Adaptive Climate Response Cost Models for Infrastructure

The climate in Alaska is changing with temperatures and precipitation increasing, and experts say those changes will continue. The changing climate will, among other kinds of social and economic effects, damage public infrastructure throughout Alaska, adding to maintenance costs and shortening the life span of everything from schools to sanitation systems. The question of how much cost impact this change will have and when the costs are expected to occur is the question of significant concern and debate. One approach to answering this question is addressed in this paper. Based on a probabilistic model, the results in this paper estimate how much climate change will add to future costs of public infrastructure in Alaska. Based on a database of public infrastructure throughout Alaska, and the estimated replacement costs and life spans for the various types of infrastructure statewide, the model provides a first analysis of climate impact on infrastructure costs. As an additional refinement of the model, the paper analyzes the potential mitigating impacts of adaptation in the replacement process. Although many variables can change during the next 75 years, this paper presents a first attempt at combining cost and climate models to project cost impacts on public infrastructure. 1 Associate Professor, Dept. of Civil, Env. And Arch. Engineering, University of Colorado, Boulder, CO 803090428, (ph) 303-735-1063, [email protected] 2 Professor, Dept. of Civil, Env. And Arch. Engineering, University of Colorado, Boulder, CO 80309-0428, (ph) 303-492-7111, [email protected] 3 Senior Policy Advisor, The Nature Conservancy, Anchorage, AK, [email protected] 4 Research Assistant, Dept. of Civil, Env. And Arch. Engineering, University of Colorado, Boulder, CO 803090428, [email protected] INTRODUCTION Alaska has been called a “climate canary” because it is already seeing the early effects of global climate change. Climate researchers also expect future climate change in Alaska and other Arctic places to be more pronounced than it is elsewhere in the world. Alaska is the only U.S. state with vast areas of permafrost, permanently frozen ground, which makes it especially vulnerable to a warming climate. It also has more miles of coastline than the rest of the United States combined. Some places in western Alaska are facing unprecedented erosion rates where the sea ice has retreated exposing the shore to direct wave action from Bering Sea storms. Average annual temperatures around Alaska increased approximately 1oC to 3oC over the past five to six decades (UAFGI, 2006; Chapman and Walsh, 2003). Furthermore, the most recent AtmosphereOcean General Circulation Models (AOGCMs) project that both temperature and precipitation will continue increasing in Alaska (McGuinness and Tebaldi, 2006). Temperatures around Barrow are expected to rise enough by 2080 that break-up of ice will come significantly earlier and freeze-up later than today. Problems associated with thawing permafrost, including effects on the foundations of buildings and roads, are well documented and often dramatic (ACIA 2005, Nelson et al. 2003, USARC 2003, Osterkamp et al. 1998, and Romanovsky et al. 2002). Utilities have reported that telecommunication towers are settling due to warming permafrost. United Utilities, for example, has said “warm permafrost is a result of global warming” and is seeking funds for cost overruns in the Yukon-Kuskokwim Delta of southwest Alaska (Hamlen, 2004). The continued changing of the Alaskan climate will continue to affect both natural and man-made systems with many economic and social consequences. One effect will be to increase building and maintenance costs for public infrastructure. The challenge for public officials attempting to prepare for these effects is to anticipate the potential cost of climate change impact on public infrastructure. Although significant study has been done on the cost of natural disasters in various contexts, climate change presents a new challenge for public officials. Given that modern civilization has not experienced a significant climate change event, no direct cost models exist from which to anticipate potential impact. Additionally, the combination of the long-term nature of the event and the conflicting scenarios depicted for climate change create a significant amount of uncertainty in the planning process. This paper presents one approach to answering this uncertainty in the context of climate change in Alaska. The cost model presented here represents a first step toward anticipating the costs associated with climate change through an adaptive response model that incorporates learning to respond to Alaskan climate change. Limitations exist in terms of aging inventory and adoption cycles. However, the models present the opportunities that exist to reduce climate change impact by incorporating adaptive responses. This paper provides preliminary estimates of how climate change might add to future costs of building and maintaining Alaska’s public infrastructure in particular, but also presents an approach to predicting climate change cost impact that can be adapted to any geographic location. BACKGROUND Natural disaster events occur in almost every geographical sector of the United States. Earthquakes are a regular occurrence in the west, hurricanes are regular events in the southeast, and floods make annual