Abstract Seasonal statistics on underwater ambient noise in the deep area of the South China Sea (SCS) were analyzed on the basis of the experimental observation data. The depth dependence, correlation with environmental parameters, empirical expression and probability density function (PDF) of noise levels (NL) in summer were compared with those in winter. Average of NLs at receiver depths shallower than 800 m in summer are 6 dB greater than those in winter at 1 kHz. A parameterized fitting model was utilized to determine the noise spectrum levels at different wind speed (WS) conditions and was compared with the Wenz curves, which are consistent with the NLs in winter at >wind force 3. The correlation coefficient between NLs and WS or significant wave height (SWH) were described according to the reanalysis database of the National Center for Environmental Prediction (NCEP) and WAVEWATCH III. In addition, a four-parameter logarithm model was used to model the relationship between NLs and WS or SWH. Empirical expression between NLs and spectrum levels at 1 kHz was acquired in SCS. The Weibull and Burr distributions were applied for the evaluation of PDF of NLs in summer and winter. Low-frequency NLs were dominated by distant shipping noise and their PDF satisfied the Burr distribution. High-frequency NLs were dominated by breaking waves or wind agitation and their PDF satisfied the Burr and Gaussian distributions simultaneously. The experimental observation duration was closely related to the PDF of high-frequency NLs. Standard deviation, skewness and kurtosis of NLs were also analyzed. Finally, frequency correlation and spectra correlation matrices were utilized in analyzing the noise sources mechanism, and discussing the reason for the statistical difference of NLs in summer and winter. The NL statistical differences was due to oceanic environment variation and noise source mechanism simultaneously.

中文翻译：

---

南海深部水下环境噪声季节比较

摘要 以实验观测数据为基础，分析了南海深海区水下环境噪声的季节统计数据。将夏季与冬季噪声水平（NL）的深度依赖性、与环境参数的相关性、经验表达式和概率密度函数（PDF）进行比较。夏季接收器深度小于 800 m 的 NL 平均值比冬季 1 kHz 的 NL 大 6 dB。利用参数化拟合模型确定不同风速 (WS) 条件下的噪声频谱水平，并与 Wenz 曲线进行比较，这与冬季 > 风力 3 时的 NLs 一致。NLs与WS或有效波高（SWH）的相关系数根据国家环境预测中心（NCEP）和WAVEWATCH III的再分析数据库进行描述。此外，使用四参数对数模型对 NL 与 WS 或 SWH 之间的关系进行建模。在 SCS 中获得了 NL 和 1 kHz 频谱水平之间的经验表达。Weibull 和 Burr 分布用于评估夏季和冬季 NL 的 PDF。低频 NL 以远距离运输噪声为主，它们的 PDF 满足 Burr 分布。高频 NL 以碎波或风搅动为主，它们的 PDF 同时满足 Burr 和高斯分布。实验观察持续时间与高频NLs的PDF密切相关。还分析了 NL 的标准偏差、偏度和峰度。最后，利用频率相关矩阵和频谱相关矩阵分析噪声源机制，讨论夏季和冬季NLs统计差异的原因。NL 统计差异同时是由于海洋环境变化和噪声源机制。