DATE: 1997 (in press) JOURNAL: Climate Change TITLE: OPPOSING SOUTHERN OCEAN CLIMATE PATTERNS AS REVEALED BY TRENDS IN REGIONAL SEA ICE COVERAGE AUTHORS: S.E.Stammerjohn and R.C.Smith Abstract The presently available 16.9 year sea ice record (October 1978 to August 1995) derived from satellite passive microwave data shows evidence of contrasting climate patterns in the Southern Ocean as indicated by persistent opposing trends in regional sea ice coverage. Strong increasing trends in total Southern Ocean annual and seasonal ice amount concurrent with decreasing trends in open water amount within the boundary of the ice edge have been detected. Taken individually, all subregions show similar trends except the Amundsen and Bellingshausen regions, where strong decreasing trends in sea ice and open water amount were detected. These trends have important implications for the southern hemisphere heat budget and surface albedo as well as for marine ecosystems associated with various sea ice habitats. Other evidence of contrasting climate patterns with respect to southern hemisphere atmospheric circulation is explored. Due to the relatively short sea ice record, it still remains to be seen whether these trends are indicative of global climate change or natural decadel variation. However, the persistent opposition in Southern Ocean regional ice coverage is noteworthy and needs to be studied using global circulation models in order to better define potential positive and negative feedbacks for global change scenarios. Introduction Sea ice plays an important role in the global climate system by influencing the regional heat budget, surface albedo, and consequently oceanic and atmospheric circulation. Several studies have suggested that long-term variations in sea ice coverage may be indicative of possible large-scale and long-term climatic change [. fletcher 1969, sissala 1972, budd 1975, kukla garvin 1981, fletcher 1982, zwally carbon 1983, zwally satellite 1983, parkinson satellite 1987, Jcarleton antarctic 1989, jacka climate 1990 .]. Other studies have detected quasi-periodic El Nino/Southern Oscillation (ENSO) signals in both polar sea ice covers suggesting linkages between interhemispheric climate trends and sea ice [. gloersen 1995, peterson white 1996, baker stammerjohn smith surface 1996 .]. The strong seasonality and interannual variability of sea ice also impacts the polar marine ecosystem, from primary production to the survival rates and distributions of prey and predators [. hunt ice-influenced 1991, ross quetin larval 1991, Fraser trivelpiece ainley warming 1992, Ainley ecological 1994, Karl baker smith fraser 1995, Siegel and Loeb 1995 .]. To date, attempts to find possible trends in polar sea ice coverage using various satellite data have made differing conclusions depending on the time period and sea ice parameter studied. For example, early analyses of Antarctic sea ice extent, using mainly visible and infrared satellite data, indicate an increase in ice extent from 1966 to 1973 [. sissala 1972, streten summer 1973, zwally carbon 1983 .]. Using Navy-NOAA (National Oceanic and Atmospheric Administration) weekly ice maps, a decrease in Antarctic sea ice extent during the mid 1970's was observed [. kukla garvin 1981 .], which was then followed by an observed increase from the late 1970's to the early 1980's [. zwally carbon 1983 .]. Using a longer time series of the Navy-NOAA weekly ice maps, an overall decrease in Antarctic sea ice extent was observed from 1973 to 1988 [. jacka climate 1990 .]. More recently, studies analyzing the 1978 to 1987 time series of Antarctic sea ice concentrations derived from multi-frequency passive microwave satellite data have found no significant increasing or decreasing trends in Antarctic sea ice extent, actual sea ice amount or open water area [. gloersen campbell 1991, gloersen campbell cavalieri 1992, parkinson southern 1992 .]. Because of the different types of data and variables analyzed in the above studies, it is difficult to determine whether in the combined observations from 1966 to 1987, there is any long-term decreasing or increasing trend superimposed on the apparent decadel variability. Annual and Seasonal Integrated Indexes Sea ice coverage in the Southern Ocean is marked by HIGH seasonal and regional variability, although total Southern Ocean sea ice extent varies less due to the circumpolar distribution of opposing sea ice anomalies [. zwally gloersen 1983, parkinson southern 1992, gloersen campbell cavalieri 1992, parkinson southern 1994 .]. ...... To determine which regions were contributing to the trends observed in total Southern Ocean seasonal ice area and open water area, linear regression analyses were performed on the regional seasonal ice indexes. Results were tabulated (Tables 1 and 2), and the seasonal indexes of ice area for all 6 regions are shown in Figures 4-9. (Figures of seasonal ice extent and open water area can be found in Stammerjohn et al., 1996.) Most notable about the regional analysis, best illustrated in Table 1, is the general opposition in trends between the two southeastern Pacific regions (the Amundsen and Bellingshausen) and the other four regions. The Amundsen and Bellingshausen (AB) regions show decreasing trends in ice extent, ice area and open water area (most significantly in summer), whereas the other four regions in general show increasing trends in ice extent and area (most significantly in non-winter seasons) but decreasing trends