appearances in the Midwest. Although these natural events are not directly analogous to climate change, the local, regional and Federal government responses that have been documented over the last several decades provide foundational models from which to anticipate the adaptation response that may occur in the case of climate change in Alaska. The importance of these responses to the climate change scenario is the concept that individuals learn from natural events and adapt to the events over time. Adaptive learning provides the opportunity to offset the effects of natural events on infrastructure through the improvement of the infrastructure in response to the negative effects. This type of adaptation is appropriate to elements such as infrastructure which have extended useful life scenarios. In contrast, areas such as high technology focus on adapting for obsolescence. In this frame of reference, technology companies focus on improving disposable products to make existing products obsolete. From an economic perspective, this is logical since technology improvements result in new entities with anticipated lifespans of a short, or disposable-based, duration, whereas infrastructure improvements result in the maintaining of a structure which is intended to have an extended, non-disposable, lifespan. The study of natural disaster response has been a priority for agencies such as the Federal Emergency Management Agency (FEMA), as they strive to improve recovery response and improve the resiliency of structures to natural event occurrences. A leader in these studies has been the States of Florida and South Carolina where hurricanes introduce a continuously evolving set of requirements and recommendations for improving structure resiliency. Since the major events of Hurricane Hugo in 1989 and Hurricane Andrew in 1992, Florida and South Carolina have aggressively focused on improving the resiliency of structures to major storm events. These responses have been in the generation of major studies by Clemson University (Clemson, 1999) and the State of Florida (Florida, 1996; Florida, 2002) regarding the costs of adaptation and the adoption of proposed recovery solutions. The result of these studies has been significant improvements in the technology associated with infrastructure and private structures. Individual advances in areas such as window construction, roof attachments, and structural connections have resulted in significantly enhanced structural resiliency and useful life expectancies that are the increase. Unique to this form of adaptive response has been the role of private industry in developing these responses. Since the adaptations are focused on a single item such as a window pane, private manufacturers have taken a lead role in developing these adaptations. However, these advances have sometimes taken an extended period of time to be fully adopted due to the time required by local building authorities to test, evaluate, and document the advances. Similar to the efforts by South Carolina and Florida, states such as California that experience earthquakes on a regular basis have attempted to quantify the risks associated with adapting structures to resist potential earthquake damage. These efforts have attempted to balance the cost of adaptation with the potential loss associated with earthquake damage (FEMA, 1995; FEMA 1998). The difficulty in these response efforts is balancing the cost of adaptation with the risk of earthquake exposure (Stein and Tomasello 2004). As outlined by these research efforts, the question of earthquake frequency and exposure needs to be balanced with regulations that require forced adaptation through building codes. Additionally, the need for research and development through publicly funded centers is an essential element of earthquake adaptation as research facilities for earthquake simulation are limited to large research institutions. In contrast to the incremental improvements developed by private manufacturers for hurricane resiliency, earthquake resiliency has been focused on incremental building code modifications. In these regular code modifications, results from research are translated into structural enhancements with the intention of improving earthquake resistance. In this manner, earthquake adaptation emphasizes a regulatory approach rather than the market development approach that is prevalent with hurricane adaptation. The final adaptation response effort that appears as a relevant model for climate change is the response to floods that occur regularly in the Midwestern segment of the United States (Aglan, Wendt, and Livengood, 2004). This model differs from the previous two models in that flood response is primarily an event-driven model. In this form of model, adaptive responses are not put into place until an event, in this case a flood, is significant enough in terms of reducing useful life to cross a threshold that is determined by the local authority. If an event does not reach this threshold, then no action is taken with the understanding that this event is part of the natural cycle in the area and it may result in reduced lifespan for the structures. New structures are constructed with the same specifications since this is considered the standard at the time. In contrast, if an event exceeds an event threshold, then a rapid adaptation response is put into place resulting in a change of building codes that requires a