel flow with suspended sediment. Proceedings of ASCE No.16(HY8), Iwagaki, Y. (1956). Hydrodynamical study on critical tractive force. Trans. of JSCE No.41, 1 21,(in Japanese). Iwasaki, T., Y. Shimizu, S. Masuya, and K. P. Dulal (28). Simulation of a tidal creek with vegetation effect. River Flow 28 2, Knighton, A. D., C. D. Woodroffe, and A. K. Mills (1992). The evolution of tidal creek network, mary river, northern australis. Earth Sur. Proc. Land 17, Leopold, L. B., J. N. Collins, and L. M. Collins (1993). Hydrology of some tidal channels in estuarine marshland near san francisco. CATENA 2, Masuya, S., Y. Shimizu, and S. Giri (27). Simulation of morphology in the tidal environments. RCEM27, Pye, K. and P. W. French (1993). Erosion and accretion processses on british salt marshes. Cambridge Environmental Research Consultants.