How Reliable Are Climate Models?

Full Briefing available in PDF format.

Preface

Predictions of harmful global warming from human-generated greenhouse gases are largely based on computer model “experiments” intended to show what happens to the earth’s climate system when it is subjected to various “forcings” or postulated changes in energy levels. Although impressive in their complexity, the models suffer from several well-known weaknesses. The computers that operate them are not fast enough to simulate climate change on sub-global scales; hence the models cannot forecast the regional or local impacts of global warming or cooling.

More importantly, climate models suffer from the limitations of current scientific understanding of the forces that determine weather patterns and climate change. Climate modelers cannot avoid making assumptions about the operation and effects of such key variables as water vapor, clouds, ocean-atmosphere interaction, solar cycles, ocean currents, and the natural carbon cycle. Those assumptions are fraught with uncertainties and may be little better than educated guesses. It is not surprising, therefore, that the models have been notoriously poor at replicating current and past climate. For example, the models that served as the scientific background for the 1992 Rio Treaty implied that the world should have warmed 1.5°C since the late 19th century. In fact, surface measurements indicate the world has warmed only 0.5°C. The models were off by a factor of three.

As the models have improved, the amount of predicted global warming has gone down. In 1990, the UN’s Intergovernmental Panel on Climate Change (IPCC), using then state-of-the-art models, forecast a 3.2°C warming by 2100. In 1995, the IPCC lowered that forecast to 2.0°C.

The question for citizens and their elected representatives is whether even today’s models are accurate enough to guide public policy making. Seeking an informed discussion of this topic, the Competitive Enterprise Institute, on behalf of the Cooler Heads Coalition, invited climatologists Robert Davis of the University of Virginia and David Legates of Louisiana State University to address an audience of congressional staff, journalists, and other interested persons.

This joint lecture covered a lot of ground, but the high point or peak was the revelation, graphically documented by Professor Legates, that all major climate models both overestimate and underestimate rainfall by as much as 60 inches per year over huge portions of the globe. Rainfall affects the mass balance, energy balance, and water balance of the climate system. Rain (or the absence thereof) is what most people mean by “weather.” If the models can’t get rainfall right, why should we trust them to get anything else right?

Marlo Lewis, Jr.Vice President for Policy & Coalitions