Examensarbete vid Institutionen för geovetenskaper ISSN 1650-6553 Nr 283
NAO Index: An Extreme Pressure Approach
NAO Index: An Extreme Pressure Approach
Patrik Boström
Patrik Boström
Uppsala universitet, Institutionen för geovetenskaper Examensarbete E i Meteorologi, 30 hp
ISSN 1650-6553 Nr 283
The North Atlantic Oscillation (NAO) is the major mode of atmospheric winter variability over the North Atlantic. Its current state has long been described by the station-based NAO Index. This index has been shown to represent winter temperature anomalies over Northern Europe well.
Nonetheless, its positions of measurement are stationary and may not catch the moving NAO pattern’s characteristics sufficiently well to best represent the Scandinavian winter temperature anomalies. A new index based on North Atlantic maximum and minimum sea level pressure (SLP) is introduced in this study (called NAOE Index). The points of NAOE Index are therefore moving over the North Atlantic SLP-field and may better catch atmospheric processes affecting Scandinavian winter temperature anomalies.
NAOE Index correlation with Scandinavian winter temperature anomalies is analyzed through NASA’s (National Aeronautics and Space Administration) MERRA-data (Modern Era-Retrospective analysis for Research and Applications). This study shows that NAOE Index represents the Scandinavian winter temperature anomalies almost as well as NAO Index from observed values does. The indices are also well correlated with each other.
The direct difference between maximum and minimum SLP is also analyzed with regard to the Scandinavian winter temperature anomalies.
The SLP-differences are class-divided and their correlations to the class-mean temperature anomaly are shown to be very high. This correlation is significantly higher than the correlations of temperature and each index. Hence, the results from this study suggest that an index based on the direct extreme-SLP-difference is preferred for representing the NAO’s impact on Scandinavian winter temperature anomalies. This study gives additionally a comprehensive view of NAO. Studies concerning the NAO’s history of research, dynamics, temperature connections and secondary impacts are reviewed.
Supervisor: Hans Bergström Examensarbete vid Institutionen för geovetenskaper
ISSN 1650-6553 Nr 283
NAO Index: An Extreme Pressure Approach
Patrik Boström
Copyright © 1BUSJL#PTUSÚN and the Department of Earth Sciences Uppsala University
Abstract
The North Atlantic Oscillation (NAO) is the major mode of atmospheric winter variability over the North Atlantic. Its current state has long been described by the station‐based NAO Index. This index has been shown to represent winter temperature anomalies over Northern Europe well.
Nonetheless, its positions of measurement are stationary and may not catch the moving NAO pattern’s characteristics sufficiently well to best represent the Scandinavian winter temperature anomalies. A new index based on North Atlantic maximum and minimum sea level pressure (SLP) is introduced in this study (called NAOE Index). The points of NAOE Index are therefore moving over the North Atlantic SLP‐field and may better catch atmospheric processes affecting Scandinavian winter temperature anomalies.
NAOE Index correlation with Scandinavian winter temperature anomalies is analyzed through NASA’s (National Aeronautics and Space Administration) MERRA‐data (Modern Era‐Retrospective analysis for Research and Applications). This study shows that NAOE Index represents the Scandinavian winter temperature anomalies almost as well as NAO Index from observed values does. The indices are also well correlated with each other.
The direct difference between maximum and minimum SLP is also analyzed with regard to the Scandinavian winter temperature anomalies. The SLP‐differences are class‐divided and their correlations to the class‐mean temperature anomaly are shown to be very high. This correlation is significantly higher than the correlations of temperature and each index. Hence, the results from this study suggest that an index based on the direct extreme‐SLP‐difference is preferred for representing the NAO’s impact on Scandinavian winter temperature anomalies. This study gives additionally a comprehensive view of NAO. Studies concerning the NAO’s history of research, dynamics, temperature connections and secondary impacts are reviewed.
Referat
Den nordatlantiska oscillationen (NAO) är det dominerande mönstret av atmosfäriska variationer över nordatlanten under vintern. Dess nuvarande tillstånd har länge beskrivits av det stations‐
baserade NAO Indexet. Detta index har visat sig represententativt för temperaturavvikelser i norra Europa under vintern. Icke desto mindre är dess mätpunkter stationära och indexet kan vara otillräckligt för att bäst fånga det icke‐stationära NAO‐fenomenets påverkan på Skandinaviska temperaturavvikelser under vintern. Ett nytt index baserat på det högsta och lägsta trycket vid havsnivå över nordatlanten har föreslagits i denna studie (kallat NAOE Index). NAOE Index har alltså mätpunkter som rör sig efter det nordatlantiska tryckfältets extrempunkter och kan således vara bättre på att fånga atmosfäriska processer som påverkar temperaturavvikelser över Skandinavien under vintern.
Korrelationen mellan NAOE Index och temperaturavvikelser är i denna studie analyserade genom NASA’s (National Aeronautics and Space Administration) MERRA‐data (Modern Era‐Retrospective analysis for Research and Applications).
Denna studie visar att NAOE Index representerar avvikelser i vintertemperatur i Skandinavien nästan lika bra som NAO Index från observerade värden. Båda index visar sig också vara väl korrelerade med varandra.
Den direkta skillnaden mellan maximala och minimala tryck över nordatlanten är också analyserad med hänsyn till temperatur i Skandinavien. SLP‐skillnaderna klassindelades och deras korrelation till medeltemperaturavvikelsen i varje klass visar sig vara väldigt hög. Denna korrelation är märkbart högre än korrelationen mellan temperatur och de båda indexen. Resultaten från denna studie visar alltså att ett index baserat på den direkta skillnaden mellan tryckextremerna över nordatlanten är bättre för att representera NAOs påverkan på temperaturavvikelsen i Skandinavien under vintern.
Denna studie ger därtill en övergripande bild av NAO. Studier om NAO’s forskningshistoria, dynamik,
temperatursamband och sekundär påverkan granskas.
