¿Introducir error al quemar los streams?

Question

<\/HEAD>Estoy intentando modelar una cuenca hidrográfica a una resolución más fina que la que muestra mi estado en el NHD. Entiendo los procesos de Hydro pero me preguntaba por qué es aceptable quemar los streams en el DEM cuando al hacerlo no solo se altera la sección del stream del dem en 10M sino que también se suavizan las celdas circundantes. Parece que al hacerlo se introduciría error desde el principio. <\/SPAN> Después de modelar con y sin quemar el NHD en el DEM, claramente hay diferencias en las capas de stream resultantes. Los puntos de observación coinciden más con los resultados quemados. ¿Es este simplemente el compromiso? ¿Que el error total se reduce al introducir un error leve al inicio?<\/SPAN> Gracias<\/SPAN><\/BODY><\/HTML>

LornaMurison · Answer

If you burn the streams only in selected areas like at road crossings (like you are suggesting) or in particularly flat areas, where filling sinks eliminates the stream itself (like I may have to do) don't you then introduce sinks into the stream itself? Can that then be corrected by using fill sinks?

Thanks,

MarkBoucher · Answer

The goal of the DEM reconditioning using Spatial Analyst or Arc Hydro tools is to get the flow direction right. When I work on the stream burning step, I try to minimize the width of the burn by making the stream buffer only 2 cells wide. You may be able to make it just one cell wide, but I vaguely recall that not working well so I've stayed with 2. I work with a 10x10 dem. The walls have to be at least 2 cells wide or you will get "leakage" via the diagonal.

I also minimize where I put the streams to burn. If the rawdem is good for the flow direction, I don't put a line in it. I only put in lines at bridges and culverts. If I have a stream layer, I will buffer the road layer and use that to clip the stream layer so I only have short segments at the stream crossings. Of course I have to review the crossings to make sure the stream polyline is "correct".

So, does burning streams introduce error. Yes. Does the rawdem have errors already? Yes. If you are diligent enough you can make the rawdem better for flow direction determination and minimize the introduced error.