A new observational evidence of generation and propagation of barotropic Rossby waves induced by tropical instability waves in the Northeastern Pacific
Kang-Nyeong Lee1, Chanhyung Jeon2, YoungHo Seung1, Hong-Ryeol Shin3, Seung-Kyu Son4, and Jae-Hun Park1*
1Department of Ocean Sciences, Inha University, Incheon 22212, South Korea.
2 Department of Oceanography, Pusan National University, Busan 46241, South Korea.
3Department of Atmospheric Sciences, Kongju National University, Kongju 32588, South Korea.
4Deep-sea and Seabed Mineral Resources Research Center, Korea Institute of Ocean Science & Technology, Busan 49111, South Korea.
Corresponding author: Jae-Hun Park (jaehunpark@inha.ac.kr)
Key Points:
Abstract
Tropical instability waves (TIWs) in the equatorial eastern Pacific (EEP) exhibit 25–40-day westward-propagating fluctuations with seasonal and inter-annual variations, which are stronger during July–December and La Niña periods. They likely transfer their energy northward by forming barotropic Rossby waves (BTRWs). Long-term near-bottom current measurements at 10.5°N and 131.3°W during 2004–2013 revealed a spectral peak at 25–40 days, where significant coherences were found with satellite-measured sea surface height in a wide region of EEP with maxima approximately 5°N. Simulated deep currents from a data-assimilated ocean model concur with the observed near-bottom currents, and both currents vary seasonally and interannually, consistent with the typical characteristics of TIW. Further analyses using 25–40-day bandpass-filtered barotropic velocity data from the model revealed that they reasonably satisfied the theoretical dispersion relation of TIW-induced BTRW (BTRWTIW). We reconfirmed BTRWTIW propagating northward above 10°N in the northeastern Pacific by in-situ observations.
Plain Language Summary
Tropical instability waves (TIWs), which are located at the boundary between the warm pool and the cold tongue in the eastern Pacific, propagate westward with 25–40-day periods and vary seasonally and interannually, which are stronger during July–December and La Niña periods. Near-bottom velocity measured over a 10-year period at 10.5°N, 131.3°W just above the northern boundary of the waves fluctuates with 25–40-day periods, coinciding with that of sea surface height (SSH) in the equatorial eastern Pacific, especially around 5°N. We find that the wavelike pattern has wave crests oriented southeast-northwest from the model, and that this pattern appears across the study area and has characteristics consistent with TIWs including seasonal and interannual variations with the typical wavenumber and frequency. This pattern was verified to be a barotropic Rossby wave (BTRW) through a model result analysis. Thus, TIWs induce BTRWs that transfer their energy to the abyssal ocean above 10°N in the northeastern Pacific. This study provides a new observational evidence that near-bottom currents vary with BTRWs induced by TIW.
1 Introduction
Tropical instability waves (TIWs), which propagate westward and have a cusp-like shape with repetitive high amplitudes near 5°N around the boundary of the cold tongue in the equatorial eastern Pacific Ocean, can be observed using satellite-measured sea surface temperature (Legeckis, 1977; Legeckis et al., 1983) and sea surface height (SSH) (Lyman et al., 2005; Farrar, 2011; Holmes and Thomas, 2016; Tchilibou et al., 2018). It is known that TIWs result from instability by interactions between equatorial current system such as the Equatorial Undercurrent, the South Equatorial Current, the North Equatorial Current, and the North Equatorial Countercurrent (Philander, 1976; Lyman et al., 2005). Previous studies described broad ranges of wavenumber and frequency of TIWs depending on measurements utilized for them. Lee et al. (2017) summarized the previous estimates of the wavenumbers and frequencies of TIWs over the spectrum and reported that the TIWs observed by SSH measurements show peak near periods of 33 days and wavelengths of 12°–16° in the wavenumber-frequency spectrum.
The waves are representative phenomena with intraseasonal periods in the tropical eastern Pacific Ocean, although these properties are not always remarkable (Chelton et al. , 2000; An, 2008; Shinoda et al. , 2009). TIWs exhibit seasonal variations in the occurrence of intense growth from July to December, with more energetic activities during La Niña periods, linked to the strengthening of upwelling in response to strong trade winds in the equatorial eastern Pacific (Contreras, 2002; Warner & Moum, 2019).
Previous studies have focused mainly on the effects of TIWs near the equatorial ocean because it is known that the waves play an important role in regional ecosystems and the balance of heat associated with advection in the equatorial surface ocean (Willett et al., 2006; Moum et al., 2009). However, Farrar (2011) identified that TIWs can affect their energy up to approximately 20°N. The longitude-time band-pass filtered SSH shows a structure of TIW at 0°−10°N and a propagation of barotropic Rossby waves (BTRWs) induced by TIW north of 10°N. Furthermore, using both results from barotropic ocean model and newly gridded satellite-measured SSH with a mapping algorithm without latitudinal variation in its filtering properties, Farrar et al. (2021) showed that the propagation of the BTRWs continues until 35°N. However, these studies lacked in-situ observations.
Here, we used 10-year-long in-situ near-bottom current measurements that were recorded at a site located north away from the active region of TIW. The in-situ near-bottom current measurements clearly show that the energy of the TIW-induced BTRWs propagate northward. The processes of energy propagation in the form of BTRWs were also analyzed through the satellite-measured SSH as well as the results of data-assimilated numerical simulation (GLORYS12V1). In addition, the long-term in-situ measurements, satellite measurements, and results of GLORYS12V1 between 2004 and 2013 enable the verification of interannual variations according to the El Niño-Southern Oscillation (ENSO).
