Opinion Challenges with floods & forecasts Brisbane SEQ, NENSW

Discussion in 'Eastern & Marine' started by Flowin, Nov 29, 2019.

  1. Flowin

    Flowin One of Us

    Joined:
    Jul 5, 2019
    Messages:
    393
    Likes Received:
    1,961
    Location:
    Pinjarra Hills, QLD
    Discussion topic about floods and forecasts.
    In the sub tropics particularly say in domain of the NE/NSW SEQ thread our weather is diverse.
    We get summer rains typically, but not regularly.
    We can sometimes get the monsoon influence but not regularly, we can sometimes get ex TCs but not often actual TC strength, we can get ECLs, we get troughs moving through some with moisture feed from tropics, others with driving influence from south.
    Difficult rain forecasting for our region because we are on the edge.
    Then our floods are also diverse. Big floods in the record are notably January or summer, but if you dig deeper in the records you also find floods in all months, and some like June 1893 and 1983, May 1996 and many other winter ECLs.
    We also gets floods in some rivers and not the next to it. How many people know of the jan 1947 floods ( yes I mean 1947, not 1974 ) Brisbane river no flood, Logan river just to the south it was epic flood.
    Forward to now today we got the 2011 floods class action decision - which said - forecasts should have been used.
    I have already made comments here <a href="https://www.ski.com.au/xf/threads/ne-nsw-se-qld-weather.85082/page-58#post-4106314">NE NSW / SE QLD weather</a>
    It ain’t easy for our region. Meanwhile in early 2019, the Brisbane River Floodplain Management Strategy was released. It identified flood damages if averaged between no flood years and flood years are $300 million per year in Brisbane and Ipswich floodplains. Many years no flood, then once in a while a multi billion dollar flood. That will only get worse with climate change.
    Forecasts are only one part of the picture but a very important one in among the mix of other things like building better resilience.
    So that’s the theme for this thread- May the discussion live a long time cos it is not a problem we can forget, even if in 2019 we are in drought.
     
    #1 Flowin, Nov 29, 2019
    Last edited: Nov 30, 2019
  2. Nature's Fury

    Nature's Fury Hard Yards

    Joined:
    Nov 1, 2019
    Messages:
    76
    Likes Received:
    507
    Location:
    Brisbane (Western Suburbs)
    Gleno71, comet and Flowin like this.
  3. Inclement Weather

    Inclement Weather Early Days

    Joined:
    Sep 27, 2019
    Messages:
    8
    Likes Received:
    53
    Flowin, thanks for starting this thread. I read your comments in the previous thread in response to my post, and read your post here. I agree with your point that forecasting is not a precise science. However, forecasting has advanced in leaps and bounds with numeric modelling and ensembles. The PME (Probability Matched Ensemble - what we know as the WATL ensemble) was referenced frequently in the judgment. It should be noted that the BoM has in its notes the following on the PME:
    "Testing has shown that it easily outperforms a random guess or climatology, and also the forecast from any individual weather model, to at least five days ahead. This does not mean it is perfect - no forecast system ever will be - but it does mean that people can be confident that on average, the forecast will supply them information to help shape their decisions." (my emphasis)
    Indeed, Justice Beech-Jones at [16] Chp 6 of his judgment observed:
    "It can be accepted that the precise combination of weather events, the precise rainfall amounts and the distribution of rainfall were not readily capable of being predicted in the immediate period prior to 9 to 12 January 2011. However, as Professor Walsh and Mr Kane stated, there was a strong likelihood of higher than average rainfall throughout the entire period of December 2010 to January 2011. The forecasts and internal assessments that were produced during the period of the flood event pointed to a likelihood of very large falls occurring in an already saturated catchment. In those circumstances, there was clearly a reasonable possibility of rainfall in, around and below the upstream catchments in amounts higher, sometimes much higher, than the forecasted amounts and which approximated to the amount of rain that actually fell." (my emphasis)
    The significance of this passage, is that the flood engineers should have placed greater weight on the prevailing la nina event and err on the side of possible greater rainfalls that the PME suggested, which is exactly what transpired.
     
