Ana səhifə

National Park Service U. S. Department of the Interior Natural Resource Program Center

Yüklə 14.4 Mb.
ölçüsü14.4 Mb.
1   2   3   4   5   6   7   8   9   10

Figure 34. Temperature and precipitation changes in North America projected for 2050 (2040–60 average) by an ensemble of 22 climate models used in the IPCC AR4. Changes are shown relative to the 1950–1999 average. The top row is the multi-model average temperature change for the annual mean (left), winter (center), and summer (right). The second row shows the percentage change in total precipitation (data source: CMIP3 multi-model dataset, PCMDI). The bottom row shows multi-model agreement. Source: Ray et al. 2008; reprinted with permission.

Although temperatures are expected to increase across the landscape, natural short-term variation (i.e., years to decades) is still expected. One example would be the unusually cool Northern Hemisphere winter of 2009–10, which is likely linked to a natural climate variation called the North Atlantic Oscillation (NAO, Hurrell 1995), a large-scale circulation feature that can alternately block or allow cold air from the Arctic to enter the mid-latitudes of North America, Europe, and Asia. In 2009–10, the NAO was positioned to spawn cold winters as well as strong storms and heavy snowfalls. However, when considering the global average, the December 2009–February 2010 period still ranks as the 13th warmest in the last 131 years ( Rates of warming in the Northern Hemisphere have slowed somewhat in the last decade, even though the 9 of the 10 warmest years on record occurred during this period. This phenomenon has been largely attributed to a temporary decrease in the water vapor in the lower stratosphere which acts as a greenhouse gas (Solomon et al. 2010). Because of this natural short-term variability, some scientists prefer to report the averages of projections for 20 to 30 years, and this is done for the projections presented here.

Changes in the amount and spatial distribution of precipitation are still poorly understood and thus difficult to project (Solomon et al. 2007), especially against a background of substantial year-to-year variability. This is largely because precipitation is controlled by complex interactions between global and hemispheric-scale circulation features and ocean-and-land-surface atmospheric interactions that occur across a range of spatial and temporal scales. Moreover, the complex terrain of the study region will likely alter the impact of any broad-scale shifts in precipitation pattern. Predicting future precipitation is further complicated by the fact that human activities may also be altering natural controls (e.g., ENSO and PDO) on atmospheric circulation patterns and storm tracks (Barnett et al. 2008; Bonfils et al. 2008).

For total annual precipitation, the dominant pattern in North America projects a wetter climate in the northern tier and a drier climate in the southwestern United States (fig. 34, middle row), with small (≤10%) but important changes in annual precipitation in much of the four climate regions, although individual models (not shown) exhibit a range of projected changes. The importance of even modest changes relates to the timing of increased or decreased precipitation. Summer precipitation is expected to decrease for much of the western United States, causing increasingly dry warm-season conditions, whereas increased precipitation in the winter could result in dramatically higher volume streamflows in parts of the Upper Columbia Basin and northern U.S. Rocky Mountains. While models are in better agreement for projected increased winter precipitation for parts of the northern Rockies and the Upper Columbia Basin and decreases in the Southwest, overall uncertainty remains high (Solomon et al. 2007). There is only weak agreement among the models as to whether annual precipitation will increase or decrease (fig. 34, bottom row), but there is an indication of a seasonal decrease in summer precipitation for parts of the four climate regions, and an increase in winter precipitation (and more agreement among the models for the latter). In the central Rockies and GYA, model results vary widely, and it is unclear how conditions might change in coming decades. In addition, all of these areas feature pronounced natural variability at multiyear to multidecadal scales, and this natural variation may mask or enhance the regional expression of any broad-scale precipitation trends.

The CMIP3 projections, which document a broad spatial scale of warming, are at large resolution (e.g., 200-kilometer [124-mi] grid) that is of limited use for assessing impacts more locally. Hence, we present results from downscaled modeling efforts more relevant for regional or local planning: westwide at 4 kilometers, temperatures at specific sites driven by three emissions scenarios (B2, A1B, and A2) for three future periods, and projections of mid-century conditions for three climate and hydroclimate variables based on the A1B scenario (rapid economic growth, global population that peaks in mid-century and declines thereafter, and the rapid introduction of new and more efficient technologies).

