Climate Resilience and Benefit–Cost Analysis: A Handbook for Airports (2019)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

12 While evaluating climate resilience may be a new challenge, it can be incorporated as part of the overall risk management processes that most airports already have. Appendix A describes a generic management structure for assembling a team to analyze the potential impacts and responses to climate change and discusses specific risk management activities that such a team might undertake. In addition to using existing management processes and structures, airports can also take advantage of existing resources that are directly relevant to the issue of assessing climate change. C H A P T E R 2 Evaluation Methods Under Risk and Uncertainty About This Chapter Chapter 2 briefly describes the differences between benefit–cost analysis and financial feasibility studies and how climate risk enters into these formal types of analysis to support decision making. Among the important elements discussed are: • How to conduct an initial screening analysis for climate risk; • Defining Monte Carlo analysis and why it is important for dealing with analysis of climate risk; • Defining “value at risk” as a means for supporting decision making by helping to identify levels of risk that an organization is not willing to tolerate. Many of the technical details required to undertake a full risk-adjusted analysis are in the following appendices: • Appendix C provides a detailed description of the technical analyses and methods. • Appendix D describes the available climate projections and how they can be accessed and interpreted. • Appendix E describes two Microsoft Excel templates that have been developed in conjunction with this handbook: one can be used to assess potential extreme water events due to expected sea level rise near coastal airports in the United States, and the other analyzes the incidence of increased high temperatures and their effects on weight restrictions for aircraft takeoffs. Both use the methods and analytical approach discussed in the handbook. • Appendix F provides two numerical examples using the Excel templates. About the Next Chapter Chapter 3 describes climate change projections and where to obtain them.

Evaluation Methods Under Risk and Uncertainty 13 These are discussed in the following, before getting into the details of how to actually conduct climate-related risk analyses. It is also important to note that while the present focus is on using the frameworks of FFA and BCA, there are of course other contexts in which climate risks could be assessed and evaluated; some of these are discussed in Appendix B. A primary goal of this handbook is to demonstrate how components of climate risk and uncertainty can be incorporated into financial feasibility and benefit–cost analyses. Indeed, the prevalence and character of risk and uncertainty in climate forecasts have specific implications for how to address them analytically. 2.1 Benefit–Cost Analysis and Financial Feasibility Analysis At the outset, it is important to distinguish between a BCA and an FFA. In the context of this handbook, the primary difference between these approaches can be thought of as follows: • BCA: Would society (including all aviation stakeholders) be better off undertaking a proposed project? • FFA: Are the returns from a project adequate for the airport and its users to justify under- taking it? Is there a viable plan to pay for it? A BCA focuses on whether a proposed project should be undertaken after taking into consider- ation all relevant benefits and costs to all aviation stakeholders. Such benefits and costs may include items that affect the airport itself, but other entities may be affected as well, including airport users (passengers, airlines, etc.) and the surrounding community. The benefits and costs should be measured in constant (inflation-adjusted) dollars, and those occurring in future years must be discounted using an appropriate discount rate. The results from a BCA are usually presented either in terms of net present value (NPV)—measured as discounted benefits minus discounted costs—or as a benefit–cost ratio—measured as discounted benefits divided by discounted costs. An FFA focuses on whether a project can be paid for using available sources of funds. It also compares benefits and costs, but it does this by considering only those cash benefits and costs accruing to the airport itself or its users; these may be different from the benefits and costs affect- ing other aviation stakeholders. These distinctions are discussed in more detail in Sections 2.4 and 6.2. It is important to note that these two approaches answer different questions but are not mutually exclusive, so decision makers could elect to undertake both types of analysis. 2.2 Using Existing ACRP Resources ACRP has published work on climate change and its effects on airports. In particular, ACRP Report 147: Climate Change Adaptation Planning: Risk Assessment for Airports (Dewberry et al. 2015) provides information to help airport practitioners understand the specific impacts climate change may have on their airport, develop adaptation actions, and incorporate those actions into the airport’s planning processes. ACRP Report 147 discusses the climate change risks to practitioners’ airports and then con- siders a variety of mitigation scenarios and examples. Accompanying the report is an elec- tronic assessment tool called Airport Climate Risk Operational Screening (ACROS) that can help airports answer the question, “Within the entire airport, what’s most at risk from pro- jected climate change?” The ACROS tool uses a formula to compute an estimated level of risk for assets and operations at the airport. In addition, the tool uses airport-specific climate data for 489 U.S. airports to rank specific risks in order to provide an enterprise-level estimate of the relative risk posed by each asset and operation.