  4. Flowin

    Flowin One of Us

    Joined:
    Jul 5, 2019
    Messages:
    393
    Likes Received:
    1,961
    Location:
    Pinjarra Hills, QLD
    There were no instructions in the manual or even mention of La Nina. The decision makers were engineers not meteorologists. The evidence of Walsh and Kane was meteorology not dam operation decision making.
    PME at that time I think was 50km grid, and about six months or so beforehand was a 100km grid,
    The WATL maps you see on the web are downscaled from a courser grid. There aren’t actually many grid points covering the catchment.
    Big risk of errors simply from location displacement upstream and downstream. You would need a super-ensemble to deal with that. What I mean by a super-ensemble would be put the 50 ECMWF, 20 Access, 20 GFS ensemble members and other weather models, on to the catchment models to generate a true ensemble of catchment flow possibilities. Putting just one average rainfall of all the weather models onto the catchment model is not ensemble forecasting for dam operation, it is not risk management, and actually....that would be risk creation.
     
    #4 Flowin, Nov 30, 2019
    Last edited: Nov 30, 2019
    Gleno71 and comet like this.
  5. Flowin

    Flowin One of Us

    Joined:
    Jul 5, 2019
    Messages:
    393
    Likes Received:
    1,961
    Location:
    Pinjarra Hills, QLD
    Below is an example of EC 50 ensemble member precipitation variations I posted back around mid September (around page 28 of the main SEQ thread). The scale of those images north to south is Taree NSW to Rockhampton in QLD, about say 1500km.
    That example shows vast variations because it is for eight day totals, but if you look at others there is still a lot of variation even at two and three day lead time totals.
    The errors in location of heaviest rainfall are what make forecasting catchment flood flows most challenging. Displacement of only a few tens of kilometres can mean the difference between small or modest flood in one river and major in another not far away.

     
    Gleno71 and comet like this.
  6. Michael Hauber

    Michael Hauber One of Us

    Joined:
    Jul 4, 2019
    Messages:
    210
    Likes Received:
    775
    Well it was my belief the manual did not require the engineers to take into account forecasts prior to this new judgement, however this judgement seems quite insistent that the manual did require forecasts to be taken into account. I am surprised by this.

    And while the engineers may not be meteorologists they are surely smart enough to take forecasts into account, and being flood engineers you would hope they might know some basics of meteorology, similar to many of the non meteorologists on this forum.

    To not take into account the forecast rainfall amounts is to effectively forecast 0 rain, which was clearly a less accurate than what was forecast. Of course good decision making should consider a range of possibilities which can complicate things a lot.
     
    Gleno71, comet and Flowin like this.
  7. Flowin

    Flowin One of Us

    Joined:
    Jul 5, 2019
    Messages:
    393
    Likes Received:
    1,961
    Location:
    Pinjarra Hills, QLD
    That’s the most important part Michael.
    If the forecast is wrong what consequences have I caused by acting on it?
    Inconvenience yeah ok that would be unfortunate.
    But flooding people’s property and businesses that’s a lot more serious, on speculation that may be 50% chance right, or 50% chance wrong is not sensible risk management.
    Similarly relying on forecast rainfall to dump one years equivalent water supply storage, hoping forecast rainfall will fill it back up again 50% right / wrong does not seem like sensible management of risks for water supply that serves 3 million people.
    Think about it like this would you build your house, or your business in a location that floods, with a 50% chance you might be right or wrong?
    Risk management for high consequences usually demands lower chances of getting it wrong.
    the class action which has only analysed the 2011 flood, has probably found an alternative way of operation (without ensemble flows) that would fail in a majority of other flood situations. So what- well unnecessary flooding caused by use of forecasts would equally be grounds for litigation. So even if the forecasts were that good that you could get a better result 9 times out of ten. Then the one time out of ten it is wrong you still get sued.
     
    Gleno71 and comet like this.
  8. Jacko4650

    Jacko4650 One of Us Ski Pass: Silver

    Joined:
    May 15, 2014
    Messages:
    1,775
    Likes Received:
    2,791
    I am not a weather dude, rather a skier who appreciates the talent around Alpine winter forecasting and obs on this forum. Nonetheless, I'm from Maryborough where we suffer from floods, the most recent 2013 which was really, really bad and 2011.
    Data has improved quite a bit, forecasting definitely so but we still have many reservations as flood mitigation efforts have all but dried up in line with new personnel that come along with various Council and State roles.
    River height gauges are critical (too see what's coming), but so too are rainfall gauges (to predict what may be coming). We have a business that can handle water heights up to 6m but once it goes past that we have to start evacuating assets and people. Waters went to 10.6 in 2013. To go or not to go is the tricky dilemma we face as evacuations take a day at least. The biggest drama we had in 2013 was BOM holding back data because the figures didn't look right. Water started inundating well before the BOM data was live, rendering it a completely useless system.
     
