A coupled atmosphere–ocean–wave model was used to examine mixing in the upper-oceanic layers under the influence of a very severe cyclonic storm Phailin over the Bay of Bengal (BoB) during 10–14 October 2013. The coupled model was found to improve the sea surface temperature over the uncoupled model. Model simulations highlight the prominent role of cyclone-induced near-inertial oscillations in subsurface mixing up to the thermocline depth. The inertial mixing introduced by the cyclone played a central role in the deepening of the thermocline and mixed layer depth by 40 and 15 m, respectively. For the first time over the BoB, a detailed analysis of inertial oscillation kinetic energy generation, propagation, and dissipation was carried out using an atmosphere–ocean–wave coupled model during a cyclone. A quantitative estimate of kinetic energy in the oceanic water column, its propagation, and its dissipation mechanisms were explained using the coupled atmosphere–ocean–wave model. The large shear generated by the inertial oscillations was found to overcome the stratification and initiate mixing at the base of the mixed layer. Greater mixing was found at the depths where the eddy kinetic diffusivity was large. The baroclinic current, holding a larger fraction of kinetic energy than the barotropic current, weakened rapidly after the passage of the cyclone. The shear induced by inertial oscillations was found to decrease rapidly with increasing depth below the thermocline. The dampening of the mixing process below the thermocline was explained through the enhanced dissipation rate of turbulent kinetic energy upon approaching the thermocline layer. The wave–current interaction and nonlinear wave–wave interaction were found to affect the process of downward mixing and cause the dissipation of inertial oscillations.
The Bay of Bengal (BoB), a semi-enclosed basin in the northeastern Indian Ocean, consists of surplus near-surface fresh water due to large precipitation and runoff from the major river systems of the Indian subcontinent (Varkey et al., 1996; Rao and Sivakumar, 2003; Pant et al., 2015). The presence of fresh water leads to salt-stratified upper-ocean water column and the formation of a barrier layer (BL), a layer sandwiched between the bottom of the mixed layer (ML) and the top of the thermocline, in the BoB (Lukas and Lindstrom, 1991; Vinayachandran et al., 2002; Thadathil et al., 2007). The BL restricts the entrainment of colder waters from thermocline region into the mixed layer; it thereby maintains a warmer ML and sea surface temperature (SST). The warmer SST together with higher tropical cyclone heat potential (TCHP) makes the BoB one of the active regions for cyclogenesis (Suzana et al., 2007; Yanase et al., 2012; Vissa et al., 2013). The majority of tropical cyclones are generated during the pre-monsoon (April–May) and post-monsoon (October–November) seasons (Alam et al., 2003; Longshore, 2008). The number of cyclones and their intensity is highly variable on seasonal and interannual timescales. The oceanic response to the tropical cyclone depends on the stratification of the ocean. The BL formation in the BoB is associated with the strong stratification due to the peak discharge from rivers in the post-monsoon season. The intensity of the cyclone largely depends on the degree of stratification (Neetu et al., 2012; Li et al., 2013). The coupled atmosphere–ocean model was found to improve the intensity of cyclonic storms when compared to the uncoupled model over different oceanic regions (Warner et al., 2010; Zambon et al., 2014; Srinivas et al., 2016; Wu et al., 2016). Zambon et al. (2014) compared the simulations from the coupled atmosphere–ocean and uncoupled models and reported significant improvement in the intensity of storms in the coupled case as compared to the uncoupled case. The uncoupled atmospheric model produced large ocean–atmosphere enthalpy fluxes and stronger winds in the cyclone (Srinivas et al., 2016). When the atmospheric Weather Research and Forecasting (WRF) model interacted with the ocean model, the SST was found to be more realistic compared to the stand-alone WRF (Warner et al., 2010; Gröger et al., 2015; Jeworek et al., 2017; Ho-Hagemann et al., 2017). Wu et al. (2016) demonstrated the advantage of using a coupled model over the uncoupled model in a better simulation of typhoon Megi's intensity.
Mixing in the water column has an important role in energy and material transference. Mixing in the ocean can be introduced by the different agents such as wind, current, tide, eddy, and cyclone. Mixing due to tropical cyclones is mostly limited to the upper ocean, but the cyclone-induced internal waves can affect the subsurface mixing. Several studies have observed that the mixing in the upper-oceanic layer is introduced due to the generation of near-inertial oscillations (NIOs) during the passage of tropical cyclones (Gonella, 1971; Shay et al., 1989; Johanston et al., 2016). This mixing is responsible for the deepening of the ML and the shoaling of the thermocline (Gill, 1984). The vertical mixing caused by storm-induced NIO has a significant impact on upper-ocean variability (Price, 1981). The NIO are also found to be responsible for the decrease in SST along the cyclone track (Chang and Anthes, 1979; Leipper, 1967; Shay et al., 1992, 2000). This decrease in SST is caused by the entrainment of cool subsurface thermocline water from the mixed layer into the immediate overlying layer of water. This cooling of surface water is one of the reasons for the decay of cyclones (Cione and Uhlhorn, 2003). The magnitude of surface cooling differs largely depending on the degree of stratification on the right-hand side of the cyclone track (Jacob and Shay, 2003; Price, 1981).
