Climate Patterns: Origin and Forcing

This brief review described spatial-time climate patterns generated by the dynamics and thermodynamics of the Earth’s climate system and methods of identifying these patterns. Specifically, it does discuss the following major climate patterns: El Niño-Southern Oscillation (ENSO), Cold Ocean-Warm Land (COWL) pattern, Northern and Southern Annular Patterns (NAM and SAM), Atlantic Multidecadal Oscillation (AMO) and Atlantic Meridional Overturning Circulation (AMOC), Pacific North-American Pattern (PNA) and Pacific Decadal Oscillation Pattern (PDO). In view of an extensive number of publications on some climate patterns, such as the ENSO, which discussed in many hundred of publications, this review is not intended to cover all the details of individual climate patterns but intends only to give a general overview of their structure, mechanisms of their formation and response to external forcing. It is assumed that the climate patterns can be treated as attractors of dynamical systems allowing us to extract and predict some specific features of the patterns such as the origin and evolution of the climate patterns and their role in climate change. Thus the knowledge of patterns allows the climate prediction on long time scales and understanding of how an external forcing affects the frequency of occurrence of climate patterns and their magnitude but not the spatial structure.

preferred flow regimes using model clustering and joint PDFs.
Here we briefly review the knowledge of characterization and origin of climate patterns and their role in climate change. We describe a set of major climate patterns, the physical mechanisms of their generation, and the external forcing of these patterns. There is an extensive number of publications on almost all major climate patterns, such as the ENSO and Annular patterns. This brief review is not intended to cover all the details of individual climate patterns and intends only to give a general overview of their structure, mechanisms of their formation and response to external forcing.
In Section 2 we outline the methods used for climate pattern identification.
Section 3 presents a short notion of potential mechanisms generating the climate patterns. Section 4 represents a set of selected major climate patterns generated by atmospheric-ocean processes. We give a description of a climate pattern, followed by review of suggested mechanisms of its generation and the external influence on that pattern. Section 5 presents a discussion of pattern persistence and its possible role in long-term climate prediction.

Methods of Pattern Identification
There are well-developed methods of finding patterns in chaotic and deterministic systems. Here we describe the major methods used in climate studies.
The method employs the data in the form of a matrix ( ) ( ) ( ) 1 , , m x t x t = X , wherein the Earth case X is a m × n-dimensional vector usually presented by m spatial pixels in a set of data maps taken at discrete times t = 1:n. The idea is to seek a linear combination of the columns of matrix X with maximum variance.
The PCA is defined as an orthogonal linear transformation that transforms the data to a new coordinate system such that the greatest variance by some scalar projection of the data comes to lie on the first coordinate (called the first principal component), the second greatest variance on the second coordinate, and so on. The data in every pixel are typically centered, with the mean over time values where The U(x) functions are usually called "Empirical Orthogonal Functions" (EOF) and V(t) is called "Principal Components" (PCs). The term "empirical" emphasizes the fact that these functions determine the covariance (second-order correlations) between the observed spatial data points. It does not assume any physical causes of the covariance. The PCs can also be reconstructed using the data, EOFs and λs: The EOFs that result from the analysis often difficult to interpret in terms of physical processes. It might be beneficial to rotate the orthogonal basis to another basis, which can be better explained in terms of physical forces. Upon rotation, we will lose a nice property that EOFs have an orthogonal basis (no crosscorrelations between EOFs). We will perhaps also lose orthonormality of the EOFs matrix if we choose a non-orthogonal transformation of the data. It is also important to note that these rotations do not use any particular property of the EOFs (such as orthonormality) and you essentially reduce EOF analysis to noise reduction-via the reduction in the number of EOF-after performing these rotations. Some rotational methods retain the orthogonality of the modes but not the principal components or EOFs. Most commonly, the rotation has been used with the varimax code. The objective is to minimize the mode complexity by making the large loadings larger and the small loadings smaller (Jolliffe, 2002).

