Using the tool "minus" during a flow direction analysis

[MagnusAlcazar]

I'm doing a lab and I'm trying to understand exactly why I'm using the tool "minus".

I have a raster which I'm calling "heights". I also have two rasters with water surfaces and watercourse. I have changed the values in the water rasters so that water pixels has the value 10 and noData has the value 0.

I have then used the tool "plus" to put these two rasters together into a joint raster which I'm calling "water" plain and simple.

Now, I am supposed to use the tool "Minus" to put in the water raster together with the raster heights, I can't understand why I need to do that.

My first thought was that the intention is to make sure that it's only the watersurfaces that shows up in my result later on and not land surfaces, but that doesn't seem right.

Why am I using minus in this scenario? It's to lower the watercourses I guess but why is that needed?

The final goal of the lab is to find out from what area of the map a poison is coming from. To investigate this I have made an analysis of the watersurfaces in the area based on the water surfaces data and the flow direction.

[DanPatterson_Retired]

your water is either 0 or 10.

Elevation - water

where water is 0, you will end up with elevation

where water is 10, you will end up with elevation -10

    In the Elevation raster, the water/river etc probably has an actual elevation, subtracting 10 from it would 'burn' the water into the elevation file making it lower than the surroundings dry land, making watercourses a preferred destination for overland flow

There a loads of links, but this is pretty good from a visual perspective

http://proceedings.esri.com/library/userconf/proc14/tech-workshops/tw_335.pdf 

[MagnusAlcazar]

Thank you for your answer.

I have looked throught your link, but one thing I don't get is what would actually be the difference if i did not use the "burning in" thing? The height raster are already "perfect", why would I want to lower the watercourses?

Also, is there another way to do this instead of using the "minus" tool?

[DanPatterson_Retired]

If you had a perfectly formed drainage basin, maybe nothing, but when you use your 'flow' derived parameters, don't be surprised when you get to determine the 'streams' they don't lie where the actual streams are.  burning, pretty well increase the possibility that derived streams overlay actual streams.

[DanPatterson_Retired]

PS..  instead of minus you have assigned 0 and -10 to the streams, then used the Plus tool , it needs a burn.  There are alternatives in the academic literature, but understand this methodology before proceeding to other possibilities

[MagnusAlcazar]

Yes, I'm going to try and understand this. I just want to understand the entire concept so I would like to go on and understand other ways after I have figured out this. Can you give me an example in another methodology to do it?

[DanPatterson_Retired]

The pictures in the link I sent show what happens when the stream is burned and incised into the DEM, If you understand the concepts of flow accumulation and flow direction, then it will become apparent(ish) that as the kernel moves over the streams, they will become preferential locations for water to flow into.

The other methodologies are more esoteric 

[MagnusAlcazar]

When answering the question "why do you have to use minus in this scenario " What I answered was basically 

"after I have run the layer heights minus the layer water the watercourses will be lower than they really are in comparison to the surrounding surface. This happens because the pixels in the raster water i have the water values at 10 everything else at 0 (with reclassify). The height data will then get a value that's -10 lower than all the water courses. I'm doing this to get elevation in all the watercourses and in the modell the water will prefer to travel trough the water courses that's within the flow direction area. If I don't do this then some water courses may be without elevation and /or streams will be in the wrong places, even if I have a perfect height model, compared to where the actual streams are. So, I'm using this method to increase the chance that the watercourses are where they should be"

This was dead wrong. The answer I received that it will very well work with a perfect height model and that the height model should do just fine for analyzing. So the answer was wrong, that's not why I was using minus.

I'm totally lost here. No idea why of would use minus if that's not the answer. Any ideas?