Contents
1 Introduction ... 1
1.1 The NAO pattern ... 1
1.2 History of NAO research ... 5
1.3 Dynamics of NAO ... 6
1.4 Temperature connections to NAO ... 8
1.5 Ability to model NAO ... 8
1.6 Reliability of standard NAO Index ... 9
2 Secondary impacts of NAO ... 9
2.1 Nature ... 9
2.2 Economics ... 10
3 Data and methodology ... 11
3.1 Scandinavian temperature ... 11
3.2 Methods ... 11
4 Results ... 13
5 Discussion ... 18
6 Conclusions ... 21
References ... 23
1 Introduction
The North Atlantic Oscillation (NAO) is a pattern of variability in atmospheric mass between southern and northern latitudes over the North Atlantic. It is regarded as the most common pattern of
atmospheric winter variability over the North Atlantic. The state of NAO is generally described by a station‐based index. As we will see, the winter temperature in Scandinavia can be related to the standard NAO Index (Hurrell, 1995, Kushnir, 1999). The goal with this study is to test if an index described by the maximum and minimum sea level pressures (SLP) over the North Atlantic is closer connected to the Scandinavian winter temperatures than the standard NAO Index. Reanalysis data is used for winter months over 30 years.
This work also reviews previous studies of the NAO pattern to get a comprehensive understanding of the NAO phenomenon.In addition, results from these studies are connected to what is found in the present study.
Winter months in this work are always considered as December, January, and February.
1.1 The NAO pattern
NAO is called a teleconnection1 pattern and is expressed by the state of the atmospheric mass distributed between the Arctic and subtropical Atlantic. It is a zonally asymmetric pattern of meridional variability. NAO is also regarded as an important source to atmospheric anomalies in areas adjacent to the North Atlantic. The weather during winters in extra‐tropical regions can be affected diversely depending on the state of NAO (Walker and Bliss, 1932, van Loon and Rogers, 1978, Hurrell, 1995, Kushnir, 1999). The Greenland‐Scandinavia temperature seesaw is the name of the phenomenon that the winter temperatures in Greenland and Scandinavia are inversely covariate (i.e. when it is an anomaly cold winter in Greenland it is an anomaly warm winter in Scandinavia).The Greenland‐Scandinavia temperature seesaw is a result of the NAO pattern.
Recent studies have suggested that NAO affects the hemispheric‐scale circulation of the whole northern hemisphere (Thompson and Wallace, 1998, 2000). With that considered, NAO can easily be regarded as an important teleconnection pattern being intrinsically dependent of surrounding atmospheric motion.
The typical appearance of the atmospheric state over North Atlantic is governed by the Azores high and Iceland low. How the connection between north and south varies defines the NAO. In other words, NAO can be stated as a dipole pattern with two centers of anomalies with opposite signs. One center is located over polar region of the North Atlantic and the other zonally outside of Africa in between 35 and 40°N (Walker and Bliss, 1932, Hurrel, 1995, Stoner et al. 2009).
The current view of NAO regards it as an oscillation between a positive and a negative phase. The positive phase means deeper subtropical high and arctic low than normal. In contrast the negative phase represent weaker high and low pressure systems than normal. The negative phase can in
1 Atmospheric teleconnections are simultaneous variability (covariance) of meteorological and climatological variables in remote regions. They are mainly synoptic scale variability with frequencies of weekly or longer (Hurrel et al., 2003). The term was invented by the Swedish scientist Ångström (1935) to describe atmospheric connections (Stephenson et al. 2003).
extreme events mean that the pressure systems are of reversed positions (i.e. a northerly high and a southerly low) (Hurrel, 2003).
The most frequently used mathematical description of the NAO phase is the standard NAO Index. In this work, this index will be called the standard NAO Index since a new index will be suggested later on. The standard NAO Index is calculated by taking the difference between normalized surface pressure anomaly from a 30 years mean of a northern and a southern station (Equation 1). P is the mean monthly SLP, the 30 years average of every corresponding month’s SLP is used and std means the corresponding month’s monthly standard deviation. Hence this defines the positive and negative NAO phase as described above (Hurrel, 1995). Different locations and time‐scales can be used in the index. It is common to use monthly or winter averages from Ponta Delgada, Azores or Lisbon,
Portugal (as southern location) and Stykkisholmur, Iceland (as northern location) (Stoner et al. 2009).
The time‐scale of the variables used in the standard NAO Index is usually monthly. The correlation between daily SLP anomalies at Iceland and Azores is ‐0.33. Based on monthly values, the
corresponding correlation is +0.56. Standard NAO Index is for that reason preferred at monthly values (Löptien and Ruprecht, 2005). Each pressure anomaly in the standard NAO Index is normalized by its 30 years standard deviation to take the greater variability at the northern dipole into account (Hurrel et al. 2003). NAO Index’s history, a more detailed explanation and reliability will be studied in following sections.
(1)
Figure 1 and Figure 2 show comprehensive explanations of the atmospheric behavior during a winter month with each NAO phase. Winters with positive standard NAO Index correspond to enhanced westerly flow and thus more warm maritime air is advected to Western Europe. The Mediterranean part of Europe is still primarily affected by the Azores high. Therefore these areas tend to be drier and colder than normal (Bachman, 2007). Greenland gets more cold Arctic air during the positive NAO phase. Northeastern Canada experience colder weather than normal due to stronger northerly winds. Additionally, higher temperatures are brought to the coast of North‐East America (Hurrel et al., 2003). Winters with negative standard NAO Index is characterized by cool and dry conditions in North Western Europe while Canada and Greenland gets warmer. This leads to that mild winters in Scandinavia means severe winters in Greenland and vice versa. This is known as the Greenland‐
Scandinavia‐seesaw (anti‐correlation) in temperature (Hurrel, 1995). More discussion of the positive and negative NAO phase will follow in section 1.4.
The transport of air masses influenced by NAO is considered to have severe impacts on countries’
humidity and precipitation. The amount of precipitation has a positive correlation with standard NAO Index in north‐western parts of U.K. and south‐western parts of Norway. In contrast, the
precipitation over Greenland and the Mediterranean is negatively correlated to standard NAO Index (Hurrel, 1995). This is the explanation to dry winters in the Mediterranean and Greenland ice sheet declining.
Figure 1 and wa
Visualisation o ater flow, blue a
of the positive N and red lines ar
NAO phase. Sur re SLP (from Wa
rfaces represen anner et al., 20
Bern).
t temperature 01) (With perm
and sea‐ice ext mission of Heinz
tension, arrows z Wanner, Univ
s show air ersity of
Figure 2
2 Visualisation oof the negative pe
NAO phase. Sy ermission of He
ymbols are the einz Wanner, U
same as in Figu niversity of Ber
ure 1 (from Wa rn).
nner et al., 200
01) (With
1.2 History of NAO research
This chapter is a review of previous and present studies concerning the NAO phenomenon. We will start out with a brief history of the NAO research. The NAO pattern has paid scientific interest for more than two centuries (Stephenson et al, 2002). Nevertheless, the first sign of the Greenland‐
Scandinavia temperature seesaw was recognized many years before. The local weather of Northern Europe comprises great variability and this have always fascinated the people living there.