The near-inertial process can be analyzed from the baroclinic component of currents. The vertical shear of horizontal baroclinic velocities that is interrelated with buoyancy oscillations of surface layers is utilized in various studies in order to gain an adequate understanding of the mixing associated with high-frequency oscillations, i.e., NIO (Zhang et al., 2014). The shear generated due to NIO is an important factor for the intrusion of the cold thermocline water into the ML during near-inertial scale mixing (Price et al., 1978; Shearman, 2005; Burchard and Rippeth, 2009). The alternative upwelling and downwelling features of the temperature profile are an indication of the inertial mixing. The kinetic energy bounded with these components of the current shows a rise in magnitude at the right side of a cyclone track (Price, 1981; Sanfoard et al., 1987; Jacob, 2003). This high magnitude of kinetic energy is linked to strong wind and the rotating wind vector conditions of the storm. The spatial distribution of near-inertial energy is primarily controlled by the boundary effect for inertial oscillations (Chen et al., 2017). The NIO is found to decline with decreasing depth and vanishes in the coastal regions (Schahinger, 1988; Chen et al., 2017).
The aim of this paper is to understand and quantify the near-inertial mixing
due to the very severe cyclonic storm Phailin in the BoB. Phailin developed
over the BoB in the northern Indian Ocean in October 2013. The landfall of
Phailin occurred on 12 October 2013 around 17:00 GMT near the Gopalpur
district of Odisha state on the east coast of India. After the 1999 super
cyclonic event of the Odisha coast, Phailin was the second strongest cyclonic
event that made landfall on the east coast of India (Sanil Kumar and Nair,
2015). The low-pressure system developed in the north of the Andaman Sea on 7
October 2013 and was transformed into a depression on 8 October at
12
Most of the past studies on oceanic mixing under cyclonic conditions were carried out using in situ measurements, which are constrained by their spatial and temporal availability. To the best of our knowledge, the present study is the first of its kind to utilize a coupled atmosphere–ocean–wave model over the BoB to estimate cyclone-induced mixing, its associated energy propagation on the cyclone track, and the location of maximum surface wind stress during the period of the peak intensity of the cyclone. The study also focuses on analyzing the subsurface distribution of NIO with its vertical mixing potential. Further, the study quantifies the shear-generated mixing and the kinetic energy of the baroclinic mode of the horizontal current varying in the vertical section at a selected location during the active period of the cyclone. The dissipation rate of NIO and turbulent eddy diffusivity are quantified.
Numerical simulations during the period of Phailin were carried out using the
Coupled Ocean–Atmosphere–Wave–Sediment Transport (COAWST) model, described
in detail by Warner et al. (2010). The COAWST modeling system couples the
three-dimensional oceanic Regional Ocean Modeling System (ROMS), the
atmospheric WRF model, and the wind wave generation and propagation model
Simulating Waves Nearshore (SWAN). The ROMS model used for the study is a
free-surface, primitive-equation, sigma coordinate model. ROMS is a
hydrostatic ocean model that solves finite difference approximations of the
Reynolds averaged Navier–Stokes equations (Chassignet et al., 2000;
Haidvogel et al., 2000, 2008; Shchepetkin and McWilliams, 2005). The
atmospheric model component in the COAWST is a non-hydrostatic, compressible
model Advanced Research Weather Research Forecast Model (WRF-ARW), described
in Skamarock et al. (2005). It has different schemes for the representation
of boundary layer physics and physical parameterizations of sub-grid-scale
processes. In the COAWST modeling system, appropriate modifications were made
in the code of the atmospheric model component to provide an improved bottom
roughness from the calculation of the bottom stress over the ocean (Warner et
al., 2010). Further, the momentum equation is modified to improve the
representation of surface waves. The modified equation needs the additional
information of wave energy dissipation, propagation direction, wave height,
and wavelength that are obtained from wave components of the COAWST model.
The spectral wave model SWAN, used in the COAWST modeling system, is designed
for shallow water. The wave action balance equation is solved in the wave
model for both spatial and spectral spaces (Booij et al., 1999). The SWAN
model used in the COAWST system includes the wave wind generation, wave
breaking, wave dissipation, and nonlinear wave–current–wind interaction.
