1	Validation of Tsunami and Dambreak Hazard Models Modelling SIG has adopted idea of using reference tests to establish the validity of modelling techniques Reference tests should preferably be based on field or laboratory measurements Where field data is sparse and of poor quality (as with tsunamis) need to include analytical models These generally apply only in simplified situations, but these can often be matched in numerical models Where field and/or laboratory data available, what about experimental errors? Experiments need authentication to identify possible measurement errors Authentication can be established by eliminating obvious outliers, and by checks for mass conservation (e.g. masses under flow curves should balance) A quick way of authentication is to fit a good numerical model: if there is agreement, then authentication and validation both happen together If there is no agreement, then back to square 1! Illustrate by two examples: An analytical tsunami validation and a laboratory dambreak validation Propose for adoption by national standards authority for compliance testing
2	Tsunami Model Validation Solitary wave analysis is known to describe the propagation of gravity waves (such as tsunamis) without a change of shape. Can therefore run a model wave over some distance and check for change in shape Can also check that wave speed is modelled correctly This applies only in water of constant depth and negligible friction These conditions are easy to arrange numerically by ensuring the constant wave depth is much larger than the wave height The downstream boundary needs to be far enough away that no reflected waves interfere with the test. Then only need a boundary condition describing the solitary wave A 6m wave was chosen as this will reach about 10m against a fully reflecting (steep) coastline This is a typical height for design investigations 1000m depth chosen as this gives a wave speed of approximately 100m/s
3	Solitary Wave See Ippen “Estuary and Coastline Hydrodynamics” p.122 η = wave height H = peak height at x=Ct h = normal water depth C = wave speed ≈ [g(H+h)]^1/2 Table at right shows η(t) for x=0, H=6m, h=1000m Note: Wave tapered off from 0.133m at 60s to 0.0m at 0s to avoid an infinite wave length at infinitesimal heights.
4	Example: AULOS Results For Accuracy Bias B = 7 (highest setting) the wave propagates over 60km without change in shape in a time between 600s (depth=1019m) and 605s (depth=1003m) For Accuracy Bias B = 0 (default setting) the wave propagates at the same speed but with some numerical diffusion. This reduces the wave height. Development of other standard tests for validation of modelling waves on sloping beaches now proposed for NZWERF research funding
5	Dambreak Model Validation: Experimental Layout
6	WES Test Flume (1960) US Army Engineer WES (Waterways Experimental Station), Vicksburg, Mississippi Rectangular Flume 400ft (122m) long, 4ft (1.22m) wide Slope = 0.005 Manning n = 0.009 Dam removal took 0.01-0.03s Surface velocities measured by time lapse photography of floating confetti Depths measured by timed photographs
7	Selection of Test Data WES tested various combinations of dam breach openings and initial downstream base flows Test 5.1 was chosen by Sanders (2001) for validation testing of his solution. This involved a full depth (0.3048m) rectangular breach of width 0.4 ft (0.122m). The WES report does not picture the model dam for Test 5.1, but it was presumably similar to that for Test 6.1 (above), which differed only in having a narrower breach of width 0.24 ft (0.073m). This shows the slot sides were defined by sharp edges, not rounded
8	Upstream Reservoir Levels Levels at STA188 (57.3m from upstream end) Time(s) Level(m) 0 0.6096 2 0.6066 3 0.5944 4 0.5971 5 0.5944 7 0.5913 10 0.5913 20 0.5913 30 0.5913 60 0.5852 120 0.5547 150 0.5425 180 0.5304
9	Example: AULOS Results at STA188 Plotted STA188 results as per previous table. This suggests that outflows are accurately matched. The model used the energy equation at the breach, with a textbook discharge coefficient C_B=0.9 for sharp edges (Henderson, 1966)
10	Downstream Observations These were made at three points: STA225, STA280, STA350 respectively 225, 280 and 350 ft (68.6m, 85.3m and 106.7m) from upstream end of flume. Flow values calculated from mean velocity x depth, but only surface velocities available Mean velocity V assumed = kV_s Where V_s is surface velocity k was assumed to be 0.80 by WES, but this is too low except at STA225. Analytically, k=0.88 a better estimate for smooth channels.
11	Example: AULOS Depth Results The depth validation is generally good, with excellent matching of wave front arrival times
12	Example: AULOS Flow Results The validation again gives an excellent match with dambreak wave arrival times, but the flow match is only within about 10% at STA 225 (k too high) and STA280 (k too low). The close match at STA350 and with reservoir drawdown upstream (STA188) suggests main problem with evaluating k.
13	Conclusions Reference tests are an excellent way to establish the validity of modelling techniques They also indicate the effect of using various model tuning parameters Where possible, reference tests should be based on field or laboratory measurements Analytical models can partly fill the gap where field data is sparse or of poor quality (as with tsunamis) For tsunamis, a solitary wave forms a good analytical test for the ability of a model to reproduce wave behaviour in deep water. Further work on runup on sloping beaches will (hopefully) follow soon. For dambreak waves, the 1960 WES laboratory experiments are a useful source of wave behaviour under a range of conditions. Experimental scatter in level measurements can be averaged out The evaluation of mean velocity from the measured surface velocity required appeal to mass conservation plus an analytical shear model The result adjustments have been authenticated partly by successful fitting with a model which conserves mass, momentum and energy All input data required for sample validation of numerical tsunami and dambreak models is documented in this presentation These tests are therefore recommended as suitable for adoption by a national standards authority for compliance testing.