Seasonal Variability of Global Intraseasonal SST Perturbations from 1982 to 2021

The seasonal variability of the intraseasonal SST perturbations is analysed based on the high spatial-temporal resolution and long-time series (1982-2021) OISST data. There are two bands of high intraseasonal variability (ISV, 20-90 days), one band is along the equatorial eastern Indian and western Pacific Oceans, the other is along the axis of the Antarctic Circumpolar Current (ACC). For large-scale perturbations, during boreal winter, the ISV of SST is large south of the equator, especially south of 30°S. During boreal summer, the situation is opposite and the ISV is more concentrated north of 30°N. During boreal spring, the ISV is increasing north of equator. During boreal autumn, the ISV is increasing south of equator while decreasing north of equator. For small-scale perturbations, the ISV of SST is large all the time over the Southern Ocean and western boundary currents. In the tropical ocean, the small-scale ISV in the eastern Pacific Ocean is apparent in summer and autumn. Compared to large-scale ISV, the distribution areas of the small-scale ISV are apparently narrower.


Introduction
There is various intraseasonal variability (ISV, 20-90 days) in the global ocean due to many different processes, for example, Madden Julian Oscillation (MJO), baroclinic instability, tropical intraseasonal oscillations (TISOs), and so on [1].Surface fluxes passing through the interface between the ocean and atmosphere, including momentum, moisture and heat, give rise to the perturbations of the sea surface temperature (SST), which also may affect surfaces fluxes in return.Sea surface temperature (SST), as the most important fundamental parameter in the ocean, is the key to understand the interaction at the air-sea interface.Krishnamurti showed that the intraseasonal (from 30 to 60 days) SST variability was high mainly concentrated in two areas, namely the western Pacific Oceans as well as the equatorial Indian Oceans [2].Hendon and Glick, based on 7 years of SST analyses which were weekly [3], also showed the same results [4].
Several articles which were compared with in-situ data have reached a conclusion that the weekly SST analysis [3] underestimates the intraseasonal variability on a global basis [5] and the TMI performs better [6].AVHRR measurements have constrained usage in retrieving SST near the area where the weather is cloudy, and this situation is more problematic in studying regions where the MJO precipitation variability is dominant [7].Although the TMI performs better, the time series is limited and the new Reynolds product has great improvements [8].
In this paper, the organization is as below: the SST data and analysis methods are introduced respectively in section 2; the distribution of the intraseasonal variability is analyzed in section 3; section 4 presents the large-scale and small-scale perturbations of the intraseasonal variability; section 5 is the conclusions.

Data
The new OISST product [8] is used to estimate and analyse the intraseasonal perturbation of the SST.The spatial grid resolution of the new OISST product is 0.25° and the temporal resolution is daily.There are two versions of the new OISST product.One version stems from the Advanced Very High Resolution Radiometer (AVHRR) satellite SST data and the AVHRR is the infrared remote sensor.The remote sensing SST data of the other version is not only from the AVHRR, but also from the Advanced Microwave Scanning Radiometer (AMSR) which is the microwave remote sensor.The AVHRR-only product spans from January 1982 to December 2021 and the AVHRR-AMSR product spans from June 2002 to December 2011.For data consistency, the AVHRR-only product is chosen to analyse in this paper (Figure 1).

Analysis methods
First, the intraseasonal SST anomaly is estimated using a 20 to 90-day Batterworth passband filter to filter the SST time series from 1982 to 2021.Based on the high-resolution data, the ISV of SST which is caused by two different processes can be possible to separate.The two different processes respectively are the small-scale process which is caused by the internal ocean waves and eddies and the large-scale process which is caused by the atmospheric perturbations [1].
Then, a box average smoothing method is used to extract the large-scale signals, which is based on the fact that these large-scale ISV signals are mostly affected by large-scale intraseasonal perturbations in the atmosphere and that the residual intraseasonal signals are small-scale and mostly affected by the small-and meso-scale process in the ocean, such as waves and eddies.
Considering the different characteristic scale among different latitudes, the size of the box is 12× 12 degree multiplying cos φ (φ, latitude).Taking the complexity of the ISV of SST near the Polar regions into account, the research area in this paper is limited from 60°S to 60°N.

