Search This Blog

Wednesday, January 19, 2011

After-math 2

It is now long enough after the Brisbane floods for the vultures to start circling.  With 20-20 hindsight they are certain that SEQ Water (the authority managing Wivenhoe Dam) should have done things diferently, and that if they had done so, there would have been no flood worth talking about.

I disagree!

Of course, it is probable that in the light of experience, the Wivenhoe Dam flood management strategy could be improved slightly.  And it is almost certain that improvements to infrastructure and town planning could prevent a repeat of the disaster under similar circumstances.  But this is a truism.  It is true after every disaster, real or potential; and follows straight forwardly from the fact that humans are not perfect.  There is always something more for us to learn.

But suggestions that SEQ Water should have done things differently given the knowledge they had available are simply wrong.

Here are the facts.  In the early years of the millennium, Brisbane suffered from an extended drought.  The result was that dam levels declined from 100% of normal capacity (1,165 Gigaliters or 2 Sydney Harbours) to 15% of capacity (175 Gigaliters) in August of 2007.  Rainfall in 2008 was sufficient to lift that to 23% of capacity by midyear, but the dam did not reach full capacity until April of 2010.  At the time of lowest capacity, the three largest dams from which Brisbane can draw water, representing 85% of Brisbane's water supply, had a capacity of 16.7% (293 Gigaliters).  That represented enough water to supply Brisbane at then (much restricted) consumption rates for 1.5 years.

Since then it has been raining, on and of in Brisbane, and in the last few months, nearly always on.  In early October, 2010, Wivenhoe's storage was lifted by 26% in just a few days.  Controlled releases of 7.1% (80 Gigaliters) was predicted to cause localized flooding in Brisbane.  Additional peaks followed just before (11% above normal capacity) and after (23% of normal capacity) Christmas.  Releases following the former of 5.3% normal capacity (60 Gigaliters) in a day led to localized flooding in Brisbane. 

The amount of water that can be safely released from Wivenhoe without flooding depends on rainfall in other parts of the catchment (Bremer River and Lockyer Creek), local rainfall at Brisbane, and the height of tides.  These are difficult to juggle in that water from Wivenhoe takes approximately 36 hours to reach Brisbane, a time which varies depending on how smoothly the river is running.  From recent experience, releases of as little as 60 Gigaliters per day can cause localized flooding, but under suitable conditions releases of up to 100 Gigaliters per day can be made without local flooding.

On Friday the 7th of January, 2011, the heavens opened up over the Brisbane River.  In the four days that followed, 430 Gigaliters of water was released from Wivenhoe.  The rate of release was twice that which had caused local flooding in Brisbane just three weeks before.  It was managed because of lower tides, but it was pushing the limit.  Higher than expected rainfall over Brisbane would have resulted in significant flooding.  Despite this Wivenhoe increased its storage by 42% of normal storage capacity (490 Gigaliters, or about the volume of Sydney Harbour) over those four days.

On Monday 10th, Toowoomba got hit by its "instant inland tsunami".  That water flowed west, but the part of the storm which hit the Lockyer valley devastated it, and flowed east to contribute around 5% of the total water that flooded Brisbane. 

On that day, rainfalls north of Toowoomba in the Brisbane River catchment were higher than those in Toowoomba.  As a result, water levels in Wivenhoe rose by 27.5% (320 Gigaliters) in a single day.  To avoid them rising the extra 550 Gigaliters that would have caused the dam to overflow, destroying the wall, SEQ water released 645 Gigaliters over the spillway, a release that contributed 80% of the water to the Brisbane flood.

It is worthwhile thinking carefully about those figures.  Had SEQ Water not increased the release of water, then the dam would potentially have filled to 99.8% of its maximum possible capacity without destruction.  Even in that event, Brisbane would still have flooded because of the additional water coming down the Bremer River and Lockyer Creek.  Any attempt to reduce the water coming from Wivenhoe to avoid flooding altogether would have guaranteed the destruction of Wivenhoe, with the consequent wall of water containing five times the volume of Sydney Harbour sweeping Brisbane flat.

As it happened, the rainfall in the Brisbane River catchment finished 12 hours earlier than predicted by the Bureau of Meteorology.  Even so, additional water flowing down the creeks increased Wivenhoe's storage by and additional 10.6% (123 Gigaliters) on Wednesday.

So much for the 20:20 hindsight.  Now for what was known at the time.

At the time of the initial releases (Friday through to Monday), it was not known that a super storm was on its way.  BOM was predicting rain, but at no greater rates than in the days before.  While releasing 100 Gigaliters a day over the weekend, SEQ Water did not know that a storm that would generate an "instant inland tsunami" when it fell over a few square kilometers in Toowoomba was about to hit their catchment of 7,000  square kilometers.  But they allowed for it.  They allowed for it by releasing water at the fastest rate they could without causing flooding.

When the clouds burst on Monday, SEQ Water did not know it was not going to rain just as heavily on Tuesday, but they allowed for the possibility that it would.

This is what the armchair critics are missing out - the limited knowledge available to decision makers.  Having no knowledge of what was about to come, there was no reason for SEQ Water to cause flooding in Brisbane to keep the dam safe on the weekend (which is what the critics are retrospectively asking for).  But, on Tuesday they faced a choice as to how much water to release, and how much to retain.  Had they decided, for example, to reduce the flood waters at Brisbane by 40%, they would have pushed the dams storage level up over 200% of its normal capacity.  (This would not have stopped the flood, only reduced its effect slightly.)  They would then have been in a situation where a follow up storm would have given them no leeway but to release an even larger flood down the river.

In addition to the critics who have said that more water should have been released earlier, at least one well qualified critic has argued that the normal storage in Wivenhoe should have been reduced to around 60% of its current capacity.  However, even with a reduction to 75% of capacity, a similar drought to that which preceded these floods would have seen Brisbane without water.  A 50% reduction would have seen Brisbane without water for more than two years.

It is true that water recycling and a desalinisation plant have extended Brisbane's effective water supply since then, but this assumes that no drought worse than the 2000-2008 drought will occur in Brisbane.  That's an unsafe assumption even given natural variability in the past.  However, Global Warming predicts (in addition to more intensive floods) that droughts will become longer lasting and more severe in the future.

