| You Get 50% Of $67-thousands Of Keywords, Send Just 20 Clicks Right Now!! | Biggest Niche On The Internet... |
Table of contents for CSIRO/BoM DECR
- Temperature Index Drought
- Comparison of Models and Observations in CSIRO/BoM DECR
- Western Australia Future Rainfall
- DECR: The message starts to slide
- Scientists Biasing Research
There must have been some way that the models of exceptionally low rainfall (drought) were validated in the CSIRO/BoM Drought Exceptional Circumstances Report. Usually, models are checked against observations to make sure they have ’skill’ at the purpose for which they are intended. In this case, the global climate models used in the Drought Exceptional Circumstances Report must have been compared with observations, before they were used to show increasing frequency and severity of droughts in the next 30 years.
-
- My replotting of figure 10 of the Drought Exceptional Circumstances report, by region, with the last panel showing data for all regions.
-
- Figure 10 of the Drought Exceptional Circumstances report, the exceptional low rainfall observations are plotted against model projections.
In Figure 10 of the report (above right), the exceptionally low rainfall observations are plotted against model projections. However, the figure is more of a cartoon and hard to read, so I replotted it using data from the report provided by Kevin Hennessy (above left). There are a number of other difficulties with Figure 10 that make it hard to see miss-match between the models and observations. The y axis is a large 30% exceptional low rainfall area, the drought observations are averaged over 10 years instead of the usual climatological average of 30 years, and the lines of the model are different again, with a jagged appearance. The confidence intervals are not standard deviations, but yet another novel metric. In replication of their figure 10, using R for the statistics, the last panel shows data for all regions.
On the figure in red are the 30 year moving average of observed percent area with exceptionally low rainfall. In black with dashed confidence intervals (1 s.d.) are the averages of the 13 models used in the study. This is the same data as figure 10.
It looks to me that in the last half-century of observations (1950-2007) in almost all regions droughts are decreasing (red), while the models show drought increasing (black). SW-WA and VicTas are a little different as the observations are constant, while models are increasing.
These graphical observation are consistent with the numerical results in Table 1 of the report Tests of Regional Climate Model Validity in the Drought Exceptional Circumstances Report that the regions SW-WA and VicTas are the only regions where the models of drought are not of the opposite sign to the observations.
So we conclude that while the observations of drought are decreasing in the last 50 years (in terms of a climatological average of frequency and areal extend of exceptional low rainfall), the projections of drought are increasing over the same period. This is supported by the last panel in the figure where all regions are averaged, the models predict increasing droughts, but the observations show decreasing droughts. The models go in the opposite direction to observations.
I have been in contact with Kevin Hennessy a number of times about the report, but unfortunately cannot report much progress. He maintains that the authors were satisfied with the validation of the models in the report, but has not provided details of the validation procedures or results that they used. Unfortunately the validation of the models was not reported in the DECR either. The only reference I can find is in Section 4.3 where it states:
The observations are generally within the range of individual model results.
It seems like the ’skill’ they are thinking of is in the range of model results enclosing the observations (generally). But projections of linear regressions are the more usual way of projecting model results, and more statistically well defined. Moreover, their statement above is not quantitative, and is done by what is known as ‘eyeballing’. It is not a validation test, per se.
As the data provided by Kevin Hennessy as used in the report were obtained after considerable pressure from a number of blogs across the web, originally withheld due to ‘Intellectual Property’ reasons, I was hoping he would be more forthcoming this time.
Looking at the forward projections of the models, droughtedness increases in SW-WA, SWAust and VicTas, but decreases in the other models. Given that the trend of models don’t resemble the observations at all in the 1950-2007 period, I think little confidence can be put in these forward projections. It is very strange to find models performing so differently to observations, and yet the projections of the consequences of warming to be stated with such certainty, as in the quote from the summary (below). Note also that the regions where model projects decreasing droughts are interpreted as ‘little detectable change’, suggesting strong interpretive bias in order to create alarm over increasing droughts.
If rainfall were the sole trigger for EC declarations, then the mean projections for 2010-2040 indicate that more declarations would be likely, and over larger areas, in the SW, SWWA and Vic&Tas regions, with little detectable change in the other regions.
I am very curious to see the methods of validation they used and the actual results they obtained. I would also like to know why the results of the model validation were not reported.
27 responses so far ↓
David,
I’d suggest you change to ‘CSIRO/BoM’ in the heading and the first sentence. The BoM logo is on the front cover of the Report, and again on the back page. As you say, the models must have been validated in some way. The BoM authors presumably have the data that supports the validation. The BoM should have ensured that this was published in the DECR - in any case the Bureau must be in a position to publish the details and should do so.
I don’t agree that it’s for Kevin Hennessy to say yes or no to the release of this information. The DECR is a report to Government from the two agencies. I think the heads of both of the partner bodies - Greg Ayers, Head, CSIRO Marine & Atmospheric Research and Geoff Love, Director, BoM - should each be asked to publish this information.
