by Balaji Rajagopalan and Yochanan. Kushnir

SOURCE: The Climate Report, Vol. 1, No. 2, Spring 2000. Copyright 2002, Climate Risk Solutions, Inc.  For more information contact Maryam Golnaraghi via Email: maryam@climaterisksolutions.com, or Tel. 617.566.0077.

Extreme temperature events such as hot and cold spells have been related to impacts ranging from adverse effects on crop yields to increased demands for energy. The extent of these impacts depends on strength, length and persistence of these extreme events.

Various scientific studies indicate that the variability of such extreme events across the US, are linked in varying degrees to well known large scale patterns of Sea Surface Temperatures (SSTs) and Sea Level Pressures (SLPs) such as El Niño Southern Oscillation (ENSO), Pacific Decadal Oscillation (PDO) or the North Atlantic Oscillation (NAO). [For definitions of ENSO, NAO and PDO see Some definitions at the end of this article]. These large-scale climate patterns, depending on their relative strengths, change the magnitude of incursions of warm tropical air and cold polar air masses over the continental US, thereby, significantly affecting the temperature spells across the nation. Given this, it is critical to understand these linkages and the impacts of such global climate patterns on the frequency and intensity of these extremes in different regions in the US. Ultimately, this together with advancements in forecasting phenomena such as ENSO, PDO and NAO could lead to capabilities to predict future changes in climate extremes.

Here we report the results of a study in which we have examined the properties of daily temperature fluctuations in the Northeast US and investigated the relationship between these fluctuations and global climate variability.

Methodology and data sources

For this study, we have used three different datasets in our analysis. This included the daily maximum and minimum temperatures from 1951 to 1998, which we obtained from the US National Climate Data Center (NCDC). Specifically, for this study we chose a subset of 23 stations to represent the Northeast region of the US (37°- 47° N and 70°- 82° W). We also retrieved gridded monthly anomalies of global SST and SLP from the data library of Lamont-Doherty Earth Laboratories of Columbia University. [A list of data sources is provided in the Suggested references and web sites Section at the end of this article]

We have examined daily maximum and minimum temperature fluctuations in the context of the number of days per season with temperature exceeding or falling below a certain threshold. Specifically, we have focused on winter and summer seasons, defining months of December, January, and February as the winter season, and June, July and August as summer. In addition, we define extreme hot (or cold) days as days during which the maximum temperature exceeds (or falls below) a given threshold. Similarly, we define extreme warm (cold) nights on the basis of the daily minimum temperature. We also have considered temperature thresholds separately for each station and season. Extreme warm and cold thresholds correspond to the 75-th and 25-th percentile values of the temperature distribution, which we compute from the entire pool of daily temperature data (48 years). At each station, for each year and season we compute the "spell days", which we define as the number of days with a temperature above or below the threshold values.

To determine the main patterns of year-to-year spell variability within the 23 Northeast stations we apply principal component (PC) analysis to the spell variable data (which are a function of the station number and the year). We have found out that for all seasons and all spell variables, a single PC dominates the variability. This implies that all 23 stations constitute a coherent set with a large degree of interdependency. In other words, the first PC can be thought of as a spatial average of the 23 stations. Therefore the time series of the first PC provides a unique index of extreme temperature fluctuations from one year to another over the entire region. We also use the equatorial Pacific Ocean SST and North Atlantic Ocean SLP data for the 48 year period to computes indices for ENSO and NAO, respectively. We use an ENSO index based on the sea surface temperature anomaly index in Niño-3 region (5°S-5°N and 90°W-150°W) in the tropical Pacific Ocean. The NAO index is defined as the difference between the SLP of Azores and Iceland. We correlate the first PC of the hot and cold-spell days to the ENSO and NAO indices. The spatial pattern of the correlations will then reveal the type of large scale climate forcing.

What drives temperature extremes in the Northeast U.S.?

A summary of our results is provided below:
1. Wintertime hot and cold spells are strongly related to wintertime pattern of NAO in the Northern Atlantic Ocean.

Specifically, we have looked at the correlation between the PC of hot-spell days and the winter NAO index. Over the 48-year period of analysis, we find that the correlation between the PC of hot-spell days and winter NAO index is 0.65 (statistically significant at 95% confidence level). As shown in Figure1, the time series of the PC follows closely the up- and down swings of the NAO index including the positive trend in the recent decades. This means that by knowing the strength of NAO index, one could explain approximately 42 percent of the variability in the hot-spell days across the northeastern US.

The results for the wintertime cold-spell days are almost a mirror of that of the hot-spell days. Thus, overall, the Northeast wintertime temperature spells are governed by
fluctuations in the strength of the dominant pressure patterns over the Atlantic (i.e., the Icelandic Low, and in particular the Azores High).

These relationships have implications for predicting wintertime hot and cold spells in the Northeast. However, to exploit these relationships, the NAO index needs to be predicted with reasonable skill. This also has implications for predicting snow amounts in the region. The snow amounts are strongly related to the temperatures, i.e., the temperature determines the form of precipitation (rain or snow). Given the strong link between temperature spells and NAO, it is consistent that a strong relationship also would exit between the snow amounts and NAO. We have found that correlation of PC of the wintertime snow amounts in the northeastern US with the ENSO and NAO indices reveal a similar pattern to that obtained for the hot-spell days.

Figure 1: Time series of PC1 of winter hot spell days (heavy line) and the winter NAO (dashed line). The correlation is about 0.6.

