Problema con las líneas de flujo en ArcMap

Question

<\/HEAD>¡Hola a la comunidad, George aquí!<\/P> Como parte de mi doctorado tengo que mapear los ríos del Norte del Peloponeso, Grecia. He descargado dos DEMs de Nasa Reverb, que conecté con "mosaic to new raster". He comenzado el proceso habitual de "fill", "flow direction" y "flow accumulation" para poder analizar los ríos del área. Pero desafortunadamente estoy enfrentando dos problemas.<\/P> El primero es que los resultados que obtengo de la herramienta Flow Direction son incorrectos y no puedo encontrar la solución, sin importar lo que intente. Como pueden ver en la primera captura de pantalla, hay algunas líneas de flujo horizontales que no tienen sentido. Intenté editar los valores (traté de hacer que arcmap excluyera valores menores que 1) del raster original y luego usar "fill" pero eso no evita que la herramienta "fill" ignore estas áreas (tercera captura).<\/P> El segundo problema (que parece estar conectado con el primero) es que decidí pasar al siguiente paso y usar "Flow Accumulation" de todos modos ya que algunas áreas parecían estar bien y ver qué pasa.<\/P> Como pueden ver en la segunda captura de pantalla las líneas de acumulación de flujo siguen la forma de la línea costera y luego se extienden hacia el mar. Hasta donde entiendo, estas líneas de acumulación de flujo siguen a las de dirección de flujo por lo que el problema se crea cuando uso la herramienta "flow direction".<\/P> ¿Podrían ayudarme por favor? Soy bastante nuevo en esto así que disculpen mi ignorancia.<\/P> Si es un problema con los datos originales y no se puede corregir el problema, ¿hay alguna forma de solucionarlo?<\/P> <\/P> ¡Gracias!<\/P> <\/P> <\/P> <\/P> <\/P> <\/P> <\/P> <\/P><\/BODY><\/HTML>

MarkBoucher · Answer

Georgios,

I found the area you are working in using Google Maps (see attached image). The area where you have the strange lines is in the water (Gulf of Corinth). If the elevation in your DEM is level in the water area, the flow direction would be somewhat random. Since you are dealing with a coastal area, you will have some unique issues to deal with. However, I don't see anything wrong with the results. There was some recent discussion on coastal drainages. Here is the link to that includes that discussion: Re: Catchment vs watershed.​

You are right that the flow accumulation is based on the flow direction. If the flow direction is wrong (has errors) the rest of the analysis will have errors.

Since you have creeks, rivers, and streams that have to cross under roads on their way to the sea, I suspect you will want to create an agreestream layer and "burn" the streams into the rawdem at the road crossings so the flow direction will be correct at those locations. It looks like the topography is steep enough that for the most part you won't need to burn in streams anywhere else except in very flat place or where man-made drainage systems control the flow direction instead of the topography.

So, basically you are not doing anything wrong to get the "horizontal flow lines". A possible way to get rid of them would be to use the Con() function to make every raster grid below a certain elevation (i.e. in the water) "Null". I'd have to do a bit or research myself to give you direction on how to do that. I'm sure there is a tool in Spatial Analyst that would do work. This link might show you how. ArcGIS Desktop Help 9.3 - Setting values to NoData: SetNull.​ Once the raster has null (or "no data") values in the water area, it is possible that the flow direction grid will not be defined there and so you won't get flow accumulation there either.

The flow direction grid is only a foundational layer and will not show in your final analysis where I assume you will define watersheds. You might simply move on and establish your analysis process for a small area until you settle on a process you are happy with. Then perform that process on the whole dataset.  I use Model Builder in this way. I allows me to build and edit the process and they reapply it on different datasets.

Best,

Mark

GabrielUpchurch1 · Answer

Hi George,

It seems to be something related to the data but that doesn't necessarily mean the data is bad.  It may just take some preprocessing to get it to work properly.  I have never observed this behavior before but let's see what we can figure out.

For starters, what specifically do you mean by:  I tried editing the values (tried to make arcmap exclude values lower than 1) of the original raster  My first thought would be to eliminate the "sea" area from the data so I am curious what method you used when attempting this.