I have a suspicion this post is going to be a mess, much like the event itself. I thus apologise in advance.
Why is this so? In part at least it is due to the BoM practice of resetting observations at 0900 hrs. This means that rain from 0901 on day 'n' to 0000 on day 'n+1' , which to the casual observer all falls on day 'n', is all recorded on and against day n+1. Thus you may have a soaking afternoon and evening on day n, but BoM will record zip against that day if the period prior to 0900 is dry. While this was understandable when the recordings were made by human beings looking at instruments, in this age nearly all recordings are made by automatic weather stations which need no sleep.
A further example of the daftness of the 0900 reset is that at 1109 on 12 June the fall for that day is already set at 5.4 mm, being the total in the period to 0900 on that date.
Of course making the change to using a midnight reset would involve a lot of work for BoM which the ScuMo Government will not fund!
A second part to the messiness is that the initial forecast was some large amount off the mark but it is difficult to demonstrate this in a coherent way. I shall try.
This first chart shows the bracket forecasts offered by the BoM on 7 June for the dates shown and the actual rainfall on that day as reported by BoM.
The upper bound is the fall that is 25% likely while the lower bound is 50% likely. This is a pretty honest attempt at indicating the variability in the forecast but it is rarely interpreted that way, but usually presented as absolutes - "on day x there will be between a and b mm of rain". In fact the statistics are up front that 1 time in 4 there will be more than the upper "bound" and 1 chance in 2 there will be less than the lower "bound". On all the days shown in these forecasts the actual fall was less - some cases dramatically less - than the 50% forecast. Normally it could be expected that the more distant forecasts are 'worse' than the close dates but that isn't the case here with 10 June being the worst.To keep things fairly simple I have only shown the 25% series for 10 June: that is reasonably similar for 7 and 8 June but then plummets on the 9th. It is still double the actual fall (which just matched the 50% estimate) marked with a star. Both forecasts are shown for the 11th which showed a similar plummet but commencing on the 8th rather than the 9th. The actual fall on the 11th (24.4 mm - starred) was almost exactly at the midpoint between the two forecasts made on the previous day.
"With uncertainties about location persisting right up to issue time, BoM had to be fairly generous with the amount of Gippsland covered by heavy rain. The alternative would be to pick an area for flood rains, then be hung, drawn and quartered when it fell 100km farther along the coast."
It was probably a no-win situation for BoM.
Looking at the event overall the next chart shows:
- two sets of daily observations for my weather station (one for the calendar day, the other with a 0900 reset to enable comparison with the BoM data);
- BoM data for Mallacoota (ie the airport site); and
- BoM data for Gabo Island.
The pattern of heavier falls to the west on the 11th with Mallacoota and Gabo scoring more on the 12th is quite clear. The outstanding observation is the massive amount of rain at Mount Baw Baw showing how the runoff from the higher country filled the rivers and creeks which flooded Traralgon (in particular) and Sale.
No comments:
Post a Comment
Comments are welcome but if I decide they are spam or otherwise inappropriate they will not be approved.