FMA Challenge 2: Difference between revisions
Content deleted Content added
No edit summary |
No edit summary |
||
Line 155:
The 15m simulation provided an even better calibration, and gives the best overall performance of all simulations carried out. The maximum flood extent layer is provided in the ftp download for this simulation.
In conclusion, we would
[[File:Critical_Embankment3.jpg|600px]]
Line 164:
==Challenges==
No unexpected “challenges” for Challenge 2, but it is worth noting the challenges that arise in calibration exercises. Sometimes there is an expectation that models should be able to reproduce recorded levels to within a very small tolerance. In these situations there needs to be a good understanding/appreciation by both the client and the modeler of the uncertainties and reasons for discrepancies; and the acceptable tolerances between the modeling and recorded observations accordingly set.
Line 188 ⟶ 189:
<li>Structures are often poorly represented due to lack of data. For example, in the aerial photograph below for Challenge 2 there appears to be some structures (which has two calibration points in the vicinity). However, with no details provided, the structure below can only be modeled as a simple opening in the embankment with no allowance for bridge decks, etc.</li>
[[File:Critical_Embankment7.png|600px]]
==Caveats==
===TUFLOW Fixed Grid Models===
In summary, the TUFLOW 15m No Embankments simulation provided the best calibration (the shape file for the maximum inundation area for this run is included in the ftp download).
===TUFLOW FV Flexible Mesh Model===
Given the short timeframe, selection and adoption of approach was based on minimizing preparation time. This approach could be interpreted as a “rapid solution” approach, providing preliminary advice for floodplain managers.
The cell sizes in the main river channels were chosen to allow between 5 and 10 cells across the channel width. As flow in the main channels is primarily in one direction, cells length were made longer than cell widths in order to optimize computational performance. Cell sizes on the floodplains were chosen to broadly represent overland flow characteristics, without capturing finer resolution features (such as roads, etc).
Differing friction values were tested (including a N = 0.022 in the main channels and rougher values on floodplains), however results with a constant N = 0.035 value provided the most accurate model predictions. N values at 0.035 are somewhat higher than anticipated (especially in the main channels), but considered acceptable.
It could be argued that a more detailed description should be made of hydraulically significant features on the floodplains (roads, etc) and also a more detailed description of overbank friction (according to land use). However, the calibration event occurred in 1964, which questions the validity of the adopted land use, terrain and (potentially) channel bathymetry – such uncertainties may overshadow any adoption of such details in the model application.
| |||