Through older studies and documents, Stephenson et al. (2003) review that seafaring Scandinavians discovered signs of NAO. The seafarers found that severe winters in Greenland occurred during mild winters in Scandinavia and the opposite for mild winters in Greenland (i.e. the Greenland‐
Scandinavia temperature seesaw). This was common knowledge in 18th century when Greenland’s climate had been carefully studied by the Danish missionary Hans Egede Saabye (van Loon and Rogers, 1978).
Major part of the studies during the 19th century was primarily about the North Atlantic temperature patterns’ characteristics to interpret the NAO phenomenon. The temperature seesaw between Greenland and Scandinavia was later confirmed by Loewe (1937) with use of up‐to‐date historical data. It was with the groundbreaking studies by Teisserenc de Bort (released 1883) that the behavior of atmospheric SLP‐distribution started to occupy scientists’ interest (Stephenson et al., 2003). The studies by Teisserenc de Bort encouraged Hildebrandsson (released 1897) to study time series of SLP over the North Atlantic. Hildebrandsson found that the SLP at Iceland are inversely related to the SLP at the Azores (Stephenson et al., 2002). The latter study is known as the benchmark for all
investigations using SLP at Azores and Iceland to describe the NAO pattern. Additionally it brought further interest to the coupling between local weather and expansive weather systems (Feldstein, 2002).
The history of NAO research contains many attempts to describe the state of NAO with an index. One of the first indices was developed and defined by Walker and Bliss (1932). This index was the sum of surface pressure or temperature (depending on the stations location) from stations in the vicinity of the North Atlantic Ocean. All data in the index was adjusted with an early regression procedure so that every term had a standard deviation of √20 units (i.e. variance of 20 units).
To understand the behavior of the NAO pattern more studies of the dynamics of the atmosphere were needed. Rossby et al. (1939) provided such knowledge with their study of planetary waves and eddy fluxes. For additional information and understanding of Rossby waves and dynamics of the atmosphere, readers are referred to the comprehensive work by Holton (2004).
Many studies of correlation matrices of SLP over North Atlantic were studied in the early 20th
century, but it was first with Lorenz et al. (1956) that the dynamics of NAO started to be investigated.
Lorenz et al. (1956) started a new era by investigating the correlation matrix of the North Atlantic Ocean and found the major pattern of variance in winter SLP. This was concluded to be the NAO which was in agreement with the position of the earlier observed seesaw (Wallace and Gutzler (1980)).
Today’s concept of NAO as a dominant mode of atmospheric variability in the Northern Hemisphere during winters originates from the studies by Rogers and Van Loon (1979). Rogers and Van Loon
(1979) explained variability with dynamic correlations between current atmospheric states at different locations in Northern hemisphere.
The NAO phase was predominantly positive in the last decade of 20th century. This captured many scientists’ interest and the number of NAO studies increased rapidly (Stephenson et al., 2003). Many studies during the 90s performed advanced statistical analyzes of pressure fields at different levels over sea to describe the appearance of NAO. Modeling had also become very popular to simulate different NAO states. Hurrel (1995) interested many researchers by regarding pressure and temperature variability over Europe as an impact of the NAO pattern.
Walker and Bliss’ (1932) index was precursor to the more commonly used station‐based normalized pressure difference standard NAO index stated above. Different north and southerly locations have been used in the NAO Index (such as Rogers (1984), Hurrel (1995) and Jones (1997)). The most common locations were mentioned above.
An always persistent problem with NAO studies have been the lack of observations covering the North Atlantic. The phenomenon is aimed to be studied as a low‐frequency problem (low (high) frequencies are eddy periods longer (shorter) than 10 days) (Feldstein, 2003). Historical observations throughout the North Atlantic are limited to shorter periods and only from land observations. This sparseness of both temporal and spatial resolution limits NAO‐studies. Improvements in proxy‐data has helped scientists to study earlier NAO‐states from ice‐cores, tree‐rings etc. (Schmutz et al., 2000).
This gives additional sets of NAO‐states to be compared with those from measurements. Cook (2003) reconstructed standard NAO Index back to AD 1400 by using multi‐proxy data (i.e. proxy data from combined sets of different proxies). This reconstruction was shown to follow winter standard NAO Index from European instrumental and non‐instrumental data. The successful study by Cook (2003) opened up for longer studies of standard NAO Index.
Despite the lack of full understanding of NAO, many studies today focus on how the nature, human activity etc. are affected by NAO. A summary of secondary causes of NAO is found in section 2.
1.3 Dynamics of NAO
NAO is intrinsically connected to other atmospheric processes; the dynamics of NAO are not fully understood. The current interpretations of the dynamics and motions of NAO are reviewed through this section.
The positive phase of NAO represents positive geo‐potential height and pressure anomalies in the southern and northern location. The opposite is valid for the negative NAO phase. This affects location of storm‐tracks and jet streams over North Atlantic (Hurrel and van Loon, 1997, Greatbatch, 2000).
The variations in NAO dynamics are mainly caused by extra‐tropical horizontal eddy fluxes and storm‐
track shifting. Eddy fluxes are in this report referred to horizontal eddy flux on the synoptic scale (not to be mixed up with vertical eddy‐fluxes). Storm‐tracks over the North Atlantic are closely connected to the state of NAO. The storms represent the highest extra‐tropical eddy activity over the North Atlantic (Manola et al. 2013). Shifts in the storm‐tracks are thus important for the appearance of the NAO. To compute the eddy’s part of contribution to the time flow the Reynolds decomposition is applied to the momentum equations. See Kok and Opsteegh (1985) for a more detailed description.
Disturbances of the Rossby wave have been suggested to have large impact on NAO. Such
disturbances can be caused by eddy flux. By interaction with the Rossby wave, eddy flux can cause what is called a Rossby wave‐breaking. The name Rossby wave‐breaking can be misleading since it only means a disruption of the Rossby wave (Strong and Magnusdottir, 2008, Rivière et al., 2010).
The studies are provided by both models and reanalyzes and they show the same results. To the dynamics of NAO, the Rossby wave‐breaking implies that energy of high frequencies (i.e. eddy flux) is transferred to low frequencies (Rossby wave anomalies).