If Greg and Geoff take the view that the information is now ‘owned’ by the Bureau of Rural Sciences because the BRS paid for it, then the head of BRS should be asked to put the material into the public domain. Or the Secretary of DAFF - SOMEBODY must be responsible for making this important information available.
The models in question may have been some of those “validated” for various CSIRO climate reports. (I use teh term loosely for reasons that will become obvious.)
In my analysis of these reports (see http://mclean.ch/climate/EE%2017-1_03%20McLean%20ok.pdf) I showed that the CSIRO accepted climate models that were frankly woeful.
Of course it wouldn’t take much for them to better one of the early CSIRO alarmist reports about the Murray Darling Basin that consistently predicted historical rainfall at about 55% of of what it really was. (see http://mclean.ch/climate/Murra.....limate.htm).
I suspect that the major problem is the over-emphasis on carbon dioxide and the glaring downplaying of El Nino-Southern Oscillation (ENSO) events - El Nino, La Nina and conditions approaching both - despite the fact that the joint CSIRO-Bureau of Meteorology report of 2007 indicated that ENSO influenced every climate element for Australia.
John, If by validation you mean the paper by Suppiah then there was no validation. As Ian pointed out in a previous, the models in the DECR report are not the same subset as the models in Suppiah. AFAIK the full suite of 23 IPCC models were using in the Australian Climate 2007 report.
Even if the models showed skill at temperature, and rainfall, they would not necessarily show skill at exceptionally low rainfall. Penny Whetton made this point in a recent paper. Climate scientists know this.
To use a model for forecasting drought, you must determine its skill at modelling drought first.
David
The drought issue is a significant and recurring problem for agriculture in Australia – http://en.wikipedia.org/wiki/Drought_in_Australia and the drought report you’re commenting on would not be on the agenda I suspect if the Australian Federal government and various state governments had not spent billions and billions of dollars in drought aid over the last 17 years or so; and that the nation finds itself with a Murray Darling Basin with record low inflows and all the associated impacts to industries, towns and individual families. Also many Australian capital cities have also experienced drought periods and water shortages in recent years.
So although Australia has very high rainfall variability from its exposure to El Nino, some of us have wondered if things have changed climate-wise given recent drought sequences. Are the last 120 odd years of rainfall records our best information for making significant planning decisions about our agricultural and water supply investments i.e. for government, water resource managers, agribusiness and farmers themselves - whether to stay on farms or go, invest, retreat, support, withdraw, diversify, innovate or move to northern Australia.
Or has the probability distribution changed due to natural or anthropogenic influences?
Do we need to fundamentally re-evaluate our position?
So what you are discussing is of some pith and moment.
Much of the preamble much of the argument I would use is already tabled here in the lead post – and other subsequent comments on cost of drought and evaporation fine points here. http://bravenewclimate.com/2008/08/24/dr-jennifer-marohasy-ignores-the-climate-science/#comments so I’ll take that as read.
So just a few observations – the classic BoM rainfall time series for eastern Australia is here http://www.bom.gov.au/cgi-bin/silo/reg/cli_chg/timeseries.cgi?variable=rain®ion=eaus&season=0112
The wetter 1950s and 1970s stand out in that series. A period of La Ninas and negative IPO activity. There is a fair bit written about decadal and quasi-decadal phenomena – an example being http://www.bom.gov.au/bmrc/clf.....m_2005.pdf
So any trend before 1950 would be up and after 1950 down. What does this really mean if anything?
Students of Australian drought would not miss the devastating Federation drought of 1895-1902. An account here http://www.bom.gov.au/lam/climate/levelthree/c20thc/drought1.htm Indeed not going back to the 1890s in the Bureau’s 1900 onwards time series (link above) misses the development of that Federation drought which is the gold comparison standard for all future droughts - spatially and severity. http://www.bom.gov.au/lam/climate/levelthree/c20thc/drought1.htm
However there were also much wetter periods – for example Queensland is noted for its floods.
Significant floods have occurred several times since the European settlement e.g. in Brisbane region. The most significant of these events was the 1974 Brisbane Flood. Notable instances of flooding include:
* 14 January 1841 - highest flood level to date
* March 1890
* February 1893, a sequence of flood peaks over some three weeks saw the highest recorded flood level in the Brisbane central business district.
from http://en.wikipedia.org/wiki/Brisbane_River
and in detail
from http://www.bom.gov.au/hydro/flood/qld/fld_reports/brisbane_jan1974.pdf
A chart showing all the floods at the Brisbane Port Office since 1841 is presented
in Fig 2. ……. (Note in 1893, two floods bigger than 1974 flood only a few weeks apart and four peaks overall !)
The earliest flood recorded was in 1841. Its exact height is uncertain but it was
said to be the highest flood known at that time………
AND we can go even further back still and analyse the coral core records off north Queensland and calculate a likely inflow into the massive Burdekin River catchment. http://www.nrw.qld.gov.au/wrp/gif/burdekin_bmap.gif Annual (water year, October-September) runoff (mm) of the Burdekin River, Queensland, Australia was reconstructed from intensity of fluorescence measured in two coral cores from Havannah and Pandora Reefs
which lie in the Burdekin River plume. The data are here: ftp://ftp.ncdc.noaa.gov/pub/data/paleo/coral/west_pacific/great_barrier/burdekin_2001.txt
So we can estimate wet and dry seasons from 1644 and forward till 1980 – so any longer term analysis might suggest that the trajectory for north-eastern Australian rainfall at least could be downwards.
