Measurement and Assessment

measurement and assessment issues are central to debates about the nature and impact of global warming. Global warming models require information about the parameters in these models (measurement) and exploration of the meaning of the predicted results (assessment). Inference, lack of replicability, cyclical variables, and parameter attributes all shape these models and their predictions. Models can be attacked or defended, based primarily on their underlying measurements and assessments; thus, it is vital to understand the role measurement and assessment have in building and understanding these models.

The essence of science is the development of ideas that produce predictions to explain measurable phenomena. Measurement and assessment are central components of this process. The difference between these two concepts is tied to the modeling concept: measurement refers more to the evaluation of phenomena, and assessment refers more to the evaluation of models. Another link is the quality of the measured data: it is impossible to produce detailed assessments if the measurements are not detailed.

three attributes

It is axiomatic that any parameter has three attributes, only two (any two) of which can be simultaneously maximized. Furthermore, the more the third is minimized, the more the other two can be maximized. These three parameters are precision (minimizing the variance of the parameter), accuracy (the congruence between the calculated value of the parameter and the actual value of the parameter), and generality (the degree to which the parameter can be applied to a wide range of scenarios).

Two complications in the development and assessment of global predictive models are the nature of both reductionistic and experimental approaches. First, science is particularly good at a reductionist approach, that is, to isolate a situation as much as possible and manipulate or explore it to understand it better. However, the planet is not particularly amenable to attempts to isolate one segment of its atmosphere. Second, experimental approaches use multiple subjects to test the effects of changes in variables. Experimental approaches produce models that most confidently assess the predictive power of the model being tested because other variables are controlled. Situations in which experimental manipulation of variables is impossible or unethical (for example, many medical questions) are normally solved by the application of multivariate statistical techniques to extremely large groups of subjects. Planetary questions, such as global warming, are not desirable questions to be answered experimentally; the sample size is currently set at one planet.

Experimental approaches are not ethical or feasible, nor is it feasible to obtain multiple planets (unless additional planets are assigned to experimentation). However, situations such as these are still within the realm of scientific investigation, particularly through the use of inference and model assessment and calibration. The precision, accuracy, and generality of such results are both more difficult to determine and more likely to be challenged by skeptics. Evolution, global warming, and various factors influencing human health are all examples.


The modeler's ideas about relationships among variables need to be specified as precisely as possible. The actual variables have to be measured, the model implemented, and the results determined. Then, model assessment requires: the systematic testing of the degree of influence each parameter put into the model has upon the output. Typically, modeling is an iterative process, in which the assumptions of the model are sequentially considered and more powerful models that reduce the stochasticity or the probable impact of assumptions on precision, accuracy, or generality are reduced. Just as with other variables, assumptions can maximize two of the three attributes; typically, as with other aspects of the scientific method, modelers prefer to reduce generality. Thus, for example, more complicated models that consider each tree separately have replaced models of forest dynamics that treated forests as blocks of area. Models of global climate divide the globe into different areas. Each area has its own dynamics driven by the various parameters modeled, and each area impacts adjacent areas. The size of these areas within climate models has steadily decreased, and the number of parameters and the complexity of the interactions between areas have increased.

Because of lack of computing power, the tension among the three parameter attributes, and the amount of stochastity within complex systems, models themselves cannot be precise, accurate, and generalizable. Models are usually wrong in some way, but even a "wrong" model can still have substantial conceptual and scientific value. Models are especially good at identifying areas of weakness in understanding, areas that will be productive avenues for further research.

Within the context of global warming models, an additional challenge is the incorporation and assessment of cyclical variables. Cyclical variables are difficult to understand because they are constantly changing. A clear understanding of a cyclical variable should include a reasonable mechanistic explanation for the cycle, the periodicity, and the magnitude of variation. Global warming models are particularly complicated because there are multiple cyclical variables, each with its own mechanism, periodicity, and variation. For instance, sunspots, carbon dioxide, and climatic variation (both annual and era) have some degree of cycling. A comprehensive model, thus, needs to incorporate both stochastic variation and periodic variation for each of these variables. It is particularly hard to effectively incorporate a cyclical variable when the periodicity is much greater than the range of time during which measurements have been obtained.

These difficulties do not invalidate attempts to model global climate patterns. Science is not just the art of exploring easy problems. Computational power, the collection of more and better measurements, and richer forms of assessment (such as comparisons between different models through

Renewable Energy Eco Friendly

Renewable Energy Eco Friendly

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.

Get My Free Ebook

Post a comment