Reducing Uncertainties

Uncertainty in future climate stems mainly from scenario and model uncertainty, and both of these for the most part are not intrinsic to the system. In principle, we are free to choose a scenario for future emissions through our actions. Model uncertainty is different, as there is a 'true' climate system, so the uncertainty does not reflect a choice but our incomplete understanding of the system and our inability in describing it in a numerical model. For the decision relevance that climate projections provide, it is also interesting to ask how projection uncertainties will change in the future. If uncertainties are likely to disappear soon, the strategy to wait for better information before spending money may be attractive. If not, then the strategies certainly need to be robust under uncertainty (Lempert and Schlesinger 2000) and an early decision may be wise, for example to have more time to adapt.

Climate models and their projections may improve in the near future in various ways. Some of the errors in the mean state of climate simulations (for example, errors in average temperature and precipitation patterns) appear to develop very fast after the simulations are initiated, suggesting that their causes reside in the behavior of the atmospheric part of the system, rather than the slowly evolving ocean state. Thus, combining weather, seasonal, decadal and long term forecasts in the same modeling framework may improve longer term projections, allowing to develop a better understanding of processes through the verification offered by the shorter term forecasts. This idea of 'seamless prediction' (Palmer et al. 2008) is currently discussed in the scientific community, but it is certainly challenging, both technically and because many assumptions and parameterizations in weather and climate models are not valid across the whole range of spatial and temporal scales that these models would cover. Evaluating models for different climatic states (e.g., the ice age, Otto-Bliesner et al. 2006), variability and trends and on abrupt changes observed in the past can reveal limitations in the model physics.

Short-term predictions may improve through initialization with observations (Smith et al. 2007), assimilation of data or synchronization of multiple models (Kirtman and Shukla 2002). This idea is actually at the basis of the newly developing area of decadal predictions, where a climate model initialized close to the observed state (especially of the oceans, which drive the behavior of the system in the slower frequencies) could be able to generate a climate in sync with the real world, thus moving from projections to actual predictions of the climate system over decadal scales. This new area of research will have to address fundamental questions of predictability (e.g., for how long can we expect two closely initialized versions of the system to stay close?) and methodological issues (e.g., how do we observe enough of the ocean's surface and depth to have a sufficiently accurate representation of the real thing that we want to mimic in our models?). The belief though is that, if successful, these shorter term predictions aiming at simulating not only the overall trend but the decadal oscillations around it would be extremely valuable to impact researchers devising adaption solutions.

Increased computer performance will allow for higher resolution in climate models and more simulations. Resolution will help to improve certain aspects of the simulation (e.g., to resolve topography, or convection) but it does not necessarily help in the case where the processes are poorly understood (e.g., how to parameterize the effect of vegetation on climate and the water cycle). More simulations will be useful to better quantify the uncertainty in the models. Better observations, in particular long-term records of observed changes, will further constrain the models and help to understand processes critical to improving model performance.

Many important quantities in the climate system are only observed since the advent of satellites, making it difficult to separate long-term trends from natural variability. A hierarchy of models (Held 2005) with different structures and families of similar models can be used to track a behavior across models, and to identify which quantities are most useful to identify model deficiencies and to constrain future projections (e.g., Hall and Qu 2006). Parametric uncertainty in a model can possibly be reduced by calibration if computational capacity is large enough to run the model many times. The structural uncertainty, i.e. the fact that the model is unable to match all observations for any set of parameters, or that different model formulations may do similarly well at reproducing observations and cannot be distinguished, is harder to eliminate.

Given the complexity of the system we are trying to describe and predict, and the large number of processes, interactions and feedbacks occurring on different spatial and temporal scales, the uncertainty in climate projections may not decrease quickly in the near future. The present climate seems to provide only a weak constraint on the future, and models continue to improve in simulating the present (Reichler and Kim 2008) but they do not clearly converge on the future trends. Models also continue to include more processes and feedbacks interactively, giving rise to new sources of uncertainty and hitherto unknown interactions Some uncertainties are intrinsic and irreducible (e.g., the chaotic nature of short-term weather and climate variability which limits the predictability on timescales of weeks to years, or the timing of volcanic eruptions in the future).

It is therefore important that scientists specify all possible outcomes, rather than trying to reduce spread where it cannot be reduced. The situation is difficult in that overly optimistic and tight uncertainties may make society vulnerable if things turn out to be different from what was predicted. On the other hand, providing large uncertainty estimates can prevent action, and is often seen as being alarmist because extreme changes are not ruled out. Some constraints will emerge as climate change proceeds, so even with the same models and methods, we expect uncertainties to shrink somewhat. In some situations they may also grow, if the additional data reveal that the model is imperfect, and that further processes need to be considered.

Renewable Energy 101

Renewable Energy 101

Renewable energy is energy that is generated from sunlight, rain, tides, geothermal heat and wind. These sources are naturally and constantly replenished, which is why they are deemed as renewable. The usage of renewable energy sources is very important when considering the sustainability of the existing energy usage of the world. While there is currently an abundance of non-renewable energy sources, such as nuclear fuels, these energy sources are depleting. In addition to being a non-renewable supply, the non-renewable energy sources release emissions into the air, which has an adverse effect on the environment.

Get My Free Ebook


Post a comment