High-frequency fluctuation and EHO-like mode in the H-mode pedestal on the EAST tokamak

Figure 4 shows the 2D local conditional wave number and frequency spectrum of the amplitude signal (${{{A}}_{{\text{HFF}},{\text{ K}}}}$), as measured by DFR-K at two poloidally separated positions on a single surface. It is evident from the figure that the EHO-like mode exhibits a distinct mode structure within the frequency range of $11{-}66{\text{ kHz}}$, with a poloidal wave number of approximately ${{{k}}_{{\theta }}} \unicode{x2A7D} 0.12{\text{ c}}{{\text{m}}^{ - 1}}$. Consequently, the poloidal rotation velocity of the EHO-like mode is estimated as ${\upsilon _{{\text{lab}}}} = 2\pi f/{k_\theta } = 34\,{\text{km}} \cdot {{\text{s}}^{ - 1}}$ in the direction of electron diamagnetic drift (EDD) in the lab frame. These characteristics align with EHO observed in DIII-D tokamak [30], both experimental and simulation results reveal that the EHO has a lower poloidal number (${{{k}}_{{\theta }}} \unicode{x2A7D} 0.12{\text{ c}}{{\text{m}}^{ - 1}}$ for $n = 5$), with the EHO identified as a low-n (${\unicode{x2A7D}} 5$) kink/peeling mode destabilized by edge $E \times B$ rotational shear. Due to the absence of the magnetic measurement in this shot, the toroidal number of the EHO-like mode's magnetic fluctuation in a different discharge will be presented.

Zoom In Zoom Out Reset image size

Figure 5 illustrates the evolution of modes and pedestal parameters during two large ELM spikes. The typical plasma $\# 79861$ (${{\text{B}}_{\text{t}}} = 2.2{\text{ T}}$, ${{{I}}_{\text{p}}} = 650{\text{ kA}}$, ${q_{95}} = 4$) is heated via LHW and electron cyclotron resonance heating. As observed in figures 5(d) and (e), the pedestal density and temperature gradient begin to rise around 3.811 s following the burst of the preceding large ELM. At ${{t}} = 3.814{\text{ s}}$, when the pedestal reaches a threshold, the HFFs are triggered as depicted in the density fluctuation spectrogram in figure 5(b). It is clearly seen a periodic alternation of HFF and a low frequency fluctuation (LFF) with ${{f}} < 500{\text{ kHz}}$ in figure 5(b), the pedestal density and temperature increase in the HFF phase and drop when the LFFs are excited, such phenomenon has been introduced in previous report [31]. In alignment with the ELM-free period, the amplitude of HFF (${{{A}}_{{\text{HFF}},{\text{ J}}}}$) is modulated by the EHO-like mode between two LFF bursts as displayed in figure 5(c). Magnetic measurement is available in this shot, and the toroidal mode number of EHO-like mode is deduced from magnetic fluctuations $\delta {B_{\text{K}}}$ and $\delta {B_{\text{L}}}$ measured by magnetic coils at mid-plane in K-port and L-port with a toroidal separation of $22^\circ$. It can be found in figures 6(a) and (b), the coherence between density and magnetic fluctuations at EHO-like frequency is much higher than the noise level, signifying a magnetic component of EHO-like mode. The toroidal number is computed from the of cross-phase difference between $\delta {B_{\text{K}}}$ and $\delta {B_{\text{L}}}$, as shown in figure 6(c), where the toroidal number of EHO-like mode within the frequency range $18{-}80{\text{ kHz}}$ is $n = 1 - 5$, with the positive value indicates the EDD direction. The $30{\text{ kHz}}$ mode in figure 6(b) is possibly the $n = 1$ magnetic coherent mode [32].

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Figure 7(a) depicts the PSD of the density fluctuation ${{\text{S}}_{{\text{ori}}}}$ and the amplitude ${{\text{A}}_{{\text{HFF}},{\text{ K}}}}$ measured by DFR-K during the HFF phase, while figure 7(b) displays their coherence. The EHO-like mode is also discernible in the spectrum of the original density fluctuation signal. The notable coherence at multiple frequencies of the EHO-like mode suggests that the amplitude of HFF in the $0.6{-}1.0{\text{ MHz}}$ range is modulated by the low-frequency EHO-like mode spanning $10{-}100{\text{ kHz}}$.