You could look up other paleo studies for other parts of Australia – speleotherms, crater lakes, pollens, sediments etc. These may also reveal drying trends e.g. http://www.sciencedirect.com/science?_ob=MImg&_imagekey=B6V6C-430G27H-D-18&_cdi=5811&_user=613232&_orig=browse&_coverDate=06%2F01%2F2001&_sk=997539998&view=c&wchp=dGLbVzz-zSkzV&md5=b5c5263604e1409c15006f9408fbcdb3&ie=/sdarticle.pdf
Other investigations are underway http://www.ansto.gov.au/__data/assets/pdf_file/0018/25821/rainfallconf_release.pdf
SO all depends what year you want to start analysing. Starting in 1900 is VERY arbitrary.
The next point is the relevance of the area of drought – the Federation drought being a large drought area wise while the recent Millennium drought being more severe in specific areas.
But those smaller areas can be very significant – capital cities water supplies or the Murray Darling headwaters worst on record. Two current cases of river and dam inflows (Murray River and SE Queensland) are documented in references here. http://bravenewclimate.com/2008/08/24/dr-jennifer-marohasy-ignores-the-climate-science/#comments
And the reason they can be significant is movement in the sub-tropical ridge or location of high pressure cells. Again discussed in the bravenewclimate post. So small changes in the location of macroscale phenomena can have major impacts on highly productive regions. A worst on record drought in the Murray headwaters over multiple years is very bad news if you look at the productivity of the region. http://kids.mdbc.gov.au/encyclopedia/agriculture
So you ask about model validation – the literature reports these attempts at model verification.
http://atmos.es.mq.edu.au/~holbrook/pubs/perk_etal_2007.pdf
http://www.inderscience.com/search/index.php?action=record&rec_id=16108
However how many important rainfall processes are captured – like El Nino, La Nina, IPO, PDO and quasi-decadal effects. http://www.bom.gov.au/bmrc/clf.....m_2005.pdf I’d be surprised if any of the decadal effects were simulated. So without this it would be amazing if you got a statistical relationship of correlation with the models and 20th century record, even though the overall probability distribution might still be OK.
So in summary:
1. Length of record an issue and potentially misleading – 1970, 1950, or 1900 – or 1890? Or 1841 or 1644? What’s the right time period. The Federation drought is a dominant feature.
2. ENSO and decadal signals at various scales an issue
3. Small areas of droughted high value land can be critical
4. Subtle changes in the sub-tropical ridge latitude, high pressure cell location, changes in Antarctic circulation (SAM - ozone), Indian Ocean, Tasman Sea and a decreasing Walker circulation can play out with subtle but significant effects
5. Cyclical inter-annual and decadal natural variability mixed with mixed anthropogenic signals (greenhouse, ozone, land clearing, Asian aerosols) make attribution complex
6. Drought more than just rainfall alone – evaporation (function of solar radiation, wind, humidity and temperature) and antecedent conditions also major factors
7. Multi-year drought sequences an issue for river flow – antecedent conditions (bone dry catchments take lots of rain to wet up again) a big issue in generating runoff – so hydrological drought is different from agricultural drought
8. Small less significant changes in rainfall can be amplified when combined with changes in evaporative flux – more variables to plant growth than just rainfall alone.
9. How you define drought is an issue – if you declare droughts after annual rainfall is percentile 5 (1 in 20 years) you might think you would be in drought only 5% of the time, but agricultural systems need time to recover from severe drought. What could the rules for revocation be – achievement of percentile 30 or median (percentile 50) rainfall. So getting out of drought perhaps not as simple as getting in, and you may in drought conditions more than 5% of the time. And given crops take time to grow and harvest and grazing herds take time to recover and regenerate; as an economic unit the farm will still have no income for some time after good rain, so the status and support of drought declaration can persist for some time.
10. Lastly you would only expect the various global climate models to probably represent the broad probability distribution of the Australian climate at the current level of evolution, especially representation of decadal processes. So one wouldn’t be surprised if their output didn’t correlate with the 20th century patterns of drought.
So does that exclude the use of GCMs for generating a view of the future of drought or not?
I’m suggesting this issue is quite complex.
And has statistics any application for helping the farmers, agribusiness, water resource managers and government left with multi-billion dollar decisions still to make ?
In the Exceptional Circumstances Report: “impact of climate change”: To what extent is the process stationary, and autoregressive ?
Looking at the report graph projections, another Hockey stick, wrong continent, should have been a Boomerang ?.