In addition to secondary effects on NAO (as described above), eddy fluxes have also been suggested to have a direct effect (Feldstein, 2003). To investigate the dynamics behind the growth and decay of NAO, Feldstein (2003) analyzed NCEP/NCAR reanalysis data2 and forced barotropic model runs. A simple lifecycle of NAO was thereby represented. Feldstein (2003) showed that transient eddy vorticity fluxes of high and low frequencies are behind the growth of NAO. It was also shown that low‐frequency eddies can force NAO to decay (i.e. decrease the strength of NAO phase). The results were applicable to both phases of NAO but nonetheless, the study suggests different characteristics for the initiation of each NAO phase. An eastward moving remote wave packet from North Pacific initiates the anomaly known as the positive NAO phase. Interestingly, it is suggested that the negative NAO is initiated by an in situ growth of a block situation (Feldstein, 2003). Although the study shows what large scale processes that create and destroy both phases of NAO, it leaves many unanswered questions. The source of the North Pacific wave train is still unknown. The role of baroclinic processes such as eddy heat fluxes, diabatic heating to the NAO life cycle needs to be examined further (Feldstein, 2003).
The abovementioned suggestion by Feldstein (2003) may be hard to understand. The ability of extra‐
tropical eddy fluxes to influence NAO such as described above was better understood by Benedict et al. (2004). High‐frequency eddy fluxes that move eastward slow down as they enter the NAO region and consequently break down. The decay turns the eddy fluxes to low frequencies. This is by Benedict et al. (2004) suggested to be the physical entity of NAO. Thus, by recalling that both high and low frequency eddy fluxes cause the growth of NAO while only low frequency eddy fluxes causes a decay (Feldstein, 2003), Benedict et al. (2004) suggests that the low‐frequency eddies causing a NAO phase decay can be a former high‐frequency eddy which caused NAO growth and then broke down. The results then suggest then that the life time of one NAO phase event also is determined by the scale of mixing time associated with the decaying eddy fluxes.
It has long been suggested that ocean‐atmosphere couplings are strongly affecting the NAO pattern.
However, recent studies show no such behavior (Thompson et al., 2003). Models show that the NAO pattern can be created by processes in the atmosphere alone. Ocean impacts such as sea‐surface anomalies are negligible on monthly and yearly timescales. Hence, NAO can be considered an internal pattern of the atmosphere (Greatbatch, 2000). It is driven by interactions between the eddy and mean flow.
To conclude, the dynamics of NAO may not be fully understood but studies suggest that anomalies in atmospheric processes (primarily subtropic eddy fluxes) entering the North Atlantic have a dominant effect on the NAO pattern (Rivière et al., 2010, Manola et al., 2013). The eddy fluxes with high and
2 A corporate gridded reanalysis data set from National Centers for Environmental Prediction (NCEP) and National Center for Atmospheric Research (NCAR).
low frequencies can cause a growth of the NAO phase (i.e. strengthen its current phase). High‐
frequency eddy flux break down as it enhances the NAO phase and then becomes a low‐frequency eddy flux that weakens the NAO phase. This highlights that NAO cannot be considered as a pattern separated from atmospheric motion outside of the North Atlantic. Most studies expect that eddy fluxes are the main source to low‐frequency variability (Vallis et al, 2003). Although NAO is internal in the atmosphere on monthly and annual time‐scales (not significantly ocean‐coupled), it is not
internal in a region such as North Atlantic.
1.4 Temperature connections to NAO
Numerous studies of temperature connections with NAO have been made. The most evident effect of NAO is the strength of the westerly winds over the North Atlantic during the winter. Strong westerlies, such as during a winter month with positive NAO phase, transports warm humid air from the North Atlantic sector zonally to Europe. This implies winter temperatures above normal for many areas (Hurrel, 1995). During the negative NAO phase the winter temperature over many parts of Europe are often below normal due to weaker westerlies. Thus, NAO causes prominent winter temperature variations in Western, North Western Europe and North Eastern America. The
correlation between standard NAO Index and winter mean temperature over North Western Europe is 0.60 (Stephenson et al., 2000). The corresponding correlation between standard NAO Index and Greenland is ‐0.59. Through station data, Hurrel (1996) found that NAO accounts for about 1/3 of the total variance in winter temperature north of 20°N (Greatbatch, 2000).
Monthly and seasonally winter temperature anomalies have been discussed above. By using climate models, Beranova and Kysely (2013) studied daily temperature responses to NAO during winter. The daily temperature anomalies north of 45 °N were shown to be positively correlated with the station‐
based NAO Index. South of the same latitude the temperature anomalies were explained by a weaker negative correlation. Meanwhile, Beranova and Kysely (2013) also found that the station‐based NAO Index captures cold extremes more accurately than warm extremes. Hence, it is understood from this study that the standard NAO Index can also be used to capture daily temperature extremes during winters.
1.5 Ability to model NAO
As mentioned above, studies of NAO from observations are limited to a period which may be too short to capture a behavior of NAO. This makes climate models of unmatched importance. Several studies have analyzed the reliability of models to reconstruct the NAO phenomenon. Studies show that both atmosphere‐only and coupled ocean‐atmosphere general circulation models (GCM) simulate the NAO pattern realistically (Stephenson and Pavan, 2003).
Manola et al. (2013) analyzed the eddy contribution to NAO (described above) using both a barotropic climate model and ERA‐Interim3 reanalysis data. They found that the model’s behavior was similar to the eddy flux forcing observed in the reanalysis data.
Stephenson and Pavan (2003) evaluated 17 coupled ocean‐atmosphere GCMs included in the CMIP4 (Coupled Model Intercomparison Project). 13 out of these models were able to recreate the
3 ERA‐Interim is the European ECMWF’s (Centre for Medium‐Range Weather Forecasts) current reanalysis project.
temperature pattern from NAO realistically. This encourages studies of temperature connection to the NAO pattern. Frequently occurring errors among the models were however incorrect placements of the southern NAO dipole. It was often located far more eastward than what has been seen in measurements from real observations.
Models are very common to use in simulations of future climate. The appearance of the NAO pattern in a future climate has been a sidetrack of many climate studies. Beranova and Kysely (2013) used GCM simulations during the period 2071‐2100 with the requirements of IPCC’s (Intergovernmental Panel on Climate Change) SRES (Special Report on Emission Scenarios) A1B emission scenario5 and found that the temperature variations from positive and negative NAO phases are similar to model simulations of today’s climate. Hence, the results suggest a possibility how the atmosphere will react to future scenarios and may in addition contribute to other parameters of future climate.
