Fugro OCEANOR are often asked to reconstruct past wave and wind conditions in connection with marine incidents, accidents, court cases, sea trials and such like. Using our WorldWaves and World Wave Atlas data bases and modelling tools, we are able to reconstruct to the highest quality past weather conditions anywhere worldwide and at anytime over the past 50 years.
We show an example of how this is done through an example. We note that we have not been involved commercially in the example given which was chosen through a simple Internet search for the loss of a ship at sea caused by weather conditions.
The Alexandros T, a bulk carrier, sank about 450 km south of Port Elizabeth in South Africa on the night of 3rd May 2006 on its way from Brazil to China. A BBC report the day after reported that “…The ship ran into trouble due to very, very heavy seas”. Elsewhere it was reported that “MV Alexandros T began breaking up in heavy seas around 7pm on 3rd May”. The wreck site is given as about 38°S, 30°E.
In order to reconstruct the wave conditions on the night of 3rd May 2006, we first examine the WorldWaves model data for the day in question for the closest WorldWaves model grid point at 39°S, 30°E (Error! Reference source not found.). The time series for selected wave and wind parameters are shown in Figure 2. The data suggest that the significant wave height was between 3 and 4 metres on the 3rd and 4th May 2006, a relatively modest seastate (about average for the time of year). The winds were also low, decreasing from strong to moderate breeze during the 3rd. The traces of swell height, peak period and swell direction that a long period swell arrives about midday on the 3rd. The data would not in any way suggest that the MV Alexandros T would have experienced “very, very heavy seas”!
Figure 1 Overview map showing the WorldWaves model grid points and satellite passes during the incident.
Figure 2 Time series of Significant Wave Height, Wave Direction, Peak Wave Period, Wind Speed, Swell Height and Swell Direction at the WorldWaves grid point 39°S, 30°E.
Can we trust the model data? In our reconstruction, we try to answer this question. First, the WorldWaves model data are validated against long-term satellite altimeter data in the same location. The scatter plot in Figure 3 shows that the model data are on average unbiased against the satellite data (which are considered to be ground truth). The scatter we see is in part due to some events being slightly overestimated and some underestimated. For a 4m seastate, one can see that the model can be up to 1m in error (under- or overestimate). Based on this we can say that if the model suggests 4m as in the present case, the real seastate was between 3 and 5 m to a high degree of confidence.
We can be even more confident than this by examining the actual satellite measurements made in the area around the incident site on the days in question. It turns out that there were two passes of the GFO (Geosat Follow-on) satellite. First, GFO flies along Track 1 (see Error! Reference source not found.) on the 3rd May at about 2300 UTC and along Track 2 half a day later at 1300 UTC on the 4th. The satellite makes a measurement of significant wave height about each 6km along its ground track and, in this way, measures a profile of wave height off South Africa. In Figure 4, the two track profiles are plotted. On the 3rd at 2300 UTC, GFO shows a fairly steady significant wave height of around 3-4m between about 34 and 40°S.
On 4th at 1300 UTC the significant wave height is between 2.5 and 3.5m. In Table 1 below all model/satellite coincidences are compared for Tracks 1 and 2 on 3rd and 4th May 2006 (i.e., we have compared satellite data against different model points in the area of interest). There is no evidence that the model was underpredicting the waves on the day of the incident. We therefore conclude that the data shown in Figure 2 are a true representation of the actual sea conditions. The BBC report of very, very heavy seas is therefore not confirmed…
Figure 3 Long-term 1997 to 2006 validation of significant wave height (Hm0) from the WorldWaves model data at 39°S, 30°E against co-located satellite altimeter data. The red dots and line show the q-q regression fit to the data.
Figure 4 Along-track significant wave height trace against latitude for the two satellite passes on 3rd and 4th March 2006; the tracks are shown on the map in Error! Reference source not found..
Figure 5 Scatter plot of available model vs satellite significant wave heights for 3rd and 4th May 2006, confirming that the model performance was extremely good during the incident.
Table 1 Comparison of model estimates and satellite observations during the incident confirm the model accuracy.
Ref.1: BBC report after the loss of the Alexandros T: http://www.wrecksite.eu/docBrowser.aspx?737?1?1
Ref.2: Wreck details, cause given as “gale/storm” http://www.wrecksite.eu/wreck.aspx?132371