14 Climate Resilience and Benefit–Cost Analysis: A Handbook for Airports ACRP Report 147 makes a case for climate change adaptation and suggests establishing a stakeholder advisory committee to assist with setting resilience goals and identifying and pri- oritizing risks. It provides a primer on climate change and uncertainty for airports, provides national climate change projections, and suggests developing adaptations based on the vulner- ability of critical assets and refining risk assessments as new, higher-resolution data become available on a 3- to 5-year review cycle. ACRP Report 160: Addressing Significant Weather Impacts on Airports (ICF International 2016) contains another useful tool. The Airport Weather Advanced Readiness (AWARE) toolkit is designed to raise airport operator awareness about vulnerabilities caused by significant weather events and to help airports develop more robust contingency and recovery plans. The Excel- based AWARE toolkit focuses on events that are “rare but plausible”; that is, events that may have happened in the distant past or in adjacent geographic areas, but are not common event types at the airport itself. The AWARE toolkit draws on historical weather data relevant to the airport’s specific location in order to identify significant weather event types that the airport operators may wish to prepare for. AWARE also contains seven readiness modules that allow users to review best practices for preparing for these different weather events, assess their readiness for the events, and generate customized checklists for preparing for and recovering from them. The seven readiness modules are administration and finance, planning and environment, airfield operations, terminal operations, ground transportation and parking, safety and security, and a consolidated streamlined version of the full toolkit for small airports. The toolkit also contains an impacts tracking module, which is designed to help airports track the costs and other impacts of weather events (e.g., flight delays) over time. 2.3 Adding Climate Risk to an Analysis It is useful to briefly discuss the distinction between risk and uncertainty. While there is no universal agreement on the distinction, a common approach is to define them in terms of whether a probability of occurrence can be estimated. With risk, the specific outcome that may occur is unknown, but one either knows or can reasonably estimate what the outcome distribution looks like. On the other hand, uncertainty implies that there is little or no knowledge of the outcome distribution itself. For example, a game of chance like roulette involves risk, but one can calculate what the specific odds are for any given outcome. On the other hand, the probability of, say, a terrorist event occurring may be completely unknown, which implies uncertainty. But for present purposes, there is no real advantage to explicitly labelling something as a risk as opposed to an uncertainty. Regardless of what they are called, the goal here is to provide an overview of how to incorporate estimates of risks or uncertainties into a formal analysis. In practice, the primary risks and uncertainties associated with climate change will typically affect the benefits side of a benefit–cost or financial feasibility analysis. For example, when one is considering a particular infrastructure investment—say, lengthening a runway to allow higher- weight takeoffs during extreme heat—the costs of extending the runway a certain number of feet may be well understood. However, the benefits of the project may be subject to significant uncertainty because even the best climate science projections of future temperature rise cannot predict exactly when or how often such extreme heat events will occur. There is also additional uncertainty regarding actions that might be taken worldwide to reduce generation of GHGs, potentially slowing down and possibly reducing the effects of climate change. Once again, prior publications can be reviewed to help understand how climate risk and uncer- tainty can affect these types of analyses. ACRP Report 76: Addressing Uncertainty about Future Airport Activity Levels in Airport Decision Making provides a broad treatment of the factors leading

Evaluation Methods Under Risk and Uncertainty 15 to risk and uncertainty in forecasting airport activity levels (Kincaid et al. 2012). While the focus of this report is not on airport activity levels, some of the same proposed methodological approaches can be considered. Some of the material presented in the following is derived from this report. There have been tremendous advances in quantitative climate change modeling over the past 30 years. While it is not the purpose of this handbook to delve into these models, the results that can be obtained from such models will be discussed in some detail, along with how such results can be incorporated quantitatively into a BCA or FFA. Specifically, projections of surface tem- perature changes, sea level rise, and increased likelihood of flooding events can be used directly to help quantify the expected benefits of proposed airport infrastructure investments. There are four key elements to adding climate risk to a BCA or FFA: • Accessing climate projections: What types of climate projections are available, and how can their uncertainty be assessed? • Vulnerability: How likely is it that some or all airport operations will be disrupted by events due to climate change? • Criticality: If the disruptions occur, how expensive would they be? • Adaptations: What can be done (changes in operation or infrastructure), how much will vulnerability and criticality be reduced, and how expensive are the adaptations? Chapter 3 deals with climate data projections and how to access and analyze them. Chapter 4 addresses the potential impacts on airports, including vulnerability and criticality. Chapter 5 considers airport adaptation strategies, including those not involving infrastructure investments, as well as practical financial constraints that an airport may face. The remainder of this chapter focuses on the suggested methodology for dealing with airport climate risk. Two steps are suggested to evaluate climate risk at an airport, as shown in Exhibit 2-1. Exhibit 2-1. Suggested two-step method for evaluating airport climate risk.

16 Climate Resilience and Benefit–Cost Analysis: A Handbook for Airports The first step uses the ACROS tool as a screening device to rule out climate stressors unlikely to affect an airport, at least through 2060 (Dewberry et al. 2015; the ACROS tool can be down- loaded at http://www.trb.org/publications/blurbs/173554.aspx). The second step involves more formal climate risk analysis, which is the main subject of this handbook. Both steps are described in the sections that follow. 2.4 Step 1: Initial Screening Analysis for Assessing Climate Risk In the remainder of this handbook, information on how to deal with climate risk and uncer- tainty will be presented. It is important to understand that the quantitative methods are built on the same fundamental framework as a conventional BCA or FFA, which describes specific benefits and costs of a proposed project, projects these out over its useful life, and discounts the monetary impacts to estimate a net present value. However, the basic framework is then extended and modified to treat costs or benefits in a probabilistic way that reflects the uncertainty inherent in future climate change events that may affect an airport. The methods described are straightforward, but in practical terms they may require a fair amount of resources and expertise. This section provides guidance on using ACROS to screen for climate stressors that may affect a specific airport. Output from this model can also be used to assess uncertain future climate change events within the context of a conventional BCA or FFA that does not rely on the formal probabilistic methods described in this handbook. A Prior Example A relevant example of how to use the ACROS model is presented in ACRP Synthesis 13: Effective Practices for Preparing Airport Improvement Program Benefit–Cost Analysis, which cites a BCA pre- pared for Houma-Terrebonne Airport in Louisiana (Landau and Weisbrod 2009). While it pertains to oil spills, the method used can also be applied to climate change events. The project involved strengthening a runway, sections of several taxiways, and the apron in order to handle several new heavy aircraft that would be used by a company specializing in oil-spill mitigation along the Gulf coast. The entire analysis comprised environmental benefits measured as the avoidance of damage resulting from untreated oil spills. The BCA evaluated a range of specific scenarios with different numbers and magnitudes of likely oil-spill events (ranging from two to seven per year and occur- ring at different distances offshore), and the individual benefit–cost ratios ranged from 0.2 to 4.0. To support this approach, the analysis cited documentation justifying the specific scenarios selected for analysis, which included an outside oil-spill risk analysis, plus historical data show- ing the incidence and location of spills over the prior 2 years. Each scenario was evaluated as a conventional BCA with known benefits and costs. Airports could use this same approach to perform an initial screening analysis of the potential effects of climate change. Under this approach, a few different discrete scenarios would be selected for comparison using a conventional BCA or FFA with specific assumed benefits and costs. How to Identify Climate Stressors Airports may be particularly susceptible to climate stressors such as high temperatures above some predefined threshold or increased likelihood of storms and flooding events due to sea level rise. The immediate question, then, becomes one of identifying and justifying specific climate scenarios involving these stressors. In lieu of undertaking the significant effort that may

