Below, we discuss changes in future wildfire activity across the different emission scenarios and climate models (Section 4.1) and future changes in terrestrial ecosystem carbon storage (Section 4.2). In each section, biophysical impacts are presented at a national scale, summarized by region, and then monetized. The basic approach to defining the impacts of the POL3.7 GHG mitigation scenario for both wildfire and terrestrial ecosystem carbon storage relies on calculating an annual difference in projected physical impact values under the REF versus the POL3.7 scenario for a given year. These annual differences are then monetized, discounted, and aggregated. More details on historical vegetation, wildfire, and carbon dynamics are provided in Online Resources 7 and 11, with the latter also showing wildfire area burned at the regional level by decade.
Future changes in area burned
To estimate the benefit of the GHG mitigation scenario on area burned, POL3.7 scenario projections of burned areas are subtracted, by region, from similar REF projections. Thus, any reduction in area burned resulting from the POL3.7 GHG mitigation scenario produces a value greater than zero.
When using the projections for IGSM-CAM, MIROC, and CCSM, the benefits (area values > 0) or damages (area values < 0) of implementing climate change mitigation policy vary widely with annual variations reflecting millions of burned hectares (Fig. 1). For comparison, in recent extreme fire years, total area burned is on the order of 3.5 million hectares (National Interagency Fire Center 2013). The magnitude and variability of these results reflect characteristics of the model-specific climate projections and MC1. Specifically, fire ignition occurs in MC1 once threshold conditions for temperature and fuel moisture content are satisfied as ignition sources are assumed to be present. Online Resource 8 presents graphs depicting the POL3.7 impact on area burned from the different model projections nationally and by region.
As Online Resource 9 describes in detail for a single modeled cell, the IGSM-CAM method simulates realistic natural variability at the global and regional levels, as well as future changes in natural variability (changes in magnitude and frequency) (see Monier et al. 2014, this issue, for more detail). Compared to other GCMs, the IGSM-CAM projects a “wetter” future for most of the contiguous U.S., in addition to the underlying pattern of warming, which produces conditions that are generally more favorable to vegetation growth compared to today. However, as identified in Online Resource 9, this general pattern of growth is increasingly interrupted by extreme climatic conditions that drive massive burning. Online Resource 9 shows how an extremely wet period followed by an extremely hot/dry season can result in significant burning. In general, it is the repetition of this pattern that generates the large burned area projections for IGSM-CAM under both the POL3.7 and REF scenarios. Online Resource 10 provides additional context regarding the likely drivers of the wildfire results reported in this paper and a detailed comparison to findings reached in previous studies.
For IGSM-CAM, the average of results from the different initializing conditions show that implementing the POL3.7 scenario would reduce cumulative area burned between 2011 and 2100 by roughly 122 million hectares, relative to the REF scenario. The corresponding discounted (3 %) monetized estimate of reduced response costs over this period is $9.24 billion (2005 dollars). While all IGSM-CAM initializations project a net reduction in wildfire area burned from implementing the POL3.7 emission controls, timing and magnitude of net impacts vary by initializing condition (Fig. 1). This is reflected in the standard deviation (SD) for the mean area (63 million hectares) and the SD for the mean-avoided wildfire response cost results ($4.73 billion, 2005 dollars). The MIROC and CCSM pattern-scaled models show benefits of 84 and 91 million burned hectares avoided with the POL3.7 implementation, with an associated cost reduction of $7.31 and $7.77 billion (2005 dollars), respectively.
While these national-level summaries suggest general agreement across the models, the regional results show considerable variation (see Online Resource 11). For example, projections for the Southwest region show reductions in the amount of burned area and response costs with implementing the POL3.7 emission controls across all models. In contrast, the same projected results for the Western Great Basin are quite different across models. The regional results also highlight how a limited number of regions dominate national results. For example, across models, the Southern Area region’s results for area burned help determine national results (Online Resource 11).
The regional variability is also evident when examining a map of future burned area (Fig. 2). The IGSM-CAM projections consistently show burning affecting a larger region of the U.S. than the other two models, particularly in the South, Southeast, and Central region of the country.
While direct comparisons are difficult, variable but consistent increases in projections of the area burned by wildfires in future climates have been observed in other research. Online Resource 10 provides a comparison for recent studies in the contiguous U.S.
Future changes in terrestrial ecosystem carbon storage
Terrestrial ecosystem carbon storage impacts from implementing POL3.7 controls are quantified by subtracting REF from POL3.7 projections. As a result, additional carbon storage from POL3.7 implementation produces a value greater than zero.