Clustering Methods
Another approach for pattern identification is the use of clustering methods. For example, the International Satellite Cloud Climatology Project used the K-means cluster technique (Hartigan, 1975) to identify cloud and weather regimes (Rossow et al., 2005). A cluster analysis had been applied to the CloudSat data from the A-train formation of satellites for identifying the type of clouds (Sassen & Wang, 2008). Johnson and Feldstein (2010) used the K-means clustering to investigate the spatial and temporal variability of the wintertime North Pacific sea level pressure (SLP).
The K-means method uses K centers of clusters determined by minimizing the sum of squared errors of the data with a cost function are the cluster centers with n k being the number of data points in the cluster k. The second sum (the sum over j) is taken over the data points in the cluster k.
As the next step in development of methods of data clustering preserving the information containing in the original data, Ruzmaikin and Guillaume (2014) explored a more advanced algorithm called Deterministic Annealing (Rose, 1998), which is based on the minimization of the cost function relative to two independent parameters (the distance between data points and the Shannon entropy of the data distribution) and provides probabilities with which data are associated with each cluster. The method has a close and deep analog to the clas-

Mechanisms of Pattern Formation
The climate patterns are created by internal dynamics and thermodynamics of the Earth's climate system.
An important component of the dynamics is noise. Hasselmann (1976) introduced a model of climate variability driven by random noise excitation with short time scale ("weather") disturbances. Application of this model to sea-surface temperature (SST) (Frankignoul & Hasselmann, 1977) and to zonally-averaged energy (Lemke, 1977) produced a red-noise response spectrum with the most of its variance concentrated at very long periods. However uncorrelated noise forcing does not generate any correlations necessary for creating spatial patterns.
The dynamics that include spatial correlations come from the Earth's rotation and large-scale waves. Carl-Gustaf Rossby (Rossby, 1939) was the first to emphasize the importance of the two main ingredients of the atmospheric dynamics: the zonal-mean zonal wind and non-zonally symmetric deviations of pressure or geopotential heights. He described the non-zonally symmetric deviations as waves, which are now known as "planetary waves" or "Rossby waves" (Holton, 2004). The air or the ocean fluid on the Earth that moves toward the pole will deviate east; a fluid (or air) moving toward the equator will deviate west (true in either hemisphere). This effect is caused by the Coriolis force and the conservation of the potential vorticity leading to changes of relative vorticity analogous to the conservation of angular momentum in mechanics. The phase speed of the Rossby waves is always directed from West to East and is where u is the basic zonal speed, β = 2Ωcosφ/R is the so-called Rossby parameter (Ω is the Earth angular velocity, R is the Earth radius, k is the zonal wavenumber, and l is the meridional wavenumber.) Thus Rossby waves owe their origin to the gradient of the tangential speed of the planetary rotation (planetary vorticity). The lowest wave mode ( with the minimal k = 1) changes its sign once over the 360˚-long longitudinal circle.
The planetary waves are generated by winter flow over mountains and by sea-land temperature contrasts and propagate in horizontal and vertical directions (Charney & Drazin, 1961). The vertical propagation of the waves into the stratosphere along with the decreasing air density dramatically increases their amplitude. This increase often leads to nonlinear wave breaking accompanied by energy release that produces temperature anomalies and sometimes reverses the direction of the zonal wind. The zonal wind in turn affects the wave propagation by modifying the wave refraction index.
Another wave that defines the earth's dynamics is the "Kelvin wave", which  (Gill, 1982).
An example of thermodynamically generated pattern is COWL (Cold Ocean Warm Land) pattern. Wallace et al. (1996) argued that the COWL is essentially induced by the land-sea temperature distribution. Wallace et al. (1995) showed that the contrast in thermal inertia between land and ocean is responsible for the existence of the COWL pattern. The land surfaces, with their small heat capacities, equilibrate much more rapidly with the temperature of the overlying air mass than does sea surface temperature and thus experience larger temperature variability in response to month-to-month changes in atmospheric circulation patterns. It follows that hemispheric mean surface air temperature is largely determined by the temperature of the continents, even when surface air temperature over the oceans is taken into account in the averaging. By allowing the atmosphere to respond to ocean mixed layer temperature fluctuations only via heat exchanges directly aloft i.e. excluding the air-sea interaction in the coupled land-ocean model, Broccoli et al. (1998) have found that the coupling between atmospheric and oceanic circulations does not play a critical role in existence of this pattern.