In all, the hydrologists at SEQ Water did a sterling job, and Brisbane ought to gratefully thank them for their efforts.  It is only with 20:20 hindsight that some people imagine they could do better.

(I am, of course, open to being corrected on this.  People who think that I am wrong need only point to the criticisms in print of the low levels of release from Wivenhoe before the storm hit Toowoomba.  I know of no such criticism.)

Update 20/1/2011:

(Photo courtesy of wikipedia, follow link for full size image)

As an addendum to the above discussion, I have just come across this call by the Liberal National Party that an additional 25% of Wivenhoe's capacity be used for storage rather than flood mitigation.  SEQ Water's responce was that they had to carefully study the impact such a move would have on flood risk, to which the LNP's spokesman responded that the additional water should be stored before the study was complete.  This was in March last year.  Opinion by commentators is equally divided between those for and those against the proposal, including one by an Ipswich man who had lived through the 1974 floods and argued to keep the water because you can survive the floods, but running out of water is terminal.  Not one person suggested the now popular alternative of lowering the storage capacity to increase mitigation.  (Both ideas, are IMO bad ones.)


  1. TC - in the Australian blog to which you linked, did you notice the second comment by Goresh of Brisbane? If he is correct, draining the dam below 1.16 million ML would be very difficult. The only way the managers could get that water out quickly would be to have all the water taps in Brisbane turned fully open 24 and 7! He's saying no spillway opening goes low enough to quickly remove the drinking water component. I suppose SEQ water could have started a bucket brigade. How many buckets are there in Queensland? I don't know that I believe him. I once toured a dam that was drained to do maintenance on the feeder tunnels to the turbines. Biggest mud puddle I've ever seen.

  2. For convenience, I have included a picture of the spillway above. The dam wall is actually 50 meters high. The spillway gates are 12 meters wide and 16.6 meters high, and appear (in the photo above) to have a about 4 meters above their upper edge to the top of the wall. Probably maximum design capacity does not come to the top of the gates. Allowing 6 meters between the top of the wall and the level of maximum design capacity (2,600 gigaliters), then, then maximum design capacity would be about 12 meters above the base of the spillway, which allows 32 meters for the normal storage. That makes sense because as the dam gets deeper, it gets wider, so the same amount can be stored with much less depth. In fact, just 2 meters of additional storage over the current adds 25% of normal capacity, which suggests that 14 - 16 meters above the spillway adding 140% of normal capacity is about right.

    I did notice the small pipe outlet on the right of the spillway at the base, however. That would certainly allow them to drain the dam at need. So Goresh's conclusion is invalid. (Shame really, I had made a similar speculation on another forum.)

  3. I saw that little opening. It's hard to say what that is. It could just be an inspection opening. I would think dam engineers would be very reluctant to poke many holes through an earthen dam, especial the size necessary to do flood mitigation. Earthen dams seep. Best not to encourage that.

    I'm certain they can drain it, but not anywhere close to the speed necessary to have made much of a difference in the timeframe under consideration. Even on sunny days of 2010, water would be flowing into the dam from the basin, and that would offset the amount that could be released through openings that are significantly smaller than the flood gates; in other words, likely a fairly slow process.

  4. If you look closely at the large resolution photo (follow link) you can see a circular opening just inside the square outer opening. The most sensible interpretation of the circular opening is, in my mind, an outlet pipe.

    You are of course correct that so small a pipe would be of little practical use draining the lake with significant inflows. However, the idea being proposed is that the lake be drained to a lower level before the wet season to enhance flood mitigation, which it certainly could do.

    Of course, there were very significant inflows to Wivenhoe in the weeks preceding the flood. In all around 35% of capacity flowed into the dam, excluding water let out over the spillway (which was continous over this period). Had the dam been left at 75% of capacity coming into the wet season as an additional flood mitigation method, it is doubtfull so small a pipe could have dumped enough of the excess to get dam levels below 100% of supply capacity in any event.

    As a curiosity, I also notice in the photo the disturbed water at the base of the spill way, indicating some below surface outlet. I assume this is related to the pumped storage hydroelectric scheme operated at the dam.

  5. TC - first, there is a photograph that you might find interesting of water coming out of there on this website:

    That may not be indicative of its flow capacity. Earthen dams seep, and they have to do something with it. It could be that is part of the system to take seepage away from underneath the dam. I'm not greatly confident in this speculation, but getting it to the spillway would seem a logical thing to do with it. When you think about the area from which they would be collecting seepage, the size of the pipe might make sense.

    Where are the turbine generators? The turbulence you noted, I've seen that in every spillway to which I've been. When I was kid I used to stare at that. It could be they just want to keep the ponded water at the base of the spillway moving. I don't know. Maybe SEQ Water would take a call from Texas and answer these important questions!

  6. TC - don't want to drive you crazy with this stuff, but found this aerial video of our little the tube discharging on December 15, 2010. Great view of the lake side of the dam, which are very hard to find:

    Seq Water indicates the dam was 100% full on that date.

  7. You may find the lift out from the Australian interesting. (PDF) It shows a cross section of the dam, which shows there are two 1.5 meter diameter outlet pipes through the base of the dam that exit at that point. One is used for the hydroelectric station. The other, according to the Wivenhoe Dam operator's manual (PDF), is used to fine tune the flow when opening and shutting gates. Also of interest in the first link is a picture of the dam wall clearly showing the auxilliary spillway installed in 2004 (from memory) to rprevent the risk of dam collapse. The fuse plug in the auxilliary spill way is designed to contain water, but to quickly erod of water flows over it. Water levels came within 100 mm (for inches) of overflowing the central section of the fuse plug on the day after the greatest release from Wivenhoe (Wednesday).

    Also of interest may be the fact sheet (PDF) on Wivenhoe Dam released by SEQ Water. It mentions that the dam had inflows of over 700 Gigaliters of water in October 2010, sufficient in the absence of the dam to cause a 1974 style flood in Brisbane.

    The video you mentioned is very interesting. According to the SEQ Water website, Wivenhoe was actually 2.1% over capacity on December 15, which fits well with the idea that 100% of supply capacity is level with the base of the flood gates.

  8. Well, then maybe we can conclude that at 2.1% over capacity, they're fine tuning the level. Drawings are just for what I was looking. Thanks.

