Climate Science: Climate Change and Its Impacts

Studies | Environment | Global Warming

No. 285
Monday, May 15, 2006
by David R. Legates, Ph.D., C.C.M.

Computer Models Are Unreliable Due to Faulty Data and Unsound Assumptions

The claim by global warming alarmists that humans are causing dangerous changes in the Earth's climate is based both on several sets of data — temperature measurements, greenhouse gas levels and other phenomena thought to be affected by the climate (such as precipitation) — and General Circulation Models (GCMs) that attempt to predict the Earth's climate. Specifically:

  • Ground-level temperature measurements show the Earth warmed approximately 1° F over the last century.
  • Atmospheric carbon dioxide (CO2), a primary greenhouse gas, has increased by more than 30 percent in the last century and a half.

However, the link between human activities and current temperature trends and other environmental impacts is very unclear.

"Air temperature measurements do not accurately represent global patterns."

The Data Are Flawed. Our knowledge of present and historic temperature trends is limited by the quality of available data and the way in which it is used. For example, there are major problems with data gathered from direct observation of air temperatures. Air temperature measurements do not accurately represent global patterns because of changes in the location, number, distribution and development surrounding observation sites over the 20th century. For example, city temperatures on warm summer days can be as much as 8° F warmer than the surrounding countryside and annually average about 4.5° F warmer. Over the years, cities have grown so dramatically that many temperature stations are now significantly affected by this urban heat island effect. Cities are heat islands because they have:

  • more impervious surfaces — less heat is removed from concrete than soil via the evaporation of water,
  • less wind — standing structures disrupt and reduce the wind's exchange of heat by convection,
  • darker surfaces — asphalt and other materials that absorb and retain heat,
  • canyon-like clusters of structures — skyscrapers increase solar energy absorption, and
  • heat sources — human activities such as manufacturing, transportation, air conditioning and so forth generate heat.

In addition to urban observation stations being subject to heat island effects, many other observation stations have either been moved or removed, causing discontinuities in the location of measurements over time. Moreover, observation stations are biased toward mid-latitudes, coastal areas and lower elevations — where most people live and where the urban heat island effect is strongest. Oceans (covering approximately two-thirds of the Earth's surface), high latitudes (the Arctic and Antarctic) and high-altitudes (mountains) are underrepresented.

Arguably, the recorded rise in average global temperatures is due in part to inconsistent data collection, rather than actual warming temperatures. The marked fall in global air temperatures between the early 1960s and the mid-1970s coincided with an increase in the density of observing stations around the globe, whereas the rise over the past 30 years has coincided with a steady decline in station numbers.

Because greenhouse gas concentrations and temperatures were not directly measured for most of human history, much of the data input into the models are estimates based on proxy data, such as the concentration of CO2 molecules trapped in ice cores. But even when data from direct observations is available, global warming predictions are often based on unrealistic estimates regarding real-world conditions. For example, Hansen, regarded by many as the "father of global warming," recently conceded that CO2 emissions are now rising 1.0 percent per year, yet computer simulations forming the basis of the Third Assessment Report assumed that emissions would be growing almost twice as fast.3

Furthermore, GCMs predicting substantial global warming in the near future assume far greater per-capita energy use than is currently the case, and far greater future per-capita energy use than most current estimates. By contrast, when models use more realistic estimates of energy use, they produce far less alarming results.

"Predictions by climate models vary widely."

Climate Models Are Limited. To assess future climate trends, climatologists rely upon GCMs that attempt to describe Earth's climate. They include many variables (such as temperature and CO2 emissions) and make assumptions about how changes in one variable affect others. The many climate models scientists use to generate climate predictions vary widely in which variables they include and in the assumptions they make about how those variables interact. For example, there is considerable scientific debate concerning the overall impact of atmospheric aerosols on the Earth's climate and how the models simulate these effects.

Aerosols are minute particles suspended in the atmosphere. When these particles are sufficiently large, they scatter and absorb sunlight, which can reduce visibility. The resulting haze reddens sunrises and sunsets. Both IPCC and National Assessment projections assume that all atmospheric aerosols have a slight net warming effect. However, more recent scientific data provide a less certain answer.

One study — of which James Hansen is a coauthor — concluded that the warming effect of carbon black aerosols arising from human activities is about twice that used in the IPCC's Third Assessment Report.4 But another study suggested that the net effect of sulfate aerosols is to cool the Earth, not warm it.5 The net effect of different types of aerosols must first be properly resolved before we can assume a general warming impact on the climate. As the latter study concluded:

"Until [researchers resolve how] the climate system respond[s] [to aerosols], the possibility that most of the warming to date is due to natural variability, as well as the possibility of high climate sensitivity [to the effects of greenhouse gases], must be kept open."

Thus, the extent to which the present warming trend is due to natural factors, human greenhouse gas emissions, and other impacts like the emission of aerosols remains an open question.

Another variable that climate models do not take into account is the effect of changes in solar radiation on the Earth's climate. Over the past 350 years, scientists have discovered cyclical changes in the Earth's climate due to solar activity — such as increases and decreases in solar flares and sun spots. Some researchers have argued that solar variability may be responsible for about 0.45° F of warming between 1900 and 1990 — just under half of the recent warming — and about a third of the total warming since 1500.6 This is notable since approximately half of the observed 20th century warming occurred before 1940 and cannot be attributed to human causes. Others have shown that the effect of changes in solar radiation can account for 71 percent of the variation in global surface air temperature from 1880 to 1993.7 When changes in solar output are considered in climate simulations, the models predict this warming.8 However, it is still not possible to incorporate into a single model all the variables that affect the climate — such as solar variability, changing greenhouse gas concentrations, volcanic eruptions, changes in cloud type and coverage and various pollutants.

"Computer models cannot accurately predict the present or future climate."

Climate Models Do Not Make Accurate Predictions. In addition to problems with the data, computer models are limited by our incomplete understanding of how the Earth's climate responds to a variety of external forces. They are also limited by the speed and capabilities of contemporary computers.9 As a result, they do not accurately describe the current climate and have not been able to accurately describe the climate of the past 30 years. For instance, computer models consistently project a rise in global temperatures over the past century that is more than twice as high as the measured increase. As the models cannot explain what has happened in the past, it is fair to question their predictions of future warming. This is particularly true of projections regarding regional changes in temperatures and other climate phenomena. [See the side bar on the U.S. National Assessment.]

The difficulty of reconciling GCM simulations of present-day conditions with real-world observations, and the difficulty of formulating appropriate assumptions about human-caused emissions, led the American Association of State Climatologists (AASC) — a professional organization of regional and state climatologists who use local climate data every day — to conclude in their policy statement on climate change:10

"Climate predictions have not demonstrated skill in projecting future variability and changes in such important climate conditions as growing season, drought, flood-producing rainfall, heat waves, tropical cyclones and winter storms."

Therefore, relying on climate model simulations to draw conclusions about the future is very risky, since the simulations do not, and perhaps cannot, accurately simulate the present climate.

Read Article as PDF