IMPACTS Project Tasks
Recent observations show that ice sheets can respond to climate warming on decadal time scales and that the Greenland and West Antarctic ice sheets are losing mass at an increasing rate (Shepherd and Wingham 2007). The largest losses of ice are associated with the acceleration and thinning of large outlet glaciers and ice streams in Greenland and West Antarctica. Although surface atmospheric warming plays a role, ocean warming appears to be the primary driver (Bindschadler 2006). Subsurface melting of ice shelves has reduced their buttressing effect, leading to increased discharge of grounded ice (Shepherd et al. 2004). These changes suggest that recent IPCC sea level projections, which exclude large dynamic changes in ice sheets, underestimate the risk of rapid sea level rise.
The potential instability of marine ice sheets (i.e., ice sheets grounded below sea level) is of particular concern. The world’s only large marine ice sheet is the West Antarctic Ice Sheet (WAIS), the collapse of which would raise global sea level by about 5 meters. The WAIS has retreated substantially since the Last Glacial Maximum (LGM) (Conway et al. 1999) and may have collapsed completely during previous interglacials (Scherer et al. 1998). On several occasions since the LGM, sea level has risen at rates of more than a meter per century, possibly with significant contributions from Antarctic ice (Clark et al. 2002; Cronin et al. 2007). About a third of the WAIS lies in the Amundsen Sea Embayment (ASE), where outlet glaciers are weakly buttressed by small ice shelves and appear vulnerable to ocean warming.
The two largest ASE glaciers, Pine Island and Thwaites, have accelerated and thinned since the 1990s, and their grounding lines have retreated by several kilometers (Thomas et al. 2004). Early studies (Weertman 1974) suggested that marine ice sheets are unstable to small perturbations if the seabed deepens inland, as is the case for most of the WAIS. (See Figure 1.1.) As the grounding line retreats, the ice thickness at the grounding line increases, leading to increased outflow and thinning of grounded ice and further grounding line retreat. In model simulations, Thomas and Bentley (1978) found that this feedback could cause a hypothetical Ross ice sheet to collapse within decades. These authors used simple models that neglected certain stabilizing mechanisms (van der Veen 1985). However, recent theoretical work (Schoof 2007) suggests that grounding lines may, in fact, be unstable on beds that deepen inland and that WAIS instability is a real possibility. Similar concerns apply to several of Greenland’s largest outlet glaciers, which lie in deep subsurface troughs and have recently accelerated (e.g., Jakobshavn Isbrae; Joughin et al. 2004).
There is an urgent need to assess the likelihood of the abrupt collapse of marine ice sheets, especially in the ASE. This collapse would be abrupt in the sense that the ice sheet would cross a threshold to a new state on a time scale of decades, significantly shorter than the time scale of the cause (anthropogenic CO2 emissions). Collapse could be triggered by increased flow of warm intermediate water beneath ice shelves, leading to reduced buttressing and rapid grounding line retreat. Models and geologic data suggest that grounding lines could retreat by several kilometers per year, removing a significant fraction of the WAIS within a few decades. As a result, global sea level could rise at a rate of a meter per century or more, with devastating impacts on coastal cities and ecosystems.
Terrestrial ecosystems north of 45°N contain more than one-third of active terrestrial organic carbon, approximately 750 Pg C, most of which resides in the soil (Schlesinger, 1997). In addition, there are an estimated 455 Pg C in large peatland basins, and an additional 1000 Pg C in yedoma and non-yedoma permafrost (Gorham 1991; Zimov et al. 2006) This region has already experienced rapid changes in environmental conditions under current global warming, including: higher mean annual air temperatures and reduced snow cover (Euskirchen et al., 2007; Jorgenson et al., 2001; Osterkamp, 2005; Osterkamp et al., 2000; Taylor et al., 2006); increased runoff from higher precipitation and melting permafrost (Prather et al., 2001; Romanovsky et al., 2000; Vitt et al., 2000); larger CO2and CH4 emissions (Friborg et al., 1997; West and Schmidt, 1998; Whalen and Reeburgh, 1992; Zimov et al., 2006; ACIA, 2005; Davidson and Janssens, 2006; Prentice et al., 2001); and longer snow-free period and growing season (Chapin et al., 2005). Moreover, the spatial distribution of boreal and arctic vegetation types has started to shift with large consequences for snow cover and regional albedo (Chapin et al., 2005). In combination, these changes imply a strong positive feedback to increased climate warming through increased greenhouse gas (GHG) emissions, decreased albedo, and hydrology and ocean circulation changes (Chapin et al., 2005; Lawrence and Slater, 2005). These positive physical and biogeochemical feedbacks can, with high probability, cause a change in state over a period of 20 to 30 years in terrestrial ecosystems climate forcing that is potentially 2–3 times greater than is the change in radiative forcing from fossil fuel burning. The associated changes in terrestrial ecosystems composition, spatial distribution, and GHG dynamics are irreversible over millennia, comparable to the temporal scale of glacial-interglacial cycles. Although some degree of boreal/arctic feedback to warming is almost certain (indeed, has already been documented, see Chapin et al., 2005), there has been no comprehensive assessment of its likely magnitude—in part because the model capability has been lacking. Thus, we suggest there is a pressing need to extend state-the-art land surface models and apply them to a global modeling evaluation of global abrupt climate change from the biophysical and greenhouse gas feedbacks generated by vegetation range shifts, biogeochemical dynamics, and climate forcing in the boreal/arctic regions. Even if we project that these feedbacks will not achieve sufficient strength and speed to cause abrupt climate change, they will still be playing an important role in future climate and it will be valuable to quantify this amplification of anthropogenic radiative forcing.