in open water area. With respect to trends in ice area, there appears to be no one season for all regions which shows a predominance of significant trends, although summer reveals strong but opposing trends in Weddell and AB ice area, while spring reveals increasing trends in Indian and Ross ice area, and autumn reveals increasing trends in Weddell and West Pacific ice area. It also appears that those regions (not including the Southern Ocean as a whole) and seasons which have strong trends in ice area, also have strong trends in ice extent, with the exception of the autumn ice extent time series for the Weddell and West Pacific regions. Whatever the causative factors influencing these trends, they appear to be strongest in non-winter months, since none of the regional time series show strong trends in winter ice area. It is also interesting to note that all regions show decreases in open water area in at least one season, regardless of whether sea ice extent and/or area is increasing, decreasing or showing no change. ... Other Evidence of Contrasting Climate Patterns Analyses of the 16.9 year annual and seasonal time-series of Southern Ocean regional sea ice coverage show: (1) a strong increasing trend in total Southern Ocean sea ice area, coincident with a decreasing trend in open water area within the sea ice extent boundary, (2) variability in trends with respect to season, where trends are most detectable in non-winter months, and (3) variability in trends with respect to region, where a persistent opposition exists between decreasing summer ice and open water area in the AB regions versus increasing ice area but decreasing open water area in non-winter months everywhere else in the Southern Ocean. How does this 16.9 year record relate to other longer term climatic records? Are these increasing/decreasing trends in ice area during non-winter months indicative of a climate trend, and if so, why the asymmetric response? Is the asymmetry due to differential response times between atmospheric and oceanic forcing on sea ice coverage, or to persistent regional climate patterns? The zonal non-uniformity of circumpolar ice extent anomalies is a well characterized feature of the Southern Ocean [. zwally comiso parkinson campbell carsey gloersen 1983, parkinson southern 1992, gloersen campbell cavalieri 1992, parkinson southern 1994 .]. As a consequence, it would not be surprising to observe asymmetry between ice anomalies in the AB regions versus the rest of the Southern Ocean over a year or two period. However, it is noteworthy that not only has the asymmetry been maintained over a 16.9 year record, but that it reveals detectable opposing trends in ice area. Jacobs and Comiso (1993) reported a record decrease in Bellingshausen (62-100\(deW) sea ice extent from mid-1988 through early 1991 which was most evident in the summer perennial sea ice extent. It now appears that the record decrease they observed may be part of a decadel or long-term trend. They show that this decrease coincided with (1) an increase in northwest surface winds (which would explain the concurrent decrease in open water area observed here), (2) an increase in cyclonic activity, and (3) a rise in surface air temperatures in the western Antarctic Peninsula region. They suggest that the persistence of the 3 year decrease was primarily due to ocean heat storage but that an increase in northerly air flow and warmer air temperatures could have contributed to the decrease in ice extent as well. Finally, Jacobs and Comiso (1993) mention that these findings are inconsistent with global warming models [. stouffer 1989, manabe annual 1991 .] which predict a fresher surface layer and thicker ice cover. Could it be that global warming has a two-fold effect in the Southern Ocean, whereby all regions experience colder conditions except the AB regions which experience warmer conditions? There is evidence in other studies of similar patterns in regional climate variables. For example, a recurring opposition in air temperature anomalies between Antarctic Peninsula and mainland stations has been observed in all seasons but spring [. rogers 1983 .]. ... The regional circulation features of the Southern Ocean have not yet been well identified, thus it is difficult to determine what is contributing to the opposing trends in Southern Ocean regional ice coverage observed here. However, the temporal trends and spatial patterns found in the 16.9 year record of Southern Ocean sea ice coverage appear to have existed before, if not always. Perhaps the increase in ice area observed everywhere in the Southern Ocean except in the AB regions for the 1978-1995 period is part of a long-term increase in ice coverage first observed in the 1966 to 1972 record [. streten summer 1973, sissala 1972, zwally carbon 1983 .]. Or perhaps the decrease in Southern Ocean ice extent observed in the 1973-1980 record [. kukla garvin 1981, zwally carbon 1983 .] suggests that long-term variability in the Southern Ocean consists of short-term decadel fluctuations. Decadel shifts have been observed in the southern hemisphere atmospheric circulation between the 1960's to 1980's [. trenberth 1979, vanLoon 1993 .]. Finally, the paleoenvironmental study of ice cores [. mosley-thompson thompson grootes 1990, mosley-thompson 1995 .] suggests that regardless of the temporal fluctuations, the spatial contrasting climate patterns in the Southern Ocean have been persistent features over much of the last five centuries.