Policy decision tools

Cutting across uncertainty

If we are trying to determine when a certain sea level rise will occur, we can use the figures in What happens? Cutting across the 90% confidence region we get (more or less) a 90% region of years before which this level of increase is quite unlikely, and after which it is highly likely. We can of course not rule out the possibility of that level occurring earlier than our lower bound or not until after our upper bound. There is after all (at least) 10% probability of an event in the tails of the distribution. Also, we do not know which scenario will play out in reality. But the following figure and table gives guidelines that can be quite useful for the decision making. They are computed relative to the average 1970-1999 sea level.

Scenario 1 (RCP 2.6)

Sea Level (Ft) Min Year Median Year Max Year
.25 2015 2031 2046
.5 2032 2048 2066
1 2061 2081 After 2100
2 After 2100
After 2100 After 2100

Scenario 2 (RCP 4.5)

Sea Level (Ft) Min Year Median Year Max Year
.25 2015 2030 2045
.5 2032 2047 2062
1 2058 2075 2095
2 2099
 After 2100
 After 2100

Scenario 3 (RCP 6.0)

Sea Level (Ft) Min Year Median Year Max Year
.25 2015 2031 2046
.5 2033 2049 2064
1 2059 2077 2094
2 2097
 After 2100
 After 2100

Scenario 4 (RCP 8.5)

Sea Level (Ft)) Min Year Median Year Max Year
.25 2015 2030 2043
.5 2032 2046 2058
1 2056 2069 2083
2 2086 After 2100  After 2100

The cost of not deciding

When thinking about how to deal with the possibility of sea level rise, there are consequences of postponing decisions, or not making them at all. While many of the policy options for dealing with sea level rise are extremely expensive upfront, the costs of doing nothing are not zero and could potentially be even greater than the cost of protections.  Not deciding assigns an implicit value for how much you are willing to pay in cleanup costs if sea level does indeed rise and causes damage to the city. When thinking about the damage costs, the city must consider the immediate cost of cleanup after flood waters have receded, as well as costs associated with public and private property and infrastructure damage and environmental harm that are more long-lasting. An increase in sea level rise would also mean an increase in frequency of flooding events, so the city could incur these costs over and over.

The following example uses estimates from the City of Olympia’s Engineered Response to Sea Level Rise Technical Report in 2011. There are estimates for the city to begin building a number of protective options that guard against a 1ft sea level rise, as seen in Policy Options(requires link), as early as sea level rises 0.25ft and 0.5ft in various locations. We chose to look at the probability that sea level rise is below 0.5ft, above 1ft, and in between for the years 2025, 2050, and 2075. Additionally, the Technical Report has conceptual cost estimates for these protective options against a 1ft rise in sea level. These estimates will be used in our example (pg.53)

Assume that a portfolio of protective options for guarding the coastline against a sea level rise of 1ft or less costs \$8,400,000 and the clean-up cost for a flooding event is \$X. The table below illustrates the value of the probabilities for varying heights of sea level rise regarding an example of the expected costs of deciding to build protective options or not. The (very simplified) assumption is that sea level rise under half a foot is not causing damage. Calculations are approximate and in current dollars. The probabilities are computed relative to the 2010 sea level in Olympia as was done in the 2011 Technical Report. In 2025 it is very likely that sea level will not rise more than half a foot, and the implied cleanup value of not building is therefore quite high. By 2050 it has become fairly likely of having a sea level rise of more than half a foot, but still not likely to go over a foot, while in 2075 we project substantial probability of sea level rise over a foot, with values less than half a foot are very unlikely.

Contact: peter (at) stat.washington.edu