Zoom In Zoom Out Reset image size

A salient characteristic of the EHO is its capability to facilitate benign and continuous edge particle transport, serving as a preferable alternative to ELMs. To verify the impact of the EHO-like mode on divertor transport, we select the amplitude of the EHO-like mode ${A_{\left[ {10{-}100} \right]{\text{kHz}}}}$ from the $10{-}100{\text{ kHz}}$ density fluctuation measured by DFR-K to do coherence analysis with divertor particle flux in shot #89649. Figures 8(a)–(c) shows coherence of EHO-like mode's amplitude ${A_{\left[ {10 - 100} \right]{\text{kHz}}}}$ and divertor particle flux ${{{\Gamma }}_{{\text{div}}}}$ with sampling rate of $50{\text{ kHz}}$. The cross power and coherence, depicted in figures 8(a) and (b) by green solid line, show an obvious peak within $0{-}3{\text{ kHz}}$, which indicates that the EHO-like mode is modulated by the $0{-}3{\text{ kHz}}$ mode. A negative phase delay is discernible in figure 8(c), from which the time delay is deduced as ${{{\tau }}_{{\text{delay}}}} = {{\Delta {\text{phase}}}}/\left( {2{{\pi \Delta f}}} \right) = - 0.12{\text{ ms}}$, as indicated by the red dashed line. The cross-correlation, illustrated in figure 9(a), is computed using the amplitude of the EHO-like mode and the particle flux, both subjected to a low-pass filter of $4{\text{ kHz}}$, from $2.98{\text{ s}}$ to $3.00{\text{ s}}$. It is evident that the peak cross-correlation coefficient aligns with a time delay of $- 0.12{\text{ ms}}$, suggesting that the amplitude of the EHO-like mode precedes the particle flux onto the divertor by approximately $0.12{\text{ ms}}$. This time delay harmonizes with the result inferred from the slope of the cross phase, a rationale supported by the transit of particles from the outer mid-plane to the divertor along the scrape-off layer (SOL) magnetic field line on an ion transit time scale. Here the transit time ${\tau _\parallel }$ is estimated by ${\tau _\parallel } \approx {L_\parallel }/{c_{\text{s}}}$, with connection length ${L_\parallel } \approx \pi {q_{95}}R$ and ion sound speed ${c_{\text{s}}} = \sqrt {\left( {\left( {{T_{\text{e}}} + {T_{\text{i}}}} \right)/{m_{\text{i}}}} \right)}$. For the HFF phase of $\# 80485$, ${q_{95}} = 3.5$, ${{{T}}_{\text{e}}} = {{{T}}_{\text{i}}} \approx 150{\text{ eV}}$, and thus ${L_\parallel } \approx 20.3\,{\text{m}}$, ${c_{\text{s}}} = 1.20 \times {10^5}{\text{m}} \cdot {{\text{s}}^{ - 1}}$, consequently the transit time is ${\tau _\parallel } \approx 0.17\;{\text{ms}}$. Therefore, the observed time delay ($0.12{\text{ ms}}$) between the amplitude of EHO-like mode and the divertor particle flux is actually close to the time scale for a particle to move from the outer mid-plane to the divertor targets along the SOL field line.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Furthermore, figures 9(b) and (c) depict the temporal evolution of the particle flux signal, as measured by the divertor probe near the striking point, and the amplitude of the EHO-like mode ${A_{\left[ {10 - 100} \right]{\text{kHz}}}}$. Their smoothed curves are obtained via a low-pass filter of $4{\text{ kHz}}$. It is apparent that the particle flux increases after the burst of EHO-like mode. In particular, both the amplitude of EHO-like mode and the particle flux exhibit a 'bursty' behavior, with the former preceding the latter by approximately $0.1{\text{ ms}}$ in time evolution as denoted by the red dashed lines. These findings suggest the burst of the EHO-like mode could drive particle transport toward divertor plates.

