ClimateNet: Bringing the power of Deep Learning to weather and climate sciences via open datasets and architectures

Image | Atmospheric River and Tropical Cyclones | Response

ClimateNet Dataset is a dataset for atmospheric river and tropical cyclone damage assessment. It contains 219 image samples, and is introduced for semantic segmentation and multilabel classification.

  • ML task type: Semantic segmentation, multilabel classification
  • Data Source: Climate simulation data (Expert-labelled)
  • Size: 219 Samples
  • Timespan: N/A
  • Geographical Coverage: Global
  • Baseline Information
  • Evaluated on: DeepLabv3+
  • Metrics used: Mean Intersection over Union, Background IoU, Tropical Cyclone IoU, Atmospheric River IoU
  • Results as reported in original paper: Mean IoU: 0.5247; Background IoU: 0.9389; Tropical Cyclone IoU: 0.2441; Atmospheric River IoU: 0.3910

Karthik Kashinath, Mayur Mudigonda, Ankur Mahesh, Jiayi Chen, Kevin Yang, Annette Greiner, and Mr Prabhat. Climatenet: Bringing the power of deep learning to weather and climate sciences via open datasets and architectures. In AGU Fall Meeting Abstracts, volume 2019, pages GC33A–06, 2019.