Across models, the POL3.7 impact on terrestrial ecosystem carbon storage under the IGSM-CAM projections varies substantially over time. Figure 3 shows this variation with net benefits in some years and net damages (i.e., loss of carbon relative to the REF scenario) in others, often with rapid changes in sign and magnitude.
IGSM-CAM projections of carbon benefits/damages vary substantially among initializing conditions (Table 1). Across all initializing conditions, the IGSM-CAM results provide an average reduction in carbon storage in terrestrial ecosystems of 0.9 billion metric tons for the 2001–2100 period (SD = 39.2 billion metric tons) from POL3.7 implementation. In some cases the timing of carbon storage changes, combined with discounting, results in net benefits even when there is a net reduction in carbon storage. Specifically, three of the five IGSM-CAM results show a reduction in net carbon storage while four of the five sets provide a monetized net benefit (discounted at 3 %) that averages roughly $1.66 trillion dollars for the POL 3.7 implementation (2005USD, SD = $2.71 trillion).Footnote 5
Variation in the sign and value of projected impacts from the POL3.7 scenario are also observed under both the MIROC and CCSM projections (Fig. 3, Table 1). The cumulative results from these projections bound the average of the IGSM-CAM results—a net reduction in carbon storage of 1.9 billion metric tons projected with CCSM and an increase in carbon storage of 60.5 billion metric tons with MIROC. These changes are associated with a reduction in discounted monetized value of approximately $0.18 trillion for CCSM and an increase in value of $3.53 trillion (2005USD) for MIROC (Table 1).
The trajectory and timeline of national level terrestrial ecosystem carbon storage follow very different paths under the different models (Fig. 3). It is particularly striking that the largest carbon storage benefit from POL3.7 implementation accrues with MIROC despite its projecting an overall decline in carbon storage over the period, the only model where this occurs.
The timing of periods where POL 3.7 implementation generates benefits, and the magnitude of those benefits, also varies by model. For most initializing conditions, the IGSM-CAM results show implementation benefits occurring in the first part of the evaluated period (i.e., 2001–2050) with a loss of carbon storage beginning in roughly 2050 that continues through 2100. However, the benefits from POL3.7 implementation for the IGSM-CAM (WIND = 1) projections do not follow this pattern, with benefits accruing mainly in the 2050–2075 period (see Fig. 3), with loss of storage at the start and end of the 2001–2100 period.
With CCSM, small, relatively consistent annual losses in carbon storage begin in roughly 2040 and continue through 2100 (Fig. 3). In contrast, the MIROC results begin to consistently show benefits of implementing the POL3.7 controls starting in roughly 2040. These benefits then continue to increase in magnitude through year 2100 (Fig. 3, Table 1). The IGSM-CAM also shows much more inter-annual variability in national-level carbon storage than the two pattern-scaled projections (Fig. 3). Online Resource 2 describes the likely causes of these differences.
It is also interesting to note that while the net benefits accrued imply that the MIROC and CCSM model results bracket those of the IGSM-CAM, this only holds true for the cumulative results (Table 1). While the MIROC and CCSM results tend to produce fairly consistent increases and decreases in storage over time, at least from 2040 on, the IGSM-CAM paths shows much more inter-annual variability (Fig. 3).
Reviewing terrestrial ecosystem carbon storage dynamics by region provides further insights into national-level patterns (see Online Resource 12). As with wildfire, the results by region vary considerably over time across models (Table 1). Similarly, national results are largely driven by impacts from a small number (i.e., one to three) of regions. Specifically, across models, dynamics in the Southern Area and the Northern Rockies are important drivers of the national trend. However, for the IGSM-CAM results, the importance of these regions varies, with the Eastern Area and Rocky Mountain regions also playing a significant role at different points in time (see Online Resource 12). In the IGSM-CAM, it is striking that the POL3.7 carbon storage benefits shift from positive to negative in some of the regions (e.g., Northern Rockies and Southern Area) over the century.
We note that when focusing on long-term averages, critical underlying annual patterns and dynamics could be missed or misinterpreted. For example, when 30-year averages of ecosystem carbon storage benefits are examined for 2050 and 2080 (see Online Resource 13), the acceleration of benefit accumulation over time under both the MIROC and CCSM models is not evident (Online Resource 12). In addition, the IGSM-CAM averages shown in Online Resource 13 fail to capture the entire period over which significant benefits accrue for this model.