Simple stochastic cellular automaton model for starved beds and implications

CA model

CA models range from the coarse graining of sand by thinking about clusters of sand grains (sand slabs) rather of person particles. In CA models, the topographic height is taken is the quantity of compiled slabs. The topographic height h(i, j) at site (i, j) inside a two-dimensional field changes as time passes. Previous CA models for aeolian sand dunes used a phenomenological formulation of saltation that’s, it wasn’t according to fluid motion. Although there’s some variation within the formulations of saltation, the prior models determined situational-specific saltation distances, for example defining the jumping length like a purpose of the condition from the sand bed.

Allen () noticed that the variations in water and air transport stemmed from variations within the viscosity and density from the medium. On subaqueous beds under relatively slow flows, the primary transport mode of sand grains is moving and sliding, that bed contact is continuous (Allen Reid and Frostick Bridge and Demicco ). Therefore, rather of utilizing the saltation length, low transport stages of subaqueous bedforms could be better described by alterations in the migration speed of grains in continuous connection with your bed. Within the simulation used in our study, the movement of particles was considered individually for 2 processes: transport because of fluid motion, with continuous bed contact and avalanching because of gravity. Through out this paper, we’ll use “;bedload” to mean the continual bed-contact load that’s only because of the flow (not because of gravity). Other processes because of flow, for example saltation or suspension, won’t be considered to ensure that we are able to concentrate on subaqueous environments with low transport rates that form topographies for example sand patches. Within the simulation, which process (bedload transport or avalanching) occurs is evaluated deterministically in the geomorphological condition from the position. However, the movement of sand slabs because of bedload transport is really a stochastic variable that expresses finite velocities for discrete positions that’s, slabs migrate based on a probability that is representative of the rate (see below). The simulation ignores the interior structure from the flow, which becomes important once the flow is comparatively fast and also the bed roughness (bedforms) is big for simplicity, we disregard the results of strong erosional vortices and secondary flows.

Formulation of bedload transport

When simulating bedload transport inside a CA model, sand slabs ought to be limited to relocating to the adjacent downstream cell, unlike the procedure for saltation. The topographic height is impacted by bedload transport the following:

$$ hleft(i,jright)to hleft(i,jright)-_bmathrmkern0.5em hleft(i+1,jright)to hleft(i+1,jright)+_b, $$

where site (i + 1, j) is downstream of site (i, j), and q b is the amount of slabs that migrate throughout a single calculation step. Here, we consider bedload transport in just the direction from the flow this is actually the positive x-direction of these two-dimensional (x, y) system.

When the bed is perfectly uniform, sand might be transported everywhere in a constant rate, but otherwise, the speed of sand migration depends upon the place. We formulated an easy phenomenological rule for the prospect of sand movement (surrogate of velocity) that doesn’t require calculating the fluid motion. In every time step, a sand slab at first glance moves to another downstream cell or remains at same position, as determined stochastically this probability matches the rate from the sand migration velocity (that’s, a greater probability matches a greater velocity), which depends upon the condition of every site.

We made the next assumptions. Sand slabs on the stoss side move fastest in the local peak (condition s1), along with other sand slabs (for instance, partway lower the slope) slow lower because the downstream gradient increases, because of the competing gravitational pressure. Sand slabs on lee slopes exercise gradually than individuals on stoss slopes, because to some degree, the bedform shields the downstream flow. For any slab on the lee slope, the prospect of movement depends upon the level from the shielding aftereffect of the bedform we think that this relies around the gradient from the lee slope. Once the slope of the lee side is under a particular value (condition s3), slabs can move like a bedload otherwise, when the slope is more than or comparable to a threshold value (condition s4), slabs have been in a wonderfully shielded area (shadow zone), where the flow isn’t sufficient to share sand like a bedload, and therefore, the sand moves only by avalanching. From all of these assumptions, we have an easy formulation for that condition-dependent (s1 to s4) probability, B i, j , the sand at site (i, j) will migrate by bedload transport towards the nearest downstream cell:

$$ _=leftalpha hfill & mathrm mathrm mathrmmathrm mathrm, mathrmmathrm mathrm mathrm mathrm left(mathrm1right)hfill alpha / left(hleft(i+1, jright)-hleft(i, jright)right)hfill & mathrmhalfway mathrmmathrmmathrmmathrm mathrm mathrm mathrm left(mathrm2right)hfill gamma lefthfill & mathrm mathrm mathrmmathrmmathrm mathrm, mathrm mathrmkern0.5em mathrmkern0.5em mathrmkern0.5em mathrm mathrm left(mathrm3right)hfill 0hfill & mathrm mathrm mathrm mathrm left(mathrm4right),hfill finishright. $$

where α, β, and γ are positive constants. The 2nd expression around the right-hands side implies that the problem of climbing a stoss slope depends upon the gradient. The 3rd expression around the right-hands side shows the level that the bedform cuts down on the flow, like a purpose of the gradient from the lee slope.

