In this analysis, climate change impacts are quantified in terms of total kilometers of road degraded prior to lifecycle design and total cost of repairing or mitigating future damages. The impacts are determined by a “stressor–response” methodology, in which exogenous factors (i.e. stressors) have a direct effect on (and subsequent response by) focal elements. In the context of climate change and infrastructure, these factors include: changes in precipitation levels, temperatures, storm frequency and intensity, and wind speeds.
The stressor–response methodology presented here diverges from previous studies that emphasize qualitative statements rather than quantitative estimates (see section 2). A combination of material science reports, usage studies, case studies, and historic data are used to develop response functions for different infrastructure categories. Where possible, data from material manufacturers were combined with historical data to obtain an objective response function. However, when these data were not available, response functions were extrapolated based on performance data and case studies from sources such as Mozambique's Department of Transportation and other government ministries. Three climate stressors are examined in particular—temperature and precipitation changes and changes in the intensity and frequency of flooding events.
Road construction and maintenance costs were determined using both commercial cost databases and country-specific data. Six road types are examined in this study: paved and unpaved surface types with primary, secondary, and tertiary classifications within each. It is important to note that road types are treated homogenously when determining impacts of climate on the road inventory. Although differences in quality, maintenance, and construction standards exist within the country, these micro-level differences are beyond the scope of this study. Rather, a macro-level perspective is adopted that treats all roads within a classification type as similarly constructed and maintained. This perspective is necessary at a country level but should be examined in the context of specific decisions. Similarly, assumptions are in place that tonnage and traffic levels average out over the road network to provide distributed maintenance requirements. Although these assumptions require a macro-level reading of the results, they are in line with the economic and climate restrictions required for a country-level analysis.
The impact of climate change on road infrastructure is examined for each road type. Costs of new construction (or lifecycle-rehabilitation), maintenance costs to address climate-change related stressor impacts, and maintenance savings are evaluated in this study. All costs are applied with the goal of keeping a road functional for its design-life. This premise was established as a baseline requirement due to the preference for retaining infrastructure for as long as possible rather than replacing infrastructure on a more frequent basis. Achieving this goal may require a change in the design standard for new construction or an increase/decrease in maintenance for existing infrastructure.
In the current study, we consider policies that adapt to climate change and those that do not. In the “No Adaptation” scenarios, no actions are taken to mitigate the effects of climate change on paved and unpaved roads. In other words, Mozambique continues to build and maintain roads according to the design standards established without climate change. This is the engineering equivalent of the classic “dumb farmer” approach where the individual employs the same cropping patterns and techniques despite changing climate conditions. In terms of roads, this classic model is modified to a “dumb engineer” model whereby the engineers continue to design and maintain roads in the same manner despite an evolving climate; the result being that as temperatures increase and precipitation patterns change, maintenance needs often increase as the roads are not designed to withstand the changes in stressors. In contrast, under adaptation, roads are constructed to evolving design standards that anticipate climate change based on climate trends.
Stressor–Response Relationships: Precipitation and Temperature
The stressor–response relationships in the current study are divided between paved and unpaved roads based on the effects of stressors on the different surface types. For paved roads, estimates of stressor–response values are based on available data that establish threshold impact levels for precipitation and temperature. The available data suggests that levels of 10 cm increase in annual precipitation or 3 degC maximum temperature increase will result in pavement degradation (Lea International, L.D., 1995; National Oceanic and Atmospheric Administration (NOAA), 2009).
For unpaved roads, the stressor–response relationship associates impacts with changes in maximum monthly precipitation. Ramos-Scharron and MacDonald (2007) attributed about 80% of unpaved road degradation to precipitation, while the remaining 20% is attributed to factors such as the tonnage of traffic and traffic rates. Given this attribution to precipitation and the focus on retaining design lifespan, we assume that climate impact is based on maximum monthly precipitation, rounded to 1 percentage point increments. For example, if the maximum monthly precipitation increases by 10% in a given location, then we assume an 8% (0.8 × 0.1 = 0.08) impact on lifespan. Available data suggests that there is no relationship between temperature and unpaved road degradation. Our approach is summarized as follows:
where CIU is the impact in degradation for unpaved roads associated with a unit change in climate stress, MIP is the increase in maximum monthly precipitation.