    Gleno71, comet and Flowin like this.
  9. Flowin

    Flowin One of Us

    Joined:
    Jul 5, 2019
    Messages:
    393
    Likes Received:
    1,961
    Location:
    Pinjarra Hills, QLD
    @Jacko4650 i agree with much of what you say but I think you need to clarify your last sentence. Bom don’t hold data back, but it is forecasts that may sometimes be delayed until they are comfortable it is sensible
     
    Gleno71 and comet like this.
  10. Inclement Weather

    Inclement Weather Early Days

    Joined:
    Sep 27, 2019
    Messages:
    8
    Likes Received:
    53
    The judgment placed a heavy emphasis on the Manual and to its compliance. The overwhelming conclusion was that the flood engineers at the end of the day did not comply with it. An important factor that should be borne in mind is that the dam has two functions: water storage and flood mitigation. The water should be at or below the Full Supply Level (FSL 65m) during a flood event. Obviously, this is to leave room in the dam for flood mitigation. The Manuel does not explicitly state that rainfall forecasts should be used in 'predicted' inflows, but, the judge held, on its proper construction, that 'predicted' included rainfall forecasts.
    He also held that the interpretation of the manual should be referenced to its 'objectives': At [253] Chp 3, he observed: "Ultimately the conclusion that the Manual requires that rainfall forecasts be considered in making decisions on dam releases follows inexorably from the statement in section 8.4 that “[w]ithin any strategy, consideration is always given to [the flood] objectives in [their] order [of priority], when making decisions on dam releases” and from the above conclusion that strategies are determined by predictions that incorporate rainfall forecasts."
    I must say, as a lawyer, this judgment is ripe for an appeal on a number of points as it is based broad assumptions and a broad interpretation of the Manuel. That being said, it would be a brave political move by the Government to appeal it in the face of so many claimants and public sympathy towards their plight.
     
  11. Flowin

    Flowin One of Us

    Joined:
    Jul 5, 2019
    Messages:
    393
    Likes Received:
    1,961
    Location:
    Pinjarra Hills, QLD
    I am not sure if you quoted that from the judgement or interpreting somehow else.
    The full supply level for the drinking water storage (FSL) at time was 67m (that was never in doubt). The understanding of the engineers was not to go below that level. When I read the relevant version of the flood manual applicable for 2011 flood that was my understanding.
    Somehow the judge has found they could go below that level - which I find astounding. But I have not yet read the judgement.
    I am not a lawyer, but I do note your point.
     
    #11 Flowin, Dec 1, 2019
    Last edited: Dec 1, 2019
    Gleno71 and comet like this.
  12. Flowin

    Flowin One of Us

    Joined:
    Jul 5, 2019
    Messages:
    393
    Likes Received:
    1,961
    Location:
    Pinjarra Hills, QLD
    @comet, I have responded to you post here, rather than the day-to-day as this thread is specifically for the topic at hand.
    It is not quite that simple as is it is not really practical to say can't build on floodplains, because that would remove large parts of existing established areas Brisbane and Ipswich.
    If you look at the Brisbane River Catchment Flood Study Maps on the Queensland Reconstruction Authority website it shows much larger flood extents up to the 1 in 100,000 annual chance flood.
    It needs to a risk based approach, where the risk is accepted because the probability of flooding is very low.
    In the past that was roughly labelled the 100 year flood, but many people have misintrepreted that to mean a flood that only happens once every hundred years, whereas what it actually is a flood that has 1% chance ( 1 in 100) in any year.
    The 1 in 100 flood has 50% chance of occuring in a seventy year timeframe. I think the appropriate levels should be about the 1in 500 or 1 in 1000 annual chance flood.
     
    #12 Flowin, Dec 1, 2019
    Last edited: Dec 1, 2019
    one drop, Gleno71 and comet like this.
  13. comet

    comet Hard Yards Ski Pass: Silver

    Joined:
    Jul 7, 2019
    Messages:
    84
    Likes Received:
    329
    Location:
    Kangaroo Point, East Brisbane
    Thank you Flowin - I do understand that a lot better now. From what you have just said the chances are very low indeed.
    I, for one, would not be prepared to build my dream house in a flood area though. Floods are so destructive and deadly.
     