The reflectometry fluctuation amplitude measurements rely heavily on the reflection layer location. As the pedestal structure changes due to the transport, it is very likely that the reflection layer location changes with it, hence producing the illusion that the EHO-like mode amplitude changes with the pedestal profile or the divertor particle flux. The measurement of Mirnov probe (MP) does not have this issue, so the coherence between the amplitude of MP and the divertor particle flux is shown in figure 8. The strong coherence is also found below 3 kHz, which indicates that the low frequency modulation of EHO-like is not due to the movement of cut-off layer. The coherence between the amplitude of HFF and divertor flux is also shown in figure 8 by black solid line. Strong coherence can be also found below $3{\text{ kHz}}$, which indicates that the HFF could also contribute outward particle flux toward divertor. Note that weak coherence can be found at about $17{\text{ kHz}}$, which is the lowest frequency of EHO-like mode in this shot. This result suggests that the EHO-like mode could modulate the amplitude of HFF and also the particle flux at the frequency of EHO-like mode. Back to the figure 2, the transition into the high pedestal HFF phase is also accompanied by a sudden cessation of the ECM. At the time when the ECM ceases there is a large drop in the ${D_\alpha }$ signal (figure 2(c)), which is probably due to the lack of the ECM induced transport and the weaker transport from the EHO-like mode. As the EHO-like mode disappears, the ${D_\alpha }$ signal increases to the previous level. Therefore, the relatively weaker transport level caused by EHO-like mode could partly contribute to the continuous growth of the pedestal density.

It has been observed that both the DFR-K and DFR-J capture the respective fluctuations of HFF, the frequency range being 0.6–1.0 MHz for DFR-K and $1{-}3{\text{ MHz}}$ for DFR-J. A primary conceivable reason for this discrepancy is the variation in the sampling rate. To elucidate this, we resampled the data acquired by DFR-J from $20{\text{ MHz}}$ to $2{\text{ MHz}}$. Figure 10 shows the PSD and coherence as gauged by the two systems. In figure (10 a), a pronounced broadband turbulence within the $0.6{-} 1.0{\text{ MHz}}$ range is discernible in both signals captured by DFR-K and DFR-J. Figure 10(b) depicts that the coherence within the corresponding frequency range significantly surpasses the noise level. This substantiates that the corresponding fluctuations in the $0.6{-}1.0{\text{ MHz}}$ range as measured by DFR-K also represent the HFF, the discrepancy in the frequency range is attributable to the lower sampling rate of DFR-K in comparison to DFR-J. Figures 10(c)–(e) unveil the cross power, coherence and poloidal wave number of the HFF as measured by the DFR-K at two poloidally separated positions P₁ and P₂ during the HFF phase. As illustrated in figures 10(c) and (d), both the cross power and coherence within the HFF range, i.e. $0.6{-}1.0\ {\text{MHz}}$, are notably robust. Figure 10(e) reveals that the poloidal wave number of HFF lies between ${{\text{k}}_{{\theta }}} = 0.9{-}1.6{\text{ c}}{{\text{m}}^{ - 1}}$, corresponding to ${k_\theta }{\rho _{\text{s}}} = 0.14{-} 0.25$, with the positive value signifying a rotating direction of the EDD in the lab frame.

Zoom In Zoom Out Reset image size

As depicted in figure 5, the HFF is triggered once the pedestal reaches a certain threshold. To elucidate this, figures 11(a)–(c) present statistical data captured during the ELM cycle from 3.810 to 3.845 s, encompassing the gradient of pedestal density, temperature, pressure as well as the amplitude of the HFF. It is observed that the initiation of HFF fluctuation is contingent upon reaching a threshold in gradient of pedestal pressure. In this particular instance, the threshold is identified to be around 3.2 kPa, and beyond this threshold, the amplitude of fluctuation exhibits an increasing trend with pedestal pressure. Figures 11(d) and (e) elucidate the evolution of high-frequency mode density fluctuations and amplitude signals over a brief time span, revealing an intermittent bursting characteristic of the high-frequency mode. This characteristic also leads to the occurrence of points with lower amplitude during phases of elevated pressure, aligning with the phenomena depicted in figures 11(a)–(c).

