It is now widely admitted that climate change impacts on water resources are occurring and are expected to amplify in the future. However, these impacts are highly uncertain and usually scientist and decision makers prefer to have a quantitative assessment of these associated uncertainties. How can we assess future uncertainty related to climate change in the absence of observations? Then, how can decision makers consider these uncertainties in their future plan management of water resources for instance?
Haykel, you have asked a question which in many ways highlights the different "world views" of scientists and decision/policy makers. As you point out, decision makers would like to minimize the uncertainty while scientists, especially climate scientists, live with uncertainty. Part of the problem stems from the two communities not always understanding the needs, capabilities, and limitations of each other. In this particular case the problem is compounded since it is uncertainty upon uncertainty - climate projections contain inherent uncertainty and, as you point out, the impact of climate on water resources is also not well understood. Nevertheless the tools that we have available today, mainly global and regional climate models, are much more powerful and useful than simpler statistical methods of extrapolation. As our understanding of the system improves, so will the models, and thereby the uncertainty will be reduced, although it will never be completely eliminated due to the inherent nature of the system. Certain approaches such as multi-model, ensemble projections, or some type of statistical postprocessing of the model results can further reduce the uncertainty. At the same time new approaches in decision making and management need to be developed which are able to account for and deal with the inherent uncertainty. And perhaps most important, a common "language" needs to be developed that will allow scientists and decision makers communicate more effectively.
Dear Haykel, They rely on trend analysis using time series data on particular climate condition especially those involved in agricultural sector. Observation is also used for evaluation impacts of sea level rise and coastal erosion, etc.
Hi Haykel, Indeed there is huge uncertainty attached to the future and would be hard to access the impacts on any of the resources in that case. Fortunately, It is possible to access this impact on resources using Climate Model (GCMs) dataset to a higher precision that it could be done by any of extrapolation methods. There are uncertainties involved with GCMs as well but this is the best possible we can do. The models have evolved very much since they were first put to use in the community. It is wide topic to be dealt within one question though.
Nowdays, use of RCM data i.e., regional climatic models is in trend for analyzing the probable impact of climate change on water resources.... they are having high resolution than that of GCMs. Uncertainty is always there because natural systems are very complex...
Yes to all the above :-) Uncertainty is a big issue, and I agree that one of the big barriers we face as scientists is persuading policy makers and budget holders to make investments in time. There is financial risk from taking action to prepare for climate change if it is not needed. It may deprive other areas of investment that address pressing issues such as public health and education. But there are also risks to taking no action that must be considered as well. The least action each community at risk should take is to develop a response plan that considers how they can use existing resources to cope, adapt and mitigate the impact of short-term change and then what additional resources may be needed to meet the challenge of long-term change. There is often a lot that can be done in the short term that involves public education about water conservation and efficiency... where even a modest investment to repair leaky infrastructure can save as much as 30% on transmission losses. In rural areas where most water comes from wells, public education is even more important so that water table draw down and salt/brackish water incursion do not become problems. Farmers need to understand how to minimize evaporation losses from irrigation. Better agricultural practices and the use of drought tolerant varieties can make a big difference. None of these practices require an immense amount of money. Looking further to the future plans to drill systematically into deeper aquifers where available should be considered and the development of water storage capacity (underground or at the surface) to catch and store rain water and runoff when it becomes available. This kind of planning is essential. The consequences of not planning will be the permanent loss of resources that could have been conserved. The social consequences could be equally problematic leading to regional conflict over remaining resources (e.g. access to river water or aquifers that cross borders) and at worst, the impact of forced migration of climate refugees. So uncertainty is an important concept to embrace, but the consequences of action must be set against the consequences of inaction. Very often small changes that do not require major infrastructural investment can make a big difference especially in rural communities.
Uncertainty is definitely a big deal but we should be careful that it doesn't become an excuse. It is not a simple issue, but there are different options to deal with uncertainty. For example, prioritising flexible adaptation options that can be modified to best fit the conditions. I know Prof. Anil Markadya has some publications on the issue, perhaps you could check his profile for more specific info.
Haykel, you have asked a question which in many ways highlights the different "world views" of scientists and decision/policy makers. As you point out, decision makers would like to minimize the uncertainty while scientists, especially climate scientists, live with uncertainty. Part of the problem stems from the two communities not always understanding the needs, capabilities, and limitations of each other. In this particular case the problem is compounded since it is uncertainty upon uncertainty - climate projections contain inherent uncertainty and, as you point out, the impact of climate on water resources is also not well understood. Nevertheless the tools that we have available today, mainly global and regional climate models, are much more powerful and useful than simpler statistical methods of extrapolation. As our understanding of the system improves, so will the models, and thereby the uncertainty will be reduced, although it will never be completely eliminated due to the inherent nature of the system. Certain approaches such as multi-model, ensemble projections, or some type of statistical postprocessing of the model results can further reduce the uncertainty. At the same time new approaches in decision making and management need to be developed which are able to account for and deal with the inherent uncertainty. And perhaps most important, a common "language" needs to be developed that will allow scientists and decision makers communicate more effectively.
Decision-making under large uncertainty is a growing research field in social sciences and your question, I believe, point to the necessity for the research community to work more trans-disciplinary. A combination of natural and social sciences is required for tackling the problems policy-makers face. We can ellaborate with projecting uncertainties but we also have to improve knowledge transfer and demand-driven data production. Perhaps Courses including practical training in policy-making for scientists would improve our understanding of what the demands are and how we can optimize our knowledge Communication..
Indeed RCMs (Regional climate models) can be put to use for your work. I am not sure of your study area and not all study areas have RCMs in place. The way ahead would be to downscale either dynamically or statistically the data from GCMs and use for your work. Uncertainty would still be part of it.
Perhaps decision makers should heed the medical profession’s Hippocratic oath: first, do no harm. The problem with making broad-ranging decisions where there is a great deal of uncertainty is that any actions taken may well make things worse, not better. For example, there is organization known as 350.org whose aim is to reduce the CO2 content of the atmosphere from its current level of 400 ppm to 350 ppm, i.e. back to where it was around 1988.
Let us suppose that by some magical means we can reduce CO2 to this level. The first thing that will happen is that crop yields will decrease (see for example, Effects of low and elevated partial pressure on growth and reproduction of Arabidopsis thaliana from different elevations, J.K. Ward and B.R. Strain, Plant, Cell and Environment, vol 20, pp. 254-260, 1997). The world now has about 2 billion more people than it had in 1988. While I am not prepared to make a positive statement at this time (uncertainty again!), it is likely that we would find difficulty in feeding those extra 2 billion people at 350 ppm. If so, the result would be massive starvation and warfare on a global scale. Are we willing to risk this for what, at this time, are purely theoretical benefits?
Some interesting answers already. The only thing I would add is that I think scientists, with their disciplinary baggage, are uncomfortable in the climate-change space because of the inability to even quantify the uncertainty surrounding the issue - and fair enough, we, and future generations, should never have been put in the position of having to make the best of a very messy situation that is emerging following decades of inaction from government, corporations and the rest. However, putting that aside, urban planners and architechs, for example, have always needed to create their wares for future users, based on predictions of future needs - albeit, with mixed success. This thinking now needs to extends to nature conservation or ecological intervention and other areas affected by climate change that used to be informed by more objective knowledge.