    Gleno71 and Flowin like this.
  14. Flowin

    Flowin One of Us

    Joined:
    Jul 5, 2019
    Messages:
    393
    Likes Received:
    1,961
    Location:
    Pinjarra Hills, QLD
    the irony is that flood magnitude related to say height at a river gauge or, flow along a river as key metrics for flooding inundation are straight forward to map. More useful during real events also to emergency agencies to understand escalating flood levels and consequences. And Easier than mapping bushfire hazard.
    Putting probability numbers on river reach peak flood for flow and heights often relies on past data for river records, gauge flow estimate, or rain records and probability analysis of the records.
    Climate change may change some of the rain probablities, and flow probability, so the likelihood probablity may change.
    But the maps that use flow in a reach, or height at a gauge remains more useful over the course of time even if the chance numbers and labels change.
     
    comet and Gleno71 like this.
  15. Nature's Fury

    Nature's Fury Hard Yards

    Joined:
    Nov 1, 2019
    Messages:
    76
    Likes Received:
    507
    Location:
    Brisbane (Western Suburbs)
    I was reading some articles online today and they seemed to suggest that the court finding proves the Brisbane/Ipswich flood was a 'man-made' disaster and responsible for most, if not all, of the flood event that hit these two cities. I thought there were three major contributors to the flood event: inflow from catchments below the dam, water up the Bremer River and the dam releases. I though the dam releases were only about 1/2 to 1/3rd of the total water volume? Can anyone clarify this.
     
    comet, Gleno71 and Flowin like this.
  16. Flowin

    Flowin One of Us

    Joined:
    Jul 5, 2019
    Messages:
    393
    Likes Received:
    1,961
    Location:
    Pinjarra Hills, QLD
    Numbers that I am aware of are around 2,700,000 million litres into Wivenhoe and around 4,000,000 million litres flowed past Moggill not far downstream of Bremer River.
    The flood storage of Wivenhoe dam for flood mitigation was close to 1,000,000 million litres. So the inflow flood was around 2.7 times the dam’s temporary flood volume. With that inflow volume letting that go at some point probably unavoidable.?
    But these ratios are not necessarily representative of contributors to peak flow and levels, because that’s just the volume and not peak flow or levels contribution.
     
    #16 Flowin, Dec 1, 2019
    Last edited: Dec 1, 2019
    comet and Gleno71 like this.
  17. Inclement Weather

    Inclement Weather Early Days

    Joined:
    Sep 27, 2019
    Messages:
    8
    Likes Received:
    53
     
    comet likes this.
  18. Inclement Weather

    Inclement Weather Early Days

    Joined:
    Sep 27, 2019
    Messages:
    8
    Likes Received:
    53
    Yes, the figure is 67m not 65m as I had incorrectly stated.
     
    comet, Flowin and Gleno71 like this.
  19. Gleno71

    Gleno71 One of Us

    Joined:
    Jul 4, 2019
    Messages:
    252
    Likes Received:
    1,170
    Location:
    Gold Coast, Queensland
    So if im not mistaken, the 2011 floods were caused by the floodgates not being open sooner then went into panic mode? Or was it just the excessive amount of rainfall from that storm in Toowoomba? What was the original forecast for that region on that day? Did models originally forecast the amount of rainfall in the Toowoomba region ?
     
    comet likes this.
  20. Michael Hauber

    Michael Hauber One of Us

    Joined:
    Jul 4, 2019
    Messages:
    210
    Likes Received:
    775
    A big issue is whether the operation of the dam was in accordance with the requirements of the flood manual. The judge decided that it was not as forecast rainfall was not taken into account, and the judge felt that the manual clearly stated that forecast rainfall should be taken into account and referred to passages such as this in section 8.4 of the manual:

    However I think there is some ambiguity in this passage, and it is not clear from this passage how much weight should be put on the forecast rainfall amounts. Should the actual amount forecast be included in the amount of water likely to flow into the dam, or should the dam operator take a general note of the forecast, and if in doubt adjust strategy according to the forecast. Eg if its borderline whether to release or not release to protect the dam against overflowing, and heavy rain is forecast in the catchment that might shift the decision to make release.

    The general overview of the manual is detailed and explicit about how forecast rainfall should be used. It states that BOM forecasts are useful for determining when a rainfall event is over. It also states quite clearly that forecast rainfall should not be taken into account when modelling inflows:

    Seems pretty clear to me on this basis that the judge was wrong in this decision. However note I'm referring to the current version of the manual, so its possible that the manual has been changed since Jan 2011, although the change history does not make any mention of changes of this nature. It does seem odd to me that the judgement seems to make no note of the defendant trying to use this argument as a defence so I might have missed something.
     