    If brainwashed by the evil climate scientists to only think about drought, building a fuse-plug spillway in the mid 2000s is sort of damn odd dam behavior. If I had been brainwashed to think only about drought, I would have gone Manchurian on that dam and krazy-glued the radial gates.

    I'm horrible at interpreting odds. In that manual, what is your take on the 1 in 100,000 odds of an overtopping event? Does that mean in a La Nina season they saw the odds of an overtopping to be 1 in 100,000?

    Links you might find interesting:

    Note the enhanced numbers in the above.

    The simplistic schematic in this PDF supports my notion of how the dam works. I simply cannot fathom having storage behind a floodgate. A floodgate failure would mean a disaster downstream. Best to use earth and rock to hold storage. But the inquiry will end speculation, including mine:

  9. Very good point about the "evil scientists". I have read those predictions in about seven different original documents now, all predating december 2010, including, of course, the IPCC assessment report 4, and the plan for the auxilliary spillway. Strangely the deniers have missed all of them, and insist despite correction that climate scientists only ever predicted drought for Queensland.

    About the auxilliary spillway, my understanding was that two such spillways were recommended to flood proof the dam, but only one has been built todate. SEQ Water have not said if, or when they will build the second.

    Unfortunately, your first URL is just a different source for the fact sheet I linked above. Your second was interesting. At one point it mentions expanding the flood compartment capacity from 1,450 Gigalters to 2,000 Gigaliters above supply capacity in the 2005 expansion. However, that document is dated to 2002, and later documents (including the fact sheet) indicate there has been no change to the flood capacity (but see below).

    I have managed to find to discussions of the proposed expansions, both dated after the construction of the auxilliary spillway. One is by Queensland Water Commission and the other appears to have been prepared for the Queensland Water Commission by GHD, an engineering firm. Both are PDFs, the first being 99 pages long (you have been warned).

    Unsurprisingly, I have found some interesting and relevant information. First, the wall height has been raised to 80 meters, which represents 2,000 Gigaliters above Full Supply Level. But, the lowest fuse plug is set at 76.2 meters, which represents 1,450 Gigaliters above FSL. Clearly my comments in the main post about filling to maximum possible capacity are in error. The dam came within an ace (less than 100 mm) of triggering the fuse plug, but still had around 1,000 Gigaliters of capacity before the walls would have been breached.

    Also of interest, are the comments about the levels. The 1:100,000 flood is the taken to be a 1:100,000 chance of a flood occuring at that level in a given year, as determined by a BOM study. Intriguingly, the level of flood required to trigger the fuse plug is a 1 in 6,000 year flood. Considering how close this flood came, that give it a little perspective.

    It's getting kind of late here, so I'll finish now. However I have come across some more information that will find its way into another post in the next couple of days. Looking forward to your comments.

  10. TC - well, in the face of your PDF (table on page 24,) my theory bit the dust. The spillway height is 57 meters. The full supply level is 67 meters. The top of the radial (Tainter) gate is 73 meters. At 57 meters, it looks like it's a little less than half full of drinking water. So there are 10 meters of water leaning on the radial gates when the dam has 1.15 million ML of water in it.

    tainter gate PDF:

    some info on gates and spillways in Australia:

  11. Tom,where does this '80% of the flood came via Wivenhoe' figure come from? I have seen this figure bandied about by News Ltd organs as though it is solid,but it seems to be based on an assumption that the 645GL/day peak release figure indicated in press release by SEQwater is actually a 24 hour mean. There is no evidence to support such an assumption.

    If they are suggesting that 80% of the water over the entire wet period came from the upper catchment,that may be closer to the truth,but the river was not always in flood,and when it was Brisbane City itself was not always at flood stages.

    We all lack hard hourly data of inflow and discharge volumes from Wivenhoe,and I think News has tried to stitch some claim together to suit their usual presumptions about 'government' incompetence,and worst case thinking.

    The only hourly figures that suggest what was happening on the river are DERM's stream gauging data. If you look at the data for Savage's Crossing,some 16 kilometres downstream of the dam,the peak flow figure hit 597 GL/day at 1AM for no more than an hour on Wednesday 12/1. This flow figure includes the contribution of Lockyer Creek [in major flood at the time] and several hundred km2 of ungauged minor catchment which had just received truly enormous amounts of rain[over 500mm in 48 hours].

    The Savages Crossing mean flow figures tell us that at the height of the flood,in the 48 hours from midnight 10th January to midnight 12 January,a total of 870 gigalitres of water went past that spot. This included at very least 200 gigalitres from the Lockyer and an unknown amount from that ungauged catchment.

    Then once past this spot,the river collects more ungauged catchment including from Lake Manchester,and the rural district on the other side of the area,then collects the flood waters of the Bremer River system. This latter is a difficult contribution to calculate,as the low gradient of that stream leads to main river flood waters damming its waters and backing them up to Ipswich. However,over the 48hours I noted 150 to 200 gigalitres was coming down the Bremer.

    Adding the known tributary contributions together,the flow going towards Brisbane at the Bremer/Brisbane confluence for the peak two days has to be around 1270 gigalitres.I'd confidently add 80 gigalitres for the ungauged element which totals some 600km2 of the wettest part of the event,so we get 1350 gigalitres.Given the Brisbane River at Savages Crossing less Lockyer Creek is something less than 670 gigalitres gigalitres,I think this pretty persuasively indicates that the Wivenhoe contribution to the peak two day flows was more in the order of 50 to 60%.

  12. My information does for the 80% does come from News Ltd, part of which can be found here. (I believe I read the equivalent article in the Australian, but I may have got it from ABC Online rather than News Ltd as we did not recieve the Australian for several days due to a meter of water in our local newsagency.) The claim is made specifically for the water flow at the time of the peak, ie, Thursday at 4 AM.

    Since last writing on the floods, I have come to believe that at Brisbane we were downriver from three floods, the Lockyer Creek flood of Monday, the Bremer River flood of Tuesday, and the massive release from Wivenhoe also on Tuesday. I believe that means the Lockyer creek peak would have passed through Brisbane before the Bremer River peak passed through Brisbane, so that if Wivenhoe had not had to release so large a volume of water, the peak in Brisbane would have been on Wednesday, and would have been 0.5 to 1 meters lower than the peak on Thursday morning. The peak on Thursday, however, would have been primarilly due to releases from Wivenhoe.