I am not a familiar with data collection systems in Europe but I would speculate they are equal to of better than what we have in the United States. There is no question that climate change will impact water resources but the question will be where and to what degree. One measure used by hydrologists is the 7-day ten year low flow for a particular drainage basin where we have flow measurement for at least 30 years or more. In good rainfall years that low flow level is never reached but the issue is how do we know that that level is correct under differing climate change scenarios. Problems arise when the demands for water exceed the supplies and the drainage basins operating close to margin are the ones we will need to watch and have a management plan. The question then becomes, how certain are we that our predictions of future shortages are correct? So the issue can be divided into two parts; how certain are we that water problems will occur and second, how certain are we that we are correct. If there is high certainty in both parts for a drainage basin, it is worth considering investments to mitigate the problems like conservation or allocations.
Roger... Facts and more references please. Along side a small often regional increase in productivity for some crop species comes a decrease in vitamin content and of course an increase in the incidence of drought and the threat from pests and disease.
http://www.nature.com/nature/journal/vaop/ncurrent/full/nature13179.html
http://www.nature.com/nclimate/journal/v3/n11/full/nclimate1990.html
As happens too often in this general discussion you collapse a multifaceted debate into a facile argument about "feeding the world". Every major study that I have read... and I read many... has determined that the medium to long-term impact of climate change will be to harm agricultural production globally, and significantly increase the risk of famine. I can't help but notice a significant mismatch between your areas of expertise as indicated by your publications and your discussion interest as indicated by your postings on climate issues. I do not mean to suggest that we should never post outside our area of immediate expertise, but we should be aware that when we do so, the probability of sounding like an total idiot escalate exponentially. I have no more patience for this nonsense.
David,
It seems we have someone else who confuses scholarship and science. I am sure you have read widely in your chosen field: this is scholarship. Science, however, is a discipline which demands that one first observes verifiable facts, then advances a hypothesis to explain those facts, and continues by searching for further facts to support or disprove the hypothesis. Scholarship can be a useful adjunct to science, but many academics seem to fall into the trap of confusing the two.
There is a vast literature of predictions that global temperatures will increase dramatically, and many sophisticated computer models have indeed predicted this. Unfortunately, global temperatures are obstinately refusing to comply. Scholarship would say that all these predictions can’t be wrong, so let’s look for reasons why current temperatures are merely a temporary aberration. Science would say that there would appear to be something at fault with our basic assumptions, so let’s look for a new hypothesis.
There is also a vast literature predicting that climate change will have a disastrous effect on agricultural production – but shouldn’t we have seen examples of this by now? Global agricultural production has increased in the last few decades to the point where any lack of food anywhere is more a result of distribution limitations than available quantities.
Here are a couple of facts you may be interested in. First, most commercial greenhouses are operated at CO2 concentrations significantly above atmospheric levels, because experience has shown that plants grow better and faster under these conditions. Second, global agricultural production has increased as atmospheric CO2 levels have increased. Yes, I am aware in the latter case that correlation does not necessarily imply causation, but to dismiss the possibility of causation without further ado strikes me as the mark of a scholar, not a scientist.
I was prepared to toss in a few sarcastic wisecracks, but so far I am impressed with the admirable level of open, sensible and civil debate on this most charged of all current issues, so wisecracks are undeserved. You folks are doing great -- keep it up!
1.There are major publications on both observed and projected impacts of climate change on water resources (see below some listed references). I feel that at the beginning we have to take into considerations of what are already available and are being used by scientists and then refine the data and information if needed. IPCC publications in this area would be a good source to start. Please see below some references:
Arnell, N. W 2004. Climate change and global water resources: SRES emissions and socio-economic scenarios. Global Environ. Chang. 14: 31-52.
Bates, B.C., Z. W. Kundzewicz, S. Wu and J. P. Palutikof (eds.) 2008. Climate Change and Water. Technical Paper of the Intergovernmental Panel on Climate Change, IPCC Secretariat, Geneva, 210 p.
Dokulil, M. T. 2013. Impact of climate warming on European inland waters. Inland Waters 4: 27-40.
Alcamo, J., J. M. Moreno, B. Nováky, M. Bindi, R. Corobov, R. J. N. Devoy, C. Giannakopoulos, E. Martin, J. E. Olesen, A. Shvidenko 2007. Europe. Climate Change 2007: Impacts, Adaptation and Vulnerability. Contribution of Working Group II to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change. In: M. L. Parry, O. F. Canziani, J. P. Palutikof, P. J. van der Linden and C. E. Hanson (eds.), Cambridge University Press, Cambridge, UK, 541-580.
2.What are some of the observed impacts on water resources (examples)
The annual run-off in some regions experienced an increase, such as at high latitudes and large parts of the USA. And there has been a decrease in run-off in parts of West Africa, southern Europe and southernmost South America. Significant decreases in lake and river size in Africa were observed. More intense rainfall events have increased in many parts of Asia. Significant decreases in lake and river size was observed in China, for example, shrinking of the Huanghe River in China. Significant increases in the occurrence of heavy precipitation have been observed across Europe, in particular in Northern Europe
3. What are some of the projected impacts on water resources (examples)
Based on six climatic models, there is likely an increase in the number of people who could experience water stress in northern and southern Africa, whereas there is a likely a decrease in water stress in eastern and western Africa. There have been projections for earlier spring flows in North America and Europe while decreasing flows in some river basins in other countries and regions (e.g. Australia). The impacts of climate change on water resources in Asia will be positive in some areas and negative in others. It is projected that more than a billion people could be affected by a decline in the availability of freshwater, particularly in large river basins by 2050. With regard to water quality, there is a 50% chance that by 2020, the average salinity of the lower Murray River, Australia may exceed the 800 EC (electric conductivity) threshold set for desirable drinking and irrigation water standard. It is projected that climate change will have a range of impacts on water resources in Europe including decrease in seasonal snow cover at lower elevations (Alps), decrease in glacier volume and area (Europe), decrease in thickness and areal extent of permafrost and damages to infrastructure (Russia)
Dear collegues,
the question of uncertainty in statistics/mathematical modeling as a whole has a real problem. Students are thought from early on to look for theoretical ideas in data sets that are designed for learning basic concepts. Look for normal distribution in residuals of your forecast, look at RMSE. But we all know that the real world/universe does not behave in such ways. I idea of parametric statistics in the real world is absurd. IPCC relies on the same basic ideas to talk about their "projections." What are these projections, nothing but time series forecasts done mostly by climate scientist that have only had little exposure to forecasting and time series analysis. All their models are computer projections based on faulty understanding of the data. IPCC has been wrong in the last four years and will continue to be wrong in their projections because of couple things. Firstly, they rely on data that is of long term memory and between cycles and secondly, they don't really understand uncertainty quantification. Now, don't get me wrong here, climate change is real but IPCC idea of climate change is not. They are bunch of hooligan scientist that sit on a pot of gold and try to defend their existence based on THEIR "science" and not actual data. The people that talk about CO2 levels have only real measurements of modern time (since measurements began) and rely on ice core data to talk about the different levels, but fact is, ice core data is limited regionally and you can not use regional data to make assumptions about global scales. However, if you look at volcanic activities above VEI 6, which happen every few hundred years, the release of CO2 from these actually affects climate globally and so CO2 levels have always fluctuated on earth.