Zoom In Zoom Out Reset image size

Figure 12 presents the evolution of the HFF and pedestal parameters in the inter-ELM period. Figure 12(a) shows several ELM periods including large ELM and ${D_\alpha }$ spikes. Figure 10(b) is the spectrogram of density fluctuation measured by Doppler backscattering system (DBS) in the gradient region around $\rho \sim0.94$, and figure 10(c) shows the integrate power of 1–2 MHz from the spectrogram. In the early phase just after the ELM crash, LFF with ${\text{f}} \lt 500{\text{ kHz}}$ emerges, and in the later phase, the LFF disappears as the HFF becomes stronger. It is seen that the HFF soon weakens just before the onset of next ELM crash, and the LFF grows up again after the ELM crash. It is noticed that the ${D_\alpha }$ signal decreases with the disappearance of LFF and also the presence of HFF, suggesting a lower transport in the phase of HFF when compared with LFF phase. To verify the hypothesis, the evolution of $E \times B$ velocity and its shear is provided by measurement of DBS. The perpendicular velocities ${u_ \bot }$ near $\rho \sim0.92$ and $\rho \sim0.94$ are plotted together in figure 12(d), the $E \times B$ shear in this region is shown in figure 12(e). As the perpendicular velocity ${u_ \bot } = {\upsilon _{E \times B}} + {\upsilon _{{\text{phase}}}}$, here the ${\upsilon _{E \times B}}$ and ${\upsilon _{{\text{phase}}}}$ are the $E \times B$ and phase velocity, respectively. Basing on a reasonable assumption that ${\upsilon _{E \times B}} \gg {\upsilon _{{\text{phase}}}}$, the ${\upsilon _{E \times B}}$ could be qualitatively represented by perpendicular velocity ${u_ \bot }$ [33]. The evolution of edge density and temperature is shown in figures 12(f) and (g). As indicated by the purple dashed line, the $E \times B$ velocity increases with the weaken of LFF in the HFF phase, and follows the slight rise of edge density and temperature. The rise of the density profile and density gradient can be found in figure 13. According to the radial force balance equation: ${E_{\text{r}}} = \frac{1}{{{Z_{\text{i}}}e{n_{\text{i}}}}}\frac{{\partial {P_{\text{i}}}}}{{\partial r}} - {v_\theta }{B_\varphi } + {v_\varphi }{B_\theta }$. Here the ${E_{\text{r}}}$ is the radial electric field, ${Z_{\text{i}}}$, ${n_{\text{i}}}$ and ${P_{\text{i}}}$ is the charge, density and pressure of ion. ${v_{\theta ,\varphi }}$ and ${B_{\theta ,\varphi }}$ are the ion velocity and magnetic field in the poloidal ($\theta$) and toroidal ($\varphi$) directions. Here we assume that the rise of ${P_{\text{i}}}$ and ${P_{\text{e}}}$ are coincident with each other. Therefore, the rise of ${\upsilon _{E \times B}}$ means the rise of ${E_{\text{r}}}$, which is firmly linked to the term of pedestal gradient of pressure. It is noticed that an obvious increase in the $E \times B$ shear as shown in figure 12(e). As shown in figures 12(c)–(e), the $E \times B$ shear starts to increase before the rise of the $E \times B$ velocity and the stimulation of HFF, implying a potential effect of $E \times B$ shear on the HFF. The increased $E \times B$ shear suppresses the LFF and destabilizes the HFF, resulting in a lower transport level. This allows for an increase of pedestal gradient and the rotational velocity until the onset of ELM.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

In figure 2, the simultaneous presence of HFF and EHO-like mode in the pedestal is demonstrated, sparking curiosity regarding the interaction between these distinct modes. The notable modulation of HFF amplitude with the frequency of the EHO-like mode is observed, emphasizing the significance of nonlinear coupling in this cross-scale interaction. We delve into the potential nonlinear interaction between broadband turbulence and the EHO-like mode employing an auto-bicoherence function: $b{i^2}\left( {{f_1},{f_2}} \right) = \frac{{{{\left| {\langle X\left( {{f_1}} \right)X\left( {{f_2}} \right){X^*}\left( {{f_1} + {f_2} = {f_3}} \right)\rangle } \right|}^2}}}{{\langle {{\left| {X\left( {{f_1}} \right)X\left( {{f_2}} \right)} \right|}^2}\rangle \langle {{\left| {X\left( {{f_3}} \right)} \right|}^2}\rangle }}$. Here $X\left( {{f_1}} \right)$, $X\left( {{f_2}} \right)$, and $X\left( {{f_3}} \right)$ denote the Fourier coefficients of the harmonics with frequencies ${f_1}$, ${f_2}$, and ${f_3}$ respectively. The bispectral coefficient $b{i^2}$ reflects the phase correlation between these three waves satisfying ${f_1} + {f_2} = {f_3}$. Figure 14(a) shows the auto bicoherence of the density fluctuation as measured by DFR-J during HFF phase of shot $\# 80485$. A pronounced coherence is discerned in the frequency domain of ${{{f}}_1} = 1{-}3{\text{ MHz}}$, ${{{f}}_2} \lt 100{\text{ kHz}}$, with a diminished coherence in other frequency ranges. As elucidated in figure 14(b), the correlation coefficients along the lines ${{{f}}_2}{ } = \pm 55{\text{ kHz}}$ and ${{{f}}_1} + {{{f}}_2} = 55{\text{ kHz}}$ within $1{-}3{\text{ MHz}}$ are significantly higher than the noise level. It is noteworthy that the $55{\text{ kHz}}$ corresponds to the frequency of the EHO-like mode exhibiting the highest power (see figure 7(a)). Figure 14(c), portraying the bicoherence satisfying ${{{f}}_1} + {{{f}}_2} = 55{\text{ kHz}}$, unveils a significant interaction with broadband turbulence ($1{-}3{\text{ MHz}}$) which aligns with the frequency range of HFF. These findings underscore a significant nonlinear interaction between the EHO-like mode and HFF.