Evaluation Methods Under Risk and Uncertainty 17 be required to access the latest climate data (as described in Chapter 3), a reasonable first step would be to use the ACROS software tool. ACROS includes projections for different climate stressors at most commercial service airports in the United States. (In ACROS, the stressors are referred to as “vectors.”) As an example, the ACROS forecast for Pensacola Gulf Coast Regional Airport (PNS) is shown in Exhibit 2-2.3 Notice that for each stressor/vector, ACROS provides a baseline (the number of days per year where the stressor occurs as of 2013), plus projections for 2030 and 2060. The projection years report median, 25th percentile (low), and 75th percentile (high) estimates. The estimates are based on a limited number of climate forecasting models and employ projections that have since been updated. Nevertheless, they may provide three reasonable scenarios (low, median, and high) for an airport’s operators to assess whether the airport may be significantly affected by any of these stressors. This may be enough information for an analyst to determine if more detailed quantitative analysis of a climate risk is warranted. Using ACROS as a Screening Tool ACROS is particularly useful as a screening tool. Potential threats for PNS are easy to identify from Exhibit 2-2. The main threat appears to be high temperatures, where aircraft might have to take payload penalties for longer-haul flights. Source: ACROS from ACRP Report 147 (Dewberry et al. 2015). Exhibit 2-2. ACROS projections for PNS.

18 Climate Resilience and Benefit–Cost Analysis: A Handbook for Airports If, in the worst case, the climate stressor would not place a large burden on the airport or its users, then the airport could reasonably conclude that no further analysis needed to be under- taken. For example, ACROS shows that for PNS, up to 19.8 very hot days could occur in 2060. In ACROS, a “very hot day” is defined as temperatures exceeding 100°F. If an analyst knows that PNS users could easily adapt to 100°F temperatures occurring as much as 20 times annually by 2060, then no further action would be needed. Alternatively, consider a coastal airport. Suppose ACROS showed that in the worst case there were no indications of sea level rise that would threaten the airport by 2060; no further effort might then be required. If there is some uncertainty about the size of the impact or whether a possible adaptation would pay off in the worst-case scenario, an analyst could undertake a conventional benefit–cost or financial feasibility analysis using the ACROS climate data. An example of such an analysis is provided in the following. Sample Climate Resilience Analysis Using ACROS Based on the data shown previously, PNS might decide to analyze whether a certain infra- structure project should be undertaken to mitigate the potential impacts from the projected increase in the number of very hot days (temperatures above 100°F). Up to three specific scenarios could be analyzed: • Median: assume that very hot days increase from 0.3 in 2013 to 4.9 in 2030 and 12.4 in 2060. • Low: assume that very hot days increase to 2.0 in 2030 and 5.3 in 2060. • High: assume that very hot days increase to 8.1 in 2030 and 19.8 in 2060. Using this framework, a conventional BCA or FFA—assuming that these increases occur with certainty—could be carried out for one or more of these scenarios.4 One strategy for performing a screening analysis is to test the proposed project using the most extreme forecast values for climate stressors in the ACROS model. If a project does not pay off at the extreme forecast, the analyst can have some confidence recommending that the project be rejected for the time being and suggest revisiting it during the next planning cycle when more information is available. So, for example, the project might be tabled until the next master planning cycle. Steps in Creating a BCA (or FFA) The following presents a sample BCA for PNS of a 1,500-ft runway extension project to miti- gate the delays due to very hot days. This is only an illustration and is not meant to represent actual opportunities or risks for PNS. • Identify the objective: The objective of the runway extension would be to avoid current and projected commercial departure delays on days when temperatures exceed 100°F. • Define a base case: The base case is that the airport does not undertake the runway exten- sion project. In this simplified example, it is assumed that afternoon departures between 13:00 and 17:59 would have to be cancelled, thereby imposing delay costs on passengers (as measured by the value of their time) and foregone costs on airlines (as measured by crew costs and aircraft depreciation). • Define a scenario case: Under the scenario case where the runway extension project is under- taken, the analyst would have to project estimated reductions in delays (benefits), which would depend on the number of flights that could avoid delays due to the project; this would be balanced against the costs to invest in, operate, maintain, and rehabilitate the runway exten- sion.5 To illustrate the process, a fixed growth rate in airport operations over time is assumed, along with a simple assumption that the project would reduce the incidence of cancelled flights (and the corresponding cancellation costs borne by passengers and airlines) by 60%.