1.6 Reliability of standard NAO Index
Despite that the standard NAO Index is unable to capture the distribution of the actual SLP its simplicity makes it popular to describe the phase of NAO. The measurements from the stations used cover long historical records. Additionally, it is easy to add to model simulations of future climate. It has been shown that the standard NAO Index describes the oscillation of the North Atlantic pressure systems quite well (Bachman, 2007). However, the question remains: what is most important to represent with an index of NAO? Should the actual spatial structure of NAO be captured or should the index be a definition of NAO’s impact (e.g. temperature or precipitation variance) on a remote area?
Beranova and Kysely (2013) used 10 GCMs from CMIP to analyze standard NAO Index’s
representation on European winter temperatures. They found that the standard NAO Index can with relatively high certainty be used to represent European winter temperature extremes. The
atmospheric processes related to atmospheric dynamics are accurately captured by standard NAO Index. What still is unknown is if it would be better to use the value of the actual dominant pressure systems over North Atlantic.
2 Secondary impacts of NAO
Apparently, the NAO has major impacts on human and nature. This chapter is a short review of the NAO’s secondary impacts.
2.1 Nature
By the changing distribution of the temperature fields etc., NAO affects sea ice distribution. The extension of northern sea ice is different between the NAO phases (recall Figure 1 and Figure 2).
Winters during a positive NAO phase are colder than normal in Canada while they are warmer than normal in Siberia. This means a southerly expansion of sea ice at the coast of Canada and a decline of Siberian sea ice northwards. For winters with negative NAO phase the sea ice distribution at the
4 A protocol of outputs of GCMs provided by the Working Group on Coupled Modelling (WGCM) under the WCRP (World Climate Research Programme) (PCMDI, 2014).
5 The A1 scenarios describe a more integrated world and the sub‐scenario A1B describes a shared access to all energy sources (Solomon et al., 2007).
coast of Canada declines and is more northerly located than normal. In Siberia the sea ice extends farther south (Bachman, 2007).
The sea ice variation due to positive NAO phase impacts the sea water salinity outside of Canada and river run‐off in Siberia. More sea ice means higher than normal sea water salinity in Canada. In addition, the density driven large‐scale ocean circulation called thermohaline circulation is affected.
More cool water due to the sea ice extension cause more deep water formation which counteracts the thermohaline circulation. The warmer weather in Siberia causes more river runoff than normal (Bachman, 2007).
Even marine ecology is affected by the variations of NAO. For example, Fromentin and Planque (1996) found that the wealth of the North Atlantic’s major zooplankton species (Calanus finmarchicus and C. helgolandicus) is closely related to the NAO. The difference in strength of the westerlies due to NAO affects the wind stress on water surface. This affects the water mixing, upwelling and ice‐
cover that all are closely related to the kingdom of zooplankton.
It has been found that there is a stronger cyanobacterial bloom (i.e. blue‐green algae) in the Baltic Sea during positive standard NAO Index. This is explained by the lower ice cover due to a higher wind stress from the stronger westerlies. The sea is then more mixed which favors the cyanobacterial bloom (Janssen et al. (2004)).
Migrating birds are affected by the different weather conditions due to different NAO phases.
Hubálek (2003) showed that short‐distant migrants from Europe had an earlier summer migration when the winter/spring standard NAO Index was positive. Hüppop (2003) also showed that the bird migration routes within Europe were changing for different standard NAO Index.
For more comprehensive studies of northern ecosystems’, animals’ and plants’ relation to NAO, interested readers are referred to Drinkwater et al. (2003), Mysteryd et al. (2003) and Straile et al.
(2003).
2.2 Economics
Weather has strong influences on most human activities. Thus, NAO will affect the economics in one way or another. Such economic effects can be costs of extreme weather protection, the amount of energy used by the society, the amount of energy produced and agricultural activities. It can also be costs from recreational activities such as summer and winter tourism.
Studies of how the quality of wheat in the U.K. depends on NAO have been made. NAO‐impacts of one winter affect the quality of the wheat the following summer (Atkinson et al., 2005). Additionally, Kim and McCarl (2005) also found a connection between agricultural quality and NAO over both U.S.
and widely across Europe. They estimated that if the NAO phase could be predicted it would give the crop yields a total increase worth $600 million to 1.09 billion per year. These results indicate the economic effects and how the understanding and forecasting of NAO could significantly help society.
Sweden and Norway are countries in direct connection with NAO. Winters with high levels of precipitation give major advantages in Norway’s energy production by boosting the hydropower.
What makes Norway of particular interest is that it also is ranked among the largest oil exporters in the world (Hurrel et al., 2003).
So the severity of winter determines the energy consumption and thus the surplus or deficit of energy within Norway is determined. This leads to that Norway can cooperate with its neighboring country of Sweden through winters with different NAO conditions. For example, a winter with negative NAO phase results in lower hydropower conditions in Norway which then can buy energy from Sweden’s nuclear power. If the NAO phase in contrast is positive the precipitation is
anomalously high in Norway and hence the hydropower is favored. Then it may be better for Sweden to buy hydropower cheaper rather than producing that energy itself. In addition, this may be positive in a climatic point of view (Hurrel et al., 2003).
3 Data and methodology
In this study, gridded SLP and surface temperature (temperature at a height of 10 m over surface level) over the North Atlantic and Scandinavia is analyzed. The dataset is NASA’s (National
Aeronautics and Space Administration) MERRA‐data (Modern Era‐Retrospective analysis for Research and Applications). MERRA consists of reanalyzes6 from 1979 to present. The two‐dimensional MERRA outputs used in this study (single‐level meteorology) are available in intervals of 1 hour. The full spatial resolution of the data is 1/2° lat. x 2/3° lon. The vertical depth of the data extends up to 0.01 hPa containing 72 levels (GMAO, 2014). MERRA‐data contains both high and low spatial and
temporal resolution and more than 30 years of data. Hence, it encourages both weather and climate researches. The data is used by NASA themselves and is open for anyone to use it. The open data can be received online at Goddard Earth Sciences (GES) Data and Information Services Center (DISC) (http://disc.sci.gsfc.nasa.gov/mdisc/). For a thorough explanation of how to download the data readers are referred to http://gmao.gsfc.nasa.gov/MERRA.
Although giving excellent opportunities to analyze many meteorological variables over large temporal and spatial scales, reanalysis data is to be used with caution. Sources of errors may originate from satellite calibration, flux parameterizations, boundary conditions and more. The ensemble of errors may influence each other intrinsically and the final error will grow. Thorough studies of errors derived by reanalysis data can be found in Nigam and Ruiz‐Barradas (2006) and Trenbert et al. (2001).