Westwide climate: Statistically downscaled projections

For much of the West, GCMs project about a 2ºC (4° F) rise in temperatures for 2050 (the orange shading in fig. 34, top row), with somewhat less warming near the Pacific coast. As part of a project for the U.S. Fish and Wildlife Service, NOAA used statistically downscaled projections to illustrate what the projected rise in temperatures would mean for the western regional climate compared to the existing north-south and elevational gradients of climate in the West (Ray et al. 2010). Downscaled temperature data from the CMIP3 22 model average projection for the A1B emissions scenario (from IPCC AR4) were added to the PRISM climatology (~4 km, 2 mi) for the June–August season. This downscaling method makes minimal, physically based corrections to the global simulation while preserving much of the statistics of interannual variability in the climate model (described by Salathé 2005). This method is similar to the so-called “delta method,” in which the temperature changes (the “deltas”) from GCMs are spatially interpolated and added to a high-resolution climatology.

The maps depict average daily temperature for the northern and central U.S. Rocky Mountains and Greater Yellowstone Area (fig. 35) and the southern Rockies (fig. 356) for the 1950–1999 climatology and projections for 20-year averages around 2025, 2050, 2090. These graphics illustrate that at large spatial scales, by 2050 the projected changes in summer climate can be visualized as a shift of temperature zones northward and upward in elevation (3rd panel in each figure). This shift of temperature zones continues through the end of the 21st century (lower panel in each figure). These maps do not illustrate the year-to-year or day-to-day variability that will also occur. Furthermore, there are a number of unknowns about how climate effects may reduce or amplify the large-scale pattern of widespread warming that is projected over the western United States. It is unclear how the details will play out at any given location.

Figure 35. Summer observed average temperatures and statistically downscaled projections for the northern and central U.S. Rockies and Greater Yellowstone Area (left) and Southern Rockies (right). Observed average June–August temperature for 1950–1999 (top panel). Projections were calculated by adding the multi-model average temperature changes to the 4-km PRISM climatology. Observed climatological averages are from PRISM (DiLuzio et al. 2008), projected changes from the IPCC (CMIP3) 22-model average for the A1B emissions scenario. Source: Ray et al. 2010; used with permission.

Climate projections downscaled to specific alpine sites

As part of the USFWS project, NOAA also generated temperature projections statistically downscaled to 22 mountain ranges in the western United States (Ray et al. 2010). Graphics for four sites not illustrated in that report are presented below (Glacier National Park and the Gallatin Mountains, Montana; Niwot Ridge, Colorado; and Clearwater Mountains, Idaho). This analysis illustrates implications of model-projected changes for the seasonal cycle, the relationship of projected climate change to historical climate variability, the spread of the individual model projections, and the evolving nature of the ensemble of projections throughout the century. This analysis used a modified version of the statistically downscaled CMIP3 Climate Projections created by the Department of the Interior Bureau of Reclamation and the University of Santa Clara. The statistical downscaling technique is known as “bias corrected spatial disaggregation” (BCSD) and was originally developed for hydrologic impact studies (Wood et al. 2004; Maurer 2002). This dataset downscales the projections to a 1/8º (12-km, 7-mi) grid (see details in Ray et al. 2008), which we adapted to the 4-km (2-mi) PRISM climatology (DiLuzio et al. 2008) to be smaller in scale for ecological applications in mountainous regions.

The resulting estimates adjusted to PRISM are among the best inferences available for temperature at this scale, albeit representative of a 4-km (2-mi) average and based on interpolation from station observations that may be distant from the grid box (See Ray et al. 2008 p 30). The results are shown in Figures 36–37. Projected temperatures from the BCSD/PRISM downscaling for the A1B emissions scenario are shown in red (thin lines) with the average projection (heavy line). (A1B scenario = rapid economic growth, global population that peaks in mid-century and declines thereafter, and the rapid introduction of new and more efficient technologies.) For comparison purposes, the 1950–1999 PRISM climatology of the monthly average temperature (solid black line) and the 10th and 90th percentiles (dashed black lines) are also shown. These percentiles represent the five warmest and coolest months from 1950 to 1999.