As for the original question before I started my rant, in:
"Forecasting conditional climate-change using a hybrid approach
Akbar Akbari Esfahani, Michael J Friedel
Environmental Modelling and Software. 02/2014; 52(February 2014):83-97. DOI:10.1016/j.envsoft.2013.10.009"
which is available on my page, you will find that we actually define the probabilities for uncertainty with our modeling approach using semi-parametric statistics.
You may find useful to read this publication of Günter Blöschl and
Alberto Montanari ... Have a good reading!
http://onlinelibrary.wiley.com/doi/10.1002/hyp.7574/abstract
To this huge amount of comments and pub links, I'll just add that, because uncertainty is inherent at every level of climate change prediction AS WELL as climate and water links, decision makers have to adapt no-regrets approaches to water resources management. That is the only way,whether the decision makers are small village communities in mountains or the water basin management authority of the Water Ministry. Long term data on rainfall and discharge can help to some extent to gauge interannaul variability, but this data is missing for most of the developing world, and also climate change introduces new uncertainty. Hence, from a decision making perspective, its safest to assume the worst and prepare for that. Usually that translates into distributed enhanced storage capacity, reforestation, water pricing policies, the use of wetlands for wastewater treatment and so on.
Certainly assessing uncertainties in climate it is a very cumbersome task. Nevertheless I think that for assessing future uncertainties the best way is to rescale climate data using a statistical approach or statistical downscaling. In this way one may choose which climate scenario (from those of the IPCC) want to analyze and to generate climate variables from there. Then one may compare actual data with the rescaled data to see if the temperature, for instance, will increase or decrease in the time frame analyzed.
I like Steve and Charlotte's responses to this question. Making decisions in the face of uncertainty is something we all have to deal with in any long term issue. Forecasts of economic and technological changes include uncertainties, yet decisions are still made by governments and businesses despite these uncertainties, and so it is with climate change. The question is, how can one make good decisions regarding climate change given available information and prevailing uncertainties? The IPCC 5th Assessment Report has indicated that a risk management approach would be a useful framework for available scientific information to enable decision making. This would require collaboration among climate scientists, researchers and practitioners from many climate-sensitive fields (agriculture, water resources, fisheries, forestry, etc.), and planners/decision makers from various fields of practice (engineers, professional foresters, urban planners, insurance, disaster risk management, etc.). In order for climate information to be translated into information on changes in risk, different kinds of models/analytical tools would be needed so that, for example, a change in rainfall could be translated into a change in wheat yield or flood risk. Multiple model runs with multiple scenarios (individually or as ensembles) would be needed in order to get a sense of the range of projections of risks, even if we may never get a complete picture of the uncertainty range. From this, a dialogue with decision makers could proceed, assessing options for adaptation response, from local scale flood protection to larger scale issues of food production, supply chain management, etc. Decision makers would then be responding to projections that have been translated into changes in risk of relevance to them. Finally, regarding IPCC, some of the comments made earlier suggest an inaccurate understanding of what the IPCC is and how it functions. IPCC organizes assessments of published research literature from many different disciplines. Climate projections, impacts projections, and assessments of risk all come from this assessment of literature. Authors of IPCC reports are volunteers from around the world, nominated by their respective national governments. Reports are reviewed extensively. The Impacts-Adaptation report author team, for example, responded to 52000 review comments during the 2011-2013 writing period. For more information about IPCC reports and how the assessment process functions, please go to http://www.ipcc.ch
Folks,
I am impressed by all the commnets and ideas, all worth on their own. my 2 cents wisdom is the scale issue and the uncertainty. Golam gives some excellent arguments about "uneven" and often reversed trends. As a soil scientist I claim to be I always look for any climate model to assess their handeling of microscale say hillslope/small basin scale and I hardly find any to my dissapointment but understand also the challange so somewhat expected, I guess. The regional models are ok I guess for regional planning but the scale where the impacts can be seen and more importantly mitigated to a certain extend is the landscape or hillslope scale as a scale that most humans can manage to grasp a bit. Most of crop deficiencies will most likely be observed at this scale as will soil changes so I am still looking for climate forcasts at that scale and if you have come across any study of future climate impacts on soils at that scale I would appreciate it. Here is an example attached that hints upon the difficulties in addressing the scale issue but again there in no model breakdown to the finer scale I was refering to.
Thank you for the references provided. Good and interesting reading.
Article The climate velocity of the contiguous United States during ...
It was written on another matter, BUT:
"...If in our house suddenly stink of sulfur, we have no right to indulge in arguments about the molecular fluctuations (I would say - model uncertainty) we must assume that somewhere near the devil with horns is, and take appropriate measures, including the production of holy water in an industrial scale."
(Strugatsky, Arkady and Boris. Beetle in the Anthill. New York: Macmillan Pub Co, October 1, 1980, 217 pp. ISBN 0-02-615120-0. LCCCN: 80017172.)
it would be better to ascertain the deviation from normal; and its impact on the LIVES and the LIFE supporting ecosystem. and in an event be prepared to know what level of management it would require, perhaps !
The climate change impact on water resources can be devastating for millions of human lives. It is better to consider and prepare to the worst scenario.
Zamir, the problem with climate models or General Circulation Models is that they work on a scale of 300 km*300 km , or with better accuracy at a 500-1000 km grid. Thus you can imagine the coarseness of GCMs. This is because our understanding of climate is in its infancy, given the tremendously complex apparently stochastic system of differential heating/cooling, pressure systems, winds and ocean currents. Downscaling model predictions to smaller areas have worked on a case by case basis, forcing the model to recapture longterm historical data at a site. Prob is that this downscaling requires longterm historical data (of precip, temp etc), and also is site-specific. That is why there aren't any universally reliable climate models that predict soil moisture at a small catchment scale (say a 1 or even 5 km grid). Another area of huge uncertainty relevant to soil moisture prediction is accurately estimating evapotranspiration in wooded vegetation areas. Hydrological models treat ET as a black box with a couple veg parameters, neglecting the tremendous variability across species, communities and seasons. The way to improve is to be able to start collecting soil moisture, precip and temperature data and then force models to replicate that.
Dear ALL
The uncertainty in climatic models is high, these models first implemented in weather forecasting for few days ahead. These models may give some information which is better than nothing
Yes. it is an important steps to assess the uncertainty of prediction in the climate change impact studies. I have used several GCM outputs in my studies to understand the impact of climate change on water resources. But different GCM outputs have given different results in the impact of climate change on water resources. Uncertainties in prediction is now a major concern in climate change impact studies.
I agree with the two last reactions, but if we for example use climate predictions for farmers in monthly to seasonal rainfall scenario predictions, we must inform them on the skills of the raw ensemble ENSO model predictions used. I am convinced that these predictions, based on the physics of ocean surface temperatures in the Pacific, will in the end be more accurate and useful than the use of statistics of past data, how complex they may be, in advising on planting dates or other advisories.. The statistics are not able to capture the highly variable rainfall data in time and space of farmer' fields, neither are the physics mentioned, but farmers are much better served with simple scenarios with an attached skill.
I do not think that it makes sense to evaluate uncertainties in climate change impact projections, unless these evaluations or assessments are directed at improving such models. Decision making in the absence of observations and based in part in the assessment of uncertainty makes no sense at all.
For water resource management decisions we need to increase the research conducted on watersheds, and as my colleagues mentioned here earlier: in a multidisciplinary way, including potential climate change impacts predictions. We know very little about the subtleties of a watershed behaviour, the ecosystems, human activity and how they contribute in the water cycle.
