Abstract The low-frequency (1/400- 1/30 Hz) oscillations within the Hambantota Port, Sri Lanka, are investigated based on the in-situ measured water surface elevations in and outside the harbor. By combining analyses of spectra derived from the in-situ measurement and modal analyses solving the extended mild-slope equation, four natural frequencies of the port are identified and their modal structures are obtained. It is found that the low-frequency oscillations within the port were dominated by its natural frequencies (i.e., the eigen oscillations) instead of the frequencies of external waves. Subsequently, the correlation analysis is carried out to determine the generation of the low-frequency oscillations. Although both the offshore low-frequency and short waves contributed to the low-frequency oscillations, the low-frequency energy spectrum within the port is found to be more significantly correlated with the offshore total low-frequency wave energy than with the offshore total short-wave energy. Moreover, the empirical formula and artificial neural network models are presented to efficiently estimate the low-frequency waves inside the harbor in terms of the offshore wave parameters. Although both methods give excellent results, the artificial neural network performs slightly better.

中文翻译：

---

西南季风期间汉班托塔港低频振荡特征

摘要 基于现场测量的港口内外水面高程，研究了斯里兰卡汉班托塔港内的低频（1/400-1/30 Hz）振荡。通过将现场测量得到的频谱分析和求解扩展缓坡方程的模态分析相结合，确定了港口的四个固有频率并获得了它们的模态结构。结果表明，港口内的低频振荡主要受其固有频率（即本征振荡）而不是外波频率的支配。随后，进行相关分析以确定低频振荡的产生。虽然海上低频和短波都对低频振荡有贡献，发现港口内的低频能谱与海上总低频波能量的相关性比与海上总短波能量的相关性更显着。此外，还提出了经验公式和人工神经网络模型，以根据近海波浪参数有效地估计港口内的低频波浪。虽然这两种方法都给出了很好的结果，但人工神经网络的表现稍好一些。