At all four sites (as well as the other 16 not shown here), the temperature increases are largest in summer. The July temperatures from almost all the model projections at these sites lie at or above the 90th percentile of the present climate. Most of the projections suggest that typical summer temperatures will equal or exceed the extreme warm summers of the last half of the 20th century. The projected temperature changes are somewhat smaller in winter and the year-to-year variations are larger. While the proportion of warm winter months is projected to increase, most years, even by 2050, will not be above the 90th percentile of the present climatology. Winter warming will be manifest in the relative absence of months colder than the current average and in the cumulative effects of consecutive warm winters, with an increase in the number of extreme warm winter months.

The spread of the light red lines in figures 36–37 indicate the range of the individual models and the uncertainty in the projection of 20-year average climates, even for a given emissions scenario. High-end projections are approximately 1°C warmer than the multi-model average, and would indicate increased risk at a number of sites. However, the uncertainty implied by this spread may be larger than the true uncertainty due to differences in climate sensitivity among the models studied. The spread about the average projection is a result of two factors: differences in model climate sensitivity (the response of a particular model to climate forcing) and model-simulated multidecadal variability. That is why many scientists prefer to emphasize the multi-model average projection. Because the BCSD/ PRISM downscaling method is based on the CMIP3 projections, the multi-model average projections shown in these figures are consistent with the large-scale patterns of warming in the GCM temperature change maps (see fig. 34, top panel).

The overall pattern that emerges is for hotter summers and somewhat warmer winters. The 2050 summer projection is consistently about 3°C (5°F) higher than in recent climatology, which is the westwide projected increase. The low model projection (the 10th percentile of the distribution) is in most cases higher than the 90th percentile of the recent climatology, suggesting that the coolest summers of the mid-21st century will be warmer than the warmest summers of the recent past. Precipitation projections are not provided, but a recent similar downscaling effort for Colorado found that, unlike temperature projections, potential changes in precipitation are smaller than the year-to-year and decade-to-decade variations observed in the historical record (Ray et al. 2008).

Figure 36. June–August 20-year temperature projections centered on 2025 (left panels), 2050 (right panels) for Glacier National Park (top panels, elevation 1866m) and the Gallatin Mountains, Montana (elevation 2778m, bottom panels) for a 4-km grid cell (approximately 30 x 40 mile). Each graphic compares observed monthly average temperatures to projections for the period. The observed monthly averages (solid black) and 10th and 90th percentiles values (dashed black lines) are based on observations during 1950–1999. Projected monthly climatologies (thin red lines) are from the multi-model ensemble for the 20-year period centered on 2050. The average of the projections is shown as a heavy red line. Data are derived from Maurer 2007. Note that the magnitude of projected temperature change is comparable to or greater than variations in the historical record. Source: Ray et al. 2010, used with permission.

Figure 37. June–August 20-year temperature projections centered on 2025 (left panels), 2050 (right panels) for Niwot Ridge, Colorado (top panels, elevation 3267m) and Clearwater, Idaho (bottom panels, elevation 2467m) for a 4-km grid cell (approximately 30 x 40 mile). Each plot shows observed monthly average temperatures compared to projections for that period. The observed monthly averages (solid black) and 10th and 90th percentiles values (dashed black lines) are based on observations during 1950–1999. Projected monthly climatologies (thin red lines) are from the multi-model ensemble for the 20-year period centered on 2050. The average of the projections is shown as a heavy red line. Data are derived from Maurer 2007. Note that the magnitude of projected temperature change is comparable to or greater than variations in the historical record. Source: Ray et al. 2010, used with permission.