I believe that in order to make better decisions in water management resources we need to change the scope from climate change impacts projections to sustainability models, that do not consider water as a independent element in the natural system.
Focusing on climate change impact projections to make decisions on water resource management also implies that we do not need to change the way we have been using water resources.
I think it is important to assess uncertainty on the future impact of climate change and assign probabilities to the chances that one type of impact could be more likely to happen than the other.
Water resources management as practiced in irrigation, hydro-power and integrated projects usually take into consideration whether and rainfall data for the past fifty years for designing the dams, reservoirs and regulating systems. And then the performance of the regulated hydraulic structures is evaluated for the fifty year cycle. ICOLD and their national chapters as well as other global institutions have a fairly large data base as well as a analytical tools which could be possibly married to the newly developed climate models etc. Then, in engineering practices safety factors are built in to design procedures considering the inevitable uncertainties. I wonder why similar principles and philosophies cannot be made applicable in the present case as well
That is the case when past climate model reconstructions come into the picture. The reconstructed climates are inherently uncertain, of course. However, by changing the environmental conditions, like CO2, sea level, etc., we can assess how much those inherent uncertainties vary in experiments. The advantage is that we do know what had happened in the past, thus our assessment can be tested against the proxies. The base assumption is that future uncertainties would vary similarly to those we are assessing for the past climates.
We agree that there are large uncertainty in climate-change impact projections. It is so important that we have to assess the uncertainty. Model is the only tool for the projections, However, in the absence of observations, all of these models are nonsenscial. At least, after the validation of models with observation, these models can be applied to finish many experiments. It is just like another 'earth' that we can control.
The global climate models have been for long dominated by the green house effect of triatomic gases in the atmosphere. These are too global to predict local and short term deviations and uncertainties that are of great relevance in project engineering. As and when analytical tools are available at reasonable costs we will use them and until then we may continue with the current practices. And that has been the practice till now and there is no reason to change the culture.
Sure. Large uncertainties exist in outputs of the current climate models. 10 yrs ago, we thought that the models were very well reproducing the spatial and temporal patterns for surface air temperature, though they were not for precipitation. However, we realize saddly at present that the models even could not well project the increasing rate of global surface temperature for the last two decades.
And that was why the climate change debate deteriorated into a debate between the believers and non-believers! I think the best course is to reject such useless theories and take decisions based on specific problems or specific issues.
Maybe indirectly relevant
Garcia, R.A., Cabeza, M., Rahbek, C. and Araújo, M.B. (2014). Multiple dimensions of climate change and their implications for biodiversity. Science 344: 6183
Garcia, R.A., Burgess, N.D., Cabeza, M., Rahbek, C. and Araújo, M.B. (2012). Exploring consensus in 21st century projections of climatically suitable areas for African vertebrates. Global Change Biology 18: 1253-1269
You can download them from here
http://macroecology.ku.dk/by_year/
For me it makes absolute sense. The tools you use may be different to a traditional error analysis.
The first and most obvious route is Monte Carlo simulation. If You can define the system you are looking at numerically as well as the ranges and distributions of possible inputs, a Monte Carlo simulation, with sensitivity analysis, will generate a range plausible and probable outcomes. It can also identify which parameters the outcomes are most sensitive to and inform management decisions.
This is an area where there is some expertise in the business world as well as the social sciences. For the hard sciences to become more relevant, there is a need for improved communication regarding risk as well as better understanding.
Some conclusions are possible to get the answers above.
.
First: That there is not the ballyhooed consensus on climate change, becoming clearer discredit primarily on the use of (RCM) Regional Climatic Models.
.
Second: Maybe the reason to discredit the use of RCM is not uncertainty over uncertainty in many cite as therefore as the use of modeling and RCM are more vulgarized and tested by a larger number of these researchers knowing the background of the limitations their models, transfer to the latter the RCM in relation to Global Climate models (GCMs), the "guilt" of incorrect results.
.
Third: That respect for the GCMs is not so much to trust the credibility, but in the academic tradition that quoting a result from other authors (in the case of simulations GCMs) are putting these as an absolute or reliable result.
.
From these considerations above ask myself: Why should we trust in the large scale simulations using them as boundary conditions and mistrust of RCM where the mesh, the time intervals and the number of variables are best defined?
.
If we use a model with better resolution and entered as a condition of impaired contour finally ask:
.
The where is the weak link in the chain?
.
.
.
Just a personal comment. I see in various works of RCM in which their results are conflicting with the observations of an overwhelming urge researchers to extract consistent results with the observed data, many times leading to conclusions hesitant about the mistakes and the successes placing them as absolute truths, since that these items reach the events provided at study start, even though the overall result these are not statistically conclusive.
Suppose the monthly needs of a crop rainfall are 280 mm / month and we have three locations where weekly rainfall is A = (0, 0, 140, 140), B = (0, 60, 80, 140) and C = (60, 60, 80, 80). What is the city that offers more uncertainty for the crop during that month.?
The sample standard deviations are 70, 40 and 10 respectively. The town C it has very close values to the mean values, this place is appropriate to the culture while location A it has a large uncertainty for cultivate that crop.
On another level it's the same with the data that often we get to work on models for climate change. You can appreciate that by dividing standard deviations by the mean values : Acv = 1; Bcv = 40/70 = 0.571428571; Ccv = 10/70 = 0.142857143;
Ccv < Bcv < Acv
which means we obtain a certain grading the in the location of the uncertainty values in relation to the success of that crop.
Well Climate projection models are just tool to predict the possible future scenario. The accuracy of observed data , which is being used in models is very essential. Thus more and accurate observed climatic parameters and complete knowledge of dynamics of ocean atmosphere certainly give reasonable future projections.
Nina, you raise an interesting point about the communication of uncertainty. It's something the science community seems to be very bad at. I have been looking at how business strategists respond to uncertainty and there are methodologies involving scenario planning. One of the most important spin-offs is that outlining a continuum of plausible outcomes can often drive data collection to narrow the range of outcomes as opposed to simply complaining about a lack of data and looking for more and more finding to collect data which may not be relevant to identifying solutions.
Business interests inevitably steps in and occupy both sides of uncertainty and debates lose their objectivity..
This is why scientists need to develop a better understanding of the implications of the uncertainties we are supposed to quantify. We also need to learn to communicate this. For the majority of people, many policymakers and many scientists included, uncertainty is seen as something bad and a poor reflection of the quality of work. In many modelling activities it simply means that instead of a single certain outcome to be dealt with, we are predicting a range or range of results. This places us between the two comfort zones of knowing exactly what to do and having so little understanding of a situation that we can do whatever we want.
To all.
.
I do not understand two small problems:
.
If climate models can not describe the last 16 years, as it will be a tool to predict the next 80 years?
.
What is the meaning of using a tool that is not calibrated?
.
My opinion is that this issue brings attention on water problems in future. We cannot quantify accurately but scenarios will be useful for adaptation and possible remedial measures against the future consequences. No doubt, it exists and quantification is also essential.............. It can be negative or positive for a particular region. I am still new in this field......................