2 Data and Methods
2.1 In-situ and satellite measurements and GLORYS12V1 model results
Long-term, half-hour interval near-bottom current data (Uobs, Vobs) were recorded at a depth of ~5000 m in the northeastern Pacific (10.5°N, 131.3°W; black star in Figure 1a) from August 21, 2004 to July 27, 2013. The observations were conducted as part of the Korea Deep Ocean Study (KODOS). To compare in-situ data with other data explained below, the former were averaged over a day.
Farrar et al. (2021) noted that the SSH data product by Copernicus Climate Change Service causes barotropic signals with 30-day periods to disappear at higher than 20°N due to a mapping algorithm. They produced a special-purpose gridded SSH product which has latitudinally uniform filtering properties. In this paper, we used the newly gridded SSH data product (hereafter referred to as Farrar SSH) with a space-time grid of 0.5° × 0.5° × 3 days to conduct the spectral analysis and squared coherency analysis with our in-situ data subsampled at a 3-day interval. The domain used was 0°–20°N and 140°–80°W during the same period of near-bottom current measurements.
We also used the results of a data-assimilated global ocean reanalysis numerical simulation (GLORYS12V1) to investigate the characteristics of TIW-induced BTRWs. The GLORYS12V1 product is provided by the Copernicus Marine Environment Monitoring Service (CMEMS), and its component is the Nucleus for a European Model of the Ocean (NEMO) platform. The daily mean GLORYS12V1 outputs have a spatial resolution of 1/12° × 1/12°. The selected domain for the analyses is the same as that of Farrar SSH, but the data cover the period from January 1, 2004 to December 31, 2013. The velocity results at 4833-m depth filtered by using a band-pass filter with cutoff periods of 25–40 days are consistent with filtered in-situ near-bottom current measurements, showing high correlation of ~0.8.
2.2 Pre-processing of squared coherency
The squared coherency (hereafter referred to as coherence) between Farrar SSH and the time series of in-situ near-bottom current measurements was performed as follows. Spectral analysis was applied to 1088 -long time series with 3-day interval from August 21, 2004, to July 27, 2013. A hamming window of length 192 days was used on the segment, and a 50% overlap was used to increase the number of segments. The 95% significance level, determined by the number of segments and the window, is 0.137 (Thomson & Emery, 2014).
2.3 Complex empirical orthogonal function analysis
The Complex empirical orthogonal function (CEOF) analysis (Hernández-Guerra & Nykjaer, 1997) using the barotropic velocity results, calculated from the depth average of the numerical simulation, requires a preprocessing procedure. The results of the numerical simulation were filtered using a longitude-latitude-time band-pass filter (zonal wavelengths of 9°−20° in longitude, meridional wavelengths of 9°–20° in latitude, and periods of 25–40 days). The longitudinal band-pass filter has a variable cut-off length depending on the latitudes considered; however, the latitudinal band-pass filter has a constant cut-off length for all longitudes. These filtering steps were performed sequentially, first for longitude, next for latitude, and lastly for time. The three dimensions (longitude-latitude-time) filtered data were converted to two dimensions (spatio-temporal section) and the two components were concatenated along the row to consider a spatial relationship between them. The results of CEOF analyses are shown separately for zonal (Ubt) and meridional (Vbt) components.
3 Results
To compare the Farrar SSH and in-situ near-bottom current velocity (Uobs,Vobs) with each other, two time series of SSH located at different latitudes, indicated by black and red stars in Figure 1a, were used. One is located at the mooring observation site (SSHhigh), and the other is located at 5°N, 131.3°W (SSHlow). Figure 1b shows the time ­series of Farrar SSHlow, SSHhigh, and in-situ near-bottom current velocity (Uobs, Vobs) that were filtered by using a band-pass filter with cutoff periods of 25−40 days. Gray lines superimposed on the filtered data show the original time series. The time series corresponding to a red star are surrounded by a box with red dashed lines, and those to a black star are surrounded by a box with black dashed lines. The maximum speed of the original (filtered) Uobs and Vobs are 13.5 (2.8) cm/s and 16.7 (3.2) cm/s. The filtered time series of Uobs and Vobs exhibit similar variations to SSHlow in approximately a month period, which is consistent with the temporal variation of TIWs reported by Lyman et al., (2007). They are strengthened during the late summer and early winter months, with inter-annual variations. In contrast, the SSHhigh shows no resemblance to others and has substantially smaller values than the original time series. The results of the spectral analysis also show the same tendency. The spectral peak around the periods of 32 days clearly shows that the filtered time series has the similar periodicity to the TIW (top panels in Figure 1c). In contrast, the power spectral density (PSD) of the SSHhigh does not show any significant peaks around that period (gray line in Figure 1c).
Coherences between the SSHlow and the Uobs exhibit a maximum value (> 0.75) at the periods of 32 days and the Vobs show higher values (~0.4) than the significance level around the periods of 32 days (middle panel in Figure 1c). In the 32-day periods, the SSHlow leads the Uobs by 119°, and the Vobs leads the SSHlow by -7°. Conversely, coherences between the SSHhigh and either the Uobs or the Vobs appear to be much smaller than the significance level (0.137) in the 32-day periods. This disparate results seen at two latitudes will be discussed in Figure 3, by using Farrar SSH data and numerical simulation results.