A  suite of tools have been developed for use by assessment teams, including scenario-driven simulations from the Canadian (CGCM1) and UK Hadley (HadCM2) general circulation models (GCMs) as one way to provide projections of climatic conditions in the 21^st century. GCMs are physically based numerical models that utilize the fundamental laws of physics to describe motions and heat transfer in the atmosphere and ocean. Today's GCMs couple atmosphere, land surface, ocean, and sea ice to provide a comprehensive representation of the climate of the Earth.

Simulations of 20^th century climate include consideration of known changes in greenhouse gases and sulfur emissions, while scenarios of the 21st century employ projections of these emissions for the future. For the simulations chosen by the National Assessment, both CGCM1 and HadCM2 models are driven by a 1% per year compounded increase of the CO₂ concentration (to represent the effects of all greenhouse gases), and by sulfur emissions that are projected to double by the end of the 21^st century. These projections are consistent with the changes projected by the Intergovernmental Panel on Climate Change.

In response to the resulting changes in atmospheric composition, both models indicate that the climate will warm significantly, with global precipitation also increasing. For the National Assessment, the focus is on what is projected to happen over the United States. Maps of projected changes for the U.S. are viewable over the Web at http://www.cgd.ucar.edu/naco/vemap/trends.html.

The projected changes in precipitation over the US are of particular interest (Figure 1). Both models project large increases in precipitation in Southern California and the Southwest. However, the CGCM1 projects a decrease in precipitation in the Southeast while the HadCM2 projects an increase. Understanding why these changes, and these differences, are occurring is one step in seeking to project the consequences of climate change. This article focuses on key contributors to precipitation variations over the U.S..

Mid-latitude Storms

There are some robust signals of global warming that turn up in nearly all GCM simulations of the 21^st century and that have particular consequences for storms (mid-latitude cyclones), especially during winter. With increased greenhouse gas concentrations, nearly all GCMs (including the CGCM1 and HadCM2) undergo enhanced warming at high latitudes in the lower atmosphere (because of sea ice-albedo feedback effects and weakening of the near-surface inversion); enhanced warming at low latitudes in the upper atmosphere (because of how vertical atmospheric structure in the tropics is determined); and greater warming over the land than over the ocean (because of evaporation, vertical mixing of heat and the ocean's larger heat capacity).

The number of storms globally is dependent upon, among other factors, the temperature change between the pole and equator; the strength of storms is dependent upon the amount of moisture in the atmosphere. A smaller pole-equator temperature gradient results in fewer mid-latitude storms, while more moisture in the atmosphere provides more energy for storms that do occur. Furthermore, the region of storm formation off the East Coast of the U.S. is locally dependent upon the land-sea temperature gradient. Warm Gulf Stream waters and a cold land surface in winter provide ideal conditions for generating many storms. With warming of the land surface in winter, the intensity of these storms could be reduced. Both models confirm a decrease in the number of East Coast storms, though some individual storms appear to be more intense. However, recent studies have shown that there is no decrease in East Coast storms over the past 100 years, but rather an increase through the 1960s.

So what causes the different precipitation anomaly patterns that the two models project for the Southeast? One contributing cause is the difference between the two models in their projection of the position of the East Coast storm track. While the storms in HadCM2 track further north and east over the Atlantic Ocean, the storms in CGCM1 track closely along the coast itself. As a result, even though both models show a decrease in the number of storms along the East Coast, the effect of that decrease is felt over the U.S. only in the CGCM1 model. Observations indicate that the storm tracks should extend along the southeastern coast of the U.S., so in this particular region, they are more accurately located in CGCM1, although they are overrepresented to the south.

ENSO

El Nino/Southern Oscillation (ENSO) is a major driver of the tropical and global circulation. During warm ENSO events (El Nino), the waters off the coast of Peru warm up, changing the atmospheric and oceanic circulation in the Pacific region. These changes then affect global weather patterns and the position of the jet stream over North America. Although the effects of global warming on ENSO are highly uncertain because of the limited ability of GCMs to simulate ENSO variability, some models, including the CGCM1, suggest the possibility of a more persistent El Nino state. Regardless, both models produce a southward-shifted jet stream, stronger Aleutian low, and a weaker subtropical high over the Pacific, which may be the result of a warm ENSO phase (El Nino). Together, these circulation changes along with warmer sea surface temperatures (SSTs) are associated with an increase in storms that penetrate inland further south along the West Coast. It is thus not surprising that the precipitation changes projected by the models include a large swath of increased rainfall projected over the eastern Pacific Ocean, extending into the southwestern part of North America.

Land Surface

Many of the precipitation changes in the GCMs, particularly during summer when the atmospheric circulation is weaker, are the result of feedbacks from the land surface. During winter, snow cover is the mechanism for this interaction, while during summer, soil moisture is most important. Over the ocean, warming generally correlates with increased precipitation because there is greater evaporation. However, because of the limited moisture-holding capacity of land, warming over land may lead to increased or reduced precipitation, depending on a range of factors such as how soil moisture budgets are calculated. Soil moisture anomalies generally correlate with precipitation anomalies during winter. During summer, however, the increase in evaporation is sometimes enough to cause drying even in areas that get more precipitation than present, which could then limit the amount of additional precipitation. Vegetation is another important feature of the land surface, although human-induced reductions in future vegetation cover are not yet being treated in climate models.

Lessons

These examples illustrate the importance of careful analysis and taking proper cautions when interpreting results from GCMs. The storm track analysis of model results shows that the same physical mechanism can lead to opposite outcomes when imposed on a slightly different base state. The uncertainty in predictions of changes in ENSO variability also necessitates caution in projecting the nature of future changes in ENSO or the resulting effects of those changes on North America.

Even with these and related limitations, however, there is good reason to believe that there will be significant changes in precipitation, and thence in the availability of water resources. The implications of these changes in precipitation are explored in detail in the water resources sector of the National Assessment.

For more information, contact:

Benjamin Felzer, NCAR; 1850 Table Mesa Drive, Boulder, CO 80307; phone: (303) 497-1703; email: felzer@ncar.ucar.edu.