When encountering land in the MERRA‐data, the pressure is extrapolated beneath the Earth’s surface down to sea level. In this way SLP can be obtained even over land surfaces. The extrapolation features drawbacks however and will be discussed in the section 5.
3.1 Scandinavian temperature
The mean area mean winter‐month temperature is calculated over three Scandinavian areas. The areas are Stockholm and Uppsala region (59 to 60.5 °N and 16.67 to 18°E), southern Sweden (57 to 60.5°N and 12.67 to 16°E) and southern Norway (59.5 to 62.5 °N and 6 to 9.33 °E).
3.2 Methods
The goal with this work is to follow the SLP extremes over North Atlantic. The standard NAO Index is obtained from two stationary points which is considered incomplete in a dynamical view. I believe
6 The method where historical weather data are used in prognostic weather models is called reanalysis.
Therefore temporally and spatially missing data can be covered. Reanalysis Is useful in research when modelling weather events and processes (Uppala et al., 2005).
that an index that follows the extreme SLPs over North Atlantic would give a more representative index for the state of NAO than the station‐based NAO Index. NAOE Index is potentially better to represent Scandinavian winter temperature anomalies because it may capture processes that impact the temperature better than NAO Index.
To cover one climate period, MERRA data from 1979 to 2009 (winters 79‐80 to 08‐09) is used in this study. To get a regional approach on NAO, I focus the SLP‐analysis over the North Atlantic and use the coordinates 71.875°W to 38.12 5°E and 25.625°N to 75.625°N.
The data is averaged monthly (winter months December, January and February) for every year. Each winter month’s maximum and minimum SLP over the studied area is found. By finding each winter month's extreme pressure, a 30 year monthly mean of the maximum and minimum SLP is calculated for each winter month. Additionally, the 30 year monthly standard deviation is calculated for each winter month. The maximum/minimum SLP‐anomaly and standard deviation replaces the
southern/northern station‐based SLP‐anomaly and standard deviation in the standard NAO Index.
To distinguish it from the standard NAO Index, I state the new index as North Atlantic Oscillation Extreme Index (NAOE Index). NAOE Index is presented in Equation 2. The terms in NAOE Index are the same as for NAO Index but max and min SLP are used instead of SLP from a southern and northern location.
2
Standard NAO Index for the same period is used. The standard NAO Index used here is not from the MERRA‐data but from observations on Stykkisholmur (Iceland) and Ponta Delgada (Azores). I use these values to get the closest connections possible to a real case.
The correlation between each index (NAOE and NAO Index) and the mean area mean monthly temperature in Scandinavia will be calculated. Also, the correlation between the indices will be investigated.
Additionally, the direct difference between winter‐monthly maximum and minimum SLP over North Atlantic is calculated. The direct SLP‐difference’s correlation to the Scandinavian temperature is analyzed.
The Pearson test will be used to test if the correlations are statistically significant. The Pearson test is given by Equation 3 where r is the correlation and n is number of samples used. The correlation is statistically significant if Equation 3 is valid (Alexandersson and Bergström, 2009).
| | √64 √
8 3
4 Results
The influence of the suggested NAOE Index on Scandinavian winter temperatures during a 30‐year period is studied in the present work. Also, the standard station‐based NAO Index and the direct difference between the maximum and minimum SLP over North Atlantic is included.
For Pearson test correlation of variables in Figure 3, Figure 4, Figure 5 and Figure 6 the sample value n is 90 (i.e. winter months (DJF) for 30 years). That gives R.H.S. of Equation 3 = 0.55. That is, the correlation is statistical significant if it is greater than 0.55.
The results suggest that the behavior of the NAOE Index has connections to the temperature anomalies in Scandinavia. This can be seen in Figure 3 where the behavior of the NAOE Index shows similarities with the temperature anomalies in every area. The temperature anomalies in Sweden are behaving very equal while the Norwegian temperature anomalies show a somewhat different behavior (different tilt of regression line). NAOE Index shows good correlation with temperature anomalies of every Scandinavian area but none of them are statistical significant (Table 1). Consistent with Figure 3, the temperature anomalies in Norway is a bit lower correlated to the NAOE Index.
Figure 3 C regressio
Correlat Statistica
The stan behavior lines of t Their cor Index is statistica The diffe noticeab correlati Index (w (althoug
Circles are temp on to the tempe
T
ion
al significant
ndard station r of NAO Ind the Swedish rrelations ar significantly al significant erence betw bly higher for
ion of index with 6/100 un
gh only with a
perature anom erature anoma
Table 1 Correla
Southe 0.53 t no
n‐based NAO ex has simila areas are sim e shown in T
higher in Sw while the co een Sweden r NAO Index and tempera nits). The cor a difference
alies to their co ly. Southern Sw N tion between N
ern Sweden
O Index was a arities with t milar and the Table 2. The weden than in
orrelation in and Norway
than for NA ature anoma rrelation in N of 1/100).
orresponding N weden is in blue Norway is in red NAOE Index and
Upps 0.53 no
analyzed in t the temperat
e regression correlation b n Norway. T
Norway is n y’s correlatio
OE Index (Fig aly in Sweden Norway is low
NAOE Index. The e, Uppsala / Sto
d.
d winter tempe
sala/Stockho
he same way ture anomal
line of the N between tem
he correlatio ot.
on of index a gure 3, Figur n is higher fo wer for NAO
e lines are polyn ockholm‐region
erature anomaly
lm Sou
0.4 no
y as the NAO y too (Figure Norwegian is mperature an ons in both S
nd temperat re 4, Table 1 or NAO Index
Index than f
nomials of the is in black and
y.
uthern Norw 48
OE Index abo e 4). The regr
differently t nomaly and N Swedish regio
ture anomal and Table 2 x than for NA for NAOE Ind
first grade Southern
way
ove. The ression tilted.
