How to setup uncertainty analysis
    • 12 Aug 2022
    • 4 Minutes to read

    How to setup uncertainty analysis


    Article summary

    You can launch UMap from the Post-Processing Tools section of the Flood Modeller toolbox. An initial window will be displayed prompting you to specify the required input files, as shown below:

    AdditionalCalculationToolsUMapimagesimage002.png

    Initial inputs required here are:

    • Shapefile of the flood outline for which the location uncertainty is to be calculated. Marked as ‘Lower Return Period Shape Extent’ in interface.

    • Another shapefile flood outline, typically from another design flow, produced from the same model and mapping method. The exact nature of this outline is not important, although it should lie outside (i.e. represent a larger flood extent) the outline for which uncertainty is required. This is used to estimate the sensitivity of shoreline location to water level. For example, if uncertainty for the 100 year outline is required, the 200 or 1000 year outline can be used. Marked as ‘Higher Return Period Shape Extent’ in interface.

    • Digital Terrain Model (DTM), in ESRI ascii raster format (see http://en.wikipedia.org/wiki/ESRI_grid for details). Marked as ‘Base DTM’ in interface.

    These files can be specified using the adjacent browse buttons provided. Alternatively each item has a drop-down list that will be populated with compatible GIS files already loaded in your Flood Modeller map view.

    Important note; the shape files can only be polyline or polygon type shape files. They cannot be polylineZ or polygonZ type. When you run the UMap analysis the shape files cannot be opened elsewhere (including in Flood Modeller) otherwise the Umap tool will stop working.

    After selecting these files and clicking OK you will proceed to the main UMap interface as shown below:

    AdditionalCalculationToolsUMapimagesimage004.png

    The previously specified files will be loaded in this interface. Additional inputs then required here to fully specify your uncertainty analysis are:

    • Water level uncertainty (see below). Values go under section ‘Uncertainty’ in interface.

    • The flood outline uncertainty discretisation levels (Shoreline Uncertainty). These control how UMap splits the outline into lengths of different uncertainty. Three classes are produced, and these are specified by giving the two values dividing these classes. For example, if set to 50m and 100m, UMap will output polylines with uncertainty <50m, 50m-100m and >100m. There are two ways to determine these parameters:

      • By trial and error to get approximately equal numbers of lines in each class. The numbers of shapefile points in each class are displayed by UMap.

      • By determining low, medium and high classes of uncertainty with reference to the how the uncertainty mapping output will be used. Typical values might be <30m (high accuracy), 30-50m (medium accuracy) and >50m (low accuracy).

    The name and location of the output file need to be specified as well.

    Water level uncertainty

    You are allowed to specify water level uncertainty using one of the three following methods:

    1. Single value : Enter a fixed single value for the whole region. This can be derived by the user, for example from model sensitivity runs.

    2. Scoring : This method calculate the level uncertainty value using the following parameters:

      • Hydrology - Index Flood

        Possible score:

        AdditionalCalculationToolsUMapimagesimage006.jpg

      • Hydraulic Complexity

        AdditionalCalculationToolsUMapimagesimage007.jpg

      • Peak flow for flood return period

        Takes positive value in m/s.

      Based on these parameters, the code estimates the water level uncertainty for the whole region and uses this for the calculation.

    3. Modelled flows :

      The above two methods limit the input to a fixed water level uncertainty for the whole flooded region, which may not apply. A common example is where a flood outline represents flooding from a main channel and a small tributary, which may have very different flows.

      Instead, geo-referenced 1D models can be used in the scoring method above to allow different peak flow values for different reaches of the flood outline.

      Steps for creating dynamic flow input:

      Step 1 : Create a cross-sections shape file:

      • Load the sections from your 1D network into the map view either by using the Add GIS Data button to select the 1D dat file or, if your network is already present in your current Flood Modeller project, simply drag this onto the map view.

      • When the network file appears in the map view you can select it in the Layers Panel, right-click and select the export to shape file option (this tool is also available in the Toolbox - Additional Model Build Tools > 1D Cross Sections > Export Cross Sections as Shapefile).

      Step 2 : Create a flow data csv file:

      • Load your 1D network into your current Flood Modeller project. Make sure the results are available for the model run you want the flows to be extracted from.

      • From Results tab, select the ‘Tabular CSV’ tool. This should start up in a new window with your active 1D network automatically loaded, as shown below:

        AdditionalCalculationToolsUMapimagesimage009.png

      • In the above tool; set Output to ‘Maxima and Minima’ and Output Variable to ‘Flow’. Once ready, press ‘Run’.

    The csv file generated would be the second flow input needed by UMap. By co-ordinating the cross-section location from the shape file with the maximum flow seen at that cross-section (from CSV) it will be able to create a spatially varying flow input to calculate water level uncertainty.


    Was this article helpful?

    Changing your password will log you out immediately. Use the new password to log back in.
    First name must have atleast 2 characters. Numbers and special characters are not allowed.
    Last name must have atleast 1 characters. Numbers and special characters are not allowed.
    Enter a valid email
    Enter a valid password
    Your profile has been successfully updated.
    ESC

    Eddy AI, facilitating knowledge discovery through conversational intelligence