An accumulation of current research hypothesizes that if global warming continues at the current pace then sea-level rise will probably surpass the most extreme projections.
Consideration of two incontrovertible observations magnifies the veracity of this proclamation.
First, the assumption that the increase in future global warming will continue at the current pace is invalid. Warming is positively accelerating over time, implying that future measurements will continue to be increasingly greater year-over-year. This refutes the assumption that global warming will maintain the current pace. Instead, current data suggest the Earth is becoming warmer and warmer, faster and faster.
Second, it is well-known and simple to show that projections made using linear models are least accurate when predicting extreme values. This is attributable to an integral characteristic of regression analysis known as “regression toward the mean.” Regression inherently underestimates the highest and/or the lowest values of the phenomenon being modeled.
Most research focuses upon relatively short-term changes in global temperatures and sea-level, identifying issues requiring extant solution. Providing perspective, some research considers long-term cycles in temperature and sea-level which occur once every hundred millennia.
In the following article Figure 2 illustrates how sea-level projections based on linear regression models fall beneath actual conditions, and also fall beneath experts’ expectations.
Likewise, underestimation of sea-level rise is attributable to inaccurate linear regression-based models of Earth’s ice sheet melting rate.
In general the more extreme the actual magnitude of change in the phenomenon being measured, the less accurate the regression-based estimate of the magnitude of change. As the rate of change accelerates, future projections become increasingly inaccurate.
Discussed by John Englander, Ph.D. in the seven-minute video linked below, over the past 2.5-million years Earth has undergone an ice age cycle every hundred-thousand years. During each cycle the mean sea level has varied by four-hundred feet. Ocean levels have been stable over the past six-thousand years (most of recorded human history), so humans alive today are pioneers.
In the short-term there is no doubt that the level of the world’s oceans is rising. When asked what can be done, Dr. Englander suggests that mankind must limit the use of fossil fuels; prepare for more extreme flooding and for more types of flooding; and adapt to the reality of rapidly increasing ocean levels over the rest of this century.
In view of the fact that most empirical research in this area used regression-based statistical analyses that regress toward the mean and underestimate outlying data, I hypothesize that changes which are significantly greater than anticipated are coming, sooner than is expected.
- Research using linear analysis predicts 30% (likely an underestimate) of the glaciers in the Hindu Kush Himalayan region—holding the most ice outside of the Poles and providing fresh water for agriculture supporting two billion people in eight countries, are in pearl (https://www.npr.org/sections/goatsandsoda/2019/02/05/691578203/report-global-warming-could-melt-at-least-a-third-of-himalayan-glaciers).
- Regression toward the mean (RTTM) occurs in “static designs” involving measurements recorded at only one point in time: (https://odajournal.com/2013/09/20/maximizing-the-accuracy-of-multiple-regression-models-using-unioda-regression-away-from-the-mean/).
- RTTM occurs in multiple-subject designs in which measurements are recorded at multiple time points, and it can sometimes be circumvented by the use of a specific data transformation (https://odajournal.com/2013/10/23/ipsative-standardization-is-essential-in-the-analysis-of-serial-data/).
- RTTM occurs in dose-response studies (https://odajournal.com/2016/05/31/using-machine-learning-to-model-dose-response-relationships-via-oda-eliminating-response-variable-baseline-variation-by-ipsative-standardization/).
- RTTM occurs in single-subject “N-of-1” studies involving multiple recordings over time (https://odajournal.com/2013/11/07/ascertaining-an-individual-patients-symptom-dominance-hierarchy-analysis-of-raw-longitudinal-data-induces-simpsons-paradox/).
- Maximum-accuracy statistical methods accurately prospectively predict temperature and precipitation anomalies (https://odajournal.com/2013/09/19/the-use-of-unconfounded-climatic-data-improves-atmospheric-prediction/).
- Novometric analysis is the maximum-accuracy alternative to regression analysis (https://odajournal.com/2016/09/19/novometric-analysis-with-ordered-class-variables-the-optimal-alternative-to-linear-regression-analysis/).
- Novometrics is the maximum-alternative to all legacy statistical methods (https://odajournal.com/2020/06/28/what-is-novometric-data-analysis/).
Have a nice day!
Modeling Global Warming and Sea-Level Rise
Paul R. Yarnold, Ph.D.
February 24, 2021