On the block coloured graphs, the black line of actual spends a lot of time outside the error bounds of the GCMs. Is this a violation of a ststistical principle, namely that computed multivariate data cannot, on average, be more realistic than the actual data? (I suppose that assumes that the “actual” data are themselves accurately acquired).
The drought issue is a significant and recurring problem for agriculture in Australia – http://en.wikipedia.org/wiki/Drought_in_Australia and the drought report you’re commenting on would not be on the agenda I suspect if the Australian Federal government and various state governments had not spent billions and billions of dollars in drought aid over the last 17 years or so; and that the nation finds itself with a Murray Darling Basin with record low inflows and all the associated impacts to industries, towns and individual families. Also many Australian capital cities have also experienced drought periods and water shortages in recent years.
So although Australia has very high rainfall variability from its exposure to El Nino, some of us have wondered if things have changed climate-wise given recent drought sequences. Are the last 120 odd years of rainfall records our best information for making significant planning decisions about our agricultural and water supply investments i.e. for government, water resource managers, agribusiness and farmers themselves - whether to stay on farms or go, invest, retreat, support, withdraw, diversify, innovate or move to northern Australia.
Or has the probability distribution changed due to natural or anthropogenic influences?
Do we need to fundamentally re-evaluate our position?
So what you are discussing is of some pith and moment.
Much of the preamble much of the argument I would use is already tabled here in the lead post – and other subsequent comments on cost of drought and evaporation fine points here. http://bravenewclimate.com/2008/08/24/dr-jennifer-marohasy-ignores-the-climate-science/#comments so I’ll take that as read.
So just a few observations – the classic BoM rainfall time series for eastern Australia is here http://www.bom.gov.au/cgi-bin/silo/reg/cli_chg/timeseries.cgi?variable=rain®ion=eaus&season=0112
The wetter 1950s and 1970s stand out in that series. A period of La Ninas and negative IPO activity. There is a fair bit written about decadal and quasi-decadal phenomena – an example being http://www.bom.gov.au/bmrc/clf.....m_2005.pdf
So any trend before 1950 would be up and after 1950 down. What does this really mean if anything?
Students of Australian drought would not miss the devastating Federation drought of 1895-1902. An account here http://www.bom.gov.au/lam/climate/levelthree/c20thc/drought1.htm Indeed not going back to the 1890s in the Bureau’s 1900 onwards time series (link above) misses the development of that Federation drought which is the gold comparison standard for all future droughts - spatially and severity. http://www.bom.gov.au/lam/climate/levelthree/c20thc/drought1.htm
However there were also much wetter periods – for example Queensland is noted for its floods.
Significant floods have occurred several times since the European settlement e.g. in Brisbane region. The most significant of these events was the 1974 Brisbane Flood. Notable instances of flooding include:
* 14 January 1841 - highest flood level to date
* March 1890
* February 1893, a sequence of flood peaks over some three weeks saw the highest recorded flood level in the Brisbane central business district.
from http://en.wikipedia.org/wiki/Brisbane_River
and in detail
from http://www.bom.gov.au/hydro/flood/qld/fld_reports/brisbane_jan1974.pdf
A chart showing all the floods at the Brisbane Port Office since 1841 is presented
in Fig 2. ……. (Note in 1893, two floods bigger than 1974 flood only a few weeks apart and four peaks overall !)
The earliest flood recorded was in 1841. Its exact height is uncertain but it was
said to be the highest flood known at that time………
AND we can go even further back still and analyse the coral core records off north Queensland and calculate a likely inflow into the massive Burdekin River catchment. http://www.nrw.qld.gov.au/wrp/gif/burdekin_bmap.gif Annual (water year, October-September) runoff (mm) of the Burdekin River, Queensland, Australia was reconstructed from intensity of fluorescence measured in two coral cores from Havannah and Pandora Reefs
which lie in the Burdekin River plume. The data are here: ftp://ftp.ncdc.noaa.gov/pub/data/paleo/coral/west_pacific/great_barrier/burdekin_2001.txt
So we can estimate wet and dry seasons from 1644 and forward till 1980 – so any longer term analysis might suggest that the trajectory for north-eastern Australian rainfall at least could be downwards.
You could look up other paleo studies for other parts of Australia – speleotherms, crater lakes, pollens, sediments etc. These may also reveal drying trends e.g. http://www.sciencedirect.com/science?_ob=MImg&_imagekey=B6V6C-430G27H-D-18&_cdi=5811&_user=613232&_orig=browse&_coverDate=06%2F01%2F2001&_sk=997539998&view=c&wchp=dGLbVzz-zSkzV&md5=b5c5263604e1409c15006f9408fbcdb3&ie=/sdarticle.pdf
Other investigations are underway http://www.ansto.gov.au/__data/assets/pdf_file/0018/25821/rainfallconf_release.pdf
SO all depends what year you want to start analysing. Starting in 1900 is VERY arbitrary.
The next point is the relevance of the area of drought – the Federation drought being a large drought area wise while the recent Millennium drought being more severe in specific areas.