Zoom In Zoom Out Reset image size

The above analyses present the coexistence of HFF and EHO-like mode within the H-mode pedestal region of the EAST tokamak. The EHO-like mode is characterized as an MHD-like mode with a toroidal mode $n = 1{-}5$, $f < 100\,{\text{kHz}}$ and ${{\text{k}}_{{\theta }}} \unicode{x2A7D} 0.12{\text{ c}}{{\text{m}}^{ - 1}}$, exhibiting rotation in the EDD direction in the lab frame. Drawing a comparative analysis with the EHO observed in DIII-D plasma, which is identified as a low-n mode ($n \unicode{x2A7D} 5$) destabilized by rotational shear, both the diagnostic data from microwave imaging reflectometer and M3D-C1 modeling exhibit a poloidal number of ${{\text{k}}_{{\theta }}} \unicode{x2A7D} 0.12{\text{ c}}{{\text{m}}^{ - 1}}$ for n ⩽ 5, rotating in the EDD direction in lab frame with frequency below $100{\text{ kHz}}$. These characteristics closely resemble the EHO-like mode identified in our investigation. The EHO-like mode is found at different ${B_{\text{t}}}$ and a relatively high ${I_{\text{p}}}$, suggesting a closely relation between EHO-like mode and the ${q_{95}}$. Result shows that the ${q_{95}}$ is about 3.5–5.5, the value is similar with EHO-discharge in JET, AUG and the DIII-D [34, 35]. Furthermore, akin to the EHO, the EHO-like mode is demonstrated to drive particle transport toward the divertor. Nonetheless, the transport level caused by EHO-like mode is much weaker than that induced by the ECM. Previous study [16] demonstrated that ECM could cause about $25{{ \% }}$ perturbation ($\delta {{{\Gamma }}_{{\text{div}}}}/{{{\Gamma }}_{{\text{div}}}}$) in the particle flux toward the divertor, while this value is about 8% for EHO-like mode as shown in figure 9(c). Consequently, a continuous increase of pedestal density is observed throughout the EHO-like mode's duration. Yet it remains unknown what trigger the transition from ECM to EHO-like mode/HFF without the change of external power and particle source. Recent result in DIII-D presented a transition from EHO to the broad-band MHD is found when the input toque ramps down to zero, which shows a strong correlation to the E × B shear [20]. However, in our study there is no active change in NBI torque during the transition from ECM to EHO-like mode as shown in figure 2. The change of local E × B shear should be taken into account in future work.