A Short Introduction of the Major Climate Patterns
Now we briefly describe some major climate patterns formed in the Earth climate system. In view of many years of investigations and numerous publications, each of the patterns listed below deserves an extensive review so that this presentation could not be considered in any way as to complete and up to date.

ENSO Pattern
El Niño and La Niña are the warm and cool phases of a recurring climate pattern across the tropical Pacific called El Niño-Southern Oscillation, or ENSO ( Figure   1). The pattern can shift the phase back and forth irregularly on the time scale of two to seven years, and each phase shift triggers changes of sea surface temperature (SST), air temperature, precipitation, and winds. These changes disrupt the large-scale air movements in the tropics, triggering a cascade of global side effects. The first written record of the impacts of El Niño was made in 1525 when the Spanish conquistador Francisco Pizarro observed rainfall occurring in the Peru deserts. Walker (Walker, 1928;Walker & Bliss, 1932), who found a connection between barometer records on the East and West Pacific (between Tahiti A. Ruzmaikin and Darwin) and its variability, called this pattern "the Southern Oscillation." Bjerknes (1969) noticed that the default state of the sea surface temperatures at the East Pacific is remarkably cold for such low latitudes. Since the western Pacific is relatively warm, a large SST gradient exists along the equatorial Pacific.
As a result, there is direct thermal circulation in the atmosphere along the Pacific. The cool dry air above the cold eastern equatorial Pacific waters flows westward along the surface toward the warm West Pacific. There, the air is heated Bjerknes (1966,1969) associated the feedback loop of the oceanic and atmospheric circulation over the tropical Pacific as a "chain reaction", noting that "an intensifying Walker Circulation also provides for an increase of east-west temperature contrast that is the cause of the Walker Circulation in the first place." Bjerknes also found that this interaction could operate in the opposite: of 1979008 showed that the eastern tropical Pacific has undergone cooling while the western Pacific has undergone warming over the past three decades, causing an increase in the SST gradient. Since the SST trend was attributed to more frequent occurrences of central Pacific-type El Niño in recent decades, it is suggested that the decadal variation of El Niño caused the intensified Walker circulation over the past 30 years.
A comprehensive review of current understanding of the spatio-temporal complexity of this climate cluster mode and its influence on the Earth system has been given by Timmermann et al. (2018). It has been shown that the leading The oscillatory nature of ENSO requires mechanisms that include both positive and negative ocean-atmosphere feedbacks. As nicely reviewed by Wang (2001), the delayed oscillator, the western Pacific oscillator, the recharge-discharge oscillator (Jin, 1997;Timmermann et al., 2018) and the advective-reflective oscillator (Picaut et al., 1997) have been proposed to interpret the oscillatory nature of ENSO. All of these oscillator models have a positive ocean-atmosphere feedback in the equatorial eastern and central Pacific hypothesized by Bjerknes (1969). Each, however, has different negative feedbacks that turn the warm (cold) phase into the cold (warm) phase. In the delayed oscillator, free Rossby waves generated in the equatorial Eastern Pacific propagate westward and reflect from the western boundary as Kelvin waves. Since thermocline depth anomalies for the returning Kelvin waves have signs opposite to those in the equatorial eastern Pacific, these anomalies provide negative feedback for the coupled ocean-atmosphere system to oscillate. In the Western Pacific oscillator, equatorial easterly wind anomalies in the Western Pacific, which are produced by Western Pacific off-equatorial cold SST and high SLP anomalies, induce an ocean upwelling response that evolves eastward along the equator to provide negative feedback. In the recharge-discharge oscillator, equatorial wind anomalies in the central Pacific induce the meridional Sverdrup transport that recharges (or discharges) equatorial heat content. It is the recharge-discharge process that leaves an anomalously deep (or shallow) equatorial thermocline that serves as the phase transition for the coupled ocean-atmosphere system. The advective-reflective oscillator assumes that anomalous zonal currents associated with wave reflection at the ocean boundaries and mean zonal current tend to stop the growth of El Niño. The unified oscillator model, Equation (2) includes all of the physics (Wang, 2001): where T is the SST anomaly, h is the thermocline anomaly, τ 1 and τ 2 are zonal wind stress anomalies. The parameters a, b 1 , b 2 , c, d, e are constants, the parameters η, δ and λ represent the delay times, and the parameters E, R h , 1 R τ and 2 R τ are damping coefficients. Ruzmaikin (1999) suggested considering ENSO as a stochastic driver that excites the atmospheric anomaly states (Figure 2). This idea led to a concept to make 11-year solar activity forcing of climate feasible through stochastic resonance -a mechanism that amplifies a weak input to a nonlinear bistable system by the assistance of noise (Gammaitoni et al., 1998).  (Ruzmaikin, 1999).