NAO ons are
y is ). The AOE dex
Correlat Index Statistica
The NAO shown to and thus
ion with NAO al significant
OE Index sho o be well cor s it is statistic
F Table 2
Southe O 0.59 t yes
ws a similar rrelated duri cal significan
igure 4 Same a 2 Correlation be
ern Sweden
temporal be ing winter‐m nt.
as in Figure 3 bu etween NAO In
Upps 0.59 yes
ehavior as th months over a
ut for NAO Inde ndex and tempe
sala/Stockho
e NAO Index a 30‐year pe
ex.
erature.
lm Sou
0.4 no
x (Figure 5). T riod, with a
uthern Norw 47
The indices a correlation o
way
are of 0.70
Figure
The conn anomalie have bee
Table 3
∆SLP (hP Class
The resu tempera anomalie each Sca the Upps significa
e 5 NAOE Index
nection betw es in Scandin en partitione
3 The magnitud
Pa) ∆SLP
≤20 1
ults show sim ature anoma
es and SLP‐d andinavian a
sala/Stockho nt.
x plotted to NAO
ween the diff navia has bee ed into classe
e of each SLP‐d
20≤∆SLP
≤25 2
milarities betw lies over Sca difference se rea is shown olm‐region s
O Index. The lin
ference in ex en analyzed es. The magn
difference‐class
25≤∆SLP
≤30 3
ween the dir andinavian ar em to have s n in Table 4. A
hows the low
ne is polynomia
xtreme SLP o (in addition nitude of eac
s. With exceptio
30≤∆SLP
≤35 4
rect SLP‐diffe reas. This is e similarities. T All temperat west correlat
al of the first gr
over North A to the indice ch class is sh
on of class 1 inc
35≤∆SLP
≤40 5
erence and w evident in Fi The correlati tures are we
tion but non
ade regression
tlantic and w es above). T
own in Table
cludes every cla
40≤∆SLP
≤45
4
≤
6 7
winter‐mont gure 6 where ion between ll correlated e of them ar
between the in
winter tempe The SLP‐diffe
e 3.
ass an interval o
45≤∆SLP
≤50
5
≤
7 8
hly mean e temperatu n the classes to the classe re statistical
ndices.
erature rences
of 5 hPa.
50≤∆SLP
≤55 8
ure and es where
Figure 6
Correlat class and Statistica
The mea the resu 7. Figure extreme get wide correlati for every of Equat
6 Circles are te
ion between d temp. anom al significant
an value of th lts show an i e 7 also show e‐SLP before
er as the clas ion between y mean‐area tion 3 0.71
mperature ano
Table 4 Cor
Sout n SLP‐
m.
0.52
t no
he temperat increasing te ws the mean calculating t ss increases a
the temper a temperatur 1. Therefore,
omalies to their Figu rrelation betwe
thern Swede 2
ure in each S emperature w
maximum a the differenc and as stated
ature and in re (Table 5).
all correlatio
corresponding ure 3 and Figur een SLP differen
en Upps
0.49 no
SLP‐differenc with increas nd minimum ce between m
d above the t creasing SLP
8 is use ons shown in
g SLP‐difference re 4.
nce class and te
sala/Stockho
ce‐class was ed SLP‐differ m SLP of each maximum an temperature P‐difference
d for the Pea n Table 5 are
e. Colors corres
emperature.
lm Sou
0.5 no
calculated. I rence. This is h class (i.e. an nd minimum
e is also incre (i.e. increasin arson test wh e statistically
spond to the sa
uthern Norw 2
In almost all s illustrated n averaged v
SLP). This is easing. The ng class) is v hich gives th y significant.
me as in
way
classes, in Figure value of
seen to ery high hat R.H.S.
Figure 7
Correlat Statistica
5 Dis
This stud tempera Scandina favored NAO is h A station and nort be ideal.
SLP over better re The avai analyses
7 Temperature ( axis
ion with clas al significant
scussio
dy was set ou ature of Nort
avia is positiv for long time however a no nary index (s thern points . The presen r North Atlan epresented.
lability of op s of the atmo
(Southern Swed s) and mean ma Table 5
ss t
on
ut to explore thern Europe vely correlat e because of on‐stationary such as stand of measurem t study inves ntic. In this w
pen reanalys ospheric beh
den is blue, Upp ax/min SLP of e
Correlation be
South 0.95 yes
e the reliabili e. Many stud ted to the sta f the quantit y phenomen dard NAO Ind
ment respec stigates a po way the actua
is data such avior for larg
psala / Stockho every SLP‐differ etween mean te
hern Sweden
ity of the sta dies have sug ation‐based
y of historica non with mov dex with Pon ctively) that r oint‐based in al state of th
as NASA’s M ge spatial an
olm‐region is bla rence class (ma emp per class a
Uppsala 0.95 yes
ation‐based N ggested that
NAO Index. A al measurem ving poles of nta Delgada a
represents th dex that foll
e NAO over
MERRA data u nd temporal s
ack and Southe genta, right y‐a nd class.
/Stockholm
NAO Index to the winter t A station‐bas ments from c
f variability in and Stykkish he NAO’s cur
ows the max a given wint
used in this s scales. The in
ern Norway is re axis).
Southern N 0.98 yes
o describe th temperature
sed index ha hosen locati n atmospher
olmur as sou rrent state m ximum and m ter month m
study encour nvestigated
ed, left y‐
Norway
he winter in as been
ons. The ric mass.
uthern may not
minimum ay be
rages NAOE
Index do than the tempera Index ma account by Beran dynamic stationa that amp position tempera This stud tempera equally g work we Atlantic.
winter te Errors in extrapol a maxim 40 to 35°
revealed Africa ca from sur
That bot winter te That is, e
oes not show e standard NA ature in both
ay additiona when descri nova and Kys cs are accura ry locations plify the wes of the extre ature.
dy suggests t ature than th good, repres e saw that NA
Hence, usin emperatures nvolved in th
ation can be mum pressure
°W, 60 to 63 d. Greenland an also be de rface pressur
Figure 8 30 y
th standard N emperature even if the m
w to be a bett AO Index. Th h Swedish are lly represent ibing winter sely (2013), s tely capture of SLP value sterly wind fl eme SLP may
that the stan he NAOE Inde sentation for
AO affects no ng winter tem s used in this
e MERRA da e seen in App
e surface aro 3°N. Without is clearly se etected. The re to sea leve
years mean win
NAO and NA may be a sig measuring po
ter descriptio he standard N eas than the
t the wester temperature stating that t
d by station‐
s may be suf ow over Nor in some cas
ndard NAO In ex. The NAO r the actual a ot only Scand mperature in
s study.
ta have to b pendix A (Fig ound 30 to 20
t having cont en in Figure SLP‐contour el pressure.
nter‐SLP over No
OE Index hav gn of how the oints of NAOE
on of the NA NAO Index s extreme pre lies better. T es in Scandin the atmosph
‐based stand fficient to ca rth Atlantic.