stair-step increase in building costs. Similar to the response to earthquakes, the response model to floods is code-based. Rather than the private market driving structural changes, flood response results in regulatory changes. The regulatory changes are codified through building code updates. However, since the code is only updated after a significant event, the regularity of the updates does not follow the pattern seen with earthquakes. The result is an expense that occurs on a less frequent basis, but which tends to have a greater impact when it occurs. Similarly, the lack of an update often results in structures being constructed with a reduced lifespan in-between the release of new building codes. THE ADAPTATION SCENARIOS Based on the existing approaches to adaptation for hurricanes, earthquakes, and floods, three specific adaptation scenarios or cost models can be put forward as potential approaches for adaptation responses to climate change. These models reflect the cycle of adaptation that occurs in relationship to the specific type of natural disaster from which they were abstracted. The foundational models are divided as follows: 1. Rapid and Continuous Adaptation – The first scenario is based on experience with earthquakes in California. In this model, the engineering and design communities place a focused effort to respond to, and anticipate, changes due to climate change. The focused effort results in a mandatory set of changes that are implemented in all designs within a regular updating of the regulatory building codes. Typically, this update occurs every 5 years in an active event location. This change results in a gradual cost increase, but retains the life expectancy of the structures. However, continued research into the effects of the scenario continue to change the building requirements and thus continue to escalate the costs each generation, but no negligible life span decrease is noticed. After the first set of adaptation codes are developed, the cost increases begin to decrease as initial code generation changes to less severe code adjustment. 2. Phased Adaptation – The second scenario is based on experience with natural disasters in locations such as Florida where regular, but non-location specific incidents create awareness for change, but not a focused interest in adaptation for all structures. In this case, adaptation occurs over a series of generations, with generations adopting enhanced technological adaptations which not only equal life expectancy, but have the opportunity to increase life expectancy. Costs rise slowly at first as only a limited number of structures adopt the technology, and then significantly during the second generation of adaptation when all structures adopt the new technologies. However, as adaptation continues to evolve, costs decrease as the adaptations become standard practice and competition impacts the adaptation market. 3. Forced Adaptation – The third scenario represents the event-driven rather than a codedriven approach to infrastructure adaptation. In this model, adaptations occur only after significant damage occurs to infrastructure elements. This scenario is often seen in areas where floods or fires are prevalent and the local government does not adapt until a failure in the system occurs. When failure occurs, significant costs are borne to enhance the infrastructure and retain life expectancy. However, these adaptations are often only sufficient to adapt to current conditions and the cycle needs to be repeated when the next significant incident occurs. These three models represent the typical responses that may occur during the onset of climate change in Alaska and represent three possible economic impacts that may occur in response to climate change. THE ALASKA SCENARIO The methodology adopted by the authors to provide an estimate of the cost impact on Alaskan infrastructure required the development of an accurate scenario from which to develop a cost estimate. The development of this scenario included several steps: (1) acquiring climate projections; (2) creating a database of public infrastructure throughout Alaska; (3) estimating the replacement costs and life spans for existing infrastructure, (4) developing a financial scenario for long-term economic considerations. To begin the analysis, it was required to know how experts believe Alaska’s climate will change in the coming years. The Institute for the Study of Society and the Environment at the National Center for Atmospheric Research provided 21 AOGCM climate projections for the years 2030 and 2080 (McGuinness and Tebaldi, 2006). AOGCMs couple atmosphere general circulation models with ocean general circulation models-thus taking into account the complex feedbacks between the earth’s atmosphere and oceans-to provide detailed projections of future climate conditions on a regional basis. The models also include societal inputs, including projected greenhouse gas emissions. Consequently, their output depends on assumptions about future industrial growth, technology, and carbon emissions. These models are considered the most sophisticated climate models currently available. The 21 sets of projections include projected mean monthly temperatures and precipitation for six Alaska locations for the years 2030 and 2080. All the projections show Alaska temperatures rising, but they vary in how much and how fast they project temperatures will rise. To accommodate