Given these overall impacts, two basic methodologies were adopted for maintenance costs. The first approach, used for paved roads, is based on the cost of preventing a reduction in lifespan that may result from changes in climate-related stress. It is assumed that any lifespan reduction caused by an incremental change in climate stress is equal to the percent change in climate stress, scaled for the stressor's effect on maintenance costs. Miradi (2004) estimated that ongoing precipitation-related maintenance for paved roads accounts for 4% of maintenance costs and temperature-related maintenance accounts for 36% of costs. After estimating the potential reduction in lifespan associated with a given climate stressor, the costs of avoiding this reduction in lifespan are calculated as the product of (i) the potential percent reduction in lifespan and (ii) the base construction costs of the asset. Therefore, a 10% reduction in lifespan has an estimated increase in maintenance costs of 10% over base construction costs.
As shown in equations (2) and (3), we implement our approach to maintenance costs in two stages: (i) estimating the lifespan decrement that would result from a unit change in climate stress, and (ii) estimating the costs of avoiding this reduction in lifespan. It is assumed that such a reduction in lifespan caused by an incremental change in climate stress (LP) is equal to the percent change in climate stress, scaled for the stressor's effect on maintenance costs, as shown below:
where LP is potential percent change in lifespan for existing paved roads associated with a unit change in climate stress, ΔS is change in climate stress, S0 is base level of climate stress without climate change, and SMT is the share of existing paved road maintenance associated with a given climate stressor, and the set i has elements [precipitation, temperature].
The total change in maintenance costs is then as follows:
where CMP is change in maintenance costs for existing paved roads associated with a unit change in climate stress, and BMP is base original construction cost of the paved road segment.
Finally, to estimate the stressor–response values for unpaved road maintenance costs, we follow the approach outlined above for unpaved roads' new construction costs. Changes in unpaved road maintenance costs are associated with a 1% change in maximum monthly precipitation. As indicated above, 80% of road degradation can be attributed to precipitation, while the remaining 20% is due to traffic rates and other factors. This implies that unpaved road maintenance costs increase by 0.8% with every 1% increase in the maximum monthly precipitation values projected for any given year. This reflects the direct costs required to mitigate the damages to the road based on available relationship data. The authors recognize that these assumptions are broad and factors such as slope and traffic rates will impact degradation rates. Further investigation is being conducted to incorporate these issues. The general form of the maintenance equation is as follows:
where CMU is the change in maintenance costs for unpaved roads associated with a unit change in climate stress or design requirements, MIP is the percentage increase in maximum monthly precipitation, and BMU is the baseline maintenance costs for unpaved roads.
Stressor–Response Relationships: Flooding
As with precipitation and temperature, flooding damage is calculated using stressor–response functions. Road flood losses are calculated based on flood return periods projected by the four GCMs employed in this analysis for northern, central, and southern Mozambique in Strzepek et al. (2010). An assessment of the nature of the flooding is also germaine. Gradually rising and then gradually receding flood waters can often leave roads, particularly paved roads, essentially intact. Running flood waters, in constrast, quickly result in the complete destruction of underlying roads generating repair costs that are close to new road construction costs. Translating the return periods into damage costs requires two steps: (i) determining the amount of road kilometers damaged, and (ii) estimating the specific costs implied by these damages.
Figure 1 illustrates the damage functions employed for the analysis. The damage functions are derived based from the report authored by the COWI consulting group (2009), Strzepek et al. (2010), and the assessment of the authors. While minor flooding events have minor or no impact, major flooding events can have substantial impact. A 100 year flood is assumed to damage 30% and 10% of unpaved and paved roads respectively in the region in which it occurs (North, South, Center). Costs of repairing damaged roads vary by road class (primary, secondary, and tertiary) and by road type (unpaved and paved). Flood damage costs are also variable as a share of new construction costs. These cost estimates are also based on COWI (2009) and the assessment of the authors.1
Some intuition on the implications of these numbers is valuable. Combining the damage functions and the cost estimates with the shares of baseline national road extension by road class and road type, we find that a 100 year flood causes damages equivalent to about 9% of the value of the road stock in the region in which the flood occurs. Based on these figures, we would estimate that the floods in the south of Mozambique in 2000 necessitated about US$122 million in reconstruction expenditures for road and related infrastructure. While we have been unable to locate any formal ex post evaluations of the infrastructure costs of the 2000 floods against which to compare this number, the available evidence would suggest that this estimate is reasonable.2
The relationships developed above are incorporated into a dynamic road network simulation model labeled “CLIROAD”. The simulation model tracks the road stock broken by age since construction (or 20 year rehabilitation), road class (primary, secondary, tertiary), road type (paved or unpaved), region (north, south, center, and urban) for each year over a simulation period from 2003 to 2050. While the stressor–response functions are constant across regions, the climate inputs (precipitation, temperature, and flood events) are disaggregated by region. It also tracks all costs to the same level of detail over the simulation period. The model is constructed in GAMS and is available upon request.