Rogerio: Climate models do not predict- rather they project different scenarios depending on our actions today. Their projections for the past 15 years are not perfect, but what they are still very robust. Their projections for the last 15 years are within established boundaries albeit at the lower end because of the impact of PDO and La Niña. I agree that to have a better predictive value over less than 15 years, we need to have a better hold on the periodicity of atm-ocean cycles... but at that range we are starting to overlap with weather forecasting and the inevitable difficulty in predicting the behavior of chaotic systems over short periods of time.
@ David.
(Warning: I am writing this detailed response with settings that surely you already know, not for you but for others who do not work in a climate).
To me it seems strange to think que both El Nino-La Niña and PDO are strange beings something called climate. The way you are interpreting it seems much one as the other elements are outside that influence climate.
.
Both the PDO and ENSO phenomenon associated parameters are normal fluctuations in the medium and short term climate in certain regions, if we look at the oscillation of the PDO Index from 1900 to 2014 simply will see a repeat of conditions that occur from 1945 to 1975 and this 5 had the same time periods in which ENSO Index showed a remarkable series of La Niñas.
.
.
For all.
.
Now there is one important fact that passes without people notice (People do not specialize in climate) , the warm phase of the PDO does not mean that Sea Surface Temperature (SST) in the Pacific also be heated, ie the PDO does not represent the surface temperatures of the extratropical North Pacific (where the PDO is derived), so if we make a spatial correlation between the PDO and the surface temperatures of the extratropical North Pacific, arrived in regions where it is positive and one is negative.
.
This is important because many attribute an PDO in its warm phase at higher temperatures of the Pacific Ocean, which is wrong.
.
The conclusion I want to make is that negative PDO periods do not necessarily indicate colder, and the reverse is true.
.
PDO was defined by ZANG et al in ENSO-like interdecadal variability: 1900-93. (http://www.atmos.washington.edu/~david/zwb1997.pdf)
.
Great points. Of course PDO/AMO (and the host of others) are an integral part of the climate system. Although cyclical, with a degree of predicability, their appearance does suggest a degree of chaotic ordering that may allow us to project their likely net impact over 50 years, but makes it much more difficult to predict their occurrence in any one year. For example, it is June and we are still not (completely) sure if an El Niño will appear this year. The spatial impact of the PDO is complex, but one that tends to bring warmer waters to the tropical Pacific in its warmer phase. I think the diagram at the web links below may clarify this for those still confused. I appreciate the chance to clarify :-)
http://cses.washington.edu/cig/figures/englobe_big.gif
http://www.skepticalscience.com/print.php?r=346
may be one way to view the problems is thru regionalization, where we can differenciate areas with better info and certainty and those with less info and higher uncertainties. and physical mechanisms of the phenomena always could be considered to reduce uncertainities
Hello Hykel -
Several respondents have commented on the difficulties that arise from differing ideas about uncertainty in the policy and research communities. I'd add that there is also a lot of variability among climate researchers as to what constitutes an expression of uncertainty. Data-heavy disciplines tend to think much more quantitatively about uncertainty since they have the information to do so; model-heavy disciplines have a tendency to treat uncertainty much more vaguely. Modelers may, in some cases, have no better way to describe uncertainty than to compare different unverified model projections, and to take agreement among models as a sign of increased certainty, even if no one model has been verified according to observations. This happens in aspects of ice sheet dynamics and sea level rise, where even hind-casting is of limited value because events in the past that could be used for validation may not be the product of the same processes that future events will be (and thus aren't a good test of future conditions).
One surprising result of this (to me) is that researchers working on problems with little capacity for validation often seem to develop an *increased* confidence in their results when validation isn't possible. Figuring that out is a perhaps a problem for another day...
Of more immediate concern is how to interpret the wildly varying standards of expressing uncertainty across the various fields that contribute to climate change research (or water resources more specifically) to achieve a consistent and comprehensible terminology that is understandable to the policy/decision-making world. I differ from Steve Brenner's statement in a small way: While decision makers would *like* to minimize uncertainty (who wouldn't!) they are, in my experience, *more* comfortable with large uncertainty than most scientists tend to be, **provided that the uncertainty is quantified in a clear and robust fashion.** That means that quantities that come into climate/water decision-making be accompanied by traceable expressions of the statistical range of the quantity in question (e.g. ± 1 SD, ± 5/95%, etc) together with some qualitative evaluation of how likely that assessment is to be correct. This later is what the IPCC tries to do with their "evidence and agreement" matrix (see, for example, http://www.ipcc.ch/pdf/supporting-material/uncertainty-guidance-note.pdf). Unfortunately, a surprising number of stated uncertainties (i.e. ± something) attached to quantifiable variables are essentially WAGs made by the authors, not based on any explicit error analysis but rather on gut feelings, usually early on in the process ("how good do you think these inputs are?" "Oh, maybe to plus/minus 10%?" "OK - we'll go with that for the outcomes as well"). This happens more than you would like to think, and I think it tends to happen in research areas that are just coming out of the realm of "pure science" topics that have only recently entered the arena of "applied science" where not being wrong actually matters. Ice dynamics and its implication for sea level rise is an example of this: 20 or 30 years ago, this was a very speculative subject where
figuring out accurately what glaciers and ice sheets may do in the future was far less important than simply getting a handle on what happens. The solution here is to ensure that much more attention is given to consistent and traceable quantification of all relevant quantities - really a bookkeeping task more than a fundamental shift.
The same thing also happens, of course, with unknowns that (for whatever reason) can't be constrained by hard observations. (This is also an element in the glacier/ice sheet problem, to be sure). Nevertheless, there are ways of handling uncertainty here as well - much of modern Bayesian methods are applicable. Just as in the case of conventionally quantifiable uncertainty, there is a context and method for dealing with qualitative uncertainty.
So to answer your original question: Yes, it *always* makes sense to quantify uncertainties in climate change impact projections (or anything else that matters) to the best of one's ability. Planners, policy makers, risk managers, etc can deal with large uncertainties, **provided that the uncertainty is quantified in a clear and robust fashion.** That doesn't mean that uncertainties that are intrinsically slippery can't be included, but rather that the stated uncertainties have be traceable to some meaningful decision making process. And, ideally, like the IPCC's "evidence and agreement" matrix, it helps to include some estimate of just how good you think your uncertainties are!
WTP
.
@Pfeffer and Filippo.
.
In fact the uncertainty (± ie something) is something that can be set perfectly in science through a sensitivity study of the independent variables, but many climate modelers do not know how the uncertainties of the independent variables themselves, simply start to use the process as well set of "gut feelings".
.
The curves of climatic variability should clearly display the maximum and minimum uncertainties taken from the independent variables, but perhaps with this process as they would reach something like 3K +-6K they are embarrassed to show the results!
.
@David
.
I put a little more complete discussion of the problems of using NAO, ENSO and PDO Indexes in a comment in https://www.researchgate.net/post/Is_contemporary_stabilization_of_surface_air_T_an_indication_that_anthropogenic_forcing_is_probably_not_the_main_driver_of_climate_variability .... I ask you to look and correct if you have any errors.
@rogerio Thank you I had seen that. Looks god to me :-) there are certainly some similarities between the general distribution of temperature in the Pacific Ocean when comparing El Niño and PDO although the time scales are very different. Some believe they are linked in some way. Perhaps we can wait and see what happens to the PDO if we do have a large El Niño this fall.
@David.
.
Actually the link between the various oscillations are not well known, did not talk about it because the text was kinda long.
.