But those smaller areas can be very significant – capital cities water supplies or the Murray Darling headwaters worst on record. Two current cases of river and dam inflows (Murray River and SE Queensland) are documented in references here. http://bravenewclimate.com/2008/08/24/dr-jennifer-marohasy-ignores-the-climate-science/#comments
And the reason they can be significant is movement in the sub-tropical ridge or location of high pressure cells. Again discussed in the bravenewclimate post. So small changes in the location of macroscale phenomena can have major impacts on highly productive regions. A worst on record drought in the Murray headwaters over multiple years is very bad news if you look at the productivity of the region. http://kids.mdbc.gov.au/encyclopedia/agriculture
So you ask about model validation – the literature reports these attempts at model verification.
http://atmos.es.mq.edu.au/~holbrook/pubs/perk_etal_2007.pdf
http://www.inderscience.com/search/index.php?action=record&rec_id=16108
However how many important rainfall processes are captured – like El Nino, La Nina, IPO, PDO and quasi-decadal effects. http://www.bom.gov.au/bmrc/clf.....m_2005.pdf I’d be surprised if any of the decadal effects were simulated. So without this it would be amazing if you got a statistical relationship of correlation with the models and 20th century record, even though the overall probability distribution might still be OK.
So in summary:
1. Length of record an issue and potentially misleading – 1970, 1950, or 1900 – or 1890? Or 1841 or 1644? What’s the right time period. The Federation drought is a dominant feature.
2. ENSO and decadal signals at various scales an issue
3. Small areas of droughted high value land can be critical
4. Subtle changes in the sub-tropical ridge latitude, high pressure cell location, changes in Antarctic circulation (SAM - ozone), Indian Ocean, Tasman Sea and a decreasing Walker circulation can play out with subtle but significant effects
5. Cyclical inter-annual and decadal natural variability mixed with mixed anthropogenic signals (greenhouse, ozone, land clearing, Asian aerosols) make attribution complex
6. Drought more than just rainfall alone – evaporation (function of solar radiation, wind, humidity and temperature) and antecedent conditions also major factors
7. Multi-year drought sequences an issue for river flow – antecedent conditions (bone dry catchments take lots of rain to wet up again) a big issue in generating runoff – so hydrological drought is different from agricultural drought
8. Small less significant changes in rainfall can be amplified when combined with changes in evaporative flux – more variables to plant growth than just rainfall alone.
9. How you define drought is an issue – if you declare droughts after annual rainfall is percentile 5 (1 in 20 years) you might think you would be in drought only 5% of the time, but agricultural systems need time to recover from severe drought. What could the rules for revocation be – achievement of percentile 30 or median (percentile 50) rainfall. So getting out of drought perhaps not as simple as getting in, and you may in drought conditions more than 5% of the time. And given crops take time to grow and harvest and grazing herds take time to recover and regenerate; as an economic unit the farm will still have no income for some time after good rain, so the status and support of drought declaration can persist for some time.
10. Lastly you would only expect the various global climate models to probably represent the broad probability distribution of the Australian climate at the current level of evolution, especially representation of decadal processes. So one wouldn’t be surprised if their output didn’t correlate with the 20th century patterns of drought.
So does that exclude the use of GCMs for generating a view of the future of drought or not?
I’m suggesting this issue is quite complex.
And has statistics any application for helping the farmers, agribusiness, water resource managers and government left with multi-billion dollar decisions still to make ?
Luke, Thanks for your long post. Sorry it got caught by the spam filter. I will read your references.
It is worth keeping in mind that the complexity of the issues and my attempts at validation are largely irrelevant.
The issue is a report that acts as if the models have skill at forecasting drought, with no evidence of such skill supplied (or offered after a number of off-line requests). Moreover, a report that intends to be influential for the policy, by painting disaster scenarios, and advocating major overhauls of the EC program.
As you are, I am not surprised that the output of models doesn’t correlate with 20th century patterns of drought. The point is, why didn’t the authors disclose what skill they did find in the report, and why are they withholding it now?
I am used to scientists who like to discuss their research, together with the strengths and limitations of using models. I am intimately aware of such issues and don’t have an expectation of easy answers.
And this documentary presented a 1400 year cyclical drying trend from various southern Australian sediment cores http://www.abc.net.au/catalyst/stories/s1848641.htm#comments {video on web site is best for explanation}
http://ems.anu.edu.au/display_.....page=index
David - sorry - just missed your comment by a few minutes.
Just a few points: it’s a real pithy issue for Australia but one can imagine for other nations, especially Africa.
So I’m encouraging you to follow up some of the issues I’ve raised (”Baby - don’t me this way” as the song goes). Just sparring with CSIRO may be fun but doesn’t help those affected by the drought issue, and doesn’t necessarily progress the intellectual development of practical solutions.
And I’m left with overall issue of problem definition and whether models that perhaps not have the decadal features are useful for future planning.
Alas modern research seems to be very institutionalised with programs, managers, IP, contracts, policy links etc - so makes easy access problematic IMO.
(BTW - I don’t work for CSIRO or BoM)
Luke - “Just sparring with CSIRO may be fun but doesn’t help those affected by the drought issue, and doesn’t necessarily progress the intellectual development of practical solutions.”