The Model Coupling Toolkit (MCT) is used as a coupler in the COAWST modeling
system to couple different model components (Larson et al., 2004; Jacob et
al., 2005). The coupler utilizes a parallel coupled approach to facilitate
the transmission and transformation of various distributed parameters among
component models. The MCT coupler exchanges prognostic variables from one
model to another model component as shown in Fig. 1. The WRF model receives
SST from the ROMS model and supplies the zonal (
The block diagram shows the component models WRF, ROMS, and SWAN of the COAWST modeling system together with the variables exchanged among the models. MCT, the model coupling toolkit, is a model coupler used in the COAWST system.
The coupled model was configured over the BoB to study Phailin during the
period of 00:00 GMT 10 October–00:00 GMT 15 October 2013. The setup of the
COAWST modeling system used in this study included fully coupled
atmosphere–ocean–wave (ROMS
The terrain-following ocean model ROMS with 40 sigma levels in the vertical
was used in this study. The ROMS model domain was used with zonal and
meridional grid resolutions of 6 and 4 km, respectively. This high
resolution in ROMS enables us to resolve mesoscale eddies in the ocean. The
vertical stretching parameters, i.e.,
The COAWST model domain (65–105
The baroclinic current component was calculated by subtracting the barotropic
component from the mean current with a resolution of 2 m in the vertical.
The power spectrum analysis was performed on the zonal and meridional
baroclinic currents along the depth section of the selected locations by
using the periodogram method (Auger and Flandrin, 1995). The continuous
wavelet transform using the Morlet wavelet method (Lilly and Olhede, 2012)
was carried out to analyze the temporal variability in the baroclinic current
at a particular level of 14 m. The near-inertial baroclinic velocities were
filtered by the Butterworth second-order scheme for the cutoff frequency
range of 0.028 to 0.038 cycle h
The analysis of the generation of the inertial oscillations and their
dissipation was performed on the basis of turbulent dissipation rate
(
Tracks of Phailin simulated by the coupled model (black) and IMD reported (red). The 3-hourly positions of the center of Phailin are marked with solid circles, and the daily position at 00:00 h is labeled with the dates. Location of buoy BD09 is marked with a blue circle.
The WRF model-simulated track of Phailin was validated against the India Meteorological Department (IMD) reported best track of the cyclone. A comparison of the model-simulated track with the IMD track is shown in Fig. 3. Solid circles marked on both the tracks represent the 3-hourly positions of the cyclone's center, as identified by the minimum surface pressure. The daily positions of the center of Phailin are labeled with the date. The WRF model in the coupled configuration does a fairly good job of simulating the track, translational speed, and landfall location of Phailin. The positional track error was about 40 km when compared to the IMD track of Phailin. The stand-alone WRF model (not shown here) was found to simulate Phailin's track in an almost identical way to the WRF in the coupled configuration. However, the intensity (surface wind speed) in the WRF stand-alone model was higher compared to the coupled model. Figure 4 shows the comparison of stand-alone and coupled WRF model-simulated mean sea level pressure (MSLP), wind speed, and wind direction at a buoy (BD09) location (marked with a blue circle in Fig. 3). It can be inferred from the figure that stand-alone WRF simulated a larger pressure drop and higher wind speed compared to buoy measurements. In addition to the cyclone-induced pressure drop during 10–12 October, the semidiurnal variations in MSLP were observed in the buoy measurements. These semidiurnal variations in MSLP, primarily due to the radiational forcing (Pugh, 1987), were not captured by the model over the cyclone-influenced region. The WRF in coupled model configuration shows better performance in simulating the surface wind speed and pressure during Phailin. The exchange of wave parameters with the WRF model in the coupled configuration provides realistic sea surface roughness that resulted in the improvement of surface wind speed.
Comparison of coupled model (green), stand-alone WRF model (red),
and observations from a buoy BD09 (black) for the
The daily averaged sea surface temperature (SST) in
Coupled model-simulated and diagnosed variables at the on-track
(left panel) and off-track (right panel) locations.
The SST simulated by the ROMS model in coupled and stand-alone configurations
was validated against the Advanced Very High Resolution Radiometer (AVHRR)
satellite data on each day for the period of Phailin's passage over the BoB.
The stand-alone WRF-simulated parameters were used to provide surface
boundary conditions in the stand-alone ROMS model. Figure 5 shows that the
coupled model captures the SST spatial pattern reasonably well with about
The coupled atmosphere–ocean–wave simulation is an ideal tool to understand the air–sea exchange of fluxes and their effects on the oceanic water column. Surface wind sets up currents on the surface as well as initiating mixing in the interior of the upper ocean. In order to examine the strength of mixing due to Phailin, the model-simulated vertical temperature profile together with the surface wind speed, zonal and meridional components of the current, and kinetic energy at the on-track and off-track locations are plotted in Fig. 6. Comparatively stronger zonal and meridional currents were observed at the off-track location than the on-track location on 12 October. The larger kinetic energy available at the off-track location leads to greater mixing, resulting in a deeper mixed layer on 12 October compared to the on-track location. The surface wind speed at the on-track location shows typical temporal variation in a passing cyclone. The wind speed peaks, drops, and attains a second peak as the cyclone approaches, crosses over, and departs the location. The surface currents forced by these large variations in wind speed and direction at the on-track location result in a comparatively weaker magnitude than the off-track location.