    comet and Gleno71 like this.
  21. Jacko4650

    Jacko4650 One of Us Ski Pass: Silver

    Joined:
    May 15, 2014
    Messages:
    1,775
    Likes Received:
    2,791
    The data was 4 hours old (on average) for the critical river height gauges up-river from us. We have the records (screenshots). We believe it was an algorithm that prevented the publication of data if readings were outside "expected" variations (ie: to stop 'faulty' readings). Turns out there was nothing faulty, rather the variations WERE extreme - exactly what we needed to know. That lack of raw data cost us hundreds of thousands of dollars as we wasted time on stage 1 preparations (fences, low-cost equipment etc) when we should have started immediately on stage 3 evacuations (people and movable buildings) It was not a case of a person at BOM saying they would hold data back, rather software that prevented it from being displayed. We have been assured this will not occur again, along with more devices being installed. You should try contacting BOM in the middle of a disaster to find out what is happening with the data!
     
    one drop, comet and Flowin like this.
  22. Jacko4650

    Jacko4650 One of Us Ski Pass: Silver

    Joined:
    May 15, 2014
    Messages:
    1,775
    Likes Received:
    2,791
    Past records are critical but can be dangerous to use if the river has been changed since the more extreme flooding events happened. For example, the Mary River is a vast system with a catchments that extends well into the Sunshine Coast Hinterland, down to the tea tree lakes north of Noosa and out west. Barages have been installed and creating silting within the river system, whereby only extreme water flows will begin to move that silting. The volume of water "within" the river system therefore reduces, so where a rain event from the pre-barage days of X volume of rain produced A levels down river, it will now be different. By way of example, mangroves never grew on the river banks around Maryborough due to the flow of fresh water, but with much of that held back by barages for irrigation, the mangroves are taking over.
     
    Flowin, one drop and comet like this.
  23. Flowin

    Flowin One of Us

    Joined:
    Jul 5, 2019
    Messages:
    393
    Likes Received:
    1,961
    Location:
    Pinjarra Hills, QLD
    @Jacko4650 those points you make are correct. Natural and man-made changes occur in rivers which mean that for a given flow rate, the levels may be higher or lower depending on what changes have occurred.
    Vegetation change (either seasonal or long term change) is an influence that can be simulated in hydraulic models but not with great accuracy.
    Denser vegetation particularly like mangrove and melaleucas can change flooding quite a bit, higher levels at and upstream of the vegetated area, but it also slows down flows so can produce lower levels downstream of the vegetated area.
     
    Jacko4650 likes this.
  24. Flowin

    Flowin One of Us

    Joined:
    Jul 5, 2019
    Messages:
    393
    Likes Received:
    1,961
    Location:
    Pinjarra Hills, QLD
    For those that have commented on houses and development on the floodplain, the following chart may be of interest.
    Earlier this year the Brisbane River Strategic Floodplain Management Plan was released, together with a Technical Evidence Report of the studies done for the management plan.
    Buried deep in the technical evidence report (around page 208) is the chart below.
    It shows for a given magnitude flood (height at Brisbane City gauge on y-axis) the number of inundated buildings in 1977 (orange lines) and 2017 (blue lines)
    Dotted lines are residential buildings
    Dashed lines are non-residential buildings
    Solid lines are total (residential + non-residential).
    The flood heights at the city gauge
    Historically 1893 flood about 8 m, 1974 flood about 5.5 m 2011 flood about 4.5 m.
    From the Brisbane River Catchment Flood study, latest estimates of flood probabilities (without effect of climate change)
    1 in 100 annual chance flood is about 4.5m
    1 in 500 annual chance flood is about 7.3m
    1 in 2000 annual chance flood is about 10m

     
    Jacko4650 likes this.
  25. Jacko4650

    Jacko4650 One of Us Ski Pass: Silver

    Joined:
    May 15, 2014
    Messages:
    1,775
    Likes Received:
    2,791
    Data is good. I purchased a house in Hill End, inner Brisbane, perhaps 60m from the river that was halfway up a hill. I was VERY surprised when I was given a flood reading on the purchase to a level that inundated the bottom level but not the top in 1974. I was surprised, but nonetheless planned for a small chance of inundation (eg: 5%) of the lower level and 1% for the top level. Everybody talked about the 1974 floods and the new dams that would ensure it never happened again. Old neighbours, back when Hill End was a series of diary farms, told me about boats going over the houses at the bottom of our street (Grey St) during the 74 floods. Wow!
    A few months after the 2011 floods I drove around my old neighborhood to see what had happened. The devastation was immense, piles many meters high on side streets but almost entirely due to "under house" renovations, with practically no damage on top levels. There was a reason for this. The so called "security" of dam walls after 1974 lulled people and insurance agencies into believing floods would not happen again.
     
    Flowin likes this.