The Climate Pattern Generated by Sea-Land Thermal Contrast (COWL)
One of the dominant modes of natural variability in the Northern Hemisphere is called the "Cold Ocean-Warm Land" (COWL) pattern. The COWL pattern was identified by partitioning the observed winter season time series of monthly mean surface air temperature into a very slowly varying radiative component, and a component exhibiting rapid year-to-year fluctuations, the latter comprising the COWL pattern (Wallace et al., 1995(Wallace et al., , 1996Quadrelli & Wallace, 2004), Figure 3. Broccoli et al. (1998) demonstrated that the COWL pattern appears to be a robust feature that can be extracted from both the observations and coupled model. Wallace et al. (1996) pointed out that the COWL pattern does not appear as a single EOF of the 500 hPa heights. Quadrelli and Wallace (2004) showed that the COWL pattern can be reconstructed as a linear combination of the first two EOFs of monthly mean December-March sea level pressure. Using the Northern Hemisphere land station data, it was determined that roughly half of the temporal variance of monthly mean hemispheric mean anomalies in surface air temperature during the period 1900-1990 were linearly related to the amplitude of a distinctive spatial pattern in which the oceans are anomalously cold and the A. Ruzmaikin Figure 3. (a) The COWL pattern obtained by regression of the monthly mean geopotential heights at 500 hPa upon hemispheric-mean surface temperature anomalies for cold season months in 1946(Wallace et al., 1996. continents are anomalously warm poleward of 40˚N. Apart from an upward trend since 1975, to which El Niño has contributed, the amplitude time series associated with this pattern resembles seasonally dependent white noise. It is argued that the variability associated with this pattern is dynamically/thermodynamically induced and is not necessarily an integral part of the fingerprint of global warming (Wallace et al., 1995).
The internally generated variability in the COWL pattern identified in the coupled model integration was used to assess the importance of the upward trend in the amplitude of the observed structure-function over the last 25 years.
This trend, which has contributed to the accelerated anthropogenic warming of Northern Hemisphere temperature over recent decades, may not be purely random (Broccoli et al., 1998).

