Long-term Polar research initiatives such as NEPTUNE, as well as EMSO, FIXO3, and the current JERICO EU funded projects, have already yielded, longer, more comprehensive public-domain time-series datasets, that typically embody very high-frequency datasets as gathered by multiple embedded real-time environmental sensors. Their infrastructures, both in open-sea and in-shore coastal contexts, are all located within the Polar Regions (i.e. those areas wherein the mean annual air temperature never exceeds the critical 10° C value).
Indeed, Polar Regions such as the High-Arctic have already been shown to exhibit the highest rate of climate change anywhere on Earth. Therefore, environmental scientists are in urgent need of optimized analytical tools and techniques in order to provide politicians and leading authorities with suitably reliable policy recommendations that timelier inform policy making both within the Arctic Council and beyond.
However, these time-series often display a rather too high degree of variability, that is, they exhibit nonlinear, non-stationary, multi-scale, and unacceptable noisy characteristics aggregated over a large range of spatial and temporal scales, ranging from seconds to many thousands of years. Classical methods of analysis have hitherto already proven to be partially or completely ineffective. Namely, Harmonic analysis, Fourier analysis, Blackman-Tukey Method, Wavelet Analysis, etc.
The PhD candidate will explore the following aspects:
Leverage the Hilbert-Huang Transform (HHT) to perform optimal statistical and spectral analyses.
Create an HHT variant that offers significant added value as compared with previous approaches.
Analysis of data comprising two principal components: an enhanced decomposition algorithm that will be coined as the Empirical Mode Decomposition (EMD+), and a fully optimized and tested enabling spectral analysis tool would be created and named as Hilbert Spectral Analysis (HSA+).
Create an Integrative (HHTI) Toolbox encapsulating a suitably ordered multi-stage decomposition wherein, any suitably complex, inherently noisy real-time derived dataset is exposed to a series of tailed Intrinsic Mode Functions (IMFs), that seek to extract, and fully visualize the intrinsic patterns present in the original aggregated observations.
Identify a suite of Python enabled data-analytic routines that exemplify an intrinsically high degree of User-Experience, and data visualization capabilities.
HHTI Toolkit is intrinsically “Smart”, that is it would be designed to be leverage the state-of-the art Machine Learning Algorithms.
HHTI is intended to support environmental scientists in their day-to-day work, that is, to fully support and enhance the decision-making capabilities reduce response times to the absolute minimum by removing for human intervention and interpretation wherever possible.
The candidate should have earned an MSc degree, or equivalent. He or she should have a strong background signal processing that will be applied for oceanography. The candidate should be familiar with Matlab. R and Python language. The qualified candidate should also be knowledgeable in mobile app implementation. Some knowledge in cybersecurity would be an advantage. Preferably full-time students, however, part-time applicants are welcome!
Statistical properties and time-frequency analysis of temperature, salinity and turbidity measured by the MAREL Carnot station in the coastal waters of Boulogne-sur-Mer (France) by Dhouha Kbaier, Pascal Lazure and Ingrid Puillat.
Investigation of Turbulence Behaviour in the Stable Boundary Layer Using Arbitrary-Order Hilbert Spectra by Wei Wei., S. Zhang, Francois Schmitt and H. Zhang.
Advanced Spectral Analysis and Cross Correlation Based on the Empirical Mode Decomposition: Application to the Environmental Time Series by by Dhouha Kbaier, Pascal Lazure and Ingrid Puillat.