The ArcGIS Desktop Help leads to a wrong understanding about the order some basic hydrologic functions are supposed to be performed. It should be urgently revised. I’m new to GIS and tried to follow the help flowchart (“An overview of the Hydrology toolset / Deriving runoff characteristics”) to define a stream network.
1st problem: starting with a DEM, it shows [flow direction] as the first function. That’s right, but if the DEM has sinks we need to make it a depressionless DEM and, after that, perform [flow direction] AGAIN (what is not clear!)
2nd problem: if the [flow direction] output has sinks, we can use it as an input to the [sink] function in order to identify them. But this is only an optional step, especially if we want to fill the sinks without specifying a “z” limit. The [sink] is not a pre-requisite to run [fill], as the flowchart show us.
3rd problem: the input to run [fill] is not the output of sink, as the flowchart lead us to think. Neither is it the output of [flow direction], used in the previous action (performing [sink]). The input to [fill] is simply the original DEM.
All other functions needed in hydrologic analysis, shown in the flowchart (and some which are not shown), need, as inputs, the outputs of either or both: [flow direction] and [flow accumulation].
Then, a better understanding of the hydrologic analysis process using ArcGIS could be reached substituting this flowchart by two very simple ones:
a) Without sinks: DEM => [flow direction] => [flow accumulation] => Stream network defined; and
b) With sinks: DEM => [fill] => Depressionless DEM => [flow direction] => [flow accumulation] => Stream network defined.
The only thing that needs to be added is a clear explanation: “We start with the flowchart (a) and run the [flow direction]. If the output raster has any cells with a value which is not 1, 2, 4, 8, 16, 32, 64 or 128, this is a sink (a numerical error or a natural sink, like a pond or a lake). Then we need to go back and use the flowchart (b).
Just that simple! But it took me a long trial and error process to arrive at such conclusions, not to mention the extensive discussions in the hydro forum, which almost led me to install a complementary software (ArcHydro) which, actually, it’s not essential to the initial analysis I’m performing. I hope this thread can help the ESRI developers to improve the ArcGIS Help; or can help other beginners to save time when performing hydrological analysis.