Seminars of the Center for Earth Sciences

Hydrophysical Processes Induced by Tides in the Waters of the Pacific Ocean Adjacent to the Kamchatka Peninsula and the Northern Kuril Islands, and Their Influence on the Early Stages of Pollock Development

This talk focuses on how weather and climate conditions affect people in urban environments. The spotlight is on the urban heat-island phenomenon and its repercussions—from microclimate changes to public-health impacts. The presentation spans both historical developments and modern research methods, including numerical modelling, field measurements in Arctic cities, and bioclimatic indices. It emphasizes the need for an integrated approach to assessing thermal comfort and to planning urban spaces, especially under a changing climate.

In the context of global climate change and anthropogenic pressure, it is essential to identify and protect unique biological resources, such as invertebrate endemics of Lake Baikal, which serve as indicators of ecosystem health. Population studies are crucial for the effective management of natural resources and for planning environmental protection efforts. This work is devoted to the development of new methods for analyzing genetic polymorphism data from natural populations of Baikal’s invertebrate endemics, aimed at providing a more detailed characterization of both ongoing and historical population processes. The primary goal is to develop and validate deep learning models that can uncover scenarios leading to observed patterns of genetic diversity. The study introduces the following models: * a model for identifying population structures that change their reproductive strategies depending on environmental conditions; * a model for detecting climate change based on polymorphism data; * a model for clustering genetic sequences according to the organisms’ habitat. These models are expected to enhance the effectiveness of population genetic research in the future.

At the seminar, based on the PhD thesis titled “Sensitivity of Ocean State Forecast Accuracy to the Localization of Assimilated Observational Data,” innovative methods for the optimal placement of measuring instruments were discussed. The focus was on two novel approaches: the Mutual Information Maximization method and the Concrete Autoencoder with LSGAN. The first approach involves strategically placing sensors in locations that provide information about the largest possible area of the ocean, achieved through analysis of spatial correlations in the physical field. The second approach employs surrogate machine learning models to optimize sensor placement and reconstruct the physical field, using a Least Squares Generative Adversarial Network for training. These methods aim to improve forecast accuracy and reduce computational costs in oceanographic modeling.

In his talk, Takaya Uchida presented a novel application of DMD to extract geostrophically balanced velocity fields from sea surface height (SSH) data, including recent high-resolution observations from the SWOT satellite. Addressing the limitations of traditional spectral methods due to SWOT’s coarse temporal resolution, Uchida demonstrated how DMD offers a data-driven, interpretable alternative capable of separating balanced and unbalanced motions such as eddies and internal waves. His results, based on idealized models, realistic simulations, and early SWOT data, suggest that while DMD performs well in identifying mesoscale dynamics, its accuracy remains sensitive to temporal sampling constraints. This work represents a promising step toward improving satellite-based velocity estimates essential for understanding ocean dynamics.