Annular Patterns (NAM and SAM)
The planetary wave-zonal wind interaction of the atmospheric dynamics generates the major climate patterns in the middle-high latitudes called annular modes   It was suggested and demonstrated in numerical simulations that the excitation of the first EOF (i.e. the NAM), which characterizes the zonally-symmetric anomaly of atmospheric circulation, involves interaction between the planetary waves and the zonal-mean flow in the atmosphere (Limpasuvan & Hartmann, 2000). The second EOF (PNA-type patern) reflects the non-zonally symmetric structure of the planetary waves (Quadrelli & Wallace, 2004).
The nonlinear wave-zonal flow interaction (Holton & Mass, 1976) can be envisioned as a dynamical system with two basic states in its phase space corresponding to positive (negative) anomalies (Chao, 1985;Yoden, 1987;Ruzmaikin et al., 2003Ruzmaikin et al., , 2006a i.e. positive (negative) NAM. The system spends some time in residence at one or another state wandering between the two states (Ruzmaikin et al., 2003(Ruzmaikin et al., , 2006a. Thompson and Wallace (1998) noted that the strengthening of the polar vortex over the 30 years , unrelated to any known tropospheric forcing, has led to speculation that anthropogenically induced temperature changes at stratospheric levels might somehow be responsible. However the analysis of the residence time distributions for the Northern Annular Mode shows that the large difference of the tails of the residence time distributions for positive and negative phases of the NAM (characterized by the kurtoses) points to a temporal dominance of one of the phases in rarely occurring events (Ruzmaikin, 2009).
This suggests an unrelated to the global warming explanation of the dominance of the positive NAM in mid-1960s to the late 1990s indicated by Thompson and Wallace (1998).
It has been shown that the NAM index at different heights of the atmosphere is statistically significantly affected by the solar variability (proxied by solar 10.7 cm flux, Figure 5) (Ruzmaikin & Feynman, 2002). The effect varies depending on the time in the winter and the direction of the tropical stratospheric winds (the QBO), see Figure 5. Response of the stratosphere to solar variability, in particular at 30 hPa, and dependence of this response on the QBO phase was first discovered by Karen Labitzke (Labitzke, 1987) and further investigated by Labitzke and van Loon (For summary of their results see Labitzke (2004).) The most interesting extra finding by Ruzmaikin and Feynman (2002) was that at the  It has also been shown that the reconstructed sensitivity of the sea level temperature to a longer-term (multi-century) solar forcing in the Northern Hemisphere is in very good agreement with the empirical temperature pattern corresponding to changes of the NAM Ruzmaikin et al. (2004b). The temperature pattern (cold in Europe-warm in Greenland) associated with this mode was dominant during the Maunder Minimum.
Time evolution of the annular spatial patterns can also be traced. Data analyses (Kodera, 1995;Baldwin & Dunkerton, 1999) and modeling (Shindell et al., 1999;Gray et al., 2003) show that wind anomalies in the upper-middle stratosphere move poleward and downward during the winter. A greater fraction of stratospheric perturbations penetrates to the Earth's surface during solar maximum conditions than during solar minimum conditions (Hameed & Lee, 2005).
These anomalies are affected by the variable solar UV flux that impinges on ozone and temperature at the top of the stratosphere (Haigh, 1994

