Downscaling Global Climate Models
Home
Seminar Slides
Introduction: Motivation
Introduction: Key Ideas
Background
Background: Related Work
Background: Gaps in the Research
Research Design: Project Stages
Research Design: Sites of Interest
Research Design: Evaluation Methods
Research Design: Quantifying Uncertainty
Research Design: Modelling
Research Design: Projections
Data Preprocessing: GCM Ensemble
Data Preprocessing: GCM Grid Selection
Data Preprocessing: GCM Time Scale
Data Preprocessing: Partitions
Model Development: Baseline Models
Model Development: Proposed Models
Model Development: Architecture Search
Model Development: Candidate Search
Results: Single Point Comparison
Results: 2d Grid Comparison
Results: MPI
Results: Projections
Results: Discussion
Conclusions Summary
Conclusions Limitations
Dissertation Chapters
Abstract
Acknowledgements
Chapter 1: Background
Chapter 2: Literature Review
Chapter 3: Materials and Method
Chapter 3: Region of Study
Chapter 3: Data Required for the Study
Chapter 3: Data Preparation
Chapter 3: Evaluation Metrics
Chapter 3: Model Development
Chapter 3: Summary
Chapter 4: Model Evaluation and Comparison
Chapter 4: Local Downscaling Comparison
Chapter 4: Universal Downscaling Comparison
Chapter 4: Summary
Chapter 5: Uncertainty of Models
Chapter 5: Assessment of Local Model
Chapter 5: Assessment of Universal Model
Chapter 5: Summary
Chapter 6: Projection of Climate Pathways
Chapter 6: Local Model Projection
Chapter 6: Universal Model Projection
Chapter 6: Summary
Chapter 7: Conclusion, Limitations and Further Work
Model Development
Proposed Model - Candidate CNN-LSTM grid search.
Use set of candidate CNN architectures.
Use configurations resulting from comparisons.
Search LSTM units 12, 32, 64, 128
Vary Learning Rate, Momentum, Loss Function.
Back
Next
Deep Dive: Chapter 3.
Downscaling Global Climate Models with Convolutional and Long-Short-Term Memory Networks for Solar Energy Applications
by
C.P. Davey
is licensed under a
Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License
.