Constrained full-river tests

I have added the valley boundary to the MNRR simulation, and fixed the location of the inlet point at Gavins Point Dam. The downstream end of the river is free to move (although it could also be fixed, if required. In reality, the tail of the river is constrained by the artificial channel and wing dikes of Kensler's Bend). The river and boundary can be displayed alone or on a 1 km grid:

(Click for larger image)

The boundary forms a closed polygon around the river. All river points are now checked:

1. To make sure they are within the boundary.
2. To make sure they are at least 1/2 river width from all boundary points and edges.

During the simulation, after each step, all points are rechecked and:

1. Are reset to their pre-step positions if they go outside the boundary.
2. Are moved away from the boundary (along a vector from the nearest boundary point, usually on an edge) if they are within 1/2 river width of the boundary.

This process allows the river banks to become very flush along the boundary perimeter, irrespective of their current width. The boundary was originally digitized at a resolution of 100 m, but that made the boundary-checking code too slow. I have decreased the boundary resolution to 1 km per edge, which is sufficient for containing the river geometry (digitized at 100 m) within the valley, but is fast enough for practical purposes.

The simulation can also be overlayed on a map or satellite mosaic. Here is the initial river centerline alone:

(Click for larger image)

The water:

(Click for larger image)

The water on the satellite mosaic:

(Click for larger image)

Here are the 100 year results of a circumferential speed (sigma) simulation, showing new centerline:

(Click for larger image)

water:

(Click for larger image)

and cutoffs:

(Click for larger image)

Here are the 100 year results of a Johannesson-Parker (u1b) simulation, using a fixed-width of 400 m:

(Click for larger image)

Here are the 100 year results of a variable-width sigma simulation:

(Click for larger image)

Here are images from last week, some of which you've already seen:

Elevation of the MNRR valley:

(Click for larger image)

Elevation between 330 and 360 m:

(Click for larger image)

All points between 360.76 and 370.76 m (1200'):

(Click for larger image)

This may give a better indication of the valley boundary, but unfortunately it is in a different projection than I am using from the topo maps and satellite images.

The valley elevation in 3d:

(Click for larger image)

A 100 yr simulation of the river inlet area:

(Click for larger image)

Next to do is a statistical coverage map of the river over time. That is, on either the 1 km grid, or one of higher resolution (e.g. 1 ha), track the location of the river (including width) at each step, accumulating counts in each of the grid bins. At end of simulation, normalize total to 1.0. An image or surface plot of the result will indicate both the maximum extent of the river over time, and the most likely locations of the river at any time. Hopefully, the accumulation of coverage maps for ~1000 simulations will indicate with some confidence the potential 'migration corridor' at different probability levels.