Evaluation Methods Under Risk and Uncertainty 19 • Identify analysis period: It is assumed that the runway construction could be completed in a single year (2019) and then would be available for use starting in 2020. It is also assumed that the analysis itself was undertaken in 2017 even though the effects from the mitigation project are not felt until 2020. The effects then run out through 2060—the 40-year life of the runway. • Apply decision criteria: – Benefit–cost ratio ê 1: If the net project benefits (measured as the discounted present value of the reduction in flight cancellations because of the runway extension) exceed the costs (the discounted present value of constructing, maintaining, and operating the runway extension), then the project has merit and the analyst would recommend further analysis of climate risk, along the lines described in Chapter 4 and beyond. – Benefit–cost ratio < 1: If the costs exceed benefits, the analyst can conclude that the project does not currently have merit because the analysis has assumed the maximum number of very hot days in the ACROS model.6 Key parameter assumptions for this example are shown in Exhibit 2-3. All dollar figures are measured in constant (inflation-adjusted) dollars. The BCA is shown in detail in Exhibit 2-4. The calculations for each relevant column are as follows: 1. Annual passenger delay costs shown in Column E are computed as the number of very hot days (B)  daily passengers (D)  average hours of delay per passenger  cost of delay per hour. 2. Airline costs, shown in Column F, are equal to the number of very hot days (B)  daily flights (C)  average block hours per flight  (crew cost per hour + depreciation per hour).7 3. Total base case delay costs shown in Column G are the sum of passenger plus airline costs. These are the costs that would be incurred under the base case, where the runway project is not undertaken. 4. As mentioned previously, for this example it is assumed that the runway project reduces the incidence of cancelled flights by 60% starting in 2020. Thus, the scenario case delay costs shown in Column H are 60% below the base-case costs in Column G. Column I simply subtracts Column H from G to represent the delay benefits from undertaking the mitigation project. Note: FAA Economic Values document is FAA 2016b. Assumptions Source Construction cost for 1,500-ft runway extension $6,562,500 Assumes $175/sq yard for 150-ft wide runway + 75-ft wide taxiway 20-year rehabilitation cost % construction cost 50% Assumed value Annual O&M expense % construction cost 3% Assumed value Affected Flights: Avg daily flights 1300-1759 in 2017 10.3 Avg block hrs per flight 1.8 Avg seatsize 91 PNS annual departure growth rate 1.1% FAA TAF Forecast 2016, ITN_AC + ITN_AT ops annual growth rate at PNS, 2017-2045 Passenger Impacts: Avg load factor 84.6% FAA T-100 Domestic Segment report -- PNS load factor for May-Sep 2016 Avg daily pax per flt 77.0 = PNS Avg seatsize * Avg load factor Avg hrs of delay per passenger 3.0 Assumed value Passenger delay cost per hr $44.30 FAA Economic Values, Table 1-1, All Purpose Intercity Air and High Speed Rail Airline Impacts: Crew cost per block hr $349 Aircraft depreciation per block hr $144 FAA Economic Values, Table 4-6, RJ more than 60 Seats Official Airline Guide (OAG) -- PNS departures for May-Sep 2017, 1300-1759 hrs Exhibit 2-3. Parameters used in conventional sample BCA for PNS.

