major cause, the alternative explanation will be obvious fire, sinking, etc.. We also note that adopting a probability measure at a higher mortality level does not imply acceptance of
greater risk. A single voyage will have different probability of 1 and 5 mortalities, but both will be a snapshot of the same risk profile. We note that the adoption of risk standards
is not the role of this report, neither do we comment on the variation of risk standard with mortality level.
The calculation of probability for a given mortality level in one stocking entry is more straightforward than that for expected mortality. The drawback is that combined results, to
give a voyage average across different lines, are not necessarily meaningful. Consequently, these figures are given only for each closed deck stocking entry and not for the voyage as a
whole. To find the probability of exceeding 5 mortality, the cumulative distribution of animal
response is first used to find the wet bulb temperature corresponding to 5 mortality. This wet bulb temperature is then compared to the cumulative probability curve for wet bulb
temperature on the particular deck to find the probability of wet bulb temperature exceeding the 5 mortality value. As before, the wet bulb probability on the deck is taken as the
ambient wet bulb probability shifted along the wet bulb scale by the deck wet bulb temperature rise.
6.2.3 Duration of Exposure
An early ambition for the statistical assessment was to allow, in the estimation of risk, for duration of exposure in a particular zone. This would have worked by adjusting the beta
distributions of animals such that they become more susceptible to heat following some exposure, and to carry a progressive risk calculation along the voyage route. Several
problems emerged with this approach. The largest problem is that, statistically, the weather in adjacent zones is strongly correlated and the weather, particularly wet bulb temperature is
very strongly auto-correlated over time. This means that the probabilities of wet bulb temperature on successive days are not independent of each other. A far more
sophisticated model of the weather involving comparison of weather time scales and ocean zone transition time scales would be required. The statistics then would most probably
require a Monte-Carlo type simulation for each stocking entry as it was completed, requiring significant computing. In addition to the difficulty of implementation, there are very real limits
on the benefits which may accrue from this approach. In particular, with heat at extreme levels, risk increases with duration, while heat at lower levels may generate some level of
acclimatisation and protect against a subsequent, more severe, episode. That is; it is by no means clear how the animal parameters should be adjusted with duration. Other problems
include:
Uncertainty about final route, with multiple ports of discharge changing during the voyage.
Relaxation of stocking density after the first discharge port changes deck parameters. In the northern summer, it is apparent that the greatest risk occurs in the southern areas of
the Gulf and the Red Sea. Transiting through those zones creates a risk. Transiting slowly creates higher risk, but only marginally so due to the strong auto-correlation of wet bulb
temperatures, and the small increase in duration in those zones. If an appropriate mathematical allowance can be defined, the first duration related risk
increase to be allowed for would be where the first port of discharge is in the hottest zones. The additional duration of exposure due to tying up is probably the most significant duration
effect.
Project: LIVE.116 – Development of a Heat Stress Risk Management Model Revision F
Maunsell Australia Pty Ltd Page 50 of 129
Final Report December 2003
7 Open Deck Risk Management
7.1 Overall Approach