this variability, projections from three climate models—projecting less (i.e. “warm”), mid-range (i.e. “warmer”), and more warming (i.e. “warmest”)— were adopted as inputs for the analysis (Smith and Wagner, 2006). The AOGCMs selected for this analysis are: 1. Warm Model: CSIRO Atmospheric Research, Australia, CSIRO-Mk3.0 2. Warmer Model: U.S. Department of Commerce, NOAA, Geophysical Fluid Dynamics Laboratory, GFDL-CM2.0 3. Warmest Model: Center for Climate System Research (University of Tokyo); National Institute for Environmental Studies; and Frontier Research Center for Global Change, Japan, MIROC3.2(hires) The next step in the analysis was assembling a database of public infrastructure in Alaska. According to researchers at the Congressional Research Service, critical infrastructure is a term used to describe the “material assets” that are essential for society and the economy to function. In line with that definition, the cost calculations in this analysis are based on public infrastructure—assets owned by local, state, and federal governments that are critical for delivering goods and services communities depend on (Moteff et al, 2003). Based on an analysis of public records, a database of nearly 16,000 individual elements of public infrastructure was developed and divided into 19 categories. Each element was identified by a location and assigned a useful life and replacement value. Each element was also assigned a set of values associated with local permafrost conditions, susceptibility to flooding, and proximity to the coast (USACE, 2006; USGS, 2006). The infrastructure the database has an estimated price tag of around $40 billion today. Much of that is in various types of transportation infrastructure— especially roads—which are expensive to build and maintain in Alaska. The database undercounts and undervalues some types of infrastructure, especially defense facilities and power and telephone lines. Information about the extent and value of defense facilities is often suppressed for reasons of national security. Whenever possible, replacement costs were obtained from public agencies. When no replacement cost was reported, average insured value or other available information was used as an estimate. Data on the projected useful life of various types of infrastructure was based on information from the Alaska Division of Finance and personal communications with employees of government agencies and academic researchers. Finally, to develop a long-term financial consideration, the net present value of the infrastructure replacement over time was adopted as a base financial consideration. Calculating the base case replacement costs is simply a matter of taking the present value of the annualized replacement costs and aggregating them. In the current effort, a 7.25% nominal, or 2.85% real, discount rate was adopted. Particular attention was paid to the selection of a defensible discount rate. Specifically, the real discount rate was calculated by subtracting the 30-year average Producer’s Price Index (PPI) from the 30-year average of the Natural Resources Conservation Service’s nominal discount rate for water resources projects (USBLS, 2007; USDA, 2007). The Alaska branch of the U.S. Army Corps of Engineers consistently uses the NRCS discount rates for its assessments of possible relocation projects, including its estimates of relocation costs for the communities of Shishmaref, Kivalina, and Newtok in western Alaska. A market-based discount rate was chosen following the lead of the Corps of Engineers and after carefully considering the context of evaluating the costs of building public structures with public funds. There may be implicit benefits to society from building these structures, but this analysis narrowly focuses on the additional construction costs due to rapid climate change. ALASKA CLIMATE CHANGE ADAPTIVE MODEL In selecting the appropriate scenario as the basis for the climate change event in Alaska, consideration was given to the regularity of the event and the primary party responsible for driving the changes. With these drivers as the selection criteria, it was determined that climate change had the greatest similarity to earthquake adaptation. Specifically, the need to drive change through building codes and the regular occurrence of the climate change makes climate change amenable to the adoption of a continuous adaptation cost model. As a further refinement of this model, continuous adaptation was divided into two possible outcomes, an event-driven model and a code-driven model. This division allowed the adaptation analysis to demonstrate the differences between using continuous code changes (a code-driven model) as a response mechanism and the use of periodic code changes with an event threshold (an event-driven model). The following introduces these models based on the climate change scenario. Code-Driven Model The concept of this model is that continuous research is being conducted to respond to changing climate conditions. The focus of the research being to retain useful lifespan amounts through continuous, incremental updates of the building codes. The result of this research is continuous modifications to existing codes in an attempt to keep infrastructure resilient to the effects of climate change. In the code-driven model, code updates are provided on a five-year cycle. The result of the codes is that each structure is able to start its