2. Summertime warm spells are related to the preceding winter ENSO phenomenon and the cold spells are related to the preceding winter NAO.

Summer hot-spell days have a strong correlation with the SST index of ENSO (Figure 2), indicating that an El Niño winter is followed by a summer with more hot days. Examination of the relationship between the time series of ENSO SST index and PC of summertime hot-spell days indicates a correlation of about 0.5. This means, that 25 percent of variability in the summertime cold spells can be explained by the ENSO index in the preceding winter. This lag relationship of more than one season provides hopes for predictability. On the other hand, there is a correlation of 0.4 between the summertime cold-spell days and wintertime NAO index, which suggests that in the northeastern US, there is an increase in the number of cool summer nights following a winter during which the NOA index is negative (Negative NAO index means positive anomaly over Greenland and negative anomaly over the Azores).

Figure 2: Time series of PC1 of summer hot spell days (heavy line) and the preceding winter NINO3 index (dashed line). The correlation is about 0.4.

3. We also have found that summer temperature fluctuations are less coherent with global climate variability. However, we find convincing evidence for a lingering effect from winter into summer in which a wintertime El Niño results in an increase in the number of warm days during summer, and a negative NAO winter leads to an increase in the number of cool summer nights.

4. Finally, the summertime temperature spell relationships to winter ENSO and NAO indices are quite interesting and certainly offer hope for some predictability. However the dynamical links for these relationships are far from clear and require further investigation.

Some definitions

ENSO: El Niño-Southern Oscillation is a coupled ocean-atmosphere phenomenon centered in and over the tropical Pacific and refers to large-scale fluctuations in a number of oceanic and atmospheric variables (e.g., SST, sea level pressure, rainfall, etc.). El Niño and La Niña episodes are the opposite extremes of the ENSO phenomenon. During an El Niño, above normal sea surface temperatures extend across the central and eastern equatorial Pacific Ocean. During a La Niña, below normal SST extend through much of central and eastern tropical Pacific.

NAO: North Atlantic Oscillation is referred to the pressure sea-saw between Azores and Iceland. Generally the pressure in the subtropical Atlantic Ocean is high (centered around Azores) and low in the mid-latitude region centered on Iceland. This phenomenon is active in the winter season, and the strength of these high and low pressure centers control the climate over the North American continent and Europe. The NAO index is defined as the difference in pressure between these two centers

PDO: Pacific Decadal Oscillation is often described as a long-lived ENSO -like pattern of pacific climate variability. Extremes in the PDO pattern are marked by widespread variations in Pacific Basin and North American climate. In parallel with the ENSO phenomenon, the extreme phases of the PDO have been classified as being either warm or cool, as defined by ocean temperature anomalies in the northeast and tropical Pacific Ocean.

Relevant Readings

Following is a suggested list of technical references:

S. Hartley, and M. J. Keables, 1998. Synoptic Associations of Winter Climate and Snowfall Variability in New England, USA, 1950-1992. International Journal of Climatology, 18, 281-298.

J. W Hurrell, 1995. Decadal Tends in the North Atlantic Oscillation: Regional temperatures and Precipitation, Science, 269, 676-679.

E. M. Kalnay, M. Kanamitsu, R. Kistler, W. Collins, D. Deaven, L. Gandin, M. Iredell, S. Saha, G. White, et al., 1996: The NCEP/NCAR 40-Year Reanalysis Project. Bulletin of American Meteorological Society, 77, 437-471.

J. M. Keables, 1992. Spatial Variability of Midtropospheric Circulation Patterns and Associated Surface Climate in the United States During ENSO Winters, Physical Geography, 13, 331-348.

G. N. Kiladis, and H. F. Diaz, 1989. Global Climate Anomalies Associated with Extremes in the Southern Oscillation, Journal of Climate, 2, 1069-1090.

D. W. LeBoutillier, and P. R. Waylen, 1988. Stochastic Analysis of Cold Spells, Journal of Applied Meteorology, 27, 67-76.

M. Macchiato, C. Serio, V. Lapenna and L. La Rotonda, 1993. Parametric Time Series Analysis of Cold and Hot Spells in Daily Temperature: An Application in Southern
Italy, Journal of Applied Meteorology, 32, 1270-1281.

N. A. Rayner, E.B. Horton, D.E. Parker, C.K. Folland, and R.B. Hackett, 1996. Version 2.2 of the global sea-ice and sea surface temperature data set, 1903-1994. Clim Res. Tech. Note 74. Unpublished document available from the Hadley Center for Climate Prediction and Research, Meteorological Office, London Road, Bracknell, RS12 2SY, U.K.

C. F. Ropelewski, and M. S. Halpert, 1996. Quantifying Southern Oscillation: Precipitation Relationships, Journal of Climate, 9, 1043-1059.

Data used in this study was retrieved from the following data sources:

Daily maximum and minimum temperatures from 1951 to 1998 and gridded monthly anomalies of global SST were retrieved from the data library of Lamont-Doherty Earth Laboratories at: http://ingrid.ldeo.columbia.edu/SOURCES/.KAPLAN/

Gridded monthly anomalies of global SLP was retrieved from the data library of Lamont-Doherty Earth Laboratories at:
http://ingrid.ldeo.columbia.edu .

NAO prediction is an active area of research. For more details on NAO and its impacts see:

http://www.met.rdg.ac.uk/cag/NAO/index.html
http://www.ldeo.columbia.edu/~cullen/NAO_figures.html
http://www.ldeo.columbia.edu/NAO/