While I am inclined to agree with you, I also think that real progress is not possible in an intellectual climate where any old model simulation passes as truth. When the observations of drought trend in the opposite directions to the models, and that is ignored, then it suggests an empiricism basic to science is threatened.
Why don’t you suggest to the authors of the DECR some of your concerns, e.g. that the output of models don’t correlate with 20th century patterns of drought? They say they are satisfied with the validation of their climate models for drought work. They have stated that increasing frequency and severity of droughts are likely. They didn’t base these conclusions on any of the sources you mention so they are not relevant to assessment of the models.
Well I’m just a hobbyist - and they’re all seriously qualified guys - so don’t want to get too beaten up. You see you can get into conspiracy theories pretty quickly here depending on your point of view on things AGW - but I reckon these guys are professional and didn’t come down in the last shower.
I’m saying that the model suite has been put through some levels of validation - the paper I cited above and I imagine various other internal evaluations. So they have a suite of models which has some degree of validation - but is it enough.
Is it specific to drought trends as you have asked.
How well does that validation simulate past drought trends I guess is the question.
Even if the models had all the decadal and quasi-decadal periodicity built in would you still expect the same sequences. Chaos? The 1950s and 1970s wet periods skew the whole trend sequence.
(and that’s if that decadal periodicity really exists as a cycle with a mechanism and isn’t an artifact
- e.g.
- “Even if the various “model initialisation” problems can be solved, is it really possible to predict how the oceans will behave so far in advance? According to David Battisti at the University of Washington in Seattle, who specialises in studying natural variability, there’s a growing consensus that the PDO is just the mid-latitude “debris” left by the past two or three El Niños or La Niñas. If this is right, it means the PDO cannot be predicted long in advance. There is no predictability in the Pacific,” he says. “If there’s any hope for predicting natural variability, it’s in the Atlantic.” Even there, Battisti thinks it will only be possible to make accurate decadal forecasts for tropical regions where there is far less variability from year to year than in higher latitudes.” from http://www.climatechangenews.org/archive.php?p=NS_next-10-years
WOW !
AND
http://www.clivar.org/organization/pacific/meetings/presentations/SPower.ppt
AND from slide 14 here http://www.clivar.org/organiza.....acific.ppt
“The existence of decadal teleconnection patterns in low pass filtered records does not imply decadal predictability
The existence of decadal signals in low pass filtered records does not imply decadal predictability.”
So statistical voodoo? Yoiks…
So I’m left with the quandary of how much GCM validation is possible?
Might getting the broad probability distribution of the 20th century about all you could hope for.
If this enough to give one a future predicted view of drought probability?
Is rainfall enough even – might a full water balance analysis reveal that non-significant changes in rainfall might be amplified to statistical significance by associated changes in evaporation.
Faustian bargains perhaps.
As I said – I’m just a hobbyist and as an Australian tax payer interested in where my billions of farm support dollars and more billions to sort out the Murray River are being spent.
New ideas are needed. Regressing on something more fundamental, not yet determined ? String theorists need to kick the imagination ball ?
You cannot have wind below the land, less rain than no rain, more moisture content than 100%, suggesting braines, surfaces in numerous dimensions.
More fundamental than space, time, temperature, rain ? Holograms are Australians ? Black Hole War for the statisticians .
Luke,
You read as far from an amateur and you should step up and be counted even if you initially feel shy. I do stupid posts all the time but I still sleep at night.
A person with a name like yours should have special knowledge of future climate, especially if your middle name is ‘Sky’.
Remember Admin’s comments above - even though you acknowledge that the CSIRO/BOM have some “seriously qualified guys”, Admin has claimed a lack of model skill that has disturbing implications. The point of your involvement is to contribute good maths, good science and improvements when you can offer them.
If I might add to Geoff’s comment I think when dealing with ’seriously qualified guys’ too not get too wrapped up in the science. Just check the data, the models fit the data, the methods are valid, normal and sound, the results are interpreted objectively, the sources of uncertainty are quantified. Honestly I think a lot of the time climate scientists are just making stuff up as they go.
Geoff - I commented above: “Is this enough to give one a future predicted view of drought probability?”
You could construct a stochastic weather generator that might describe the statistical elements of a climate, yet never produce the exact climate sequence. i.e. May 1963 would never look like May 1963.
The issue is whether they have enough the elements of the climate system included to make any simulation realistic - El Nino, La Nina, quasi decadal, IPO, PDO (if the latter three truly exist!).
i.e. and with respect - that David’s criticism isn’t valid. It’s setting a bar that cannot be reached even with “perfect knowledge”.
So I’m still thinking about it myself ?
It raises of course the question of the appropriateness of modelling as a solution.
In any case given the long term paleo trends, combined with the papers detailed in the bravenewclimate link i.e. documenting changes in Walker circulation, SAM and STR - I think I’d be erring on the drier side policy wise.
There is of course a response that government can take - abandon all drought relief payments and go laissez faire - let the market work it out (or not!).