20 Climate Resilience and Benefit–Cost Analysis: A Handbook for Airports Note: PAX = passengers. A B C D E F G H I J K L PNS MAX Very Hot Days Interpolated MAX Very Hot Days Daily Flts Affected per VHot Day Daily Pax Affected per VHot Day Annual Passenger Delay Costs Annual Aircraft Crew plus Depreciation Costs Base Case: Total VHot Day Delay Costs Scenario: Total VHot Day Delay Costs Benefits (Reduction in Delay Costs) Runway Extension Investment Annual O&M Total Annual Costs 2013 0.3 0.3 2014 0.8 2015 1.2 2016 1.7 2017 2.1 10.3 793 $225,025 $19,517 $244,543 $244,543 $0 $0 $0 $0 2018 2.6 10.4 802 $276,326 $23,966 $300,292 $300,292 $0 $0 $0 $0 2019 3.1 10.5 810 $328,706 $28,510 $357,216 $357,216 $0 $6,562,500 $0 $6,562,500 2020 3.5 10.6 819 $382,185 $33,148 $415,333 $166,133 $249,200 $196,875 $196,875 2021 4.0 10.8 828 $436,778 $37,883 $474,661 $189,864 $284,796 $196,875 $196,875 2022 4.4 10.9 837 $492,504 $42,716 $535,220 $214,088 $321,132 $196,875 $196,875 2023 4.9 11.0 846 $549,381 $47,649 $597,030 $238,812 $358,218 $196,875 $196,875 2024 5.3 11.1 855 $607,428 $52,684 $660,111 $264,045 $396,067 $196,875 $196,875 2025 5.8 11.2 864 $666,662 $57,821 $724,483 $289,793 $434,690 $196,875 $196,875 2026 6.3 11.3 873 $727,103 $63,064 $790,167 $316,067 $474,100 $196,875 $196,875 2027 6.7 11.5 883 $788,771 $68,412 $857,183 $342,873 $514,310 $196,875 $196,875 2028 7.2 11.6 892 $851,683 $73,869 $925,552 $370,221 $555,331 $196,875 $196,875 2029 7.6 11.7 902 $915,861 $79,435 $995,296 $398,118 $597,178 $196,875 $196,875 2030 8.1 8.1 11.8 912 $981,324 $85,113 $1,066,436 $426,575 $639,862 $196,875 $196,875 2031 8.5 12.0 921 $1,039,664 $90,173 $1,129,837 $451,935 $677,902 $196,875 $196,875 2032 8.9 12.1 931 $1,099,148 $95,332 $1,194,480 $477,792 $716,688 $196,875 $196,875 2033 9.3 12.2 941 $1,159,794 $100,592 $1,260,386 $504,154 $756,232 $196,875 $196,875 2034 9.7 12.4 952 $1,221,620 $105,954 $1,327,574 $531,030 $796,545 $196,875 $196,875 2035 10.1 12.5 962 $1,284,645 $111,421 $1,396,065 $558,426 $837,639 $196,875 $196,875 2036 10.4 12.6 972 $1,348,887 $116,993 $1,465,879 $586,352 $879,527 $196,875 $196,875 2037 10.8 12.8 983 $1,414,364 $122,672 $1,537,036 $614,814 $922,222 $196,875 $196,875 2038 11.2 12.9 993 $1,481,098 $128,460 $1,609,557 $643,823 $965,734 $196,875 $196,875 2039 11.6 13.0 1004 $1,549,105 $134,358 $1,683,464 $673,385 $1,010,078 $196,875 $196,875 2040 12.0 13.2 1015 $1,618,408 $140,369 $1,758,777 $703,511 $1,055,266 $3,281,250 $196,875 $3,478,125 2041 12.4 13.3 1026 $1,689,025 $146,494 $1,835,518 $734,207 $1,101,311 $196,875 $196,875 2042 12.8 13.5 1037 $1,760,976 $152,734 $1,913,710 $765,484 $1,148,226 $196,875 $196,875 2043 13.2 13.6 1048 $1,834,283 $159,092 $1,993,375 $797,350 $1,196,025 $196,875 $196,875 2044 13.6 13.8 1059 $1,908,966 $165,570 $2,074,536 $829,814 $1,244,721 $196,875 $196,875 2045 14.0 13.9 1071 $1,985,046 $172,168 $2,157,215 $862,886 $1,294,329 $196,875 $196,875 2046 14.3 14.1 1082 $2,062,545 $178,890 $2,241,436 $896,574 $1,344,861 $196,875 $196,875 2047 14.7 14.2 1094 $2,141,485 $185,737 $2,327,222 $930,889 $1,396,333 $196,875 $196,875 2048 15.1 14.4 1106 $2,221,888 $192,710 $2,414,598 $965,839 $1,448,759 $196,875 $196,875 2049 15.5 14.5 1118 $2,303,775 $199,813 $2,503,587 $1,001,435 $1,502,152 $196,875 $196,875 2050 15.9 14.7 1130 $2,387,170 $207,046 $2,594,215 $1,037,686 $1,556,529 $196,875 $196,875 2051 16.3 14.8 1142 $2,472,095 $214,411 $2,686,507 $1,074,603 $1,611,904 $196,875 $196,875 2052 16.7 15.0 1154 $2,558,575 $221,912 $2,780,487 $1,112,195 $1,668,292 $196,875 $196,875 2053 17.1 15.2 1167 $2,646,632 $229,549 $2,876,181 $1,150,472 $1,725,709 $196,875 $196,875 2054 17.5 15.3 1179 $2,736,290 $237,326 $2,973,616 $1,189,446 $1,784,170 $196,875 $196,875 2055 17.9 15.5 1192 $2,827,575 $245,243 $3,072,818 $1,229,127 $1,843,691 $196,875 $196,875 2056 18.2 15.6 1205 $2,920,510 $253,304 $3,173,813 $1,269,525 $1,904,288 $196,875 $196,875 2057 18.6 15.8 1218 $3,015,120 $261,509 $3,276,630 $1,310,652 $1,965,978 $196,875 $196,875 2058 19.0 16.0 1231 $3,111,431 $269,863 $3,381,294 $1,352,518 $2,028,777 $196,875 $196,875 2059 19.4 16.2 1244 $3,209,469 $278,366 $3,487,835 $1,395,134 $2,092,701 $196,875 $196,875 2060 19.8 19.8 16.3 1258 $3,309,259 $287,021 $3,596,280 $1,438,512 $2,157,768 $196,875 $196,875 Present Value @7%: $13,951,697 $6,050,134 $7,901,562 $8,156,393 NPV -$254,831 B/C Ratio 0.97 Base Case Mitigation Scenario Exhibit 2-4. Sample BCA results using ACROS projections for PNS.

Evaluation Methods Under Risk and Uncertainty 21 5. Construction and annual operation and maintenance (O&M) costs are shown in Columns J and K, respectively, using the assumptions shown in Exhibit 2-3. 6. Total project costs shown in Column L are the sum of Columns J and K. The relevant discounted present value benefits and costs are shown at the bottom of Columns G, H, I, and L. The overall NPV of the project is simply the difference in present values between Column I (benefits) and Column L (costs). The benefit–cost ratio is the ratio of these two values. Applying the Decision Criteria What does this simplified BCA show? The standard decision criterion for a conventional BCA is that if the discounted present value of total benefits exceeds the discounted present value of total costs, then the project has merit. In this case, as shown at the bottom right of Exhibit 2-4, the costs exceed benefits (or equivalently, the benefit–cost ratio is less than 1), and the conclu- sion is that the proposed runway extension does not make economic sense. This conclusion is valuable because it is based on the maximum ACROS forecast for very hot days. If all other factors in the analysis are the same, the median or low ACROS forecast for very hot days would show an even lower level of net benefits. Therefore, if the ACROS model forecast is reasonable, one can conclude the project does not make sense now and that it can be reconsidered at a later date, when more or better information on climate change or costs may be available.8 It is important to note that if the NPV for the runway project were found to be positive using the maximum ACROS forecast for very hot days, forming a conclusion about its merits would be more difficult. One option would be for the analyst to repeat the process using the median and low ACROS forecasts for very hot days. If the project were found to have merit under all three forecasts, then the analyst would have more confidence concluding that the project had merit. On the other hand, if the project failed assuming the median or low forecast but showed a posi- tive result for the high forecast, perhaps a more formal risk analysis (described in the following sections) would be warranted. This discussion represents a simplified screening approach to analyzing the potential impacts of climate change. In practice, airports would also need to adequately assess vulnerability (the likelihood of adverse events), criticality (the costs of the adverse events), and adaptation pos- sibilities in order to identify a relevant investment project before actually undertaking a BCA or FFA. Chapters 4 and 5 provide a detailed discussion of these topics. In sum, one can use the ACROS model to develop standard benefit–cost or financial feasibility analyses of climate mitigation projects. If the projects fail the economic tests at the high ACROS forecast level, one could reasonably conclude that they do not (currently) have merit. However, if the analysis shows a positive result, more formal risk analysis may be warranted. Differences in Undertaking an FFA In an FFA, an airport is interested in determining if a proposed project (like the runway exten- sion at PNS) would produce returns (or benefits) to the airport and its users that exceed the costs that the airport and its users pay. Compared to the previous example, the main distinctions between a BCA and an FFA would be: • Benefits: Airports and their direct users (aircraft operators) would be interested in the cash costs of delays and cancellations in the base case and how much they will be reduced in the scenario case. The impacts on passengers would typically not be considered in an FFA. • Costs: Airports would count only the net out-of-pocket costs they or their users would incur to pay for the project; for example, if the airport received an Airport Improvement Program