existence with a full lifespan. However, to obtain this benefit, a cost increase is applied with the update every five years. In the case of climate change, since there is no existing code specifically addressing this adaptation, an increase of 0.8% is added every five years to the construction of a facility. Thus if it has been 20 years since the last structure was built, then the new structure will incorporate an increase of 4 updates times 0.8% each, for an increase of 3.2% attributed to climate change. After 2030, this cost increase is reduced to 0.45% to reflect the fact that an existing code is being modified rather than created. The lifespan of the structure is affected by the impact template based on precipitation and temperature increases affecting the full lifespan. For example, a hospital with a 40-year lifespan that is affected by both temperature and precipitation increases will have its lifespan shortened based on the appropriate database entries. In the current database scenario, the worst case is a hospital losing about 5% of its lifespan for each 1 degree of temperature increase and 1 inch increase in precipitation. Given that the average model shows a temperature increase of 2.1 degrees by 2030 and a 4.3% increase in precipitation (.688 inches in Anchorage), that results in a total lifespan reduction of 23% (17.5% for temperature and 5.5% for precipitation). The lifespan reduction results in the hospital lasting only 31 years rather than 40 years. At that time, due to continuous adaptation, the hospital will be rebuilt with a full 40 year lifespan, but will cost an additional 4.8% to build (.8% times 6 updates). Event-Driven Model The concept of the event driven model is that adaptation research is being conducted, but nothing is implemented on a structure until a critical threshold is reached. The idea is that structures would incorporate additional repair money to retain a reasonable useful lifespan. The threshold that is being used in this model is 20% which represents a typical planning threshold for new structures. For example, if it is perceived that a 20 year lifespan is being reduced by 4 years, then that exceeds the planning window for a new structure, so it is considered a significant loss of useful life. In the model, the initial lifespan of the structure is impacted by climate change in the same manner as above. For example, the first generation of the 40 year hospital is reduced by 23% or nine years, resulting in a useful life of 31 years. It is at this point where the approach differentiates from the code-driven approach. Specifically, no intermediate code updates have been issued for the hospital, so no cost increases have been absorbed. Rather, the increase in cost is only absorbed when the critical threshold is reached. If the threshold has not been reached, then a new structure is built with a reduced life expectancy. In this case, it would be 31 years to start rather than the original 40 years. If the threshold has been reached, then a cost increase of 5% is absorbed and the structure is built with the original lifespan. In the example, the hospital has exceeded the 20% threshold so it is built with a 40 year lifespan with an increase in cost of 5%. Although the impact of the event-driven model is not apparent in the 40 year example, it becomes apparent in a shorter lifespan example. For example, in the case of a water treatment facility with a 20 year lifespan, the difference becomes apparent in only three generations. In the example, the first generation is affected by climate change by a reduction in lifespan of 14.3% or 3.5 years. Since this loss is below the 20% threshold, a difference arises in how the second generation is implemented. In the code driven model, three updates of 0.8% are absorbed resulting in a new structure with an original 20 year life span and an increase in cost of 2.4%. In the event driven model, the threshold has not been reached so the same structure is built with no climate change adjusted costs, but with an expected useful life of only 16.5 years. When the scenario is extended to the third generation, the code driven model results in a structure with a 16.5 year lifespan that is rebuilt to the original 20 year life span at a 2.4% cost increase. In contrast, the event driven model results in the second generation structure lasting only 14.3 years, since it started with an anticipated useful life of only 16.5 years. However, at this point the useful life reduction now exceeds the 20% reduction threshold. Thus, the third generation structure incorporates a 5% cost increase to return to the original 20 year life expectancy. The overall difference at the beginning of the third generation is that the code driven model requires a third structure in 33 years with a total cost increase over the two new structures of 4.8% while the event driven model requires a third structure in only 30.8 years at a total cost increase of 5%. The following sections introduce the application of the code and event-driven models to three specific infrastructure elements in the infrastructure database; bridges, runways, and water/sewer facilities. The difference between no adaptation, code adaptation, and event-driven adaptation are presented together with the base case of no climate change. COST ADAPTATION MODEL APPLICATION To illustrate the impact of the two scenarios on the climate change application, four cases were run through