And you will notice that CSIRO and BoM are only one input to the EC review - there is extensive industry and community consultation.
So Geoff your advice would be?
and admin - I disagree - I think you’ve only consider 20% of the problem. Hence my inputs. So I’m hoping you might immerse yourself in the actual problem rather than just chasing CSIRO around demanding satisfaction. It’s starting to get that yechy political more than science feeling.
Luke, real observations, standard methods, significance tests, these are the tools for defeating those who want to make stuff up to suit their agenda. Its the power of numeracy.
davidss last blog post..DECR: The message starts to slide
So you want to use the most of objective of tools to test a hypothesis and motives on which you’ve made a subjective judgement.
“defeating those who want to make stuff up to suit their agenda” - so is this your agenda?
And what is their agenda? Do we know?
And “DECR: The message starts to slide” - when was the interview recorded? Is it back-sliding or just what Hennessy wanted to say?
Meanwhile the farmers are still in the dark. Invest or retreat? hmmm ….
Luke, asking why is a good and relevant question. My motive is to be an advocate of numeracy. The blog tends to address how science is arrived at, rather than the science itself. The DECR is just one current project.
As to the interview, the important point is if its consistent with the report or not (its not), and if its an admission by the lead author that exceptionally low rainfall is not predictable by the models (it is).
David Stockwells last blog post..DECR: The message starts to slide
Luke,
Your personal philosophy possibly reflects your background and experience, but at the head of it should be a commitment to the presentation of the best science you can muster and a correction of the science of others that fails your standards. (This generalism applies to all science).
Climate science commentary has more than its share of subjective opinion, interespersed with occasional “hard” papers that have the rigour to promote professional discussion. We need more of the latter.
To the extent that you can contribute the latter, I suggest that you should, not as a recreation, but almost as the duty of a professional. If you condiser that the moderator or any contributor here has made a wrong statement, put the reason in detail.
As a personal explanation, I ceased too long ago to manipulate the nuts and bolts of numbers. In my latter years in management, I was more used to the evaluation of the standard of work of others seeking allocations and in the arguments of outside people afflicted with dogma. So, I feel frustration that so much poor science is slipping through because I can see it. I lack the skill now to dissect it in detail, but I applaud those who do.
In addition to your comment about what Hennessy wanted to say, the title is more of a possible explanation for the inconsistency (that he has changed the message) as I can’t see how you can rationally get from the interview (no skill at long term exceptional low rainfall trends), to the report (likely increased exceptional low rainfall). If there is some other way of resolving it, let me know.
David Stockwells last blog post..DECR: The message starts to slide
And, “Meanwhile the farmers are still in the dark. Invest or retreat? hmmm ….”
Don’t you think people would like to know when scientists are blowing smoke?
http://www.petergallagher.com.au/index.php/site/farmers-unhappy-about-csiro-drought-alarmism/
Association president Jock Laurie says while the Climate Report does say ‘exceptionally high temperatures’ are likely to occur frequently, this does not equate to drought. Alarmist reporting has added confusion and pressure to farm families at a time when they can least afford it. ‘We have received a number of calls from members who were extremely agitated, confused and upset about the reports of drought every second year in future.’ Mr Laurie said.”
Good, thought provoking observations Luke.
David Stockwells last blog post..DECR: The message starts to slide
Yes but scientists might not be the only ones that can potentially blow smoke.
So rural groups are lobbying for ongoing support of assistance measures - what some might regard as agrarian socialism – capitalising gains and socialising losses ((in good seasons they keep the profits and the taxpayer bails them out in drought years). Small business owners in cities don’t get Exceptional Circumstances funding – panel beaters, grocers, hardware shops etc simply go broke and that’s that.
So “poor farmers” is a relative concept – is it a social condition or simply yet another business with risks of operation. Obviously that Exceptional Circumstances schemes exist at all means that there is some “sympathy” for and “shared ethos” between bush and city for farmers and graziers; and ancillary arguments also exist for national food security and the viability of entire regional rural communities and associated towns. Given subsidisation of agriculture, such as in the USA and Europe, some have also argued that EC perhaps just levels the playing field in agricultural commodity trading - as Australia has among the highest rainfall variation in the world.
But many think that farmers have had more than their share of 1 in 20 or 1 in 25 year assistance. It has been perhaps impertinently suggested that some have had their next 200 years of assistance in the last 20 years. Which might imply the climate has changed, or that our knowledge of multi-year droughts is poor, or that we don’t understand the interaction with revocation rules, or simply that the scheme is rorted? All heresy of course.
So David, the areas declared and frequency of declarations are another data set you could put your statistics to use on. Have some producers had “too much” assistance. Is there a case to be put? You might be able to tell us.
I reckon there is a hard economic rationalist line afoot to ditch the assistance scheme and go laissez faire. Which means some farmers without adequate cash reserves would go the wall in the recent droughts. But do we subsidise other businesses who find conditions tough?