The HFF is discerned by DFRs operating at varying sampling rates: DFR-J, with a sampling rate of $20{\text{ MHz}}$, delineates a frequency range of $1{-}3{\text{ MHz}}$ for HFF, while DFR-K, operating at $2{\text{ MHz}}$, identifies a frequency range of $0.6{-}1.0{\text{ MHz}}$. As illustrated in figure 10, pronounced coherence within the $0.6{-}1.0{\text{ MHz}}$ range underscores the identity of these fluctuations as HFF, the discrepancy in frequency being attributable to the differing sampling rates. The poloidal scale of HFF, as inferred from the poloidal mode number of the $0.6{-}1.0{\text{ MHz}}$ mode measured by DFR-K, spans ${{\text{k}}_{{\theta }}} = 0.9{-}1.6{\text{ c}}{{\text{m}}^{ - 1}}$ (or ${k_\theta }{\rho _{\text{s}}} = 0.14{-}0.25$), which is close to an ITG scale. However, due to under-sampling, the actual poloidal scale of HFF could be substantially larger. Moreover, the HFF cannot be ETG as the DFR is measuring density fluctuations with ${{\text{k}}_{{\theta }}} < 3{\text{ c}}{{\text{m}}^{ - 1}}$ in the pedestal [36], while the ETG has a small-scale (${k_\theta }{\rho _{\text{s}}}\sim10$). In the DIII-D tokamak, it is observed the ITG-scale mode (${{f}} \lt 1{\text{ MHz}}$, ${k_\theta }{\rho _{\text{s}}}\sim0.3$, $\rho \sim0.98$) increases immediately after each ELM crash and is quickly suppressed by increased local E × B shear, instead a TEM-scale mode ($1 \lt {{f}} \lt 4{\text{ MHz}}$, ${k_\theta }{\rho _{\text{s}}}\sim1$, $\rho \sim 0.95$) become dominant in the gradient region [23]. In our experiment, the HFF could also detected by the DBS reflectometry as shown in figure 12, while the detected wave number is about ${{\text{k}}_ \bot } \sim 2{\text{ c}}{{\text{m}}^{ - 1}}$. The destabilization of HFF is accompanied by a rapid increase of the $E \times B$ shear, which is consistent with the phenomenon of the high frequency TEM turbulence. The TRANSP simulation suggests the MTM or the TEM could be the potential candidate for the observed high frequency modes [22].

The nonlinear cross-scale interaction observed between the EHO-like mode and the HFF unveils a notable interplay, as delineated in figure 12(b). Herein, the EHO-like mode predominantly interacts with the background turbulence within the frequency range of $1{-}3\;{\text{MHz}}$, coinciding with the frequency domain of the HFF. Such interaction could have indirect effect on the evolution of pedestal and the onset of ELM. For instance, in the JET tokamak, a distinct interaction between the type-I ELM precursor and the washboard mode (WBM) has been delineated. The onset of type-I ELM precursors is correlated with a weakening, or even an inhibition of the WBM as cited in [8]. Moreover, in the DIII-D tokamak, the nonlinear coupling of pedestal modes, associated with radial distortions propelling out of the pedestal, is posited as a primary trigger for the low-frequency ELMs. This nonlinear interaction emerges as a plausible candidate for the triggering of ELMs, as outlined in [37]. However, the impact of the interaction between HFF and EHO-like modes on the pedestal evolution remains an uncharted territory. The nonlinear energy transfer between these modes could potentially modulate the transport level, affect pedestal saturation, and influence the onset of ELMs, meriting further analysis.

In conclusion, the coexistence of HFF with frequency about $1{-} 3{\text{ MHz}}$ and EHO-like mode $\lt 100{\text{ kHz}}$ are observed in pure RF heating for the first time on EAST tokamak. In the study, the HFF is an ITG/TEM scale mode fluctuation rotating in the direction of EDD in the lab frame, and it is found modulated by the EHO-like mode. Measurements show the EHO-like mode has a mode structure similar to EHO, e.g. $n = 1 - 5$ and ${{{k}}_{{\theta }}} \unicode{x2A7D} 0.12{\text{ c}}{{\text{m}}^{ - 1}}$ rotating in the EDD direction in the lab frame with ${q_{95}}\sim3.5 - 5.5$. Analysis shows that both EHO-like mode and HFF could drive outward particle flux toward divertor. Statistical data analysis demonstrates that a threshold of pedestal gradient of pressure exists for the triggering of the HFF, and the fluctuation amplitude shows a fast increase with the pedestal pressure beyond the threshold. During the inter-ELM period, a significant decrease in the ${D_\alpha }$ baseline is observed whenever the LFF weakens and the HFF grows, prior to each large ELM. One possible explanation is that the rapid increase of $E \times B$ shear stabilizes the LFF and destabilizes the HFF, which lowers the pedestal transport and enables the further growth of the pedestal until the onset of the ELM. Moreover, further analysis indicates a strong nonlinear interaction between the EHO-like mode and HFF. So far, both the ELM and EHO are efficient vehicles to drive particle transport for impurity removal in ELMy H-mode and QH mode, respectively. The observation in this paper actually expands the operation space of the EHO plasma in ELMy H-mode without NBI input, and it could be a possible candidate for the low external momentum input condition of ITER. In the next step of the work, the nature of HFF should be identified by the 'fingerprint' method [7]. The velocity and its shear should be considered to study the transition from ECM to EHO-like mode. And the role of these fluctuations in the triggering of ELM would be studied. We believe the present results will be beneficial for further understanding of a turbulence transport in the H-mode.