Lab Surface Runoff

Watersheds, also known as catchment areas or drainage basins are the basic unit in hydrological studies. They can be thought of as a funnel collecting all water within the area and channeling it into creeks, streams, and rivers driven by the topography and gravity. This overland flow of water coming from precipitation, snowmelt, and even groundwater is referred to as surface runoff and its analysis is fundamental in water resource management for water allocation and hydropower generation. Surface runoff is also a hydrological process involved in floods, pollution transfer, soil erosion, and mudslides, hence its continuous monitoring is critical to provide hazard warnings and to develop mitigation strategies. In this Story Map, I present the workflow to generate an automatic watershed delineation, a stream network and a distance-area diagram which serves as an indicator for runoff timings and interpretable as a simplified version of a unit hydrograph.

The streams and catchment layers in Salzburg were loaded and the multi resolution terrain layer was added as the digital elevation model (DEM) with a maximum spatial resolution of 1 meter in Austria. The selected study area is the Imlaubach catchment in Salzburg near the Kreuzbergmaut Hydroelectric Power Station (Fig 1).

Initially it is essential to fill the sinks in the DEM by adding a "thick layer of water" and then draining the excess water to create a continuous surface  (Planchon & Darboux, 2002) .

Secondly, the runoff flow direction is established using the D8 algorithm  (O’Callaghan & Mark, 1984) . This method determines the flow towards the cell with the steepest downslope but has the disadvantage that the flow direction is discretized into only one out of eight potential directions constrained by regular squared grid DEMs (Fig 2), where the matrix of points are equally spaced in orthogonal and diagonal directions thus resolving the flow direction too coarsely  (Tarboton, 1997) .

Thirdly, the accumulated weight of all cells is calculated to aggregate all the downslope flow as seen in Fig 3  (Jenson & Domingue, 1988) .

The next step is to automatically delineate the catchment boundary. This was done using the watershed tool, which requires a pour point and the flow direction raster obtained with the D8 algorithm. The pour point is derived from the accumulated flow layer by placing a feature class in the lowest elevation of the stream right before confluence occurs (also known as the outlet). Once the watershed raster is obtained, it is converted into a polygon.

Then, a stream network is created by reclassifying the cells from the accumulated flow raster using the reclass tool to find those having throughflow of at least 15% of the total catchment (Fig 4). Subsequently, the cells with the lowest accumulation are set to null and finally, the stream network is displayed using the stream to feature tool.

For the last step, the flow length is calculated to obtain a distance-area diagram of hypothetical rainfall and runoff events. Once the output raster is classified it provides the number of cells (area) as a function of distance from the outlet, referring to the amount of water reaching the outflow point.

To automate the previously described procedure, the ArcGIS Pro ModelBuilder environment was used to create the geoprocessing workflow (Fig 5).

The generated automatic watershed boundary differs to the manually delineated. The reason is that the latter was created using 1:50000 topo sheets (500 m resolution), and the former was processed with a finer spatial resolution of 10 m. The automatic watershed has a total area of 18.35 km² whereas the total area in the manual watershed is 18.04 km² (Fig 6).

In Fig 7, the accumulated flow is shown in comparison to the stream feature layer accounting for at least 15% of the throughflow in the catchment. 

The flow length map within the watershed and the contour lines are shown in Fig 8. This map show increasing distances upstream from the pour point divided in 8 classes. In addition, the distance-area diagram is displayed in Fig 9, and can be interpreted as the amount of water that would flow through the pour point from these varying distances.

There are two critical zones in the catchment area where deforestation should not occur to help temper runoff peak flows. Those areas are between 11,000 - 13,000 meters and between 7,000 - 8,000 meters from the pour point.

To end this Story Map, a GIF comparing two flow accumulation layers is presented in Fig 10. The point of adding this animation is to highlight the importance of filling the sinks in the DEM. It is clear that the DEM without prior filling generates an inaccurate and discontinuous accumulation of surface runoff.

Jenson, S. K., & Domingue, O. (1988). Extracting Topographic Structure from Digital Elevation Data for Geographic Information System Analysis. Photogrammetric Engineering, 8.

O’Callaghan, J. F., & Mark, D. M. (1984). The extraction of drainage networks from digital elevation data. Computer Vision, Graphics, and Image Processing, 28(3), 323–344. https://doi.org/10.1016/S0734-189X(84)80011-0

Planchon, O., & Darboux, F. (2002). A fast, simple, and versatile algorithm to fill the depressions of digital elevation models. Catena, 46(2), 159–176. https://doi.org/10.1016/S0341-8162(01)00164-3

Tarboton, D. G. (1997). A new method for the determination of flow directions and upslope areas in grid digital elevation models. Water Resources Research, 33(2), 309–319. https://doi.org/10.1029/96WR03137