Model projections of future climatic and hydrologic conditions

As part of a larger U.S. Forest Service and U.S. Fish and Wildlife Service project evaluating future climate conditions, the Climate Impacts Group (CIG, University of Washington) is modeling future climatic and hydrologic conditions (e.g., SWE, soil moisture, and potential evapotranspiration) for much of the western United States. A selection of preliminary results for a few key variables from this study is shown below (figs. 39–41). These projections will be refined and evaluated before publication in a CIG/USFS/USFWS report scheduled for later in 2010.

Downscaled model methodology

The CIG project applied a range of climate change projections from the WCRP CMIP3 multi-model dataset used for IPCC AR4 (Solomon et al. 2007) to hydrologic model simulations and evaluated the impact of climate change on the hydrology of the region (after Elsner et al. ). These models were drawn from a common set of simulations of 21st century climate archived from 21 GCMs (Mote and Salathé 2010), using greenhouse gas emissions scenarios as summarized in the IPCC’s Special Report on Emissions Scenarios (SRES) (Nakićenović and Swart 2000). CMIP3 simulations were archived predominantly for three SRES emissions scenarios (A1B, B1, and A2) for most of the 21 GCMs, with A2 following the highest trajectory (most warming) for future CO2 emissions at the end of the 21st century. This work focuses on A1B (moderate warming) because it was simulated by the most GCMs, and our study focuses on mid-21st century change, at which point none of the scenarios is consistently the highest and for which a larger source of uncertainty is the variability in GCM models. We chose to use the eight GCMs for this study that have the best fit to the observed seasonal cycle of climate as well as the lowest bias for the observed precipitation and temperature records in all three modeled basins: Columbia, Missouri, and Colorado.

The spatial resolution of GCM output is generally too coarse to be meaningful for hydrological studies. Therefore, we downscaled the GCM output to 1/16º (~6 km, 4 mi) spatial resolution and applied a delta method approach to develop an ensemble based on the average of the eight scenarios (see e.g., Hamlet and Lettenmaier 1999; Snover et al. 2003). In the delta method, projected changes in precipitation and temperature as determined by GCM simulations are applied to the historical record at the resolution of hydrologic models. We performed hydrologic simulations using the historical record perturbed by these monthly changes in the Variable Infiltration Capacity (VIC) model (Liang et al. 1994; Nijssen et al. 1997) at 1/16º (~3.7 km, 2.3 mi) latitude by longitude spatial resolution over the entire region. The VIC model is a macroscale model intended for application to relatively large areas, typically from 10,000 km2 (3,861 mi) to continental and even global scales. A key underlying model assumption is that sub-grid scale variability (in vegetation, topography, soil properties, etc.) can be modeled rather than represented explicitly.




Figure 38. Estimate of (a) mean SWE, (b) mean soil moisture (June–August), and (c) mean potential evapotranspiration (June–August) for 1916–2006 and 2030–2059. Data derived from an ensemble of 8 GCMs that perform the best over the four climate regions. The temperature and precipitation data from the GCMs is used to drive the VIC hydrologic model, and this gives the SWE values. Great Basin and lower Colorado are modeled at 1/8º (~12 km), the rest at 1/16º (~6 km). Source: University of Washington Climate Impacts Group (USFS, and USFWS, in prep.); used with permission.
Climate projections for the western United States

Climate conditions

Increasing temperatures (figs. 34–37).

Increased but highly variable precipitation for parts of the Upper Columbia Basin and the northern and central U.S. Rocky Mountains (figs. 34, 38).

Highly variable annual precipitation for parts of the central and southern U.S. Rocky Mountains, with possible decreases; however, recent changes in most regions are not significant (figs. 34, 38).

Increased evapotranspiration for most of the western United States which is unlikely to be offset by increased precipitation (figs. 31, 38; Hoerling and Eischeid 2007, Seager et al. 2007).

Surface hydrology

Larger proportion of winter precipitation falling as rain rather than snow (Knowles et al. 2006; Bales et al. 2006).

Decreased snow season length at most elevations (Bales et al. 2006).