Now what surprised me most was that when I was writing the text, I tried to establish a timeline and found that the definition of these oscillations which all utilize indexes as if they were known long ago they did not have much more than a few decades.
.
The El Niño and the PDO were known by the fishermen of Peru and the North American coast community for over 100 years, but for meteorologists, climatologists and scientists in general, has become a novelty in the second half of the twentieth century. I remember that a congress of the Brazilian Water Resources Association (ABRH) in the early 80s, the final section was a plenary discussion on the influence of El Niño on the climate of southern Brazil, was a great novelty for the great most participants!
.
With the extensive discussion that skeptics are causing in the climate science, who is winning with this is all science and climate scientists (skeptical or not). I'm even thinking that opened a new editorial necessity textbooks for scientists in general to describe the weather and climate, books that do not take a position, but to clarify all the basics of meteorology and climatology for those with scientific knowledge. Nowadays we have a galaxy of scientists like economists, biologists, physicians etc.. working on climate and do not know what they are writing!
.
Dear Nina.
.
For over eight years I have been worrying about the exact effects of climate on human life and health, and the more I research on the subject encounter more challenges in defining exactly what is real and what imaginary as well as the contradictions that exist.
.
Generally all works that are written on this topic are punctual checking a single hypothesis without caring about other effects that may arise, ie lack a more holistic view. Summary questions the contradictions that currently comes to mind.
.
1) What is natural climate variability?
.
2) The proposed scenarios are beyond the natural variability?
.
3) What is healthy for humans (we are also a species to be conserved)?
.
4) What are the historical examples of change and resilience of the man with the weather?
.
5) What is better, the adjustment or attempt to change the climate?
.
6) What should be prioritized, or when the precautionary principle can have opposite effects in terms of the mankind?
.
I seek to develop answers to questions, but will give a long text with many references, if you want I can develop it and send to you. Put as an answer in ReseachGate exceeded its target and would not want to bore the other participants of the gate.
.
What would be more write personal opinion based on scientific knowledge of others, do not know if it would be much utility?
.
You are right. Assessment can help mainly for planning and decision making. You can know future projected scenarios. No model is perfect but useful. You can generate future scenarios using downscale data and appropriate models, this is not bad at all.
Yes, get parameters involving climate change impacts on water resources and use FUZZY logic or AHP for uncertainty. It may not give you a good accuracy, but probably it helps you.
Applying a Reliability-based Probability Analysis, you cannot predict only future occurrence but get reliable solutions to assess a climate change. Let me tell how it started: When there was no global scientific analytical approach yet, non-domestic animal will run into homes, shelters for cover as they sense the future occurrence of earthquake around their area - animal prediction (AP) we say - but today, we can predict more accurately than AP the future occurrence of the so-called earthquake at any place. So it goes.
First of all, we need to determine the factors affecting the climate change and then we need to assess the effects of that. For the uncertainties, I would use past data with a predictive filter (Wiener filter, Kalman filter) to get the future uncertainty. If you are using Kalman filtering, you need to use a constraint (in this case, maybe the climate change model). The uncertainities you will get, will also include the errors due to the approximations made in the model.
Here is a slightly different take. Let us try and determine the impact of climate on humankind. There are two variables, the first more uncertain than the second. We have some idea with broad uncertainty regarding climate change. It is a fickle beast with anthorpogenic impacts and, of course natural variability. Who could have forecast the steadiness in warming over the last 15-17 years. Maybe on centennial time scales we can say something but much work is to be done. The second factor is more certain and that is population growth. Pol[opu,atin ahs doubled in Pakistan since the mid-1980s. Other countries (Indai and the African nations) are showing similar growth. Thus even in a constant climate, the impamcts of a growing population
Please forget the first post:
Here is a slightly different take. Let us try and determine the impact of climate on humankind. There are two variables, the first more uncertain than the second. We have some idea with broad uncertainty regarding climate change. It is a fickle beast with anthropogenic impacts and, of course natural variability. Who could have forecast the steadiness in warming over the last 15-17 years. Maybe on centennial time scales we can say something but much work is to be done. The second factor is more certain and that is population growth. Population has doubled in Pakistan since the mid-1980s. Other countries (India and the African nations) are showing similar growth. Thus even in a constant climate, the impacts of a growing population are devastating: water resources, the need to farm in hazard prone regions, urbanization and etc, have impacts on society. Now add climate change which may emphasize the problems associated with climate change. I think that we tend to think of climate change first and population second. But I think we should not forget that anthropogenic climate change is a people problem: more people, more CO2: more people more environmental degradation.
Just some thoughts:
PW
rst post: by acident:
When dealing with the possible forseen effect of cilmate change, regardless of the different contrastic models, a very important critical factor is very often completely overlook. How different surface properties would be sxpected to respond to a given climate change. This factor is tremendousely important in dryland areas. It is obvious that surface properies play a critical role in the above areas. Rocky, sandy or loess covered areas would be expected to respond differently to the same climate change. Sometimes opposit effects should be expected.
.
Everyone.
.
I'm sorry, but I will displeasing most who write here, but I'm tired of listening fallacies about the problem of overpopulation in the world, as they always are on the head of my African and Asian brothers.
.
If we are researchers and men of science, we use data projections and correct, I intend to talking first about Africans, then I'll talk about Asians.
.
The African population density is currently about ¼ of population density of France and 1/13 of the population density of the Netherlands and have never seen a reference to the countries of Western Europe should make one-child policy to relieve overcrowding problems.
.
Projections of population growth do not take into account the process of rapid urbanization in Africa, when we all know that a rapidly urbanizing region that the population growth rate decreases dramatically, Brazil, for example, went from 2.99 to 1960 to 1.11 in 2010. As there was in Brazil at this time wars, famines (quite the contrary decreased hunger) or increased infant mortality or decrease in mean life is attributed to the effect of urbanization, then Africa will follow with population growth rates, as are designed for more than ten or twenty years. Moreover forecasts for Asia are declining population in the coming 15 years.
.
If we look at European countries such as France (115 inhabitants / km ²) and the UK (255.6 inhabitants / km ²) despite they present much higher densities than the African (30.51 inhabitants / km ²) and larger than the continent Asian (89.07 inhabitants / km ²), nobody talks about public policies to control population in these countries. Including projections of population growth both France and the UK are constant until the end of the XXI century (and is not by an aging population).
.
If we ponder the population of OCDE countries by per capita consumption of natural resources, we see that the problem of the future of the world is not in the hands of Africans and Asians (or spoken in South America because there is no comparison).
.
Soon I beg that when you talk about problems of overpopulation, lack of natural resources and generation of greenhouse gases act like scientists, query data.
.
Thanks for adding that Rogerio.
What is interesting is to look at the carbon intensity of different economies and, in the case of developing countries, the development strategy proposed. There's also the question of vulnerability to climate change, where some African countries are particularly vulnerable.
Rogerio, you say: ".nobody talks about public policies to control population in these countries". Referring to UK and France. The reason is that the population is already in decline and has been for some time. The problem in the richer parts of Europe is the future lack of people in the long run (who is going to take care of the eldery?). Not the opposite. Also, richer, countries can easier deal with environmental issues. Sweden, as an example, has decreased its CO2-emissions since 1990 with some 20-30% and still shows a continous economic growth since then. Outside Europe few countries have managed to decrease emissions and show economical growth at the same time. From an environmental viewpoint its better to concentrate people on fewer localisations since its less costly and more efficient to build an infra-structure that can cope with environmental issues in comparison to a widespread, less dense, population.
