As a civil engineer and data scientist, my work focuses on developing efficient climate data analysis tools to aid in making the best decisions for Climate Change Adaptation and Mitigation.
Climate Change Adaptation -the process of adjusting to changing natural hazards - requires rational data-driven approaches that consider the non-stationary characteristics and serial (temporal) correlation of the climate process to assess the changing risks.
The energy transition to a sustainable clean source is the primary task of Climate Change Mitigation. Ocean-based renewable energy receives much attention in this endeavor. Still, associated costs of this energy need to be reduced significantly. Some innovative ideas have the potential to reduce costs but are still in their early stages with many missing validated rational solutions.
Climate Change Adaptation
Greedy Copula Segmentation of Multivariate Non-Stationary Time Series for Climate Change Adaptation
Climate Change Mitigation
Optimizing Operations and Maintenance for Offshore Multi-Purpose Platforms in Variable Weather Conditions
Sustainable Reuse of Decommissioned Jacket Platforms for Offshore Wind Energy Accounting for Accumulated Fatigue Damage
Learning Generative Embeddings using an Optimal Subsampling Policy for Tensor Sketching
Data tensors of orders 3 and greater are routinely being generated. These data collections are increasingly huge and growing. For instance, a North American Regional Reanalysis (NARR) has been collecting 70 climate variables every 3 hours from 1979 to the present, and it is currently at a total size of 29.4 Terabytes. Directly accessing such large data tensor collections for information has become increasingly prohibitive. We learn approximate full-rank and compact tensor sketches with decompositive representations providing compact space, time and spectral embeddings of the original tensor. All subsequent information querying with high accuracy is performed on the generative sketches. We produce optimal rank-r Tucker decompositions of arbitrary order data tensors by building tensor sketches from a sample-efficient sub-sampling of tensor slices. Our sample efficient policy is learned via an adaptable stochastic Thompson sampling using Dirichlet distributions with conjugate priors.
Tensor Dynamic Mode Decomposition for Continental United States Scale Climate Forecasting
The tensor-based decomposition and dynamic mode decomposition is studied to reduce required computational resources for analyzing the huge size of climate variables. Tensor-train decomposition, Tucker decomposition, and randomized sampling-based SketchyCore approach are studied. The result shows that the climate variable tensors are also low-rank structures, and thus compression technique can be applied to reduce computation cost without loss of information.