    The information I have about releases from Wivenhoe are also from News Ltd, and are summarized in this table.

    The 7,500 cubic meters per second flow from Wivenhoe on Thursday is equivalent to 648,000. As the river was near its peak for at least 12 hours (ignoring the effect of tides), and as the peak will spread out as the water flows down stream, that is consistent with the 645,000 megaliters quoted for Wednesday being the total daily releases (as I have interepreted it) rather than the peak release in the day.

    There is more information from The Australian here. That article cites leaked emails to quote specific rates of release, and cites a peak release rate of 8000 cubic meters per second (690,000 megaliters per day) commencing at 8:30 on Tuesday evening.

    That is inconsistent, however, with the figures you quote from the Savages Crossing flood gauge. If that is accurate, then the flows quoted are peak flows only, and the releases quoted for Wivenhoe have been significantly overstated. It also means the leaked emails are significantly over stating release rates, which seems very unlikely.

    However, your reported flows past Savages Crossing for the Tuesday and Wednesday excede the reported releases from Wivenhoe. However, for consistency we would have to assume the peak from the Lockyer Valley flash floods had already passed.

  13. TC - you've probably seen this, but just in case, this article has rainfall amounts for several locations.

  14. JCH, sorry, but the link is not working for me.

  15. Sorry:

  16. Tom,don't take anything from News Ltd at face value. The table you cite is wrong. They have taken transient flows expressed confusingly as ML/day figures from press releases and mistakenly assumed/implied they can be read as 24 hour mean flow figures. 645 GL/day was NOT a 24 hour mean release,but a transient over less than an hour.

    The leaked email talked of a POSSIBLE release of 8,000cumecs at 8.30 PM.As it turned out the maximum transient release volume was 7465 cumecs [mentioned in press release AM 12/1] three or four hours later,and the journalist should have been able to cross-check this discrepancy by referring to the press release archive at WaterGrid:at

    Note also the 7465 cumec(=645GL/day equivalent) figure is the spillway release peak.The peak at Savages Xing that reflects this release figure maxs out at 6945 cumecs an hour or so later.You have to pull the Lockyer and ungauged catchment out of this,so it is in reality less than 6000 cumecs of Wivenhoe water. This means that the News Ltd purported 9000 cumecs peak flow at Brisbane gauge contained under 6000 cumecs as Wivenhoe release[65% or less],regardless of whether the total through-putting volume over say four days breaks down to a larger percentage from the upper Brisbane catchment. The transient release peak is what reflects what dam management managed to keep dam contributions to for a transient peak later at Brisbane. Water below the dam slows down and spreads out as it moves past any point in a lower gradient,natural bed profile stream.

    A way to calibrate this is to look at what SEQwater did after the flood peak had passed. Following the manual,they drew down the flood buffer as quickly as possible in order to anticipate further upstream floods.
    To do this they set the release at 300GL/day PER DAY unvaried for the next 96 hours or so.During this period the Savages Xing gauge measured around 210GL/day PER DAY once the diminishing Lockyer contribution is factored out. Hence you can see any figure at the spillway has to be downscaled in the river bed to reflect the differing physical characteristics of spillway and river bed.

    Hindsight may well say that too little water was released in the early days of the rain event,but I do not think that The Australian can claim that 80% of the hour long or even highest six hours at Brisbane came from Wivenhoe releases.

    My two day figures past Savages cannot exceed the equivalent Wivenhoe releases [no all day figures were reported] because of physical factors as I have explained,and my readings of the DERM Lockyer gauge [Rifle Range Road] suggests that peaks were pretty coincident.

    The various peaks pretty much aggregated by the time they were passing Jindalee,slowed down in the broader,tidal river environment.

  17. JCH, the QT link is again throwing in figures without required detail or contexy,and one thing they failed to note when they mentioned a figure for total rainfall at Wivenhoe Dam for January 8-14 was that the recorder failed for the 12/1. That day probably saw around 200mm+ as well,judging by the radar RF accumulation graphic and available figures for nearby stations.600mm was quite likely at the dam itself for 8-14/1

    I kept the data for the period,and they show that a considerable area from Marburg across Wivenhoe and Mt Glorious and up the D'Aiguilar Range to the upper Stanley saw falls of 500 to 700mm in the three days 10-11-12/1,so if anything the QT article understates the deluge. Five day accumulations above Somerset Dam approach 900mm,and in the upper Brisbane River from 200 to 450mm

    This may explain why dam operators may have been further behind the curve than they may have projected at the start of the event.

  18. Nick - my hunch has always been that a release to avoid overtopping could not have been forced by a run-of-the-mill rain event in the catchment(s) and basin(s) that impact operating decisions. When all the numbers are tabulated and reviewed, I suspect the dams will be found to have endured one big mother of a rain event, and in a fairly compressed period of time. I pick that over drought dreading zombies, brainwashed by climate scientists, hoarding water.

    On the question of whether or not the dam's original design called for a ~1.16 million ML drinking water storage, do either of you know the answer? I believe a SEQ Water official, when pressed, said it had always been that way.

  19. JCH, I find Nick's comments on the rainfall interesting. The rainfalls indicated in the article were on a par with rainfalls in 1974, but all indications are that, absent Wivenhoe, flows downstream of Wivenhoe would have been much greater. I had put this down to a greater peak intensity of the rainfall. I had heard that the rainfall was record breaking in quantity as well, but did not have firm evidence.

    The claim in the article that Wivenhoe was recieving 2 Sydney harbours a day as inflows is certainly wrong, and is probably a garbled account of the Premier's claim that Wivenhoe had recieved over 2 Sydney Harbours worth of inflows over several days (which is almost certainly true).

  20. Tom,that's the way I see it: Wivenhoe's 1/2011 catchment rainfall very much on par with 1/1974 in toto,but turning up in three days rather than five,so more intense.

    1974 saw higher totals and intensities only around Mt Glorious,and the totals in the west of the catchment from Crows Nest to Cooyar Creek were a little lower than 2011. '74 saw more rain in the Bremer and the Brisburbs,as well as down on the Gold Coast and Tamborine Mtn but that's another story.