the adaptation cost model for each infrastructure component; a base case of no climate change, a no adaptation case, an event-driven case, and a code-driven case. No Climate Change – The first case models the infrastructure classes with no impact from climate change on useful life. Each infrastructure element lasts for the predetermined useful life and then is replaced with the same useful life, but at a cost that is discounted by the predetermined discount rate. No Adaptation Case – The second case models the impact of climate change on infrastructure with no adaptation or learning taking place. In this case, the impact of climate change is calculated each year on each infrastructure component. The impact reduces the lifespan each year, resulting in a final useful life that is less than the anticipated lifespan. Compounding this problem is the lack of learning. Since no learning is occurring, the next generation of structure begins with an anticipated useful life equal to the ending useful life of the previous structure. Over several generations, the useful life continues to be reduced, creating a scenario that additional generations of structures are required for each infrastructure component. Code-Driven Case – The third case implements the code-driven model described previously. In this case, the structures are impacted by climate change on a yearly basis just as in the No Adaptation Case. This impact results in a lower lifespan than he anticipated useful life. However, during the lifespan of the structure, code updates are occurring every five years that are designed to offset the effects of climate change for new structures. Thus, when the useful life of the structure is over, a new structure is constructed with the full useful life restored as the starting point. The additional cost incurred from climate change is the cumulative incremental cost associated with each code update. Event-Driven Case – The final case implements the event-driven model described previously. In this case, the structure is once again impacted by the climate change on a yearly basis which results in a lower actual lifespan. The key to this scenario is how much useful life has been lost compared to the anticipated lifespan at the end of the useful life for a structure. If the lost lifespan is less than 20%, then a new structure is built with no adaptation since the event threshold has not been met. Although this new structure carries no additional costs associated with climate change, it follows the No Adaptation Case by applying a starting useful life that is only equal to the reduced useful life that existed at the end of the previous generation. However, if the lost lifespan is greater than 20%, then the event threshold is achieved and the new structure is built with a full useful life value with an additional 5% cost associated with climate change adaptation. The effect of these four cases is combined with the low, medium, and high climate change models to create a matrix of 12 overall entries, with the three infrastructure types in each cell. Scenario: Water Treatment Services with Single Climate Change Exposure To illustrate how the cost model indicates potential cost scenarios, the water treatment plant introduced previously provides an exemplar in the context of a single climate model, medium, and a single climate impact zone. In this example, a water treatment plant with an anticipated 20 year useful life is exposed to coastal flooding and minimum permafrost melting. The base cost of the plant is $5 million per structure. Figures 1 and 2 illustrate the effect of climate change on remaining lifespan. As illustrated in the graphs, the baseline case without climate change requires four generations of structures to be built, one in 2006 and then one every 20 years. When climate change is factored in to the scenario, the structure loses useful life. The No Adaptation and Event-Driven cases have the same reduction in lifespan during the time period through 2040. However, this changes as the scenario is taken further along the timeframe and the event threshold is met in later generations. In this scenario, the No Adaptation case results in seven generations of the structure being built, with the last generation having a useful life of only 6.4 years when it is built. However, the Event-Driven case deviates from this pattern as the useful life threshold is met in 2040 and the structure is returned to its original useful life in this and later generations. Examining this same scenario from the perspective of cumulative cost per structure, Figures 3 and 4 provide an illustration of the costs associated with climate change. Once again, the base case with no climate change requires the building of four generations of the same structure, each lasting 20 years. Bringing a discount factor into the scenario, the total cost of the four generations of the structure for the base case is $10.3 million. The addition of climate change changes this base cost. The No Adaptation case requires three additional generations of the structure, but no adaptation investment. The result of these extra generations is a cumulative cost of $14.8 million. In contrast, the event-driven case requires additional cost components due to the cost of the event adaptation. In this scenario, the event-driven case results in the construction of five generations of the structure with an overall cost of $12.3 million. Putting this