Impact of oceanic mesoscale eddies on the productivity of the western Bay of Bengal: contribution of new EO data and machine learning

背景和目的：海洋中尺度的可变性和涡旋(尺度从~10到100公里)在调节区域和全球的物理和生物地球化学过程中起着关键作用，包括热量输送和营养物质的混合。这些中尺度漩涡影响浮游植物的生产力，从而影响海洋物种，以及依赖它们的当地种群。最近发现，热带海洋的中尺度变率总体上在减少。尤其是热带北印度洋和孟加拉湾西部，是一个涡旋丰富的地区，经历了显著的浮游植物水华。以往关于孟加拉湾西部生产力机制的研究主要集中在热带气旋上，热带气旋会影响表层水域的营养物质供应。对于一些气旋事件，研究了海湾中涡旋的存在及其对生物生产力提高的贡献。然而，在强层结存在的情况下，中尺度涡旋的可变性以及风应力强迫的变化对中尺度过程和初级生产力的影响仍有待详细探讨。此外，到目前为止，涡旋的存在及其对区域生产力的影响仅限于利用常规卫星数据或少数稀疏的现场数据进行研究，没有采用特定的涡旋探测方法，也没有使用改进的沿海卫星产品。这项研究的目的是利用地球观测与数值模型输出和机器学习相结合，对这种变化的原因和后果进行全面研究。这项研究最终可能采取的方向有相当大的灵活性。博士将探索以下关键研究问题：1)过去几十年，孟加拉湾西部的涡旋场是如何变化的？2)这种变化是如何受风强迫的变化影响的，它是如何影响区域生产力的？3)涡旋及其对生产力的贡献在年际和季节之间是如何变化的？这将包括研究厄尔尼诺南方涛动和印度洋双极现象的影响方法：本博士将研究无监督机器学习技术对一组历史和新的EO数据和数值模式输出的适用性，以揭示中尺度涡旋和风对区域生态系统生产力的影响。中尺度特征的非线性相互作用使得无监督机器学习(ML)方法非常适合于客观地确定涡旋的时空变化。将探索的主要ML方法是自组织映射、K-均值聚类和变分自动编码器。将利用历史和新的卫星数据集。这些数据将包括高分辨率卫星海洋颜色、由叶绿素a得出的数据和海表面温度(SST)、风、测高得出的海表面高度和海流。新的SSH卫星高度计观测可以更好地捕捉海洋中尺度过程，例如将于2022年12月发射的SWOT数据。此外，最近的哨兵3A和3B卫星搭载了沿航迹分辨率更高的合成孔径雷达(SAR)高度计，也提供了更好的靠近海岸的数据。ARGO浮标测量提供了不同深度的物理和生物参数，也将用于了解孟加拉湾西部的生物和水文特性。新的和历史的地球观测数据将与覆盖卫星数据期的高分辨率海洋模型(NEMO)的输出进行比较，该模型包括生物地球化学过程(Medusa)。利用从数值模式输出中推断的环境因子，将可获得额外的物理和生物参数(例如，混合层深度(MLD)、次表层叶绿素)，这有助于进一步探索变异性变化。

Background and objectives: Oceanic mesoscale variability and eddies (scales from ~10 to 100 km) play a key role in regulating regional and global physical and biogeochemical processes, including heat transport and mixing of nutrients. These mesoscale eddies affect the phytoplankton productivity, hence marine species, and local populations dependent on them. It has been recently found that tropical oceans mesoscale variability is decreasing overall. The tropical north Indian Ocean and the western Bay of Bengal, in particular, is an eddyrich region which experiences prominent phytoplankton blooms. Previous works on the mechanisms of the productivity in the western Bay of Bengal have mainly focused on tropical cyclones, which impact the supply of nutrients to the surface waters. The presence of eddies and their contribution to biological productivity enhancement in the Bay have been examined for some cyclone events. However, the mesoscale eddies' variability in the presence of strong stratification and the effect of changes in wind stress forcing on mesoscale processes and primary productivity, remain to be explored in detail. Additionally, eddy presence and influence on the regional productivity has to-date only been investigated using the conventional satellite data or few sparse in-situ data, and no specific eddy detection method was applied, nor improved coastal satellite products used. The aim of this research is to carry out a comprehensive study of the causes and consequences of such variability using Earth Observation (EO) in synergy with numerical model outputs and machine learning. There is considerable flexibility in the direction that the research may ultimately take. The PhD will explore the following key research questions:1) How has the eddy field varied over the past decades in the western part of the Bay of Bengal?2) How this variability is affected by changes in wind forcing and how does it affect the regional productivity?3) How are eddies and their contribution to productivity changing between years and seasons? This will include examining the effects of the El-Niño Southern Oscillation and the Indian Ocean DipoleMethodology: This PhD will investigate the applicability of unsupervised machine learning techniques to a set of historical and new EO data and numerical model outputs to unravel the impact of mesoscale eddies and wind on the regional ecosystem productivity. The nonlinear interactions of the mesoscale features make unsupervised Machine Learning (ML) methods well suited to objectively determine the eddies spatiotemporal variation. The main ML methods that will be explored are Self Organizing Maps, K-means clustering, and variational autoencoders. Both historical and new satellite datasets will be exploited. These will cover highresolution satellite ocean colour derived chlorophyll-a data and Sea Surface Temperature (SST), winds, altimetry derived Sea Surface Height (SSH) and currents. New satellite SSH altimeter observations can better capture oceanic mesoscale processes, such as data from SWOT which will be launched in December 2022. Additionally, the recent Sentinel 3A&3B satellites, that carry higher along-track resolution synthetic aperture radar (SAR) altimeters, also provide improved data close to the coast. Argo float measurements, which provide physical and biological parameters at different depths, will also be used to understand the biological andhydrographic properties of the western Bay of Bengal. The new and historical EO data will be compared to output from a high resolution ocean model (NEMO) that includes biogeochemical processes (MEDUSA), covering the satellite data period. Using environmental factors inferred from numerical model outputs, additional physical and biological parameters (e.g., mixed layer depth (MLD), subsurface chlorophyll) will be available which can help further exploration of the variability changes.