As to the temperature issue, papers by both Cai and Nicholls have talked about the high temperatures and high evaporative demands levels as being higher in the Millennium drought (2001-) compared to the Federation Drought (1895-1902). That’s all discussed in the bravenewclimate link comments.
But back to original issue on rainfall – I’m still not sure your statistical complaint is reasonable as you’re confounding a boundary prediction issue with an initial value prediction issue. So you might for example have an excellent tidal model of a beach (~ long term climate) – but at any moment your ability to predict the waves beyond upper and lower bounds is very poor. So the tides are the analogy with long term climate and the minute by minute waves analogous of annual and decadal noise. However, you still have knowledge of the tidal forcing boundaries.
Of course one of the best adaptations “our poor farmers” might be use of various forecasts of rainfall using ENSO related indices. 3-9 months ahead. So one might roll with the seasonal conditions. But I shudder to think what a statistician might make of all that - http://portal.iri.columbia.edu.....mode=2#mme But the worry for all seasonal forecasters is that their ENSO indices may not be climate change proof. i.e. see the decrease in the Walker Circulation.
And just to make it interesting on longer term trends – while De Deckker (see graphic in video - http://www.abc.net.au/catalyst/stories/s1848641.htm ) might be tracking a longer term drying cycle in southern Australian lakes, the runoff onto the Great Barrier Reef is not showing a trend to wetter or drier but increased variability to extremes – making it difficult for land management and creating erosion prone conditions. Data here Hendy 2003 and Lough 2007. http://www.ncdc.noaa.gov/paleo.....rrier.html
So one size doesn’t fit all. Different trends and interactions in different parts of the continent.
PALEOCEANOGRAPHY, VOL. 22, PA2218, doi:10.1029/2006PA001377, 2007
Tropical river flow and rainfall reconstructions from coral luminescence: Great Barrier Reef, Australia
Janice M. Lough, Australian Institute of Marine Science, Townsville, Queensland, Australia
Rainfall and river flow in northeast Queensland, Australia, are highly seasonal and show high interannual and decadal variability that is modulated by El Niño–Southern Oscillation (ENSO) events and the Pacific Decadal Oscillation (PDO). Reconstructions of October–September freshwater input to the Great Barrier Reef lagoon and October–September Queensland rainfall are developed from visual assessment of the occurrence and intensity of luminescent lines in massive Porites from up to 25 coral cores from 15 nearshore reefs regularly influenced by river flood plumes. Separate reconstructions are developed for four rivers (Herbert, Burdekin, Pioneer, and Fitzroy), and these are used to reconstruct total annual freshwater flow into the Great Barrier Reef (69–74% variance calibrated) and an index of Queensland rainfall (53–57% variance calibrated). The reconstructions extend back to 1631 but are most reliable from 1661 and capture significant decadal variability. The reconstructions provide insights into long-term tropical rainfall and river flow variability and the behavior of ENSO and the PDO over several centuries. Significant, though weak, relationships are found between these reconstructions and an independent reconstruction of ENSO. The reconstructions highlight that observations from the instrumental records of high interannual and decadal rainfall and river flow variability in northeast Australia also characterize the past few centuries.
Although there appears to be no overall trend toward wetter or drier conditions, the reconstructions suggest that the variability of rainfall and river flow has increased during the twentieth century with more very wet and very dry extremes than in earlier centuries, as projected for the region as a consequence of global warming.
Luke, any statistical significance to these results of increased variability?
David - Janice Lough’s paper is 16 pages long with numerous statistics tabulated and significance tested listed.
The wettest 30-year period was reconstructed in the twentieth century and the driest in the mid nineteenth century. Importantly, neither series shows a significant trend toward either wetter or drier conditions over the past 3 centuries. The wettest and driest years and 10-year periods were, however, all reconstructed during the twentieth century.
4 ten year periods and one 30 year period were significantly different (5%) to the long term mean for maximum extreme rainfall and runoff, and one ten year period and one 30 year period were statistically different for minimum extremes of each runoff and rainfall respectively.
In addition, the percentage of coefficient of variation (cv) of the reconstructed river flow series for the 3 subperiods was 50.1%, 48.1%, and 64.3%, respectively, again suggesting an increase in variability in the most recent period.
Two additional analyses were performed to test whether this suggested recent increase in river flow/rainfall variability was real and not an artifact of the varying number of coral series through time. First, a reconstruction of river flow from the 4 longest continuous coral series (Pandora 08B, Havannah O1A, Hook O1B and Humpy OlB), 1724—1983, showed the same increase in variance (though similar median flows) between 1749—1866 and 1867—1983 with the percentage cv increasing from 56.1% to 64.3%. The anomaly of percentage of years for very high and very low flow events changed from —5%, 1749—1866
to +7%, 1867—1983 (see Table 10). Second, the frequency of luminescent lines at three midshelf reefs, which only occasionally experience the largest flood events, was examined. Both these analyses support an increase in river flow/rainfall variability since the late nineteenth century.
A graph towards the end of the paper of the 21 year standard deviation of river flow shows a marked increase in the 20th century.
Leave a Comment