This lab practices watershed analysis in multiple applications and methodologies. This lab is split into two parts, Delineation of Watersheds, and Finding Areas at risk of flooding in a Cloudburst. In the first section, watersheds of Adirondack Park in New York are delineated given a DEM, and a park boundary shapefile. In part two areas at risk of flooding in Gentofte, Denmark are found given a DEM and the buildings that are in those area are found. This data is run once in a simple model that only accounts for sinks found with the DEM file, then it is run a second time through a more complicated model that accounts not only for the sinks found with the DEM, but also the area of the watershed that the sinks are found in (resulting in more runoff in a strong rain event), and the volume of the sink. The second model is a stronger model for finding areas that are prone to flood risk, and for finding the buildings that are prone to flooding. The context of this analysis is that in recent years Denmark has fallen victim to a number of extreme rainfall events.
Methods and Results:
Part 1 Methods:
The Adirondack Park boundary file was downloaded from gis.ny.gov. The hydrology file for the area was found at cugir.mannlib.cornell.edu/index.jsp The DEM file used in this part was added from the add data drop down menu, then with the selection of Add Data from ArcGIS Online. The DEM is called 30 arc-second DEM of North America.
All data that was created within this project was stored in a new file geodatabase.
As all data that is worked with should be in the same geographic coordinate system, and the same projections should also be used, the hydrology and DEM layers were projected to the same coordinate system as the park boundary file. This was the (UTM) Zone 18N projection of NAD 1983 in meters. This projection was chosen because it is a very common local projection to use for areas in North America. Projections for the layers other than the DEM were made using the Project tool under Data Management Tools. A park boundary buffer was made using the Buffer tool under Analysis tools to use later for watershed creation (issues can arise if not allowing outside topography to affect the watershed analysis). Before projecting, the DEM was clipped to the area of interest, the buffered park boundary, with the box checked to use input features for clipping geometry. When the DEM was projected it was done by using the Project Raster tool under the raster subgroup of the Projections and Transformations folder in Data Management Tools. When projecting, the DEM was also resampled from 30 by 30 meter spatial resolution to 60 by 60 meter resolution. Streams were next clipped to the park boundary for display in final maps using the Analysis Tools contained clip tool.
Now the data was all ready and reformatted for watershed analysis. First, flow directions for each cell were created using the new DEM and the Flow Direction tool (Spatial Analyst Tools » Hydrology »
Flow Direction). The output of this operation, however, was not what was desired, as sinks had not been filled yet (the leveling of lakes and other sinks). A fill operation (Spatial Analyst Tools » Hydrology » Fill) was used with the DEM, and then the Flow direction tool was used again, in order to get the result that was desired. This flow direction feature class was then used with the Flow Accumulation tool (Spatial Analyst Tools » Hydrology » Flow Accumulation) in order to get an accumulation layer, a layer in which each cell has the value of the total uphill cells that flow into it. This layer showed a spectrum of steams. From this layer, a source raster was created. This layer was begun to be made with the Con tool (Spatial Analyst Tools » Conditional » Con). The parameter chosen for this tool was value > 50000, but 100,000 and 500,000 were also run, further limiting the streams allowed in this layer. The output of this process was then used in the Stream Link tool (Spatial Analyst Tools » Hydrology » Stream Link) along with the flow direction feature class in order to create the source file for the watersheds. Flow direction and source were finally selected for the inputs in the Watershed tool (Spatial Analyst Tools » Hydrology » Watershed) and the watershed delineations were created, clipped down to the size of the boundaries of the park, symbolized, and mapped.
Part 1 Results:
Part 1 yielded 109 watersheds, then after clipping to the boundaries of the park 96 unique watersheds remained. With a smaller spatial resolution such as 120 meters (larger cell size), this value would have been lower. With less cells there would have been smaller flow accumulations, and this smaller value would have precipitated down the chain of processing, resulting in less sources, and therefore an equal number less watersheds (each source resulted in one corresponding watershed). Also decreasing the number of watersheds delineated was the amount of streams remaining after running the Con tool. By increasing the threshold value when running this tool to 100,000, the number of watersheds dropped to 35 watersheds. By increasing this threshold to 500,000, the number of watersheds decreased to just 7.
The map below shows the delineated watersheds with a DEM spatial resolution of 60 meters, and a flow accumulation threshold of 50,000. This was the process that yielded 96 unique watersheds. As you can see, each watershed has a clear source stream.
Part 2 Methods and Results
Section A: Find bluespots and affected buildings
Data was downloaded from learn.arcgis.com/en/projects/find-areas-at-risk-of-flooding-in-a-cloudburst/. Instructions were followed from the same site.
The site explained the circumstances of the GIS analysis in the first lesson, then in the second lesson showed the student how to download the data and showed the student the extent of the data that was
handed over before any analysis. This included looking at metadata, projection and coordinate system, and other vital pieces of information about the data. In the third lesson analysis started. The lesson, "Find bluespots and affected buildings," had the student use the DEM and other data provided, along with the model provided, to run and find the bluespots of the the town of Gentofte, Denmark. The model fills the sinks that are less than the vertical accuracy of the LiDAR derived DEM, then finds all sinks. Using the minus tool, it removes the sinks that could be due to inaccuracy. The model then takes the buildings data and the bluespot data and performs various functions to format the data to get it ready for the final steps. This ends in the creation of a buildings and a bluespots feature layer. Finally the buildings that are within bluespots are selected by location, and then copied to a new feature class. This model finds the buildings that can be affected by flooding, but does not incorporate the size of the watersheds that water can run off into a bluespot from, nor does it investigate the volume of the bluespots. The geoprocessing output included 14,579 buildings that were touching or fully within bluespots. This value is 46.544 % of the 31,323 total buildings. The result of this analysis can be seen in the figure below:
Section B: Assess flood risk to buildings
The model that was used in this section did take into consideration the areas that would be most prone to flooding based on the size of watersheds (leading to more runoff into the contained bluespots) and the volume of the bluespots. After the model was run, a symbolization scheme was imported that was provided by ESRI that showed the level of risks in the map showing the output data of the model. This symbolization scheme included an offset so that the 40 mm of rainfall per hour that could be taken away by the sewer system (in perfect sewer performance) would be accounted for. The map below shows this classification scheme, with additional cartographic elements as suggested by ESRI.
The next figure shows a map output of the buildings with the most risk of flooding. These were the buildings in the areas predicted to flood by the model with only 40 or less mm of rain. Theses building were found with the attribute query "FillUp >= 0 AND FillUp <= 20" performed on the bluespots touching buildings layer, then with a spatial selection on the intersecting buildings with the selected bluespots. The map showing this area is shown below:
In the final map seen below, the watersheds can be seen overlaid upon the watersheds layer that was created when the model was run. As can be seen, some watershed lie directly over, and in the same shape as some watersheds. This is as expected.
Sources:
Sources:
- Instructions from Dr. Caitlin Curtis
- Data from gis.ny.gov
- Data from cugir.mannlib.cornell.edu/index.jsp
- Data from ArcGIS Online Community
- Instruction and data from ESRI: http://learn.arcgis.com/en/projects/find-areas-at-risk-of-flooding-in-a-cloudburst/