There is lingering concern that vast methane stocks locked in a stable ice-like state at the bottom of the ocean (known variously as methane clathrate and methane hydrate) could be released in the future due to a warming ocean and/or ocean circulation changes. The current abundance of carbon stored in hydrates is generally believed to be greater than the recoverable stocks of all the other fossil fuels combined (Buffet and Archer, 2004; Gornitz & Fung, 1994), and methane is 72 times more potent as a greenhouse gas than is carbon dioxide over 20-year time horizons (IPCC, 2007a). There is evidence that methane hydrate releases have caused abrupt climate changes in the past, such as the Paleocene-Eocene Thermal Maximum 55 million years ago when the planet abruptly warmed 5-8K (Dickens, 2003). There is also disputed evidence that hydrate dissociation greatly amplified and accelerated global warming episodes in the late Quaternary period (Kennett et al., 2000). Whether or not a consensus exists regarding paleoclimate data, the stability of the contemporary hydrate inventory to the unprecedented temperature rise from anthropogenic emissions requires a careful assessment. Fortunately, our estimates for ocean-floor warming indicate that, for most of the ocean, hydrates are stable under the influence of moderate temperature changes (Reagan and Moridis, 2007; Archer, 2007) and massive methane releases remain a long way off (centuries to millennia). However, there are some regions with large hydrate abundances that are far less stable to climate change, notably the Arctic, which contains hundreds of Gton of methane with a time scale for release of decades (review by Archer, 2007; Reagan and Moridis, 2007), and the release is predicted to be abrupt at each location because the hydrates lie close to the edge of the gas hydrate stability zone defined by temperature and pressure (see figure 4.1). Plausible scenarios could lead to methane becoming more important than CO2 as a greenhouse gas on a time-scale of decades, with the associated warming leading to further hydrate dissociation, as well as terrestrial permafrost melting, that will release additional methane and be self-sustaining. In this project we will assess the strength and consequences of methane hydrate dissociation this century, including the likelihood we will cross a tipping point, and address the risks of shifting storm tracks, increased surface ozone, reduced stratospheric ozone, increasing the strength and frequency of the Arctic ozone hole, and the formation of oceanic dead-zones (hypoxia). (Wuebbles & Hayhoe, 2002)
Drought is a recurrent feature in many parts of the U.S. Multi-year droughts, in particular, are very devastating and costly. As stated by the Western Governors Association Report (2003), drought causes between $6 billion and $8 billion a year in direct estimated losses to the U.S. economy. Planning for multi-year droughts is the top concern of most regional, State, and Federal water managers. Recently, the IPCC Fourth Assessment (IPCC 2007a) projected warmer and dryer conditions for the subtropical regions in the future. Based on the IPCC climate simulations, Seager et al. (2007) discussed a tendency for more arid conditions in the southwestern U.S. in the 21st century that parallels the severity of the 1930s Dust Bowl. They described the warming/drying (or climate change type droughts) as associated with a northward shift of the storm tracks and expansion of the subtropical high in the mid-latitudes, which may be related to increased tropospheric static stability and stabilization of the jet stream in a warmer climate (Walker and Schneider 2006). This is in contrast to North American droughts in recent instrumental records that have all been attributed to atmospheric response to SST anomalies in the tropics (Hoerling and Kumar 2003). The possibility of climate change type drought in the mid-latitudes that intensifies as the climate gets warmer and warmer in response to increasing concentrations of greenhouse gases leads to the prospects of mega droughts that could last over a decade or more. Although greenhouse warming is a gradual change, the transition to mega drought could potentially be abrupt, as seen in paleoclimate records (Overpeck and Webb 2000; Meko et al. 2007).