In Eq. (), the health of the neighborhood condition (s1 to s4) is decided the following. If h i − 1,j   h i,j , the website (i, j) is on the stoss slope or in a local peak. If it’s also correct that h i + 1,j  − h i,j   0, then your website is an optimum or perhaps a flat top (s1) otherwise, it’s somewhere along a stoss slope (s2), and also the velocity is inversely proportional towards the gradient. However, if h i − 1,j  > h i,j , the website (i, j) is on the lee slope. If it’s also correct that h i − 1,j  − h i,j   β, then site (i, j) isn’t inside a shadow zone (s3) otherwise, it’s inside a shadow zone (s4). Observe that β may be the threshold value for that shadow zone. The above mentioned resolution of the website condition for that bedload (s1 to s4) could be summarized the following:

$$ lefthfill mathrmkern0.5em _leqq _kern0.5em mathrmlefthfill ifkern0.5em _-_leqq 0hfill & hfill :(s1)hfill hfill ifkern0.5em _-_>0hfill & hfill :(s2)hfill finishright.hfill hfill mathrmkern0.5em _>_kern0.4em mathrmlefthfill mathrmkern0.5em _ – _leqq beta hfill & hfill :(s3)hfill hfill mathrmkern0.5em _ – _>beta hfill & hfill :(s4)hfill finishright.hfill finishright.. $$

Formulation of avalanching

Avalanching occurs because of gravity, and therefore, sand could be communicated by avalanching inside a shadow zone, even if bedload transport doesn’t happen. Avalanching takes place when the lee slope is steeper compared to position of repose the probability that avalanching occurs is unity, which means deterministic and immediate movement. For avalanching within the streamwise direction at site (i, j), the topographic height at (i, j) develops the following:

$$ hleft(i,jright)to hleft(i,jright)-_amathrmkern0.5em hleft(i+1,jright)to hleft(i+1,jright)+_a, $$

where q a is the amount of slabs that move by avalanching during once step. Based on flume experiments, avalanching doesn’t happen on the stoss side thus, we don’t need to consider avalanching within the negative x-direction. Avalanching always occurs when the following condition is content:

$$ left mathrmright>_r mathrm mathrm mathrm(s4), $$

where A r is really a positive constant that is representative of the position of repose, and also the slope within the downstream (positive x-direction) is understood to be

$$ left mathrmright=hleft(i, jright)-hleft(i+1,jright). $$

In contrast to bedload transport, due to gravity, avalanching also occurs within the lateral direction:

$$ hleft(i,jright)to hleft(i,jright)-_amathrmkern0.5em left_amathrmhleft(i,j-1right)to hleft(i,j-1right)+_aright. $$

Observe that oblique movements between adjacent cells are overlooked in our simulation. Lateral avalanching occurs underneath the same condition as with Eq. (), but here, the slopes of great interest are individuals within the good and bad y-directions, that are understood to be

$$ left mathrmright=hleft(i, jright)-hleft(i,j+1right),kern0.5em hleft(i, jright)-hleft(i,j-1right), $$

correspondingly. When the lateral slopes both in the good and bad y-directions are steeper compared to position of repose, only one of these simple two directions is chosen, and they’ve equal probability:

$$ _=_=frac, $$

where A l + and A l − would be the odds of avalanching within the good and bad y-direction, correspondingly. In the past CA models for aeolian dunes, avalanching sand moves within the steepest direction (e.g., Werner ). In our model, for simplicity, we adopted the idea mentioned in Eq. (). When averaged with time, this rule is the same as the distribution because of avalanching of sand inside a sandpile (e.g., Bak et al. ) that’s, the sand is shipped equally in most possible directions.


The simulation in our study was performed the following. Each and every time step, the movement of sand slabs within the x-direction, because of both bedload transport and avalanching, was resolute for every cell and updated in parallel. Next, lateral avalanching (within the y-direction) was resolute for every cell and updated in parallel. Throughout a single step, avalanching happened just once per cell (i.e., it wasn’t repeated until perfect relaxation). It was accomplished for simplicity and since we consider bedload transport, and therefore, there is no need to acquire a static slope (inside a strict sense) after perfect relaxation.

When deciding on values for that phenomenological parameters, we made the next assumptions. The topographic height was taken is the quantity of sand slabs, and for that reason, β and A r were positive integers. Since the slope of the stoss side of the bedform is definitely gentler compared to corresponding lee slope, the inequality 1 < β ought to be satisfied. Additionally, we impose the inequality α > γ(β − 1) since the migration of grains is quicker on the stoss side than around the corresponding lee slope (see Eq. () for that gentlest lee slope). The need for α is chosen arbitrarily, susceptible to the problem it be sufficiently smaller sized than unity this probability is really as a surrogate for that velocity of sand particles. The simulation answers are insensitive to α, in addition to the rate of development for any with time step. For any given worth of α, the of γ ought to be sufficiently little the above inequalities are satisfied otherwise, the model won’t adequately reproduce the shielding aftereffect of lee slopes. For simplicity and also to take account of the aforementioned inequalities, we used the next values for that parameters: α = 0.05, β = 2, γ = α/8(=6.25 × 10− 3), and A r  = 4. We let q b  = 1, addressing the idea the bedload is really a thin layer. We let q a  = 2 to ensure that there is a quick relaxation from the avalanching however, we observe that when q b  = 1, the simulation outcome was much the same, supplying α was sufficiently smaller sized than unity. The first condition from the bed was almost flat, with a few small random variation, for that available sediment (sand slabs), as the substrate was perfectly flat. The calculation space would be a 200 × 40 grid of cells, although we used a 400 × 80 grid of cells to create particular comparisons. Calculation runs were performed in excess of 105 time steps, utilizing a periodic boundary condition. Since the simulation would be a minimal model and used the easiest formulae and boundary condition, the primary discussion is going to be limited to the situation of sparse available sediments. We observe that when sediments are all around, the interference between opposing limitations can’t be overlooked, and therefore, it’s not appropriate to utilize a periodic boundary condition.


H:Simple stochastic cellular automaton model for starved beds and implications about formation of sand topographic features when it comes to sand flux

Key:Simple stochastic cellular automaton model for starved beds and implications about formation of sand topographic features when it comes to sand flux

What is cellular automata modeling?