I have a high resolution DEM and want to do a flood simulation starting at a river. My hope is that the water runs into historical (now dry) river channels (in which I'm interested in) at first and step by step. In my imagination the whole old channel system appears at the end. But the problem is, the higher the water raises the area which is covered by water gets bigger and bigger. Like in this pic (on the right):

enter image description here

But I would love to get a result like on the left side. Now I hope i can describe this understandable: My idea is to do a flooding step by step and every step will be "saved". At the end you have several flood phases which you can intersect.

enter image description here

Red, green and yellow are the different flood phases and the brighter blue is the intersection. I dont know how to manage this (even after internet research) or if it's realizable at all. Maybe you have an idea or even a better problem solution!? I use ArcGIS with extensions.

  • An illustrated step-by-step solution is given on my Web pages at quantdec.com/SYSEN597/studies/flood/index.htm.
    – whuber
    Jan 30 '15 at 18:06
  • This is confusing, you are asking for water to raise up in your dry channels but not from the main stem (your first image) then you show that happening in your desired image, spilling out into the floodplain of the channel to the right of the main stem. To reach up to the limits of your side channel (that yellow bit) it would be spilling out of the main river into its floodplain,
    – Hornbydd
    Jan 30 '15 at 18:07
  • @Hornbydd, sorry for confusing you! My fault! At the first image the red colour describes the water expansion during the flooding - starting from the main stream (in blue). On the right the result i don't want. And I know that by raising the limit the water runs into the floodplain, but this is what i want to dismiss at the end. The area between the brighter blue lines matters to me (on the second image). I want a result were only the old/dry channels are getting flooded without flooding the floodplain. [at]whuber, thanks for the hint, I'll give it a try!
    – Reen
    Jan 31 '15 at 9:23
  • So you are asking to treat the old dry bank heights as "infinitely high" so no water ever spills out of them but you are showing flood progress up these channels? If I have understood you correctly then I would approach this as a network tracing problem as you are tracking along the network and not considering anything else.
    – Hornbydd
    Jan 31 '15 at 13:55
  • You may want to check with a geomorphologist if this is a sensible approach as rivers do tend to burst their banks when flooding.
    – Hornbydd
    Jan 31 '15 at 14:02