I agree that the connection between population and environment is not straight forward. See this piece by H. Rosling:
http://www.ted.com/conversations/39/why_do_so_many_think_that_popu.html
I can recommend all to take part of Hans Rosling excellent universal database (se below) on most parameters for the world. It will give some surprising answers and very often far from the perceived reality of things. (More perhaps in line with what Rogerio states)
http://www.gapminder.org
The world is actually improving and about climate change, there are not many real world observations (so far) supporting that the changes are outside a normal envelope or cycle of some sort. (Exception: heat waves). That, however, doesn´t mean that there will not be in the future. So to connect to the original question, it is very important trying to assess the uncertainties. We have limited resources and it is very important that they are not thrown away on something that could look as politically correct right now but really is based on emotions or political motifs ("we HAVE to DO something!"). But personally I don´t see a case for panic just yet. The growing economies will eventually deal with the environmental issues synchronised with other issues in society. To me, that is the most sustainable way forward. Forcing people to stay in poverty due to perceived (or "modelled") environmental issues does not lead anywhere than to conflicts. And, that would surely be the results if the fossile path is closed to them.
Does it make sense to assess uncertainty in climate change impact projections?
Yes! the main issue in climate change impact study is uncertainaity of prediction. Different GCM output give differenct projecetions of change in temperature and preciptaion. Some times it is difficult even to understand the direction of changes . It is an important issues to assess uncertainaity of impact of climate chagne on water resourses and other sectors.
It's also important to look at where people live. High population densities need the mechanism to support them. South Africa, for example has areas of very high population density, surrounded by lower population densities, particularly in arid areas.
For example, Gauteng Province, where I am sitting writing this has a population density of 675.1/km^2 which is supported by industry, services, government etc. In contrast, the Northern Cape is more than 20 time the size of Gauteng, but supports a density of 3.1/km^2, being largely unindustrialised and arid to semi-arid.
Problems in this analysis also arise when developing countries start to emulate the developed world's growth path, starting out with a fossil-fuel powered period of industrial growth. In many cases, the growth model followed is based on exports, with post-industrial societies being the target markets of choice. Stated crudely, developed countries can lower their CO2 outputs by exporting industries to developing countries with access to abundant cheap fossil fuels. Attempts to mitigate this will often be resisted in the developing world, as the need for export-driven growth is seen as the only way to grow economies.
Dear Jan
.
My opinion is more of an emotional outburst than anything else.
.
I'm not too worried about the CO2 levels in the atmosphere, because I think the effects of greenhouse gases on Earth's temperature are overvalued. The coefficients give the climate sensitivity to increased CO2 are generated by Bayesian models (soon, statistical) and it seems that their coefficients have not been adjusted to the data of the last 15 years, that proves the extent that global models do not are following the trend of the measured data.
.
In hydrology was used for many years models in which the coefficients are recalculated when you have larger temporal series. In climate models the results follow other trends than those provided and they remain with the same degree of sensitivity.
.
So, my main concern is with the use of non-renewable natural resources such as fossil fuels. Just to give an example, in the same period in which it doubled the efficiency in internal combustion engines was increased fleet of SUVs practically nullifying the entire efficiency gain. Also the use of improper fuel for large cities, such as diesel, which generate highly particulate pollutants and damaging to both the environment and the health of people, countries like France subsidizes its use for reasons of competitiveness of its industry.
.
Germany in its program to eliminate the presence of nuclear plants is using energy from coal ( old and polluants plants ) also for purely economic reasons (used its reserves).
.
Could follow to cite inconsistencies in OECD countries, which not to harm their industry and not reduce the consumption patterns in their countries take unilateral decisions that go the opposite direction of sustainability.
.
European countries and the United States have ecological footprints ranging from over 150% to 50% its biocapacity, while most of Africa the situation is reversed, ie, its biocapacity is between 50% to 150% or above their ecological footprints.
.
To make clearer the explanation, if we analyze the 2007 data on the balance of the ecological footprint, we see Europe (considering European Russia) with a deficit of -1.8, if removal Russia (+1.3) of accounting , sure the deficit would approach the deficit of the United States (-4.1). Africa this same accounting still has a surplus of 0.1. If we compare Europe with Latin America the surplus reaches +2.9. A country like Brazil has a surplus of 6.1! Important to note, even with a tremendously comfortable situation as the situation in Brazil, I see more reports in the world press about the problem of preservation of Brazilian forests than population pressure from countries like France (-2.0), United States (- 4.1) and the UK (-3.6) that despite being in debt to the nature insist on following increasing its population by more than 50 years.
.
Despite all these data have never seen a citizen or an Latin American NGOs say Europeans should adopt a policy of middle child per couple to maintain the sustainability of Earth!
.
This all has a clear political bias and due to all I am extremely upset.
To Steve Brenner,
Your answer starts out strongly in recognizing that there is uncertainty in models regardless of how the models are used, or developed. However you seem to miss the important point that any regulatory or policy discussion should make every effor to reduce uncertainty in forecasts (i.e. make them as accurate as possible) but then should also provide the resource managers or policy makers with our best estimates of the levels of uncertainty the probabilities of a range of potential outcomes and the seriousness of their consequences.
This lattter step has been greatly discounted in many scientific disciplines where mechanistic models require a great deal of computational time. With this difficulty quantitatve statements of hte precision (i.e. uncertainty) in forecasts or other predictions have been relegated to "statistics" minmizing their importance.
The promise of "predictive" models justified through developement from basic phycical processes will only be fully realized when the majority of input paramters are fully known, which is rartely the case. As such, physics based models, while better than purely empirical models remain little more than regression models that require a great deal of computational time. Absent the computational bottle necks one would fit uncertain model parameters by maximum likelihood and assess model fit and test hypotheses as is done with purely empirical or simple mechanistic models.
In efforts to remedy this situation, the methods known as model emulation have been developed wherin the mechanistic model is treated as a computer experiment and a statistical representation is developed as a means to side-step the computational difficulties. Under such circumstances, much more complete description of hte error distributions surrounding predictions are readily available.
It is my belief that in all situations, resource and policy managers should be provided both the model forecasts as well as a statment of the liklihood of a range of potential alternative outcomes, or alternativley admission that the accuracy precision of the model outputs are unknown.
@Filippo,
I think we are on the same page other than me being a little more cynical than you, which may be a function of age?
Well me too..I thought you might have been a bit younger than me...By the way I passed on your paper on concept maps to a group with which I regularly work on design of large contaminates sediment remediation projects where we have the problem of engineers, scientists (field and lab), resource managers and political liasons all talking past one another. I think your approach will be helpful ...
