Data Sampling and Analysis for the Improvement of Estimating Heating Loads in Alaska

• Main Author: Gaumer, Madelyn Elizabeth
• Other Authors: Kutz, Nathan
• Format: Thesis
• Language: English
• Published: 2022
• Subjects: Climate Change; Energy; Heating; Machine Learning; Applied mathematics; Arctic; Climate change; Alaska
• Online Access: http://hdl.handle.net/1773/49595

Thesis (Master's)--University of Washington, 2022 Due to the accelerated effects of climate change over the past 10 years, Alaska and the larger Arctic region are in need of decarbonization far more than the rest of the world does. Over 75% of the energy utilized in the Arctic region is for heating houses and businesses. However, a key barrier to the switch to renewable energy is the absence of extensive and accurate heating load estimates in Alaska. This research builds upon previous work to establish a geospatial-first methodology using satellite data to estimate heating loads in Alaska. In this work, we analyze building data and climate data, including ERA5 and Daymet. We also use modern data sampling techniques to combat imbalanced data and show that random sampling performs well compared to other techniques.