22 Climate Resilience and Benefit–Cost Analysis: A Handbook for Airports (AIP) grant that paid for 90% of the runway extension, then in the FFA, the airport would count only 10% of the investment costs. • Discount rate: The airport would use its actual cost of capital (e.g., the interest rate on a recent bond issue). The remainder of the analysis, including applying the decision criteria, would be identical. Using ACROS for Sea Level Rise Projections For airports near coastal areas, the same sort of approach as described previously could be carried out for analysis of future sea level rise using ACROS projections. In this situation, the two key ACROS climate stressors to look at are labelled: • Sea level rise: This refers to the number of days per year where the runway elevation is projected to be inundated by tidal flooding. • Sea level rise base flood elevation (BFE): This refers to the height to which floodwater is anticipated to rise during a 100-year flood event, measured in feet relative to the North Ameri- can Vertical Datum of 1988 (NAVD88). (See Appendix D for an explanation of the use of a vertical datum in sea level climate analysis.) Airport operators could identify the critical elevations for each piece of infrastructure and determine if the airport would be exposed in the worst-case sea level rise scenario in ACROS. If there are a number of important facilities exposed, it might then pay to evaluate the ben- efits and costs of undertaking a mitigation project. The exact climate assumptions and models employed to generate the ACROS projections are not based on the same sea level rise climate data described in Appendix D. Thus, one must be cautious if trying to compare the ACROS projections of sea level rise to the projections described in this handbook. 2.5 Step 2: Risk-Adjusted Analysis for Assessing Climate Risk The conventional analysis described previously would likely require fewer resources and less effort from the airport than would a risk-adjusted effort, but it would be unlikely to reflect the full range of potential risks faced by the airport. If, in the worst ACROS case, a climate stressor could impose high net costs on the airport or its users, more analysis beyond ACROS could be warranted. Or it could be the case that the specific climate stressors used in ACROS were not completely relevant for a specific airport—for example, ACROS’ definition of 100°F for a “very hot day” may not be relevant for an airport with an 11,000-ft runway that can easily handle takeoffs of large aircraft on long-haul routes. However, it may be that temperatures above, say, 110°F would in fact start to necessitate weight restrictions on certain flights. In these cases, it may be reasonable for the analyst to proceed to Step 2 to obtain more detailed climate data. Aside from the specifics of climate stressor definitions, another important piece of informa- tion not included in ACROS is the likelihood of different future outcomes. What is the distribu- tion of future high temperatures? How likely is it that temperatures will exceed 100°F, 110°F, or 115°F in 2030, 2060, or other years? Or take the case of a coastal airport, where certain pieces of infrastructure would be at risk if flooding exceeded 5 ft while other infrastructure would be exposed at 7 ft. The analyst would want to know the likelihood of these events, recognizing that the probability might increase over time due to sea level rise. Developing estimates of these kinds of risks is central to good decision making, but ACROS was not designed to provide this kind of information. The primary limitation of using a conventional