Less spring snowpack (fig. 38; Pederson et al. submitted; Mote 2006, 2003; Mote et al. 2005).

Earlier snowmelt runoff (Stewart et al. 2005, 2004; Hamlet et al. 2005; Clow 2007).

Increased frequency of droughts and low summer flows (Gray and Andersen 2010; Meko et al. 2007).

Extreme conditions: droughts, floods, heat waves

More episodes of extreme temperatures (Parson 2001; Karl et al. 2009).

Increased frequency of extreme precipitation events, rain-on-snow events, and consequent winter and spring floods in mountains (Madsen and Figdor 2007; Groisman et al. 2005; Kunkel et al. 2003).

More frequent dry periods as a result of increased temperatures, evapotranspiration, and changes to surface hydrology (Gray and Andersen 2010, Meko et al. 2007).

Productivity and phenology

Earlier blooming dates for many plant species (by as much as two weeks; Cayan et al. 2001; Schwartz and Reiter 2000).

Longer growing season and increased productivity where moisture/soil fertility and other factors are not limiting (Bales et al. 2006).


Higher frequency of large fires, longer fire seasons, and increased area burned by wildfires in the western United States (Westerling et al. 2006; Morgan et al. 2008; Littell et al. 2008, 2009a; Spracklen et al. 2009; Higuera et al. 2010).

Greater drought stress will likely result in more insect infestations and disease affecting forests (Black et al. 2010; Nordhaus 2009; Romme et al. 2006; Logan et al. 2003; Carroll et al. 2004; Breshears et al. 2005).

Planning for the future

Planning for future conditions that are highly uncertain presents a significant challenge for land managers. However, techniques developed for business, finance, and military applications offer a roadmap for planning in the face of large uncertainties. “Scenario planning” is one such approach, and it uses a combination of scientific input, expert opinion, and forecast data to develop alternative scenarios for the future (Schwartz 1991; van der Heijden 1996). This contrasts with more traditional attempts at developing precise, quantitative assessments of future conditions, which are often of limited value for understanding climate change because of compounded uncertainties. In scenario planning, a suite of alternative scenarios can be used as a starting point for exploring species or ecosystem vulnerabilities under a wide range of possible future conditions, and as a means for examining how management strategies might address multiple drivers of change.

Jackson et al. (2009a) developed an example to illustrate this process in which alternative futures are arrayed along two axes: integrators of potential climate change (drought frequency) and potential changes in disturbance regimes (fire size). In concert with monitoring and modeling, studies of past climates can define the range of drought frequency we might reasonably expect, and studies of fire history can place bounds on potential fire size. This exercise yields four quadrants, each comprising a distinct combination of climatic and fire-regime change (fig. 39). Each quadrant provides a contrasting scenario or “storyline” for exploring potential impacts on species or ecosystems and examining the relative costs and benefits of various mitigation and adaptation measures.

At one extreme, major climate change and altered disturbance regimes interact to drive emergence of novel ecosystems. Given limited experience with ecosystem turnover in many of the climate regions, consideration of long-term paleoenvironmental records serves as a primary means for adding context to scenarios. It also helps to determine the likelihood of any of the four quadrants. For example, transition to “novel ecosystems” is analogous to the transition evident 11,000 years ago in Yellowstone when open tundra vegetation was replaced by closed forests (Millspaugh et al. 2000), while the transitions to “inevitable surprises” are analogous to the late Holocene changes in fire regimes (Whitlock et al. 2003; Romme and Despain 1989). The greatest value in scenario planning comes from uncovering vulnerabilities and potential responses, particularly those common to the range of conditions characterizing a set of scenarios. Hence, despite high levels of uncertainty, scenario planning may reveal that management response may be similar for a wide range of possible outcomes. Managers can then move beyond the challenges presented by an uncertain future to identify management responses that address ecological responses to a range of climatic conditions.

1   2   3   4   5   6   7   8   9   10

Verilənlər bazası müəlliflik hüququ ilə müdafiə olunur © 2016
rəhbərliyinə müraciət