Surface Hydrology

Geospatial Technology has become a topic of interest in multiple domains to execute the work efficiently. Especially in Meteorological Science and Water & Engineering Management. It has been widely used by experts in their modelling process. As a Geospatial enthusiastic, I attempted to understand Surface Hydrological Process. 

In this task, I intend to do hydrological modelling using a geospatial tool and prepare a hydrologically consistent delineation understanding the steps involved in this process. I believe that with these tools I will be able to identify peak flow at the outflow points.

For my study, I chose a catchment area around Weissenbach which flow into the Salzach in the Salzburg. The basis for the analysis was a DEM with a 10m Resolution fed into Hydrology toolset of ArcGIS Pro.

A model builder as shown in Fig 2 was generated to execute the work efficiently rather than running each step separately. I want my model to behave like a tool so that in future I can browse it to any project and use it. For this, I defined the input and output as user-defined parameters. Also, allowing a user to make changes in the environment settings of the predefined tool they were also set as a parameter by assigning them initially to a variable.

As the name suggested, Flow direction tool produces a raster that gives an idea about the direction in which water will flow on every bit of the surface. In this process, D8 algorithm used that will calculate the flow direction from each cell to its steepest downslope neighbour, assigning then values 1, 2, 4, 8, 16, 64 or 128 as shown in Fig 3. Each value is coded as 1: East; 2: South-East; 4: South; 8: South-West; 16: West; 32: North-West; 64: North; 128: North-East.

[Ref:  https://pro.arcgis.com/en/pro-app/latest/tool-reference/spatial-analyst/how-flow-direction-works.htm ]

Sinkholes occur naturally in certain types of landscapes, such as karst (limestone) where the rock is soluble. Outside of these areas, we will find sinks in our DEM data which are simple errors in the data, a place where the scale of the data does not adequately represent an existing drainage channel or some other sources. Generally, geospatial tools do not deal easily with sinks, whether natural or an error in the data. We will need to remove the sinks before we look at flow direction and flow accumulation. Sink filling is not for correcting of data but is filling to make the low point higher and get a hydrologically consistent data.

When compared with the flow direction that is created based on DEM without filled sink, additional values are seen in the legend. Not much difference in the map view will be noticed but variable in the legend will suggest that widely scattered pixel are incorrect and they donot aggregate at the common area. 

Flow accumulation gives you the total area (number of cells times the area of each cell) that is upslope and/or upstream of any given cell. Cells that lie on drainage networks have substantially higher values than cells on hill slopes. It is computed, for each point in the terrain that showed the mainstream network showing the flows of water in an individual pixel downhill. It will give an idea of where to create a pour point as the outlet point of the water flow. Flow direction is given as an input in this process weight is left blank that means it will be assigned as 1. In the map higher the value darker are the lines. 

[Ref:  http://resources.arcgis.com/en/help/main/10.2/index.html#/How_Flow_Accumulation_works/009z00000062000000/ ]

The map in the left is the flow accumulation raster obtained when DEM without filling the sink was used to generate flow direction whereas the map on the right is after filling the sink. It clearly shows, that when sinks are not filled then the stream network was not continuous with several breaks due to an artefact.

It was observed that that the flow accumulation initially shows only the mainstream but, when we change its visualization based on different ways of classification, it will show more defined stream networks as shown in Fig... Here, classification is done with geometric interval whereas previously it was Natural Break Classification. Hence, it depends on the depth of visualization required at the moment. [Noted: Here sinks were filled.]

A watershed is an area of land that drains rain water or snow into one location such as a stream, lake or wetland. Watersheds are important because the surface water features and stormwater runoff within a watershed ultimately drain to other bodies of water. It is essential to consider these downstream impacts when performing water management actions. 

To generate the watershed, flow direction along with the pourpoint is taken as input where pourpoint is the outcome drainage of an entire watershed. It is manually created based on the highest point in the flow accumulation. Snap Pour Points tools available in the Hydrology tool is used so that the point is right over the pixel.

There are clear differences to the nominal catchment outlines shaded as brown. The generated watershed does not coincide with the catchment boundary because of the lower resolution of the DEM used in this process. Furthermore, the generated Flow Accumulation raster has the highest flow at the pourpoint shown on the map but if we observe the existing watershed boundary, the suitable location for the pourpoint could be a bit more towards the Salzach. Because of this, some of the areas near Salzach has been omitted. Moreover, some area on top of the hills are also omitted may be due to the flow direction at the top of the mountain was not appropriately calculated due to clipped DEM within a region.

Previously generated map of Flow Accumulation gives a visualization of only those pixels that have higher values but if anyone shows an interest in generating a stream network having of at least certain percentage of the total catchment then a simple conditional raster calculation is done i.e. output = con(flow_accumulation_raster (operator) area_of_AOI). This will generate a stream that runs out of a defined number of pixels in the condition. 

In this case, I am interested in identifying a stream having a through-flow of at least 15% of the total catchment. For this I followed the following steps:

• For calculating the total area of the catchment: When the raster is converted to feature I created a new column called area and calculated its geometry for an area in km². So I have a catchment area of 16,15 km² which is equivalent to (10*10*16.15) 161500 pixels.
• Next, as I was interested only in the streams that have 15% of the total pixels flowing in them =24225 pixels.
• Using the raster calculator to filter out only the stream layer to all values above 24225, a new raster is created. The expression is: (con(Flow_Accumulation) >= 24225, 1, 0).The resulting layer contains only the main streams which cover an area of more than 15% of the catchment area. There were 1242 pixels having flowed from at least 15% of the catchment area.

For identifying the most critical zones in the catchment we generate a raster that shows the flow length to the outlet point. Calculation of flow length will show the distance of each cell along the flow path to the pourpoint. 

With the input of flow direction in the downstream, we can see the result as shown in the map. It is a time indicator of how long the water flow needs to reach to the pour point. 

The histogram for the flow length layer shows zones within a certain distance in x-axis from the pour point within the catchment, corresponding to the amount of water reaching the outflow point at approximately the same location. 

In the unit hydrograph, an increase in values can be seen. However, this is not linear, but occurs in two parts, because there is no increase between the distances of approx. 1095 to 3000 meter, the trend there is rather decreasing. Afterwards, a strong increase can be seen from 7000m to 1000m.

For the critical points in the area, we can refer to the highest values and which is located in the first catchment with the HZB CODE: 2 8272239 1 0 0 0 0 having an area 10.60 km2. When looking at this area in the Open Street Map, we can see that it is forest and barren grassland. Thus the type of land use in the critical area corresponds to a forest area, which is suitable during heavy rainfall, the trees will prevent the flooding and landslide. The reason for this is that the precipitation can penetrate through the soil and the trees will prevent it from runoff.

In addition to this, when determining the flow length, the user can also provide raster to give weight to be considered when carrying out the process. Hence, since the river flows through the forest, it can be assumed that protection might already exist.

Cost Path tool is used to identify the flow lines originating from a point defined in the map as a drop point to the pourpoint. It shows the best ways for the water runoff down the hill and ultimately reaching the highest flow accumulation point. In computing the flow path, I used a manually created Drop point as an input feature point, FlowLength as 'Input cost distance raster' and FlowDirection for 'Input cost backlink raster'. 

Hence, I have been able to meet my objective and have a good understanding of using geospatial tools for hydrological modelling. I was able to create a tool, that would ask for input for different parameters and hence can calculate the watershed. Also, Landuse should be well managed in order to prevent from surface runoff but in my study due to required landcover risk of runoff was minimum.