Evaluation Methods Under Risk and Uncertainty 23 BCA or FFA is that the approach does not directly consider the uncertainty of climate change and the risk it imposes on the airport. A more thorough and robust analysis is available through the use of Monte Carlo simulation. This approach involves: • Defining or assessing probability distributions for one or more variables of interest, such as extreme water levels or high temperatures; • Using simulation techniques to make a large number of random draws from the distributions to cover the likely range of outcomes; • Evaluating each draw to obtain a value for the variable of interest (water rise or high tempera- ture); and • Combining all of the draws to obtain estimates of the expected or most likely values. Exhibit 2-5 further describes Monte Carlo simulation. The basic procedure used in Monte Carlo simulation is to draw a random number (by con- vention between 0 and 1) for each relevant time period of an analysis, with the value determining whether an uncertain climate event occurs in that time period corresponding to its probability.9 Consider again the previous example for PNS. In the benefit–cost analysis, it was assumed that the annual counts of very hot days (above 100°F) followed the ACROS projections and increased deterministically from 0.3 in 2013 to 8.1 in 2030 and then to 19.8 in 2060. This could be transformed into a Monte Carlo analysis by repeating the analysis multiple times but using different counts of the very hot days for each repetition in order to reflect the uncertainty inher- ent in these projections. Chapter 3 and Appendix D contain detailed information about obtaining future climate pro- jections for the incidence of high temperatures. Suppose one obtained such projections from, say, 10 different climate models, each with its own projection of daily high temperatures occur- ring in each year from 2020 through 2090. For reasons discussed in Appendix D, the number Monte Carlo simulation (or the Monte Carlo method) is a computerized simulation technique that makes use of randomization and probability statistics to investigate problems involving uncertainty. Typically, it involves a computer model of a system or project (e.g., air traffic at an airport). The inputs to the model, instead of being fixed numbers or variables, are specified as probability distributions. For example, rather than traffic growth being set at X% per annum, it may be defined as having a normal (bell-curve) distribution with a mean of X% and a standard deviation of 1.0%. Using computer software, the model is run multiple times, each time randomly sampling from the input distributions, resulting in different outcomes each time. Often, the model will be run thousands or tens of thousands of times (known as iterations), and the results will be collected from each run. With enough iterations of the model, the output can demonstrate the range of possible outcomes and provide statistical estimates of the probabilities of various outcomes. Depending on the complexity of the model and input distributions assumed, the range of outcomes can be large and not always linear. Expected or most likely values can also be generated. Monte Carlo can be seen as a powerful what -if or scenario-generating exercise where every possible what-if or scenario is generated (within the confines of the model specification), including interactions between the various input factors. Another way of looking at it is that each iteration of the model represents one possible future for the system being modeled. By running the model thousands of times, the user can view whole sets of possible futures, assess which are most likely to occur, and identify areas of greatest downside or upside. Monte Carlo is used extensively in a wide range of fields. One of its first applications was in designing the shielding for nuclear reactors at the Los Alamos National Laboratory in the 1940s. (The name Monte Carlo was coined as a code name by scientists at the laboratory in reference to the Monte Carlo casino resort.) Monte Carlo simulation has since been used in finance, proj ect planning, engineering studies, traffic modeling, cancer radiation therapy, and telecommunications network design, among many other applications. Source: ACRP Report 76 (Kincaid et al. 2012). Exhibit 2-5. Introduction to Monte Carlo simulation.

24 Climate Resilience and Benefit–Cost Analysis: A Handbook for Airports of data points to handle a long projection period could be very large. But as shown there, it is relatively straightforward to summarize the data into temperature bins and count the number of days each year temperatures are forecast to be in each bin. Exhibit 2-6 is an extract from such a tabulation of individual forecasts from four geographic locations near an airport. Each column labelled Hxxx in Exhibit 2-6 is a 2-degree temperature bin, while each row is a forecast from one model for 1 year and one geographic grid point, showing the number of days where the high temperature is at the indicated level. The further one looks into the future, the wider the dispersion of forecasts would be from the different models. This dispersion represents the inherent risk and uncertainty of climate change. For example, a summary of results for temperatures exceeding a critical level (e.g., 110°F) might look like what is shown in Exhibit 2-7. To use the data in a risk analysis, one could set up a straightforward sampling plan where, say, drawing a random number between 0 and 0.025 for a given year would mean using the count from Model #1 and Grid_ID #1 (in Exhibit 2-6); a random number between 0.025 and 0.050 would use the count from Model #1, Grid_ID #2, and so forth. After going through all the relevant years, one will have completed one simulation showing a possible future path for very hot days. The process then could be repeated again and again, thereby generating new simulations representing many different possible futures, and each simulation could be independently evaluated from a benefit–cost perspective. Suppose 1,000 simulations were performed, representing 1,000 different estimates of benefits and costs. Recalling Exhibit 2-4, the NPVs shown at the bottom of Columns G, H, and I (repre- senting base-case delay costs, scenario case delay costs, and scenario case benefits, respectively) would change in each simulation. Under the assumptions used for the analysis, the present value Exhibit 2-6. Forecast count of days reaching indicated high temperature.

Evaluation Methods Under Risk and Uncertainty 25 of scenario case costs (bottom of Column L) would remain constant because they do not depend on the climate forecasts. To be clear, Exhibit 2-8 illustrates how the results from 1,000 different simulations could be laid out. Each simulation would relate to a different path for very hot days over time. The baseline outcomes in Column A are the cost of damages that would be incurred over the life of the proposed project if the project were not built (entered as positive numbers). These would be delay (and perhaps other) costs incurred by users due to payload restrictions. Column B contains the cost of damages incurred if the project were built, while Column C contains the costs of building and operating the project (say, a runway extension). Again, these values would be entered as positive numbers. With this layout of the results, the benefit–cost ratio for each simulation would simply be the reduction in damages (A – B) divided by project cost (C). An alternative but equivalent way of looking at the results is provided in Column D, which contains the NPV of the project for each simulation, equal to the damage savings (A – B) less the cost of the project (C). The benefit–cost ratio will be greater (lesser) than 1 whenever the Exhibit 2-7. Forecast range of count of days above 110°F. Baseline (without project) Scenario (with project) NPV A B C D Simulation PV $ of Damages PV $ of Reduced Damages PV $ of Project Costs PV $ of (A – B) – C 1 2 … … … 1,000 Exhibit 2-8. Results from 1,000 Monte Carlo simulations.

