I am trying to create flow accumulation grid in ArcMap using Arc Hydro. I have the DEM (25mX25m) and the stream network shapefile.

I have done the following

1)Resampled the DEM to 5m;

2)Burnt the stream network in the DEM using DEM reconditioning;

3)Filled the sinks;

4)Created flow direction grid; and

5)Created flow accumulation grid.

The flow accumulation grid I have created was matching well with the reality in most of the catchment. However, in the middle part of the catchment, two streams are connected while they are not connected in reality. In addition, the catchment outlet created by Arc Hydro was not matching with the reality as illustrated below. enter image description here

Any recommendations how can I make the flow accumulation grid created by Arc Hydro match the reality would be appreciated?

Thanks in advance for your help and time.

  • 2
    Much of the strange behavior may be artifacts of your resampling procedure: exactly how did you resample the DEM?
    – whuber
    Commented Sep 8, 2013 at 16:48
  • dropbox.com/s/fpdyphno60tve4s/Resample.jpg
    – shiny
    Commented Sep 18, 2013 at 1:49
  • 2
    I thought so: don't use nearest-neighbor to resample a DEM. It turns it into a blocky mess. Your red lines reflect that. Always use bilinear or cubic convolution resampling. (Nearest neighbor is almost never appropriate except for categorical rasters.)
    – whuber
    Commented Sep 18, 2013 at 13:40
  • Flat topographies may cause ArcHydro to give straight segments. FillSinksPlus overcome this limitation.
    – user65122
    Commented Jan 12, 2016 at 18:58
  • Welcome to GIS SE, thanks for your contribution. It would help to elaborate on your answer. Have a look at the gis.stackexchange.com/tour tour page for more tips.This seems more like a comment than an answer Commented Jan 12, 2016 at 19:32

5 Answers 5


Your stream network is only ever going to be as good as your DEM is. If there are issues in the DEM, the results may be less than optimal.

Besides that, the methods for sink / pit removal in ArcGIS seem to be less than optimal in specific cases (e.g. relatively flat wide areas in the DEM). Stephen Jackson of the University of Texas, Center for Research in Water Resources, developed an alternative approach and ArcGIS extension that may give more accurate results.

Have a look here:

Optimized Pit Removal tool by Stephen Jackson.

And see this ESRI blog post: http://blogs.esri.com/esri/arcgis/2013/03/05/optimized-tool-for-dem-pit-removal-now-available/

  • Many thanks for your answer. I think the Optimized Pit Removal Tool is what I'm searching for. I couldn't install it as I'm working on 32-bit operating system
    – shiny
    Commented Sep 18, 2013 at 1:33

Topography is not the only factor that controls stream location. Steams can meander and cut into higher ground. This is particularly the can where the ground is made of alluvial sediments that are easily remobilised.

I should also say that SRTM data does not always locate the lowest ground due to its spatial resolutions and slope determination issues inherent in the technology. In short, your DEM may be out and therefore your predicted stream may not agree with reality. I have experienced both the above issues.

  • 1
    This is an insightful and informative comment, Andrew. However, it falls short of answering the question. Would you be able to expand on it with some recommendations as requested in the question?
    – whuber
    Commented Sep 20, 2013 at 15:53

Burning the hydro network into the DEM is not ideal. It alters the DEM in subtle ways & does not create a proper hydrologically enforced DEM. Burning is basically enforcing a downward elevation change in the DEM to route flow downstream. Certainly, the resampling will have an effect depending on how it was done, I'm unclear why 5m? A mention of the DEM data source data type, ie integer or float would help.

I would recommend using the ESRI Topo to raster tool to create your DEM. It requires you going back to source data such as contour/spot heights & the river network. ESRI's implementation is based on an older version of ANUDEM. This software is used exclusively for creating hydrological enforced DEM's for surface water analysis. When it constructs the DEM, it factors in the river network & how it must flow downstream. Derivatives created from the DEM (flow direction/flow accumulation) will honour the stream network.

  • Thanks for your answer. I have selected 5m to match other datasets as I'll be using all of them later in raster calculator. Unfortunately, I don't have an access to the DEM source data
    – shiny
    Commented Sep 18, 2013 at 1:32

In general, the methods for solution implemented through two main steps: (i) calculating a flow direction matrix (8d) from elevation data and then , (ii) measuring the area that contributes to each DEM cell and delineating the channel network based on an empirical threshold value that should be determined according to landscape characteristics.

You can also use integration of image precessing of satellite imagery.


You may want to try using FILL SINKS PLUS instead of the conventional FILL SINKS operation in ArcHydro. FILL SINKS PLUS was developed to overcome the limitations that FILL SINK has for relatively flat topographies. You can download this tool freely from http://www.lago-consulting.com/fill_sinks_plus.html. You can ask for support or send comments to [email protected] or [email protected].

  • 5
    Welcome to our site! Because we like answers to be self-contained, could you please add a few words to explain what limitations you are referring to and how this tool overcomes them? We also request that people promoting their own tools clearly identify their affiliation with them, especially when such promotion consists of generic assertions like this answer. When that identification is not in your user profile, please make it explicit in your posts. Otherwise readers who are jaded by spammers may have difficulty distinguishing your efforts to help from more overtly commercial pitches.
    – whuber
    Commented Oct 28, 2013 at 22:24

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.