Fig 1.
The highest projection or scenario of global mean sea-level rise (GMSLR) for the year 2100 for the five IPCC reports (red bars) and other key studies published after IPCC AR4 (blue bars).
Note that that these numbers are not strictly comparable, as they are based on different assumptions regarding for instance emission scenarios, characterization of uncertainty (e.g. probability of exceedence) and reference years. See S1 Table for more details and sources to the numbers.
Fig 2.
Schematic of Port of LA container ship terminal showing height (H) above mean sea-level.
Fig 3.
Simplified representation of Port of LA’s decision regarding whether or not to harden its terminal at its next upgrade and the costs resulting from its choices.
Table 1.
Parameters affecting Port of LA’s decision whether or not to harden terminals at next upgrade and the treatment of the uncertainty in those parameters.
Height and hardening cost values for a decision regarding PoLA terminals is discussed in Sections 3 and 4.
Fig 4.
Observed annual global sea-level change based on Jevrejeva et al. [78] (gray curve), Hay et al. [81] (blue curve and shading), Church and White [76] (red curve and shading), and the polynomial best fit to the Jevrejeva et al. [78] data (black curve).
Shading represents plus or minus one standard deviation.
Fig 5.
Observed annually and globally averaged sea-level anomalies from Jevrejeva et al. [78] (green circles), the polynomial model best fit to the observations (black line) and model hindcast scenarios (grey lines) that sample the unresolved variability (left panel). Bootstrap projections to 2100 highlighting potential future acceleration due to melting land ice (right panel). Projections are fitted to an idealized distribution of 2100 sea-level rise based on Pfeffer et al. [24] with an additional expansion to account for uncertainty in thermosteric sea-level rise [5,83].
Fig 6.
Black line shows the General Extreme Value (GEV) distribution fitted to the hourly sea-level anomalies from the annual mean values observed near the Port of LA [94].
The interpolated GEV distribution parameters are given in Table 1. The blue line shows the GEV distribution with an expanded scale parameter of y considered in the decision analysis described in the main text.
Fig 7.
Histogram of model results generated with 500-point Latin Hypercube sample over deeply uncertain parameters in Table 1.
Positive values indicate cases in which hardening at next upgrade passes a cost-benefit test.
Table 2.
Parameter ranges defining the Harden at Next Upgrade scenario.
The center column shows the conditions under which a decision to harden at the next upgrade would pass a cost-benefit test.
Fig 8.
Parameter estimates resulting from the model calibration to the: (top) extended scenarios of Pfeffer et al. [24], and (bottom) the CO-CAT [88] scenarios.
The former uses a beta distribution and the latter a uniform distribution, and both begin with a uniform prior, as described in the text.
Fig 9.
The red line shows estimates of the likelihood of the first condition describing the Harden at Next Upgrade scenario using the: (a) Beta distribution fit to the extended projections of Pfeffer et al. [24] and (b) the uniform distribution fit to the projections of CO-CAT [88] shown in Fig 8.
Dots represent c*/t* pairs that are jointly sampled (along with polynomial parameters a, b, and c) based on the idealized sea-level rise projections (represented as beta and unform distributions). Green and blue lines show an analogous condition for two other Port of LA facilities, Berths 206–209 and Alameda and Harry Bridges Crossing, respectively.
Fig 10.
Probabilities of a long terminal lifetime (L> 50 years) and significant increase in the daily anomaly (ψ> 533 mm) required for decision to harden terminal bottoms (H = 2804 mm) (at next upgrade to pass a cost-benefit test.
Dark and light shaded regions show probabilities required using high and low estimates, respectively, of likelihood of condition on c* and t*. The dashed lines show boundary of probability regions for decisions to harden at two other Port of LA facilities, Berths 206–209 and Alameda and Harry Bridges Crossing, respectively.
Table 3.
Comparison of RDM and Probabilistic Risk Analysis (PRA) methods.
Fig 11.
Results of full probabilistic analysis showing expected cost of hardening at next upgrade and probability of passing a cost-benefit test as a function of the terminal lifetime.
Fig 12.
Assessment of the evidence and level of agreement underlying the scientific information used in this analysis, following the characterization method of Mastrandrea et al. [108].
The size of the text reflects the importance of information to Port of LA’s decision. Italics show that the factor was considered deeply uncertain in the Robust Decision Making analysis.