Science Briefs

Northern Hemisphere Sensitivity to Sea Surface Temperature Change

Fig 1
Fig. 1: Marine (red circles) and coastal (yellow squares) sites where late-glacial oscillations have been measured.

The sensitivity of our modern and past climates to changes in ocean temperature is not well known. In contrast, sensitivity tests to other known climatic forcings such as ice sheets, albedo, carbon dioxide, and methane have been performed more frequently in the global climate modeling community. For the experiments described here, we tested the sensitivity of the northern hemisphere climate to a cooling of -2°C in the North Pacific Ocean. This particular forcing was selected because of evidence for such a change within the paleoclimate record about 11,000 14C years ago, during the Younger Dryas (YD) climatic reversal.

Oceanic records collected since 1987 have indicated that the North Pacific Ocean may have experienced either a cooling or a salinity/meltwater effect during the YD. Figure 1 shows marine and coastal terrestrial sites where late-glacial oscillations have been found. Very detailed, well-dated records from the Gulf of California to the central North Pacific support the evidence of North Pacific circulation changes and widespread climatic variability, not only during the YD interval but also in previous millennia. High-resolution records from coastal Alaska also support these rapid oscillations in climate.

The general circulation model (GCM) used for these experiments is the GISS Model II originally described by Hansen et al. (l983). The model solves the equations of mass, energy, momentum and moisture and it calculates radiative fluxes, surface fluxes, latent heat and cloud cover and it has been used in previous GISS experiments to simulate the effects of a colder North Atlantic during YD time (Rind et al. l986). The control run is represented by a five-year simulation of the modern climate.

As the simplest experiment for testing the sensitivity of the model to North Pacific temperature depression, we changed the sea surface temperatures (SSTs) north of the equator by -2°C relative to the current values and by -1°C relative to current values south to 8°S.

Fig 2
Fig. 2: Change in annual surface air temperature due to colder North Pacific. Blues indicate a decrease and yellows and reds an increase.
Fig 3
Fig. 3: Change in percentage annual snow cover due to colder North Pacific. Blues indicate a decrease and yellows and reds an increase.

The primary result of the experiments is that colder North Pacific SSTs have a major effect on northern hemisphere surface air temperature (see Figure 2). In contrast to the earlier experiment in which only North Atlantic SSTs were cooled and a downstream effect over land was noted, here the cooling extends throughout the northern hemisphere. The cooling over North America is 1.5-2.5°C in most locations; this is 3-5 times the GCM's grid point standard deviation.

A variety of positive feedbacks contribute to the cooling, such as increases in reflective low level and middle level cloud cover, decreasing amounts of water vapor, and increasing snow cover in mid-latitudes (see Figure 3).

The key element in explaining the meteorological processes that cause the signal in the North Pacific to propagate around the globe is water vapor. The colder SSTs cause evaporation from the ocean's surface to decrease. Thus, drier air blows across the North Pacific, and evaporation over North America also decreases. A decrease in precipitation over the northern hemisphere ocean also occurs, and there is less latent heat release, which cools the atmosphere. Increased sensible heat fluxes over land further lower the temperature over North America.

Along the east coast of North America, the cooling decreases because of the maritime influence of the North Atlantic. East of the Gulf Stream, sensible heat flux and evaporation changes are either not statistically significant or weaker, yet marginally significant surface cooling over Eurasia nonetheless occurs, mostly equatorward of 45°. This effect appears to be associated with an increase in cloud cover and reduction of sunlight absorption over most of Eurasia. However, the differences in cloud cover over North America and Eurasia must be interpreted with caution due to the primitive nature of cloud parameterization in this version of the GCM.

A second experiment in which we changed the ice age SSTs (as determined by the CLIMAP project) by -2°C in the North Pacific resulted in a similar pattern of weakened Pacific evaporation and humidity and cooler air temperatures which led to a cooling and increased snowfall over North America.

In contrast to previous sensitivity studies to a colder North Atlantic (Rind et al. l986), this North Pacific SST change produces an entire northern hemisphere response. The geographic extent (80 degrees of longitude) is almost twice that of the Atlantic at 50°N, thus a vast and extensive surface area. If this large area were subject to rapid fluctuations in SST, as is suggested by the paleoclimatic data, then according to the GISS model result, the northern hemisphere would have been significantly affected. Even when evaporation is less, colder northern hemisphere oceans consistently produce increases in snow cover in temperate latitudes immediately downwind of the ocean. This model result appears to be a valid one, based on marine and glacial evidence from the YD and from the last ice age.

To the extent that the North Pacific contributed to the great ice sheets of the last ice age, this result is also significant. Increased snow cover at mid-latitudes produces a positive feedback for ice sheet growth. The large sensitivity of the GISS model to a 2°C shift in SST suggests that we should reevaluate our estimates of cooling during the last ice age, as recent marine and terrestrial evidence points to significant cooling of the subtropics which CLIMAP (l981) did not portray.



Please address all inquiries about this research to Dr. Dorothy Peteet.