Disaggregating Daily Precipitation Data 1990 to 2022 into Half-Hourly Intervals Using LSTM Models

Download

• Paper
• Presentation Slides

Abstract

We present a machine learning approach for disaggregating daily precipitation data into half-hourly intervals, specifically tailored for the Australian climate data collection context. The basis of our model is long short-term memory (LSTM), a type of recurrent neural network that can capture temporal correlations in sequential data over an arbitrary timeframe. We demonstrate the effectiveness of our approach in four Australian capital cities using BoM data.

Citation

Oates, H., Arora, N., Oh, H., Lee, T., 2024, ‘Disaggregating Daily Precipitation Data 1990 to 2022 into Half-Hourly Intervals Using LSTM Models’, in: Proceedings of the Asia-Pacific Solar Research Conference 2024, Australian PV Institute, Dec. 2024. ISBN: 978-0-6480414-8-1.

@inproceedings{oates2024disaggregating,
  author    = {Oates, Harrison and Arora, N. and Oh, H. and Lee, T.},
  title     = {Disaggregating Daily Precipitation Data 1990 to 2022 into Half-Hourly Intervals Using LSTM Models},
  booktitle = {Proceedings of the Asia-Pacific Solar Research Conference 2024},
  publisher = {Australian PV Institute},
  year      = {2024},
  month     = {December},
  isbn      = {978-0-6480414-8-1}
}