Pacific North-American Pattern (PNA)
PNA (Figure 7) had been identified as a pattern of the mid-tropospheric geopotential height field extending from the mid-Pacific to eastern North America (Wallace & Gutzler, 1981). Long-sustained winter regimes of alternating high and low pressure in Greenland with effects on climate in Europe were found to be associated with the pattern of the long waves in the upper westerlies showing a general reversal over the Northern Hemisphere and winter climate variability along the Atlantic coast of North America (Dickson & Namias, 1976). The positive phase of the PNA pattern is associated with above-average temperatures over western Canada and the extreme western United States, and below-average temperatures across the south-central and southeastern U.S.
The positive phase is also associated with an enhanced East Asian jet stream  Figure 7. The PNA pattern for January, April, July, and October, displayed so that the plotted value at each grid point represents the temporal correlation between the monthly standardized geopotential height anomalies at that point and the teleconnection pattern time series valid for the specified month (NOAA Center for Weather and Climate Prediction).
and with an eastward shift in the jet exit region toward the western United States. The negative phase is associated with a westward retraction of that jet stream toward eastern Asia, blocking activity over the high latitudes of the North Pacific, and a strong split-flow configuration over the central North Pacific. The positive phase of the PNA pattern tends to be associated with El Niño, and the negative phase tends to be associated with La Niña.
The mechanisms that form and drive PNA pattern were extensively investigated by Dai et al. (2017). The main mechanism involves a poleward-propagating Rossby wave train that has been excited by tropical convection. It also involves The PNA is affected by long-term solar variability (Ruzmaikin, 2007). Figure   8 shows the projection of the Total Solar Irradiance (TSI) on the PNA pattern.
The 21-century deep minimum of solar variability and the extended solar activity minima in the 19th and 20th centuries (1810-1830 and 1900-1920) are consistent with minima of the Centennial Gleissberg Cycle (CGC), a 90 -100 year variation of the amplitude of the 11-year sunspot cycle observed on the Sun and at the Earth (Feynman & Ruzmaikin, 2014). The Earth's climate response to these prolonged low solar radiation inputs involves heat transfer to the deep ocean causing a time lag longer than a decade. It had been found that the Pacific North American pattern (PNA) is a dominant spatial pattern of the climate response to CGC, which allows distinguishing the CGC forcing from other climate forcings (Ruzmaikin & Feynman, 2015). The CGC minima, sometimes coincidently in combination with volcanic forcing, are associated with severe weather extremes.
Thus the 19th-century CGC minimum coexisted with volcanic eruptions, led to The PNA pattern reconstructed by Trouet and Taylor (2009), see World Data Center for Paleoclimatology, https://www.ncdc.noaa.gov/). The characteristic features of the PNA pattern, such as the temperature anomaly of opposite signs over the USA, are seen in both images (Trouet & Taylor, 2009

Pacific Decadal Oscillation Pattern (PDO)
The PDO was first introduced by Mantua et al. (1997) as the leading EOF of North Pacific (20˚N -70˚N) monthly-averaged anomalies of Sea Surface Temperature (SST). The anomalies were defined as departures from the climatological annual cycle after removing the global mean SST (Figure 9). During a "positive" (warm), phase, the west Pacific becomes cooler and part of the eastern ocean warms; during a "cool" or "negative" phase, the opposite pattern occurs.

Influence of Climate Patterns on the Global Trend
Global warming experienced a pronounced hiatus during the period 1998-2013, which started from a very warm El Nino year 1998 (Trenberth & Fasullo, 2013;Trenberth et al., 2014;IPCC, 2014

Persistence of Climate Patterns
Here we will assume that climate patterns can be treated as the attractors of general dynamical systems, to which the climate system does belong. The assumption that the climate patterns are the attractors of the climate dynamical system is justified by a number of previous researchers such as Corti, Palmer, Ghil and other authors referred in this paper. A simple representation of attractors is potential wells (Khatiwala et al., 2001;Ruzmaikin, 2009 (Ruzmaikin et al., 2003) also indicates weak changes. Observational evidence of pattern change is limited so far but for example, Kodera (1995) found that during low solar activity the NAO pattern is confined to the Atlantic sector, while during the high solar activity the NAO-related anomalies extend over the whole Northern Hemisphere.
Introducing an external forcing would change either the phase states or the residence times (occupation frequencies) of the states. According to Rossby (1941), forcing does not change the states (i.e. the spatial structure of the climate patterns characterized by the EOFs) but only affects the mean residence times of the states. The Rossby conjecture was further advocated by Palmer (1999) and Corti et al. (1999). For visual illustration (Figure 11) Palmer presented a picture with two cups representing the phase states: a ball is randomly thrown from above for simulating occupation of the states, and a fan imitating the external force ( Figure 10(a)). He also supported the hypothesis by using as an example the Lorenz dynamical system, which (for a certain range of parameters) has two basic states. However, further analysis of the forced Lorenz system (Khatiwala et al., 2001) showed that the change in the mean residence times is a small effect compared with a more dramatic change in the tail of the probability distribution of the residence times, meaning the increase in the frequency of occurrence of extremely persistent events. The main reason for this is that the Rossby-Palmer conjecture missed an extra and critical feature of this no-linear system: the energetic barrier ∆U separating the system (the ball) in one of the states from transition to the other state (Figure 10(b)) (Ruzmaikin, 2007). As known from the 20th-century studies, the residence time in a state is exponentially proportional to the height of the barrier (the Kramers formula (Kramers, 1940)). An external forcing affects the depth of a state thus effectively increasing (or decreasing) the barrier. And, due to the exponential sensitivity of the residence time, even a small change of the barrier may induce noticeable effect on the time spent in that state. A numerical study of a model double-well potential system with stochastic transitions between the wells showed that when one of the wells is made deeper (by changing a parameter in the potential) the probability distribution of residence times in this well displays a longer tail (Khatiwala et al., 2001).