Nothing is perfect and absolute in nature, its changing and living beings always be living in uncertainity. Having said that it should also be taken into considerations there are over estimations and poor prepectives on the nature behavior. The frequency of changes if its fast it may be difficult to assess and safeguard, if its medium to slow a fairly adjustable time is required. The time scale in the nature lasts from a few minutes to thousand years to seek the change. The desire to govern the nature is always the sole logic of understanding the climate change debates by predicting the nature as an after effect of human actions. The lack of data on many counts and unaccessible natural phenomenon remains unexplained contribute towards the uncertainity. The school of thoughts traditionally differs on the logical understanding of nature through systems, which relies on the theoritical principles accepted in various sciences. It would be harsh not to belive some of them, whereas unless there is multi-disciplinary access to information across the globe there is going to uncertanity. The involvement of the institutions across the globe is on merit and micro assessment of their own regions, global prespectives also calls for scientific collaborations rathan than policy understanding. The models are limited and should not be made scapegoat on our poor understanding, scientists work hard to acheive their goals but to all of us this not going to make serious dent in understanding the nature. Its full of surprises and turns dailys to effect the micro climates in our regions. The current approach assumes uniformity to larger extant which may not be true at regional and local scale. Those of us who got opportunity to work with policy must understand the nature is major stakeholder and we do not effect them through papers but actions on the ground. The decision makers needs to be convinced that phenomenon explained about the climate change is real and certain. Not to panic but have some affirmative action. Are we certain than about the concern expressed by the decision makers? We cannot pinpoint the objects which is near and far in our own timescale. Thats why they ask where and when? Macro modelling did not results beyond policy making and limit us to the level which does not help decision makers.
Policy makers on the other hand, would love a hard answer that says 'this is what will happen.' With completely opposite approaches to the issue, it is hard for scientists and policy makers to come to a mutual understanding on this topic. Climate modelling, and modelling of individual Earth processes may provide a link between these two parties on climate change. With scientists contributing their understandings and observations to the models, and political leaders outlining the aspects, facts and figures they are able to accept, a model may be produced. This model will then, not be a product of scientists telling political leaders what will happen, or of political leaders simply predicting the future. Instead, the model will show a scenario resulting from accepted facts, figures, processes and ideas from both these parties. Also neither party can be held individually responsible for the model yielding unfavourable results. With some diplomatic liaison between political leaders and scientists, a model integrating both view points and accepted figures may be produced. This can then be used to project Possible outcomes of applied policy and changing processes. As a mutual creation, it puts both parties on the same page when discussing and assessing the impacts and effects of various government policies and environmental processes. While it is important to have reliable inputs, often too much focus is placed on minute details, instead of the broad-scale impacts of a process. Political leaders are also easily overwhelmed by many small details, for which they care little. By producing a model, all of these details are considered and included, but do not distract from the final outcome. Models are a useful tool that should be mode widely utilised to generate productive discussion between scientists, policy makers and the wider community.
A little cynicism never hurt anyone.It even sometimes works on policymakers. If I look through old presentations, I'll find the one I did in 2012 where I explained that a predictive model presented in 2010 didn't have the benefit of 2011's data and had since been updated and refined.
Yesterday I was pointed at this article: http://www.theguardian.com/science/2013/dec/02/scientists-policy-governments-science, which makes a lot of important points. Most important for me is that policymakers do understand uncertainty. I was recently pointed at some business literature about dealing with uncertainty and found that there's a lot that we can learn in how we communicate it.
In essence, there's uncertainty that we can minimise with more research or better research and there's uncertainty that is inherent to a problem and can't be addressed. This gets dealt with using scenario analysis, rather than rigid planning. What's happening in a lot of science-informed debates is that scientists withdraw from these processes and so don't take the opportunity to analyse scenarios. This detracts from the quality of the scenarios and the relevance of the science. We also have a lot to learn.
Of course the cynic in me always feels that there's not a hell of a lot you can do when scientists recommend a course of action, based on the best information available, bean counters and engineers disregard it and remove the safety factors and then don't even implement what they claimed to be implementing. This happens with amazing regularity. Luckily when everything goes pearshaped and natural systems are involved, the politicians can always claim an act of God.
Everything surmises in to the fact that we as homo sapiens must ensure that our place is sustainable to live in for the future generations to come and treat each phase of development with caution. The greatest one word phenomena which keeps life icking is 'Change' be it in any field. So, we should always reason ourselves in whatever lies ahead instead of taking a blind route for our 'human aspirations' in the form of development. Part of this reasoning is what you study for climate change.Rather than asking whether it will change, be ready for whatever possible changes that could occur in future.
'New knowledge can expand uncertainty' this is an excellent comment from Michael. So where do we draw the line with uncertainty in climate change predictions. It remains an on - going iterative process in every sphere of influence and limits the prediction time frame from long-term to short-term thus leaving room for emerging knowledge. Another possibility is to re - incorporate uncertainty at periodic future intervals, should current assigned probabilities change in the future.
There are a lot of interesting answers, comments, suggestions regarding the question related to the uncertainty assessment in climate change projections as well as to the decison-making process under uncertainty. I would like to thank all of you for your interesting comments and for this nice debate. Please, let's continue talking about this issue as it is with rising concern for all of us. Thanks!
I remember to have responded this or a similar question early: I had pointed out that it does not make sense to start afresh and the best course look at the current practice and improvise. The current practice in water management projects (irrigation, hydroelectric, transportation, fishery development, flood control etc) is to study the impact of variations over a century, 50 years into the past and fifty years into the future. The best way is to evaluate the efficacy of predictions made in the past. We may review this practice of half century cycles, increase or decrease period of cyclical analysis. Is there any better and more rational approach?
Less predictability and knowledge about the upcoming events promotes uncertainty. But can be overcome by 'new knowledge' as pointed by one of the respondent. If thats related to space and need of hour than would be to have spatial prediction model. The development models i.e., transportation, industries and housing etc needs to be over layed over regions with the prediction. The time and scale matching of the human and ecological activites can give some amount of certainty. All for the future actions and estimations, macro, meso and micro planning by the governments needs to understood and over layed. This can give the estimated vulnerability and adaptation planning for any future extreme events (which is highly unpredictable and uncertain). Need a good amount of time and research to meet the SPM targets.
Hi, a few resources on global historic climate recording as fuel for the debate:
ACRE initiative (http://www.met-acre.org/) there was a web meetup last month I'll see if I can post links to
20th Century Reanalysis (20CR) project (http://www.esrl.noaa.gov/psd/data/20thC_Rean/)
My paper on data dissemination and policy is posted here on ResearchGate
In a recent commentary in the nature climate change by Katz et al., (2013), the authors highly suggested that there is a need to include at least one author on all chapters of IPCC with expertise in uncertainty analysis to enhance the quality of the treatment of uncertainty. The statement reveals that despite the importance of incorporating uncertainty, its present treatment is viewed as quite inadequate. For the case of climate sensitive systems, to the best of my knowledge, uncertainty quantification is treated poorly in the current studies especially when climate change impacts on water resources and hydro systems are concerned. Apart from only a few exceptions, uncertainty analysis is overdue in relation to precision, model and measurements errors as a whole; Current studies have referred mainly to one type of the mentioned errors in isolated and fewer have attempted to consider all in one study. This may root in the fact that water resources engineers alone cannot solve challenging problems in quantification of uncertainties associated with climate change. Rather, increased collaboration between statisticians, climate scientists and water resources decision makers is compulsory.
Husain,
I agree, uncertainty analysis has been poorly handled. The characterization of uncertainty levels in the IPCC reports are generally amateurish.
In order to characterize uncertainty one must first of all develop a model of how the uncertain quantity is expected to behave. Is it likely to be limited to a narrow range? Can it spread equally on either side of a median value, such as a gaussian distribution? Is it likely to be a long-tailed distribution such as lognormal?
Once a behavioral model has been established, various techniques such as Monte Carlo analysis can be used. Then, and only then, can quantitative statements of uncertainty be made.