Do You Know How Bad Climate Models Are?

Statistics Is Not Science:

Do You Know How Bad Climate Models Are? 

by Laurence Hecht

Editor-in-chief, 21st Century Science & Technology 

The accompanying graphic of actual data from the National Oceanic and Atmospheric Administration (NOAA) shows how far off the best available models are when they attempt to predict the amount of water vapor in the air 48-hours into the future. Similar errors are found in attempts to predict surface temperature and other key parameters of weather. 

If you accept the theory of human-induced global warming, you believe that similar models of atmospheric circulation, based on guesses as to the amount of CO2 that will be in the atmosphere far into the future, can predict the weather 50 or 100 years from now. Study the actual results of a NOAA global model forecast which attempts to forecast weather parameters just two days into the future, and see how silly such a presumption is. 

The graphic shows four maps of the continental United States, with superimposed  color shading and numbers showing the difference between the actual and projected dew point temperature for 12, 24, 36, and 48-hour forecasts for the morning of Jan. 15, 2009. (The temperatures are in degrees centigrade; to convert to the more familiar Fahrenheit degrees, you can roughly double the displayed values.) The upper two maps show the discrepancy for 12- and 24-hour forecasts, while the lower two maps show the discrepancy for 36- and 48-hour forecasts. As you see, the discrepancy between the forecast and verifying analysis grows with forecast projection time.  The 48-hour forecast is off by more than 6° C (10.8° F) for most of the territory of the continental United States, and by as much as 11° C (19.8° F) over some areas. 

Although the period in question is that of an incoming cold wave, the forecast errors are typical for almost any 48-hour period in the year. A similar analysis generated two days later showed discrepancies as great as 20° C (36° F) over parts of the country. The color-shaded values displayed on the maps are generated by subtracting an analysis created from actual measured values, as of the morning of Jan. 15, from the values which had been forecast 12- to 48-hours earlier. What is being measured is the dew point temperature of the 30-millibar layer, that is a layer of air which reaches typically up to about 300 feet above the surface. The dew point temperature is the temperature at which the water vapor in the air condenses into a liquid or freezes.  The dew point temperature is also a measure of how much water vapor is in the air, higher values of dew point temperature meaning more water vapor in the air. The verifying analysis is a weighted average of the data obtained from satellites, aircraft, weather balloons and ground-sensing stations. 

The forecast and analysis values come from a sophisticated global atmospheric model which is used to predict weather over the world. This model is a close relative of the models used to predict the climate for the global warming scares.  

Statistics and computer models are not science, no matter how sophisticated the mathematical functions employed. Frequently, the more elaborate the statistical technique, the less truthful is the model. The recent episode of superstitious belief in modeling the behavior of stock and derivative values should serve as a poignant lesson. The modelers had never learned the ABCs of physical economy.  

Still believe in Al Gore?  

Jan. 17, 2009 

For further reading: 

      “Is Nature Warning Us of a New Ice Age?”