- 95 - 



curve. This is accomplished by selecting one of the CO2 curves 

 as a starting point. (Tests have demonstrated that starting from 

 any one of the four curves will yield the same result.) The GISS 

 model is then run to obtain a corresponding temperature curve 

 from 1980 to 2100. This temperature curve is compared with the 

 four temperature curves originally generated by the GISS model 

 and transmsitted to the ORNL model. The new temperature curve is 

 composed by interpolating among the four previous curves, and 

 a new CO2 curve is estimated corresponding to the interpolated 

 temperature curve. (In essence, this is a two-way interpolation 

 for each 10-year interval.) The whole process is repeated until 

 two successive temperature curves closely approximate each other. 

 Usually, this takes no more than two iterations. 



To check the "accuracy" of such an iterative approach, the 

 final temperature-time curve from GISS was used as input to the 

 ORNL model for a small sample of runs. The resulting C02 - time 

 curve from ORNL was compared with the final C02~time curve in GISS. 

 In all cases, the two agreed within 2 ppm for each 10-year interval 

 Estimating the Contribution of Thermal Expansion to Sea Level Rise 



Once the final projections of expansion to sea level rise are 

 obtained, the equations in GISS relating ocean volume, temperature, 

 and pressure described previously are employed to estimate thermal 

 expansion. The increase in ocean volume is then translated into 

 global mean sea level rise. 