    This flood was quicker and potentially higher upstream from the Bremer.Lockyer Creek peaked over 60cm higher than '74,Mt Crosby peaked lower by about similar,showing the mitigating effects of Wivenhoe,whether hindsight optimised or not. The big relief was the Bremer falling well short of 1974s level and much lower totals in the catchment downstream of Moggill. Back in '74 ,five day totals of up to 800mm fell in the Oxley Creek catchment,for instance.

    1974 had a larger volume as well as higher peak,but given the amount of damage a faster flood [sans Wivenhoe]of similar or greater height regardless of duration would have brought,Brisbane still dodged a bullet.

    The 2 Sydney Harbours in one day thing is quite possible,as one Sydharb is 500GL, There is a lot [c 500km2] of catchment off the D'Aiguilar Range that is unmetered-Reedy Creek,Sandy Creek Northbrook Creek,Kipper Creek,etc.,200GL/day came from the upper Brisbane for 24 hours of the 10/1 and similar or more from Somerset,and there is the surface area of the dam itself that saw rain rates of over 250mm/day.

    I think the technical assessment will be very surprising for some who think the engineers failed.

    The post flood buffer draw-down also reveals inflows quite accurately because of the fixed outflow figure of 300GL/day. In the 4 days from 9am 14/1 the dam dropped 73% to some 106%, or 850GL in volume. But total discharge was 4 x 300GL= 1200GL,so some 1550GL entered in 96 hours,120 hours after the rains had well ceased. That's 3 Sydharbs in 4 days well after the event

  21. JCH, the designed supply storage has always been 1,165 gigaliters of water.

  22. Nick, that sounds like a very fair summary. I do not think the 2 Sydharbs in a day is accurate. Taking the release figures from the Australian, plus known increases in levels, the greatest inflow in a single day was approx 1.5 Sydharb. Apparently I may have to reduce that estimate of a peak daily inflow because the figures from the Oz are peak flows, not daily summaries.

    To that end, do you have more accurate figures for the total volume released over Monday, Tuesday and Wednesday?

    With regard to the relation between spillway flows and Savages Crossing flows, something is still amiss. If spillway flows are constant, and more than Savages Crossing flows, then the level of water between the Spillway and Savages Crossing must be rising. But if it is rising, then flows at Savages Crossing will increase.

    This is a sperate issue from what happens when you release a pulse of water. A pulse will of course dissipate because not all of the water has the same velocity. But a constant inflow must be matched by a constant outflow of the same volume unless water is being stored (ie, water levels are rising).

  23. I think it is down to river bed profile at the Savages Xing gauge,and pressure factors and profile at the spillway.

    The gates are 12m wide x 16m high,so the water is escaping steeply downhill under some pressure through a sixty metre wide aperture.The bottom elevation of the lip is 57m,the water level above was up to 74m and more,so the weight of 140-150% of total volume is forcing the water out. This water is emerging under enormous pressures.

    At Savages, it's a different story of course. At high [over major flood] volume the water is well over 21 metres deep flowing in a bed profile much wider than the spillway,with higher friction and less back pressure [it is not emerging from an aperture],so rates per second cannot be directly equivalent to spillway conditions.

    So 300GL/day all day=3470 cumecs exiting from five 12m wide apertures of unknown height[presumably some metres] at the spillway transforms to a flow something like 2430 cumecs when passing across the river bed profile at Savages,a moderate flood over 15m deep and probably 150m wide, moving more slowly. The river backs up side creeks including the Lockyer and also back up its own bed to the downstream side of the dam wall in the original river bed.

    At Savage's,the total mean volume for the seventy-two hours from midnight 9/1 to midnight 12/1 was 1064 gigalitres. Estimating how much of this is from Lockyer Creek is difficult because of lag between Rifle Range Rd gauge and Savages which is over 40 river kilometres distance,and because the gauge was overtopped and clipped the peak off the actual flood: once it got past 16.43 m the gauge stopped measuring. It was the second highest flood ever recorded on the Lockyer,from data from the BOM gauge at Lyons Bridge,but I don't know how to translate that site height into volumes.

    I guesstimate the three day volume at at least 300 GL flow at Rifle Range,so that makes Wivenhoe release total at most 760GL/72 hrs. It will be less because of ungauged catchment streams like England Creek,but I'd again guess minimum about 700GL at Savages.Working backwards to the spillway,this may make actual three day discharge around 1000GL for the period you nominate.

    This is why I think News Ltd attribution s are wrong,and their speculations are lazy and hurtful. The wise-after-the-event types will argue that that hypothetical 1000GL should have been started earlier and the peak lowered by means of the extra time gained before the inflows peaked,but they are not dealing with the information the operators saw hour by hour and they overemphasise the certainty of the forecasts without seeing them all in the sequence they were generated and received. They also have no hard knowledge of the hydrology of other flooding streams,or reliability and consistency of gauge data.

  24. I'm not sure if your aware of it, but BOM has released a discussion of the floods in Special Climate Statement 24. The key passage for our discussion reads:

    "Peak rainfalls from the 1974 event were substantially heavier than those in 2011. A number of stations had three-day totals from 25-27 January 1974 in excess of 1000 mm, the highest being 1215.0 mm at Mount Tamborine, compared with the 2011 event peak of 648.4 mm. Many stations in the 1974 event experienced daily totals which exceeded 400 mm; the highest were 563.2 mm at Mount Tamborine and 561.5 mm at Wundurra, in the Gold Coast hinterland, while in the Brisbane area 475.8 mm fell on 26 January at Enoggera Reservoir. 1974 also saw much heavier rainfall in metropolitan Brisbane than 2011, with Brisbane’s three-day and peak one-day totals of 600.4 mm and 314.0 mm in 1974 comparing with 166.2 mm and 110.8 mm in 2011. However, in 1974 the heaviest rains were close to the coast, whereas in 2011 heavy falls spread further inland, and on the western fringe of the Brisbane River catchment and on the Great Dividing Range 2011 was the wetter of the two events (Figure 5, right). The weeks prior to the 1974 event, whilst wetter than normal, were also less wet than the equivalent weeks prior to the 2011 event. Over the Brisbane River catchment as a whole, average three-day rainfall in the 1974 event was 348.5 mm, compared with 286.4 mm in 2011, and all four major sub-catchments were also wetter in 1974 than in 2011, although by small margins in the cases of the Bremer (1974 442.1 mm; 2011 417.1 mm) and Lockyer (1974 331.3 mm; 2011 292.0 mm) sub-catchments.