scenario in terms of the total number of structures that have to be replaced in this category of exposure, the cumulative effects of lifespan reduction can be seen. If 11 structures of this type need to be replaced in Alaska, the resulting cost without climate change is $113 million with a 2.85% discount rate. Climate change increases the cost of this replacement with the No Adaptation case resulting in a 44% increase and the Event-Driven case resulting in a 20% increase. TOTAL INFRASTRUCTURE REPLACEMENT COST WITH ADAPTATION Using the cost adaptation models described previously, the 3 categories of infrastructure identified for the Alaska climate change model were analyzed for the impact of adaptation on the total cost. Table 1 includes the data for the 3 infrastructure types under each of the climate models. The data in Table 1 provides an indication of how using different climate models and different adaptation strategies can significantly alter the potential cost of infrastructure replacement. In each of the models, the “No Climate Change” state is the base case for the scenarios. In this column, the total cost for building and replacing a structure in from 2006 through 2080 is presented with the assumption that no climate change has occurred and the structures have lasted an entire expected lifespan before requiring replacement. As with each of the columns, this is the total cost of all generations of structures required for that single original structure. For example, the “water/sewer” entry in the “No Climate Change” scenario indicates the cost for building 4 generations of a single plant, each with a lifespan of 20 years. The remaining columns indicate the percentage increase over the base case that each approach results in for the indicated infrastructure type and adaptation approach. The “Low” climate change model provides an indication of what occurs when a minimal degree of climate change is applied in the model. In this scenario, code adaptation provides the greatest benefit for adaptation. In contrast to the continuous, but minimal investment in adaptation, the low degree of climate change causes the event adaptation model to have minimal number of times where the threshold limit is met to achieve structure adaptation. Therefore, the event adaptation approach results in a greater number of generations to be built than the code adaptation approach. The “Medium” climate change model follows similarly from the “Low” climate change model in that code adaptation once again is the preferred approach to reducing costs. However, the “Medium” model does indicate a lower overall increase than the “Low” model. This is a result of a decrease in precipitation from the “Low” model and thus a decrease in affect on the structures from excess erosion. Finally, the “High” model provides the greatest indication of how adaptation can reduce the cost of infrastructure replacement. For example, the water/sewer infrastructure category will have a 10.5% cost increase if no adaptation is applied to the new construction process. However, this cost is reduced to a 7.75% increase if event-driven adaptation is applied, and the cost is further reduced to a 7.03% cost increase if a code-driven approach is adopted. Although the cost reduction is not as significant for bridges due to their longer expected lifespan, the cost savings are consistent over every infrastructure category. Conclusion Even without climate change, maintaining and replacing infrastructure in Alaska is an expensive proposition—costing an estimated $32 billion between now and 2030 and $56 billion by 2080. Projected climate change could add 10% to 20% to infrastructure costs by 2030 and 10% to 12% by 2080, under different climate projections and taking design adaptations into account. The additional costs are relatively higher in the short run, because agencies haven’t had as much time to adapt infrastructure to changing conditions. It is important to note that strategic design adaptations have much more potential to reduce extra costs in the long run. Between now and 2030, adaptations might reduce costs related to climate change by anywhere from zero to as much as 13%, depending on the extent of climate warming. But between now and 2080, adaptations could save anywhere from 10% to 45% of costs resulting from climate change. Transportation infrastructure—especially roads and airport runways—will account for most of the additional costs between now and 2030. That’s because transportation infrastructure is expensive to build and maintain in Alaska under any circumstances, and many airports and some roads are in areas that will be most affected by a warming climate. But water and sewer systems—which are very expensive to build and difficult to maintain in areas with a lot of permafrost—will also account for nearly a third of the costs resulting from climate change by 2030. Until now, a majority of the studies detailing the potential effects of climate change have focused on how natural systems are likely to be affected. But the preliminary estimates generated from this study indicate how climate change might increase future costs for Alaska’s public infrastructure as well. The results illustrate that the potential risks for man-made systems are also considerable. This documentation should provide a foundation for further research to continue into the potential economic costs of climate change for the man-made environment.