Designing for Global Warming

1
Designing for Global Warming

• Orson P. Smith, PE, Ph.D.

School of Engineering
2
Evidence of global warming continues to accumulate
Combined global annual land-surface air and sea
surface temperature anomalies
3
Strongest signals are in the North
4
Projections are Scattered
Source Intergovernmental Panel on Climate
Change, 2001
5
Global Circulation Model (GCM) Predictions Vary
Average, minimum, and maximum air temperatures
predicted from 27 GCMs for Fairbanks, Alaska
(with permission from Vinson and Bae, 2002,
Probabilistic Analysis of Thaw Penetration in
Fairbanks, Alaska, ASCE Cold Regions Engineering
Conference, Anchorage)    
6
Other Trends Complicate Predictions
Figure from EPA website http//www.epa.gov/globalw
arming/
7
Climate change impacts involve spatial variables

• Permafrost changes
• Thaw subsidence, onshore and offshore
• increased flux of sediments into steams and the
  coastal ocean

8
Alaskas Permafrost Foundations are at Risk
9
Engineers' Viewsfrom Prior Meetings

• Proven responses to most warming problems exist
• Accurate knowledge of change saves money
• Synthesize existing data
• Monitor changes statewide
• Improve data transfer
• Refine predictions
• Revise codes, manuals, and design software

10
Strategies for Climate Change Design Criteria
Development

• Designers may address climate change by
• Subjective factor of safety
• Deterministic apply a trend
• Probabilistic Monte Carlo simulations
• Hybrid e.g., apply fuzzy set methods

11
Monte Carlo Simulations

• Random sampling, interpreted by assumed
  continuous distributions of independent variables
• Many repetitions results in a derived
  distribution of dependent variable

12
Apply a Trend

• Designers focus only on extremes
• Trends apply to entire data set
• Accelerated change is not resolved by
  conventional criteria development methods
• Additional information is necessary
• More sophisticated historical data analysis
• Predictive simulations (GCM results, Monte Carlo,
  )

13
Accelerated Trend
Storm-related extreme conditions may have
accelerated trends from more frequent and intense
storms due to global warming
14
Projections from history
Threshold of extremes
Consider the first half of the previous time
series as a hypothetical historical record
15
Conventional Extremal Analysis
Cumulative Probability
Return period
16
Anticipate a Linear Trend
17
Anticipate a Linear Trend
Remove the trend and identify extremes
Fit extremal distribution
18
Trend-adjusted Extrapolations
19
Anticipate an Accelerating Exponential Trend
20
Exponential Trend-adjusted Extreme Values
100-year return period value
50-year return period value
21
Summary of Proposed Analysis

1. Derive trend from complete data set
2. Remove trend from data set
3. Apply conventional statistics of extremes
4. Adjust extrapolated extremes with trend

22
Cycle Superimposed on an Exponential Trend
23
Options for Addressing Climate Cycles

• Remove cycles with low pass filter (10
  - 20 year period)
• Ignore cycles
• Decades of good data are required to define a
  regional climate cycle

24
Questions

1. What are the fundamental trends, cycles, and
   distributions of engineering parameters?
2. How do we best anticipate a trend in forecasting
   secondary variables (floods, storm surge,
   erosion, thaw depth, etc.)?
3. How do we best anticipate a trend for design
   criteria development (extremal analysis)?
4. How do we best anticipate a cycle for design
   criteria development?