    Insufficient rainfall data exist for a comprehensive assessment of the 1893 event. However, the available station data indicate that peak rainfalls in the region during the 1893 event were much heavier than those during either the 1974 or 2011 events. Crohamhurst, in the Glasshouse Mountains inland from the Sunshine Coast, received 907.0 mm on 3 February 1893, which remains an Australian daily record, whilst three-day totals included 1715.0 mm at Mooloolah and 1680.3 mm at Crohamhurst."

    I am a bit confused by some details of this statement. In particular, the BOM report into the 1974 flood divides the Brisbane River catchment into six major subcatchments, not four. Does this mean that in two of the catchments rain was heavier over the three day period than in 1974, and if so in which ones? Also, the claim of heavier rain on the western fringe of the Brisbane river catchment seems to contradict the claim that rainfall in all four major subcatchments was less in 2011 than in 1974. Wasn't the upper Brisbane one of the major subcatchments?

    Nick, you say you recorded the station data over that period. Would you mind posting it so I can have a better idea of the rainfall pattern.

    As it stands, I still believe the single day rainfall in the Lockyer and Upper Brisbane subcatchments exceded that in any single day in 1974, and that the 2011 flood was preceded by flooding rains just a fortnight (Upper Brisbane) or week (Lockyer) before; both factors contributing to a very high run off.

  25. Tom,the BOM statement ,in the detail that interests us, is pretty consistent with what I have observed for the Wivenhoe catchment,though I think it is a pretty preliminary report,as they have only cited non-automated recorders. Once they integrate the data yet to be verified,things will change. I compared 1974/2011 data in the upper Brisbane and Stanley catchments and I think that 2011 often comes out ahead.

    In agreement with your last paragraph,another couple of points will emerge in more detailed analysis to come from BOM: in quite a few cases [not all] the heaviest consecutive 36- 48 hour falls were higher in 2011,and the week preceding this flood event was markedly wetter than the week preceding '74 at all comparable stations. These points give this event a 'leg-up' with pre-conditions of higher soil moisture levels and higher base stream flows.

    The Lockyer flood peaked 63cm higher than '74s mark at Lyons Bridge,second only to 1955,so even if the catchment average was slightly less,the speed and 'ease' of run-off was higher this time round.

    The big one day and five day totals they mention for 1974 have no impact on the Wivenhoe in the case of the Mount Tamborine/Gold Coast data,and I did state that falls in the Brisbane suburbs were higher in '74. Once again,those falls,while critical to elements of the '74 flooding,have little bearing on what enters Wivenhoe.

    The only small area that has some input to Wivenhoe that experienced much higher falls over five days in '74 was a small area around the Mt Nebo to Mt Glorious ridgeline. Some of this area would have fed very high flows into the river above where the dam wall was built,but most enters downstream,or on the other side of the watershed into the Pine River and Brisbane Creeks catchments. In this event this area received 400-800mm over 4 days,in 1974 a phenomenal 800-1300mm over 5 days. The latter falls played a big part in levels in the city in '74

    This time round we also have the radar rainfall measurements,which show how very heavy rain was falling in many areas without automated or manual recording stations.

    There is too much data to post,but some comparisons for the four consecutive highest rainfall days,which are usually 25-28/1/1974 and 9-12/1/2011:

    Peachester 709mm in '74/741mm in 2011

    Woodford 587/632

    Jimna 311/258

    Blackbutt 329/305

    Monsildale 202/337

    Yarraman 255/393

    Gregor Creek 409/347

    Somerset Dam 313/ over 400 [one day missing,but conservatively estimated from radar]

    Crows Nest 270/350[inc 310mm in 48hours]

    Mt Binga/Emu Creek [closest stations for comparison] 315/279

    Lowood 385/491

    Mt Glorious 1306[manual]/719[alert]

    What I have seen of 1893 data shows a hell of a lot of rain over the month,mainly collected into two humps,each of a similar scale to more recent floods

  26. Tom,I have to correct my lousy maths back up the thread where I claimed the inflow during the constant rate into Wivenhoe during the constant rate drawdown period post-flood must have equalled the outflow PLUS the difference between outflow and volume drop.This of course is rubbish!

    The estimated inflow was the outflow MINUS the volume drop 1200-850 =350 gigalitres over 4 days.

    Using the DERM data and closely reading the SEQwater graph and press releases from WaterGrid,I think I've got reasonable estimates for the event.

    As the dam rose sharply over the four days to midnight Tuesday 11th's peak of 191%,using Savages Xing data less the Lockyer and allowing for the ratio between spillway figures and Savages figures,I estimate the dam discharged 700-800GL while rising nearly 80% which is the equivalent of 930GL. 930 + 800 = 1730GL entered the dam in the mitigation phase. Round it to 1700GL

    Once the dam hit peak and discharges were boosted to match inflows,the dam stayed at around 190% for the 24 hours to midnight 12/1.The press releases suggest mean daily discharge at around 350-450 GL,so same out as in.

    The next day to midnight 13/1 releases were held back to allow downstream tributaries to drop and were pegged at 220GL for the day,while the dam level dropped about 8% or 90GL. So inflows were 130GL,and the mitigation phase was over. the following 4 days saw the dam raise discharge rates and draw down 1200GL, with inflows of 385GL.

    Inflows balance outflows at the spillway over ten days at a bit over 2600GL The mitigation phase held back 900GL,to release over the back five days after the peak. Maybe not perfectly played,but certainly did some use.

  27. Interesting!

    I have been approaching it from another angle. Using the reports of the leaked engineers emails plus press releases, I have estimated discharges over given periods from Friday 7th through to 9:00 am of Wednesday the 12th (chosen to coincide with the publicly available data on Wivenhoe's levels). Using what I consider to be a very conservative approach, I have estimated total inflows to the dam to be about 1320 gigaliters from Midnight of 9/1 to Midnight of 12/1. Adding Sunday to the period increases the estimated inflows 1410 gigaliters. I emphasise this is meant to be a conservative estimate, so I consider your slightly larger estimate as a good reality check on my assumptions.

