Hosted by Decisioneering, Inc. July 19, 2007 Global Warming To listen to the session on your phone , follow the instructions in the “ Join Teleconference ” pop up dialog box which will appear in a few moments. To listen to the session on your computer speakers instead of your phone, follow the instructions in the “ Join Internet Phone ” pop up dialog box which will appear in a few moments. Please DO NOT join both, as this is redundant. Guest Speaker Gaetan ‘Guy’ Lion
Objective: simulate temperature increase over next century using IPCC timeframe <ul><li>Starting point: temperature 1980 – 1999 period. </li></ul><ul><li>Ending point: 2090 – 2099 period. </li></ul>
Basic model structure <ul><li>If not for IPCC timeframe intricacies, the model would be simple. </li></ul><ul><li>Record most current temperature (14.6 degree Celsius). </li></ul><ul><li>Next, simulate CO2 concentration by 2099 (i.e. 600 ppm) </li></ul><ul><li>Using regression, convert CO2 concentration into temperature level (16.3 degree Celsius). </li></ul><ul><li>Calculate temperature increase: 16.3 – 14.6 = 1.7 deg. Celsius. </li></ul>
The Regression Models: Log & Linear Until recently, climatologists debated whether the relationship between CO2 and temperature was logarithmic or linear.
The Natural Log Model Simulation Year CO2 conc . The model simulates temperature level by 2090 – 2099 two ways. One way just picks a year at random within the decade. The other way calculates the average over the decade.
Temp. Increase. Models vs 3 IPCC scenarios Among the IPCC scenarios, B1 is the low scenario, A1B is the mid level one, and A1FI is the high one. Watch carefully for the scale of the Y axes here.
Testing our regression coefficients vs IPCC scenarios The coefficients of the natural log model replicate reasonably well the IPCC best estimates up to CO2 concentration of 850 ppm. Linear model LN model
Why is volatility so much higher for IPCC Scenarios?
Another view of volatility. LN model vs B1 The natural log model and scenario B1 (IPCC) generate about the same best estimate in temperature increase (~ 1.8 degree Celsius). But, the confidence interval for the B1 scenario (green) is much wider at 1.8 degree Celsius vs only 0.68 degree Celsius for the log model (orange). B1 scenario LN model
Why volatility is higher in IPCC scenarios <ul><li>The IPCC estimates rely on numerous model sets that feed into each other. </li></ul><ul><li>Algorithms capture all gases mentioned earlier. Each gas radiative forcing is associated with uncertainty (random variable). </li></ul><ul><li>They capture many physical phenomenons such as cloud formation, precipitation, ice melting, ocean heat absorption, convection, radiation, etc… </li></ul>
Generating higher Volatility In this log model, I used the standard errors of the intercept and slope as random variables instead of the standard error of the regression. To moderate excessive volatility I used a high negative correlation (-0.95) between the two standard errors. But, resulting volatility was still way too high with a Confidence Interval that is too wide including large decrease in temperature.
Thank you for attending the Web Seminar Global Warming Gaetan “Guy” Lion E-mail: firstname.lastname@example.org July 19, 2007 Decisioneering, Inc. 1515 Arapahoe St., Ste 1311 Denver, Colorado 80202 303-534-1515 www.crystalball.com