ses not affect
ndex is a bett E Index may atmospheric dinavia but a n other areas
e taken into gure 8). The m
0°W, 30 to 3 tour lines of 8 and parts rs of land ma
orth Atlantic ba
ve medium‐
e large scale E Index are m
AO’s impact o howed slight essure‐based The westerlie navia. This is heric process dard NAO Ind
tch the sour Considering t anything th
ter represen y nonetheless
state of NAO almost every s could be a c
account. Cle mean winter 35°N and a m land masses of Europe, M asses derive f
ased on the ME
high correlat e systems are moving they
on Scandinav tly better co d NAOE Inde es are import
also consist es related to dex. This imp ces to the pr the NAOE In hat has impa
tation for Sc s still be a be O. In the intr ything adjace
complement
ear signs of e r‐SLP for 30 y minimum pre
s some of the Mediterranea from extrapo
ERRA data used
tion with the e consistent o
seem to cap
vian winter w orrelation wit ex. The stand tant to takin tent with the o atmospher plies that the ressure grad ndex, the mo acts on Scand
candinavian w etter, or at le roduction pa
ent to the No t to the Scan
errors from t years is desc ssure surfac em are neve an and north olation error
d in this study.
e Scandinavia over large ar pture the
weather th dard NAO ng into e study ric
e ients oving dinavian
winter east
rt of this orth
dinavian
the SLP ribed by e around rtheless hern
rs, going
an reas.
Scandinavian winter temperature almost as good as NAO Index. The method to describe the Scandinavian winter temperature from North Atlantic atmospheric behavior may be closer to the westerlies rather than the phase of NAO. Therefore, the standard NAO Index may be preferred for this purpose.
Despite above results, the suggested extreme‐SLP‐difference‐classes shows a very high correlation with mean monthly mean area winter temperature averaged over every extreme‐SLP‐difference‐
class. Increased pressure‐difference‐class is shown to be clearly followed by increased temperature (Figure 7). The correlation between the extreme‐SLP‐difference‐classes and the mean Scandinavian winter temperature anomaly per class (Table 5) is strongly higher than indices correlation to the Scandinavian winter temperature anomaly.
The synoptic scale circulation systems over North Atlantic enhance or weaken the westerlies which in addition affect the European weather. The extreme pressures (and pressure centers) over North Atlantic are non‐stationary. Hence, they may in many cases miss processes connected to the westerlies. The extreme pressures may therefore not always be the best representation for the strength of the westerlies. They may however still be a good representation of the state of NAO. This suggests that an index describing the NAO phase should be separated from an index describing NAO’s impact on Scandinavian winter temperature. Then the results from this study suggest that a SLP difference Index could better represent the Scandinavian winter temperature than a difference based on normalized anomalies from a 30‐year mean value (NAOE or NAO Index).
Why are the extreme‐SLP‐classes higher correlated to the Scandinavian winter mean temperature anomalies?
This can be because of that the classes are the direct strength of SLP difference and not anomalies from a 30‐year mean. The direct SLP difference may be more linearly connected to Scandinavian winter temperature than the anomaly. It may also derive from the normalization with standard deviation. On the other hand, it can simply derive from the class‐averaged temperatures.
The results showed us that the Norwegian winter temperature was almost as well represented with NAO (correlation of 0.47) as with NAOE Index (0.48). The difference between the NAO and NAOE Index’ representation of Swedish winter temperature (correlation of 0.59 and 0.53) was more significant. This suggests that the NAO Index represents the Norwegian winter temperature nearly as good as NAOE Index (difference of 1/100). However, for the Swedish areas the NAO Index may be the preferred index. In addition, Norwegian winter temperature anomaly has a lower correlation than the Swedish with both NAO and NAOE Index. This is consistent with other studies (e.g. Hurrel, 1996).
The correlation between NAO Index and Swedish winter temperature anomalies was shown to be statistical significant while the same correlation for NAOE Index was not. This shows that in addition to the higher correlation is NAO Index also more statistical reliable for describing Swedish winter temperature anomalies. For Norway’s correlation however, none of the indices was statistical significant.
A lack of number of winters available has been a problem in this study. Since I am covering only one climate period, the temporal scale may be insufficient to observe such phenomenon as analyzed in the present study. The reanalysis data covers the lack of spatially distributed historical observations
over North Atlantic. Nonetheless, to cover the temporal scales a climate model could be used. With verified climate models, runs over longer time scales could be executed.
I am using the mean pressure over every month and simply find the grids with largest and smallest SLP for every month. This can lead to maximum and minimum SLPs incorrectly located to the NAO phenomenon. Also, limiting the area over southern and northern location of the North Atlantic could miss the important pressure systems. A more complex algorithm to find the SLP in center of the major pressure systems could be a better suggestion. Then using these values in the same way as the NAOE Index could lead to an improvement.
Recalling the dynamics section (1.3), we have seen how external forcing such as transient eddies entering the North Atlantic can influence the variations characterizing the NAO pattern. To a station‐
based index, such processes are completely missed and may describe the noisy appearance of standard NAO Index.
There’s only a small probability that the NAO pattern can be predicted. The parameters of NAO are intrinsically connected to many other atmospheric processes and without fully understanding the dynamics of NAO there is little hope to be able to forecast the NAO pattern. Studies suggest however that because of the tilt of the pressure field with height, the NAO phase can be predicted with one month in advance during winters (Greatbatch, 2000). That is, the stratospheric pressure field is tilted forward so that it carries the information of the next tropospheric NAO phase.
Note that only low‐frequency variations are covered in this study. To cover the dynamics more comprehensively, a study of high‐frequency variations would be preferred. Many processes can be lost when limiting the study to low‐frequency but still, the variations may be too noisy if using high‐
frequencies. Additionally, when relating pressure variations over North Atlantic to temperature variations in Scandinavia may low‐frequencies be a better representation. High‐frequencies may be too short to capture the pressure feedback on Scandinavian temperature.
6 Conclusions
The reliability of the standard station‐based NAO Index to represent the impact of the NAO pattern on monthly mean winter temperature in Scandinavia has been of focus in this study. The monthly maximum and minimum winter SLPs over the North Atlantic for a 30‐year period have been analyzed. The difference between them has been connected to the monthly mean winter
temperature in Scandinavia. A new index of the NAO pattern has been suggested to better describe this temperature.
The temperature dependence of standard station‐based NAO Index and the new extreme SLP‐based NAOE Index has been analyzed. It is found that both NAO Index and NAOE Index capture the
temperature anomalies equally well. Additionally the indices are well correlated with each other.
However, the suggested NAOE Index does not show to be a better representation for the Scandinavian temperature than NAO Index.
The interesting extreme‐SLP‐difference correlation with Scandinavian winter temperature suggests that this could be a better representation of the North Atlantic pressure fields’ impact on
Scandinavian winter temperature. That is, the NAO phase should be described with one index but the
