Climate Science, Global Warming and The Scientific Method

Summary: Climate science investigations related to global warming are undertaken entirely according to the scientific method, which precludes developing preconceived conclusions and seeking biased data in support of those conclusions. Of two possible ways of conducting scientific investigations, most climate science research can be characterized as being descriptive in nature, and generally cannot be carried out with the use experimental controls that can be set by the investigator. The resulting data enhance our understanding of the global climate, and serve as the basis for models predicting future climate behavior.


Public perceptions of global warming. The American public is dramatically divided over the question of global warming. Many accept the findings that have been developed by our scientists as being accurate, and being based on sound scientific principles. Others oppose the notion that global warming, due to human activity and originating primarily in the burning of fossil fuels, is “real”, i.e. that the planet is actually undergoing an abrupt change in its climate due to greenhouse gas accumulation.

A simplified classification of scientific investigations. Climate scientists, as members of the broad community of physical and biological scientists, undertake their studies according to well-established principles of scientific investigation. Some of these principles are described here. This characterization is necessarily simplified here for the benefit of the general public.

Science is undertaken as an objective, unbiased inquiry. First, all scientific investigation is undertaken as an open inquiry. What this means is that a scientist engages in an investigation with an unbiased mind, avoiding any prior expression of what the result of the investigation should reveal. Valid science is not conducted by adopting a conclusion at the outset and seeking out those particular findings or results that bear out the preordained conclusion stated at the outset, or by designing a study in such a way as to provide those results.

Hypothesis-driven experimentation. In the simplified view presented here, scientific investigations in general can be classified in one of two distinct ways. In the first, a scientist will initially formulate a hypothesis, and conduct experiments to confirm or deny the accuracy of predictions based on the hypothesis. Hypotheses usually build on experimental results already in hand, and are constructed around particular unproven statements or assumptions about the experimental system being studied, using experimental variables subject to the control of the experimentalist. Experiments are then devised that change one or another of the variables to characterize the effect of the variation on the outcome of the study.

Hypothesis-driven experimentation is readily devised to include a second, reference, condition of the system, which is intended to characterize the system in the absence of the variation imposed in the first experiment. This reference is usually called a “control” condition, or the experimental manipulation involving the reference is called the “control” experiment, or just simply the “control”. Then two parallel experiments are conducted in which the variable is changed in the first experiment but is held constant in the control system. The results obtained in the first, varied, system are compared with those of the control system, and an objectively valid conclusion is drawn based on the finding of a difference, and/or on the amount of the difference, between the experimental system and its control. At this point, the investigator will conclude that the hypothesis is correct if the results bear out the predictions made according to the hypothesis, or else will conclude that the hypothesis was incorrectly drawn if the results contradict the prediction of the hypothesis.

Hypothesis-driven experiment, with a control for the variable C.

Descriptive investigations. A second class of scientific investigation can be called a descriptive study. Descriptive studies in general will examine an experimental system, providing a description of its nature without a preliminary formulation of a hypothesis. The scientific contribution consists in developing the new information, the new characterization, that was not available before. Generally descriptive investigations cannot be carried out using parallel controls, for any of a variety of reasons. Frequently, for example, a system cannot be contained in a sufficiently controlled environment for the scientist to knowledgeably and definitively to alter a single, particular variable. In such a situation, it may be necessary to consider the state of the system at some earlier time, and compare it with an altered state at some later time, to compensate for the inability to use controls. This is to be contrasted with the hypothesis-driven investigation, where single variables are changed under the control of the investigator, and parallel control and varied experiments are readily carried out. Examples of descriptive investigations include studies of volcanic activity, or of earthquakes, or studies in the changes observed in a series of paleontological fossils, or in current biological species variation.

Descriptive experiment showing the time dependence of the value of a variable.

Climate science is primarily a descriptive undertaking. Clearly climate studies such as those that characterize global warming fall into the class of descriptive investigations. Considering climate as a globe-wide phenomenon, the experimental system in question is necessarily the entire planet. In essence the excess burning of fossil fuels since the beginning of the industrial revolution has converted our entire planet into a single massive scientific experiment, one being conducted in the absence of a control. Since climate scientists cannot conjure up a parallel planet on which burning of fossil fuels is not occurring, much of climate science cannot be classified as being hypothesis-driven. In addition, studies undertaken in the framework of global warming investigations are precisely descriptive in nature. Various characteristics of the climate and of its effects are measured, catalogued, and trends with time are characterized. Some examples of measurable variables include land surface temperatures measured at locations across the planet, ocean water temperatures at the surface and at various depths, wind speeds and direction near the surface and at higher altitudes, and CO2 concentrations in the atmosphere and dissolved in ocean waters. Being descriptive, climate science and the study of global warming are necessarily objective in nature because the data collected are obtained in a totally naïve fashion. If conducted correctly, there cannot be any preconceived or prejudged result created before the fact, which may be sought to bias the outcome.


The time dependence of properties measured by climate scientists. Since climate science is a descriptive process, and control systems are not available, climate scientists draw conclusions from the results of the data collected based on the time development of those data. Thus, if there be a control experiment in climate science, it is the state of the planetary variables, such as temperature, atmospheric concentrations of CO2, and so on, that existed at an earlier time, and the trends for those variables as time advances up to the present. Here again, it is important to emphasize that all such measurements are conducted in an unbiased way, i.e., that the data collection is a naïve process that is not influenced by preconceived ideas about what those data should be.

Data-driven models are used to predict future trends. The contentious aspect of global warming and climate science, to the extent that it may indeed be contentious, is developing predictions of the future behavior of the climate in response to the burning of fossil fuels, and the consequent accumulation of atmospheric greenhouse gases. The models used in these predictions are extremely complex. They are formulated in advanced mathematical terms that require massive computational power to implement them, and to arrive at predictive results. The models use existing experimental data as a basis, and embody various ways of characterizing future climate changes that assume that certain climate processes will prevail over the predicted time frame. Climate scientists recognize that the details of these predictions may differ based on the assumptions made, and that there may be a range of probable outcomes from one model or another. What is certain, however, is the overall conclusion that continued accumulation of greenhouse gases in the atmosphere will have detrimental effects on the global climate as the decades pass. Importantly, sound principles employing the scientific method have been used to characterize our planet’s climate to date, and the best scientific models are being implemented to try to understand the future trends of our climate.