National Geographic : 1990 Oct
~F~b- ;; -~ _c~ Gradually increasing CO 2 Temperature difference (°C) Greater than -2 - 2toO Oto2 ] 2to4 Greaterm than 4 To manage their data, modelers divide the earth'ssurface into boxes. Typically each box is about 300 nautical miles on a side and is sliced into layers of atmosphere and ocean. Based on average conditions within each layer, such as wind, temperature, sunlight, soil moisture, and relative humidity, the computer calculates how internalprocesses will affect the sur rounding boxes. As the calculations are modified again and again, weather pat terns emerge. Simulated seasons change just as real seasons do. PAINTINGSBYMARKSEIDLER CONSULTANTS: KIRKBRYANANDSYUKURO MANABE,GFDL;ALANROBOCK NGSCARTOGRAPHICDIVISION NOAA/GEOPHYSICAL FLUIDDYNAMICS LABORATORY (GFDL) o--- 0 . 6 OE_ 120 Plate CarrieProjection Complexity challenges climate modelers. Clouds both warm and cool the planet, and scientists continue to investigate the net effect. Oceans, with their deep currents, play a large- but poorly understood- role. They are known to warm and cool much more slowly than does land, in effect storing up change. METEOSATIMAGEPROVIDED BYEUROPEANSPACEAGENCY SRather than programming a computer simply to double CO2 levels, recent models attempt to track change over time, more accurately depicting the gradual way heat-absorbinggases accumulate. Researchersat the Geophysical Fluid Dynamics Laboratory told their supercomputer to increase CO by one percent a year from a startingdate of 1958. After 30 computer years (below) global surface temperature had risen by an average of 0. 7C. (Remember that these are computer simulations and do not give results identicalto what actually happens in the real world.) r After 50 simulated years, the temperature had risen by 1. 4 C. And after 70 years, when COz had doubled, it had risen by 2. °C (3.8°F). The biggest surprise: The waters around Ant arctica failed to warm as expected. These models reveal new patterns, clues that scientists may soon be able to check against field data to see how their predictions hold up.