A Role of Patterns in the Long-Term Climate Prediction
Ed Lorenz in his celebrated paper on chaos Lorenz (1963) wrote: "When our results concerning the instability of nonperiodic flow are applied to the atmosphere, which is ostensibly nonperiodic, they indicate that prediction of the sufficiently distant future is impossible by any method unless the present conditions are known exactly. In view of the inevitable inaccuracy and incompleteness of the weather observations, precise very-long-range forecasting would seem to be nonexistent". However, in response to this statement Bunimovich (2015) provided some surprising mathematical results suggesting that a long-term forecast in dynamical systems with complex behavior is not so hopeless. After rigorously proving a prediction theorem for a simple chaotic dynamical system, he shows that by collecting enough data it is possible in principle to uncover the hierarchy of states for chaotic dynamical systems. This finding indicates that combining an analysis of chaotic bursts (transitions through chaotic dynamics) between rather regular states of the atmosphere (like cyclones, anticyclones and other types of eddies) may improve a weather/climate forecast on long time scales. Figure 11. A pictorial illustration of possible mechanisms of external influence on a climate pattern: (a) As envisioned by Palmer (1999): Solid caps correspond to two states of the pattern. Random population of the cups is controlled by dropping a ball. Forcing is depicted as a fan, which tends to blow the ball toward the left-hand cup. (b) As suggested by Ruzmaikin (2007) (see also Khatiwala et al. (2001)): There is a barrier between the states. Random transitions from one state to another are controlled by internal Earth's dynamics. The external forcing (such as solar variability) slightly changes the depth of one of the potential wells for some time leading to an exponentially amplified increase of the residence time in that well and, as a consequence, a longer persistence of this state.  Lovejoy (2013) argued that there are three qualitatively different regimes in the weather-climate system: The high-frequency regime is clearly the weather and the low-frequency regime is clearly "the climate", but there is also an in-between regime had been described with a spectral plateau as "low-frequency weather". It was dubbed "macroweather" because it is a kind of large-scale weather (not small-scale climate) regime. In each regime, the standard deviation is S(δt) ≈ δt H , so that the standard deviations of the fluctuations at "weather", "macroweather", and "climate" scales are roughly power laws (scaling) and are distinguished by their exponents. This finding generalizes the stochastic approach introduced by Hasselmann (1976) but still lucks of the role of spatial-temporal correlations that form the climate patterns.

Conclusions
After all the questions arise: 1) why should we be interested in climate patterns, and 2) is there a need to advantage the knowledge of them.
• It is well excepted that specific climate patterns, for example, El Niño or PDO, greatly influence the weather conditions not only locally but also over the globe. • The climate patterns may substantially affect the global trend (see Section 5.1).
• External forcing of climate, such as solar forcing, can be well seen in climate patterns and may be deemed and difficult to record globally.
• Climate patterns allow prediction of long-term evolution of climate.
• To advance the knowledge of climate patterns it would be important to better understand the mechanisms of their formation.
• Use the arsenal of accumulated knowledge of dynamical systems that treat climate patterns as attractors of the Earth's dynamical system.
Currently, there are an extensive number of publications that include the weather effects of specific climate patterns, such as the ENSO, PDO and NAM. However, there is a gap in research and publications devoted to investigation of mechanisms of climate pattern formation and their role in climate prediction. This review is an attempt to fill this gap and stimulate research of climate patterns both from observational and modeling viewpoints.