    By breaking the data up as I have done, I also think I see evidence for three distinct peak of inflows into Wivenhoe. The first is about 8000 cumecs around 8:30 pm on Sunday, and is explicitly mentioned by the engineer. The second inferred peak is late on Monday, with a peak inflow of between 6000 and 8000 cumecs. The third occured on Tuesday evening around 8:00 pm, and had a peak over 8000 cumecs, possibly significantly so. Over the Monday and Tuesday estimated inflows average around 6000 cumecs.

    The third peak is the real kicker. It is the one that triggered the big release, and without it, Tuesday's inflows would have been closer to 4000 cumecs, and easily handled. If we scale these peaks to match your inflow estimates, then the Monday peak would have been between 7300 and 9700 cumecs, and the Tuesday peak over 9700 cumecs (the Sunday peak has a stated value so does not need to be scaled.) In other words, over Monday and Tuesday, Wivenhoe was forced to deal with two peaks of inflows in succession, both of about the same size, or greater than the peak flow in 1974 at the city gauge.

    The dam operators faults on this evidence is that:

    On Monday he did not predict a third peak of inflows in three days, each being greater in magnitude than 1974's peak at Savages Crossing, and the third one possibly being greater than the combined flows from all streams upstream of Breakfast Creek in 1974; and

    On Tuesday, when he reacted to the third peak in order to prevent the fuse plug from being triggered, he did not predict the precise timing (and hence height) of the peak so that he was still released for a short time, more water than was flowing into the dam.

    If this understanding is anywhere near correct, the man deserves a medal.

  28. Or at least a national hug. What a tough spot in which to be.

    Excellent posts from both of you. Saw in the news that the government is considering releasing some of the 1.16 million ML storage. Any bets on what percentage of full they end up at once the rain stops? 60%? 90%

  29. Tom I can certainly confirm the first two peaks and their timings: they are to be seen on the DERM gauges on the Upper Brisbane,Cressbrook Creek and the Stanley River system,and the third surge is likely too from the downpour over the dam,so maybe that was generated closer to the wall by the D'Aiguilar tributaries. I think your 4 day inflow figures are too low,though.

    The 24 hours to midnight 12th are the period when very high daily inflow and release can be closely estimated from the four WaterGrid press releases that overlap that day.

    In the previous 24 hours releases were boosted from 236GL/day equiv. or 2730 cumecs,to the midnight 11th peak of 645GL/7465 cumecs. My estimate puts that days throughput at 360GL,but it may be closer to 400. The dam level rose during this period.

    Media releases state that the dam peaked over night 11/12th at 191%, when the 7465 cumec event arrested the rise and therefore matched transient inflow. From that time on releases were cut back quite sharply [to 205GL/2305 cumecs] to allow the Lockyer peak to subside,then eased back up slightly to 215GL/2490 cumecs until midnight 12th.Over this entire 24 hours the dam barely dropped 2%,and in fact rose slightly due to the 2305 'choke' I think we can estimate this day of balance as 400-500GL in and out,probably closer to 400.

    I'd be wary of the leaked emails,as I don't trust Hedley Thomas at all. He has a clear agenda to stir,and generally attempt to portray the dam managers as incompetent. Today he is continuing his opinionating about the irresponsibility of keeping the dam at 100%.I notice he has NOT introduced any new data since his first forays two weeks ago,and continues to use the 'authority' of the two external engineers he originally cited to cast doubt on the process. He has NO new observations from these gentlemen. Why? Have they lost interest in being selectively quoted,perhaps?

    Today,Thomas is using an old tabloid trick of falsely insinuating an admission from a SEQwater spokesman.I quote Thomas' article 'Fears Full Dam will cause New Flooding':

    "[SEQwater Strategic Relations Manager Mike]Foster did not dispute SEQwater's operation of Wivenhoe Dam was responsible for most of the flooding in Brisbane,but said a commission of enquiry would examine this."

    I put it that Foster simply said what he is quoted as saying at the end of that sentence,and the rest is Thomas' construct. There is no evidence that Thomas or anyone else at the press conference yesterday directly asked Foster whether SEQwater WAS 'responsible for most of the flooding'and he volunteered no such possibility himself! Foster did not dispute it because he likely wasn't actually asked! And Thomas is on record as having decided the case in favor of his personal opinion two weeks ago,anyway.When DID he stop beating his wife,again?

    Decontextualised fragments of emails are no good,as many of them are clearly what-ifs:the 8,000 cumec peak was never released.The max was 7465,and it came later than the engineer speculated it would need to be made.

    JCH,if they decide to go below 100%,they don't need technically to do this until rain is imminent,48-72 hours is enough to lower the dam 20% with flood warnings on the river to Moggill.

  30. JCH, it is probable that if they were to drop the level of Wivenhoe down to 80% now, that it would be full again before the end of the current La Nina. However, probable and certain are not the same thing. Brisbane has gone through La Nina's before while recieving very little rain, most recently in the later half of 2007. Therefore there is no guarantee that should the level of Wivenhoe be dropped to 80% now that it will regain 100% before the next drought.

    Even if we go into the next drought with 100% supply, you can be sure that before the drought is through, voices will be heard asking why the flood compartment of Wivenhoe was not used to provide additional storage. If we go into that drought with below normal supply, those voices will be at the head of a witch hunt.

    As recently as October of 2010, the Queensland opposition were calling for the flood compartment to be reduced to enhance supply. Just one month out from the start of the wet season, with a strong La Nina active, they wanted to reduce flood mitigation to guarantee supply

    Now everyone wants the opposite. But has we followed their advice during the La Nina preceding the drought, and reduced Wivenhoe's capacity by 20% to enhance flood mitigation, then subsequent rains would have only increased storage by 10%, meaning Brisbane would have been within 5% of no water at all by the end of the drought.

    The only time I think it is sensible to reduce supply capacity in anticipation of additional rain is if, as was the case in 1974, you have a tropical cyclone bearing down on the city.

  31. I'll amend my prior post. As recently as December 20th of 2010 the state opposition was calling for a reduced flood mitigation capacity at Wivenhoe. They are trying to spin their way out of it now, but they definitely called for dam levels to be raised.

