As generally interpreted by the impacts community, there is no difference in these terms. They are synonyms. A few examples of how other communities interpret the term “projection” include:
The synonyms for the term “projection” from Encarta are: “forecast, plan, prognosis, estimate, prediction, prognostication”.
It should need no further discussion that the terms “projection” and “prediction” have the same meaning with respect to multi-decadal global climate simulations.
However, in the global climate community, there is an attempt to make a distinction between the two terms. A recent comment (#30) by James Annan prompted this weblog (thanks James!). He said
“the science of long-term climate prediction (as opposed to mere projection) is still rather in its infancy, …”
Mike MacCracken and I had an informative exchange of essays on this subject in Climatic Change. His paper is MacCracken, M., 2002: Do the uncertainty ranges in the IPCC and U.S. National Assessments account adequately for possibly overlooked climatic influences. Climatic Change, 52, 13-23.
An extract from his essay with respect to the terms “prediction” and ‘projection” state,
“2. Prediction versus Projection
Based on the failure to treat the various land cover processes that are suggested, Pielke (2002) also suggests that ‘[i]f climate prediction is not possible beyond some time scale, a focus on vulnerability is the preferred scientific approach to provide policymakers with useful information’. This statement seems to me to confound two issues that merit further attention. Regarding the first, both IPCC and the U.S. National Assessment are very careful in their usage of the word projection rather than prediction. For these groups, the distinction is meant to convey a very significant difference that is too often being ignored by critics of these reports. Acknowledging Pielke’s point that at least some dictionaries are not yet capturing these subtleties (although my 1999 Webster’s II New College Dictionary defines projection as a ‘plan for a future course of action’ rather than a prediction), I would argue that the differences in these two words are roughly as follows:
• A prediction is a probabilistic statement that something will happen in the future based on conditions that are known today and assumptions about the physical processes that will determine these changes. A prediction generally assumes that future changes in factors other than those being predicted will not have a significant influence on what is to happen. In this sense, a prediction is most influenced by the ‘initial conditions’, that is, predictions depend on the current conditions that are known through observations. Thus, a weather prediction indicating a major snowstorm will develop over the next few days is based on the state of the atmosphere today (and its conditions in the recent past) and not on unpredictable changes of other potentially influential factors that serve as ‘boundary conditions’, such as how ocean temperatures or human activities may change over the next few days. A prediction is made probabilistic by accounting for various types of uncertainties, for example, in
the accuracy of observations, in the chaotic state of the atmosphere, etc. For decision-makers, what is important is that a prediction is a statement about an event that is likely to occur no matter what they do (i.e., policymakers cannot change tomorrow’s weather).
• A projection is usually a probabilistic statement that it is possible that something will happen in the future if certain conditions develop. In contrast to a prediction, a projection specifically allows for significant changes in the set of ‘boundary conditions’ that might influence the prediction. As a result, what emerges are conclusions of the type ‘if this happens, then this is what is expected’. The simplest type of projection is to extrapolate into the future assuming all of the boundary conditions remain the same or that the same trends prevail. For projections extending well out into the future, however, this is often a poor assumption, so scenarios (or story-lines) are developed of what could happen given various assumptions and judgments. For example, IPCC (2001) projects a range of possible temperature increases for the 21st century that calculations indicate would result in the event that the world follows a number of plausible story-lines concerning population and economic growth, energy technologies and emissions, and demographics and international relationships (see IPCC, 2000) – but also assuming no agreements to
limit emissions due to concerns about climate change. By considering how the resulting changes in atmospheric composition would affect the climate using seven different climate models, each with its own particular climate sensitivity, the projections of climate change accounted, to a reasonable extent, for a wide range of possibilities of both societal development and climate behavior.
Given this approach, projections are clearly indications of what could happen if certain assumed conditions prevail in the future – they are neither a prediction nor a forecast of what will or is likely to happen. For decision-makers, a projection is thus an indication of a possibility, and normally of one that could be influenced by their actions.”
My reading of these definitions is that “a projection is a prediction if certain actions (e.g. CO2 reduction) are not undertaken. A “projection” is a “conditional prediction” (conditioned in that humans can alter the prediction based on specific actions). These are both predictions!
In the context of weather forecasting, a ‘weather projection’ would be one where we could, for example, alter a hurricane’s intensity and path by cloud seeding. This is still a prediction.
My view on these definitions was discussed in the same issue as Mike MacCracken’s essay. My article is Pielke Sr., R.A., 2002: Overlooked issues in the U.S. National Climate and IPCC assessments. Climatic Change, 52, 1-11.
I summarized the types of climate modeling in my weblog of July 15 2005 entitled “What Are Climate Models? What Do They Do?” An extract from that weblog catalogs the types of models as,
Process studies: The application of climate models to improve our understanding of how the system works is a valuable application of these tools. In my Climatic Change essay, I used the term sensitivity study to characterize a process study. In a sensitivity study, a subset of the forcings and/or feedback of the climate system may be perturbed to examine its response. The model of the climate system might be incomplete and not include each of the important feedbacks and forcings.
Diagnosis: The application of climate models, in which observed data is assimilated into the model, to produce an observational analysis that is consistent with our best understanding of the climate system as represented by the manner in which the fundamental concepts and parameterizations are represented. Although not yet applied to climate models, this procedure is used for weather reanalyses (see the NCEP/NCAR 40-Year Reanalysis Project).
Forecasting: The application of climate models to predict the future state of the climate system. Forecasts can be made from a single realization, or from an ensemble of forecasts which are produced by slightly perturbing the initial conditions and/or other aspects of the model.
Under what of these three types of climate modeling does the term “projection” best fit with respect to multi-decadal global climate simulations, such as used by the IPCC? “Forecasting” is clearly not the type as concluded by Mike MacCracken, and I assume also by James Annan based on his comment on the weblog. Obviously, they are not “diagnostic climate models”.
The multi-decadal global climate model simulations are process (sensitivity) studies. The use of the term “projection” is misleading the impacts community that uses the model results for their studies. Its use is confusing to the impacts community who almost all interpret (correctly) that the term “projection” means “prediction”.
This is a significant issue. The use of the term “projection” results in erroneous communication to policymakers on the accuracy of the multi-decadal global model simulations in being capable of accurately predicting the climate in the coming decades. The models cannot be skillful since they do not contain all of the important first order climate forcings, as identified in the 2005 National Research Council Report. Indeed, when a range of global averaged surface temperature increases for the coming decades are next presented, the question should be asked if this is a prediction, or just the result of a process modeling study with incomplete climate physics ? The honest answer will be the later.