For climate change type droughts that could possibly last over decades, terrestrial response and its feedback to the atmosphere could be particularly important in determining the severity of droughts (NRC 2002). In studying the 1998 Oklahoma-Texas drought, Hong and Kalnay (2002) showed that while SST anomaly and atmospheric internal forcing are equally important in initiating the drought, soil moisture and land-atmosphere feedback play a more significant role in maintaining the drought once it was initiated. Schubert et al. (2004) analyzed a number of century-long global climate simulations to examine the cause of the 1930s Dust Bowl. Their study suggests that the drought was mainly associated with tropical SST anomaly that generates a global-scale response in the upper troposphere that suppresses rainfall over the Great Plains. However, the severity of the Dust Bowl drought was reduced by 50% when soil moisture feedback was omitted in another set of model simulations. Breshears et al. (2005) reported rapid, drought-induced vegetation die-off during the recent 2002-2003 Southwest drought, which highlights the possibility of rapid terrestrial response that could trigger abrupt climate change.
The importance of land surface processes is further highlighted by discrepancies among model projections of future droughts in the Colorado River Basin. Milly et al. (2005) analyzed the runoff simulated by 21 GCMs and showed a 20% reduction in annual stream flow in the Colorado River Basin by the mid 21st century compared to the present. Christensen and Lettenmaier (2007) used a hydrologic model (1/8 degree resolution) driven by global climate simulations of precipitation and temperature and obtained only a 6% reduction in annual streamflow. Since the precipitation changes that drive the runoff reduction in the GCMs and the hydrologic simulations are similar, a possible explanation for the discrepancy between these two estimates hinges on land surface processes and model resolution. In the global climate simulations, warming causes higher evaporative demand that exacerbated the drought. In the hydrologic simulations, warming leads to earlier snowmelt, which reduces evaporative demand as soil moisture is available in spring when it is relatively cool, and wetter soil due to earlier snowmelt allows more water to go into runoff. Both of these processes counteract the effects of warming on evapotranspiration and runoff. Because of coarse spatial resolution, GCMs cannot capture the snowmelt response to warming in mountainous headwaters and simulated more intense drought conditions. Besides the complexity of land-atmosphere feedback, mega droughts may lead to widespread dust storms, similar to what was observed during the 1930s Dust Bowl (e.g., Woodhouse and Overpeck 1998).
Mineral dust can obscure solar radiation; through heating of the atmosphere, it can influence atmospheric stability, convection, and the general circulation. It was suggested that increased heating by dust from the deserts of western China, Afghanistan/Pakistan, and the Middle East could intensify the Indian summer monsoon (Lau et al., 2006; Lau and Kim, 2006a). Similarly, Saharan dust may have affected the climate and water cycle of the tropical Atlantic/Caribbean region (Lau and Kim, 2006b). Dust emission associated with mega droughts may potentially alter the North American summer monsoon and large-scale circulation through redistribution of heat sources and sinks, both vertically and horizontally, to affect the large-scale moisture transport to the Southwest and the Central US. Dust particles may also influence precipitation through aerosol indirect effects (e.g., Yin et al., 2000; Rosenfeld et al., 2001). Changes in the monsoon circulation and hydrologic cycle may provide a feedback mechanism to strengthen/weaken the mega drought and enhance/reduce dust emission.
This proposal element describes our goal and approach to study the potential influence of climate change on droughts. Our focus is on understanding processes that can potentially accelerate the transition to and sustain and intensify mega droughts in North America. Following the discussion above, we argue that large uncertainties in projecting climate change type droughts are related to the terrestrial response (including vegetation, surface, and subsurface processes, and dust emission) and its interactions with the atmosphere. Furthermore, models with higher spatial resolution and more realistic representation of terrestrial processes will likely provide more accurate prediction of future droughts. In what follows, we describe two hypotheses to investigate the role of the land surface on triggering and sustaining mega droughts, and outline our approach to test the hypotheses concerning mega droughts in North America.