  32. Nick, I absolutely agree that my figures are too low. However, as I am arguing that a major reason for the flood was the freak levels of inflows into Wivenhoe, and in the absence of hour by hour details, it behoves me to use a conservative estimate. Even with that conservative estimate the flooding is seen to be unusual in terms of the volume of water involved, and the dam operator's decisions are seen to be reasonable given then available information. The case just becomes stronger with your probably more accurate figures.

    I disagree with you slightly regarding the Foster quote. I suspect he was asked if SEQ water was responsible for the level of the flood, and responded on the lines that he was unable to answer in view of the establishment of an enquiry into the matter, ie, the standard responce of any pubic servant in a similar situation.

    With regard to the emails, I very carefully distinguished between stated levels of discharge, and those which were only planned or forseen. Mostly, however, they provided information about the timing of changes in discharge.

    I believe the 8:00 pm peak discharge was probably fairly accurate based on an estimate of the time it takes for water to flow from the wall to Savages Crossing. To get that estimate, I took an estimate of the time for a flood peak to travel from Mount Crosby to the bay to estimate an average velocity for the flood peak. From this and the distance between the spillway and Savage's Crossing, I estimate it would take about three hours for the peak to travel that distance, which makes the initiation of peak discharge to be around 8:00 pm, as per the expectation in the emails.

    I'ld like to draw your attention to a letter sent to the Australian by Owen Droop, in which he writes:

    "The nation has just witnessed the very reason why Wivenhoe was specifically designed and
    is very carefully operated as a flood mitigation storage, even when very recent experience in
    an extreme drought shows that any rainfall-dependent water supply is a vulnerable and
    valuable resource.

    In drought and flood, we have to carefully balance water supply with flood mitigation. We
    cannot (and must never) erode the flood storage component of Wivenhoe for water supply
    and vice versa. These decisions must be based on careful analysis and well-founded
    science, not on fag-packet arithmetic.

    The operational rules of all dams should be reviewed when extreme events occur. Anyone
    who is even remotely aware of the diligence and professionalism displayed by those in the
    Government Flood Room will want them heavily involved in the review – not the armchair
    critics who only appear when it suits them."
    (My emphasis)

    Of course, your and my estimates are "fag packet arithmetic", but we do not expect anyone to act on them. Ours only exists to show how dreadful is the "fag packet arithmetic" of those calling for action now on dam levels without bothering with the process of careful review.

  33. Nick, would you be able to post the times and peak levels for various stations on Lockyer Creek, the Bremer River and the Brisbane River. I would find it very helpfull. Just the peaks, of course, I'm not asking for the full profile.

  34. As recently as 20/12/10? Chuckle.

    Underlying this is a real need to raise Wivenhoe,though neither party seem to want to talk about this for whatever reasons. Maybe the record will show that neither party showed any interest in the appraisals that have been made.

    The options for 2,4 and 8 metre raisings of the dam wall,with commensurate raising of the bottom lip of the existing spillway,and the compensatory provision of extra spillways,have been presented a few years back. They are not cheap,but by no means unaffordable, The lower two options provide something like 200 and 500 extra gigalitres flood compartment,which provides greater flexibility to deal with Lockyer/Bremer complications as in this last flood. I really don't think Brisbane's water grid,which is pushing 3,000,000 consumers,should have to tolerate a lower safe yield from its primary water source.

  35. Tom,excellent letter from Owen Droop. Unfortunately,it does not seem to have tempered Hedley Thomas' rubbish.

    Back to the peak estimates...well you may be right,but I'm not so sure. The press releases of 8.30AM following specifically state the peak release was 645,000ML/day transient [7465 cumecs] 'overnight'. The peak in the hourly figures[597GL/6945 cumecs] at Savages,c.18km downstream, is in the vicinity of 1AM Wednesday,and flows stay above 6,800 cumecs for some 4 hours after this.I think this indicates a peak spillway release closer to midnight than to 8.30Pm Tuesday. Low as the gradient is,I think the peak could move at over 10km/hour.

    As a slight aside ,the release rate was dropped quite steeply to 205 GL/2305 cumecs within 8-9 hours of that 7465 extreme,so that means the Lockyer peak was making up its greatest proportion of the total flood, and was at its highest after midnight 11th. This squares with the Rifle Range Road data [26km upstream of the junction] which implies a peak just before midnight.I say 'implies' because the water reached heights beyond the gauges ability to measure [16.44m] around 8 or 9PM Tues 11th. It stayed above this for at least 10 hours. Because of this I have to estimate from the available hourly data that the two day volume of the Lockyer flood was at least 250GL,and possibly 300GL. The transient peak may have been 1650-1750 cumecs. The BOM gauge just up from Rifle Range Rd at Lyons Bridge,broke the 1974 mark by 63cm.

    The Brisbane at Mt Crosby hit 26.14 which fell 60cm short of 1974's peak [which then picked up a much higher Bremer than 2011] I did not keep the time,but it was well before 6AM Wed

    Estimating the Bremer is harder,because you have to aggregate the Bremer at Walloon with Warrill Creek at Amberley and Purga Creek at Loamside,as DERM provide no public volume data at any point downstream. It's not hard to add these up,but it leaves a few hundred km2 unmeasured including the substantial Bundamba Creek.

    Anyway,adding the three day totals for the three streams,we get 40GL for the 24hours of the 10th,110GL 11th,and 150GL 12th [unfortunately,I didn't keep figures after that]. So a minimum 300GL went down towards the tidal backwater of the lower Bremer over those days. Who knows the 24hr figure through the peak 11/12 may have been 200 GL.I will not guess a transient peak,but just note that the bridge at Ipswich measured 19.4m,1.3m short of 1974,but still the third highest on record.

    But any really high level like that on the Bremer at Ipswich means that the Brisbane River is playing a big part in backing up water. The 5/1996 flood at Ipswich [11.31m] was big,probably as big locally as 1/2011,but the Brisbane River above Lockyer did not contribute a drop.The Lockyer did though,with a flood peak a meter less than 1/2011. Even so the river in 1996 was six metres lower at the bridge at Ipswich.

    To sum up all the peaks were quite close together,and I believe the Wivenhoe operators were trying to hold on until more of the Lockyer had moved downstream,but were spooked by the rain rates,rain distribution,forecast of more for at least 6-12 hours and conditions at the Somerset Dam. They may have made the big transient release just at the very time the heavy rain suddenly ceased around 11PM Tuesday,which would be a cruel twist.