Intraseasonal variability
According to the percentage of intraseasonal (20-90 days) SST variance, there are two bands of the high ISV (Figure 2).One band is mainly concentrated in two regions, one of which is the equatorial eastern Indian Oceans and the other is the equatorial western Pacific Oceans.The two regions are also corresponding to the area where surface winds and convection signals which are associated with the MJO are the strongest [9].Moreover, the high ISV is also concentrated in tropical eastern Pacific and tropical Atlantic Oceans, the intensity of which is yet weaker and the area is narrower, especially in the tropical Atlantic Oceans.The other band of the high ISV is along the axis of the Antarctic Circumpolar Current (ACC), corresponding to the most energetic area in the world ocean due to the eddy kinetic energy.
Besides the two apparent bands, the notable western boundary currents, such as the Kuroshio and Gulf Stream, also generate remarkable intraseasonal variability of SST.The high ISV associated with western boundary currents is primarily due to the small-and meso-scale processes, which is caused by the eddies or rings during the period when the currents separate from the coast and are eventually injected into the open ocean which is relatively quiescent [1].The high ISV regions along the equatorial eastern Indian and western Pacific reach the minimum during spring, while the maximum during summer.The other areas along the axis of the Antarctic Circumpolar Current (ACC) have higher ISV during summer and autumn than winter and spring (Figure 3).

Large-scale perturbations
The large-scale intraseasonal perturbations of SST are mainly forced by large-scale intraseasonal perturbations in the atmosphere, and potential feedbacks to these perturbations are further assumed.During boreal winter, the large-scale ISV is large south of the equator, especially south of 30°S (Figure 4a).Between equator and 30°S, the ISV in the Indian Ocean is larger, compared to the Pacific Ocean and Atlantic Ocean.In the western part of the Indian Ocean which is sometimes called the Seychelles-Chagos Thermocline Ridge (SCTR) [10], the ocean mixed layer is shallow and is around 20 m to 25 m [11].In the eastern Indian Ocean, especially the Timor and Arafura Seas and the Gulf of Carpentaria (ITAC), the ocean mixed layer is also shallow and is around 25 m to 30 m.The high ISV is increasing north of equator during boreal spring (Figure 4b) and meanwhile is more concentrated

Small-scale perturbations
In the Southern Ocean, the small-scale perturbations are large all the time (Figure 5).The poleward heat transport which is required to hold on the steady state across the Southern Ocean is large and eddies are the dominant mechanism due to the fact that there are no meridional boundaries which are used to support a gyre circulation [1].Besides the Southern Ocean, the western boundary currents also have the large small-scale perturbations all the time.In the tropical ocean, the small-scale ISV in the eastern Pacific Ocean is apparent in summer and autumn and is corresponding to the large tropical instability waves (TIW) signals.Besides, the smallscale ISV corresponding to the "great whirl" gyre during boreal summer originated from the Somali Current and the Tehuantepec and Papagayo Eddies during boreal winter from west of Central America is also apparent.Compared to large-scale ISV (Figure 4), the distribution areas of the small-scale ISV are apparently narrower, especially in the eastern Pacific Ocean and the western boundary currents.Because of the small scale, the role of these phenome in the beginning and development of large-scale tropical ISV in the atmosphere is secondary [1].

Conclusion
Based on the high spatial-temporal resolution and long-time series (1982-2021) OISST data, the seasonal variability of the intraseasonal SST perturbations is analyzed comprehensively.In this study, the distribution of the intraseasonal SST variability in the global ocean is discussed firstly.There are two bands of high ISV, one band is along the equatorial eastern Indian and western Pacific Oceans, the other is along the axis of the Antarctic Circumpolar Current (ACC).