The thermocline, defined as the depth of maximum temperature gradient,
is usually given with reference to the location-dependent isotherm depth
(Kessler, 1990; Wang et al., 2000). Over the BoB region, the depth of
the 23
During the initial phase of Phailin, the zonal and meridional currents were
primarily westward and southward, respectively (Fig. 6c, d, h, and i).
However, on and after 12 October when the cyclone attained peak intensity and
crossed over the location, alternative temporal sequences running
westward/eastward in the zonal current and southward/northward in the
meridional current were noticed in current profiles (Fig. 6). The frequency
of these reversals in zonal and meridional currents is recognized as a
near-inertial frequency generated from the storm at these locations. The
direction and magnitude of currents represent a variability that corresponds
to the presence of near-inertial oscillations at the selected locations. The
kinetic energy (KE) of currents at various depths is a proxy of energy
available in the water column that becomes conducive to turbulent and
inertial mixing. Time series of KE associated with the barotropic and
depth-averaged baroclinic components of the current at the two point
locations are illustrated in Fig. 6e (on-track) and 6j (off-track). The KE
associated with the baroclinic component was found to be much higher than the
barotropic component of current at both on-track and off-track locations. The
depth-averaged baroclinic and barotropic current components' KE also depict
the impinging oscillatory behavior. The peak magnitude of KE in baroclinic
and barotropic currents at the off-track location was found to be
1.2 m
The power spectrum analysis was performed on the time series profiles at the
two selected locations to get a distribution of all frequencies operating in
the mixing process during the passage of Phailin. The power spectrum analysis
was performed on the zonal and meridional components of the baroclinic
current profile and shown in Fig. 7. It is clear from the figure that the
tidal (M2, the semidiurnal component of tide) and near-inertial oscillations
(f) are the two dominant frequencies on the surface during cyclone Phailin.
Under the influence of cyclonic winds, the NIO signal was stronger
(0.84 m
The power spectrum analysis (m
The second-order Butterworth filter was applied to the baroclinic current
components to get the strength of NIO in the frequency range of 0.028 to
0.038 cycles h
Daily averaged baroclinic kinetic energy (m
To investigate the energy propagation from the surface to the interior layers
of the upper ocean, we derived the rotary spectra (Gonella, 1972; Hayashi,
1979) of near-inertial wave numbers shown in Fig. 10. The daily averaged
vertical wave-number rotary spectra provide a clear picture of wind energy
distribution in the subsurface water. The anticyclonic spectrum
(
The scalogram by continuous wavelet transform (CWT) method in
percentage at 14 m depth . Wavelet scalogram shown for the zonal baroclinic
current
The daily averaged vertical wave-number rotary spectra of near-inertial oscillations. The anticyclonic and cyclonic spectra are represented by blue and dotted red lines, respectively.
The model-simulated bulk properties at the selected point location.
The vertical shear square axis is multiplied by a factor of 10
Profiles of
To examine the generation and dissipation of these inertial oscillations,
the shear generated by the near-inertial baroclinic current (
Processes controlling the subsurface mixing were evaluated under the high
wind speed regime of the severe cyclonic storm Phailin over the BoB. A
coupled atmosphere–ocean–wave (WRF
The atmospheric and ocean model forcing data can be
obtained from FNL (
KRP and TN performed model simulations and analyzed data. VP prepared the manuscript with contributions from all co-authors.
The authors declare that they have no conflict of interest.
ECCO2 is a contribution to the NASA Modeling, Analysis, and Prediction (MAP) program. The study benefitted from the funding support from the Ministry of Earth Sciences, Government of India, and the Space Applications Centre, Indian Space Research Organisation. The high-performance computing (HPC) facility provided by IIT Delhi and the Department of Science and Technology (DST-FIST 2014 at CAS), Government of India, are thankfully acknowledged. Authors are thankful to Lingling Xie for his productive suggestions. Graphics were generated in this paper using Ferret and NCL. The constructive comments from three anonymous reviewers helped to improve the paper. Tanuja Nigam and Kumar Ravi Prakash acknowledge MoES and UGC-CSIR, respectively, for their PhD fellowship support. Edited by: Markus Meier Reviewed by: three anonymous referees