26 Climate Resilience and Benefit–Cost Analysis: A Handbook for Airports NPV is greater (lesser) than 0. These 1,000 NPV results then could be sorted and placed in order from highest to lowest. Thus, in some cases the NPV could be positive, while in others it could be negative, the results depending on the overall incidence of very hot days throughout the years for each simulation. The results could be averaged across all 1,000 simulations, as shown in Exhibit 2-9.10 In addi- tion, suppose one is interested in how often the project would pay off; one could count the number of simulations (percentage of the time) the project showed a positive result or a nega- tive result. One could also evaluate, say, the worst 1% of outcomes, meaning that the associated results that would occur 1% of the time. In this example, the average expected NPV is $0.5 million, with a corresponding benefit– cost ratio of 1.07 averaged across all 1,000 simulations. But the range and standard deviations of these measures are large, and without further consideration it would be difficult to draw a definite conclusion about whether the mitigation project should be undertaken based on these results. VaR Interpretation As a natural extension, the results from the simulations can be used to look at what is known as “value at risk,” which is a concept that originated in the financial industry in the late 1980s but is well suited to assessing the uncertainty associated with future climate change projections. More details about translating Monte Carlo simulation results into a VaR analysis are provided in Appendix C. For purposes of a VaR analysis, it is appropriate to focus on the net impacts both with and without the project. This is a slightly different way of looking at the NPV results. For conve- nience, the initial results from the simplified PNS example are summarized here: • Baseline net impacts: –$13.95 million (Column G delay costs from Exhibit 2-4). • Scenario net impacts: –$14.21 million (Column H delay costs + Column L mitigation project costs from Exhibit 2-4). The impacts are shown as negatives because, under the assumptions of the analysis, the air- port or its users would bear these impacts as net damages or costs. If one were to plot these two values on a graph, the scenario impact is more negative than the base-case impact, indicating that the project is not worth pursuing if the count of very hot days were to follow the assumed path shown in Exhibit 2-4. This could be repeated for each of the Monte Carlo simulations, resulting in a new pair of net impacts under the base case and scenario case. To assess these results across all 1,000 simula- tions, they can be sorted based on the difference between the two values and then plotted along a percentage scale. The result is a VaR graph such as the one shown in Exhibit 2-10. The results labelled “baseline” reflect the different possible projected losses that could be incurred by the airport if it chose not Average Min Max Standard Deviation Avg NPV of Project $0.5 million –$5.1million $22.5 million $4.5 million Avg B/C Ratio 1.07 0.37 3.75 0.56 Exhibit 2-9 Sample Monte Carlo aggregated results.

Evaluation Methods Under Risk and Uncertainty 27 to undertake the proposed project. The alternative is to undertake the project, in which case the airport would face the construction and maintenance costs plus the reduced delay costs; this is represented by the “scenario” line in the chart. Based on the varying benefit results from the Monte Carlo simulations, the blue line in the chart shows that if the airport does nothing, it faces a 10% chance of incurring dam- ages (in the form of delay costs) of at least around $25 million (where the blue line passes the 10% point on the horizontal axis) and could incur damages of as much as $50 million or more. On the other hand, if it does undertake the mitigation project, it must pay the invest- ment costs (about $8 million from Exhibit 2-4) and will incur any remaining delay impacts; these two factors combined could total as much as about $30 million (left extremity of chart for the red line). But also note that the range of potential net impacts is much larger under the baseline case (from about $5 million–$50 million in damages) than under the scenario case ($10 million–$30 million in damages and project costs). The chart also shows that there is about a 50% chance that the NPV of the project would be positive (indicated by the point at which the two curves intersect). It is important to properly interpret these results. Facing a 10% chance of incurring dam- ages of at least $25 million means that in 100 of the 1,000 simulations, the present value of damages would be $25 million or worse. Remembering that each simulation represents a set of future outcomes running from 2020 through 2090, these 100 simulations will include many different specific outcomes that vary across the years. In some simulations, there may be a small number of unusually hot years early on, resulting in a few highly valued delays (because they are discounted less when occurring early). In many others, the high temperatures will have been estimated to occur in later years, but they are likely to occur more often, resulting in more lower-valued delays. So it is important to recognize that the 10% chance of damages includes many different potential outcomes; it does not refer to an annual probability of occurrence, but rather the overall likelihood (over the entire analysis period of 2020 through 2090) that the airport’s users would face $25 million or more of delay costs (in present value terms) under the base case. Exhibit 2-10. VaR comparison.

28 Climate Resilience and Benefit–Cost Analysis: A Handbook for Airports Overall, the VaR analysis provides a different perspective than that from simply focusing on the positive average NPV from Exhibit 2-9. Decision makers at the airport can use the results to help decide between the risky but higher potential payoff of doing nothing and the certain cost of investing in the mitigation project, which reduces but does not completely eliminate the airport’s exposure. This is the kind of information decision makers will need to manage climate risk. 2.6 Summary and Next Steps This chapter has described a two-step process for evaluating climate risk. • Step 1: Screening with ACROS: Assuming the worst ACROS case, is it likely that some airport infrastructure would be vulnerable to one or more climate stressors, and would that be costly to the airport or its users? If the answer is in doubt and there is an adaptation that might make sense, one can use the worst-case ACROS projection to evaluate whether an adaptation would make economic sense (using a conventional benefit–cost or financial feasibility approach). • Step 2: Risk-adjusted analysis using Monte Carlo and VaR methods: When the airport is at risk for large losses or if the ACROS-based analysis is not conclusive, consider undertaking a risk-adjusted analysis to determine how likely it would be for the airport to be affected and whether a potential mitigation project would make sense. While the approach described here is relatively straightforward, there are a lot of under- lying practical and technical topics that must be understood in order to successfully undertake the suggested methodologies. Appendix C provides a more detailed technical description of the Monte Carlo and VaR methodologies, while Chapter 3 and Appendix D provide relevant information on available climate data and projections. In addition, two Microsoft Excel tem- plates have been developed to allow airport analysts to perform their own Monte Carlo and VaR analyses of the effects of future sea level rise and high temperatures. These are described in Appendix E; numerical examples using these templates are shown in Appendix F. The following chapters provide relevant context and refer the reader to technical material in specific appendices as needed.