[Introduction]
Machine learning (ML) and artificial intelligence (AI) technologies have obtained unrivaled results in a vast range of problems in computer vision, speech recognition, natural language processing, and translation, among others, and have achieved human-level accuracy in less time. These achievements show that these technologies can recognize complex patterns and uncover highly non-linear relations, i.e., in the context of weather forecasting, in a data-driven way.
ML and AI methods have potential in each aspect of numerical weather prediction workflow, including pre-processing of observation data, assimilation of observations into the modeled atmospheric state, prediction of atmospheric state, and post-processing of model outputs. MAP adopts ML methods to advance radar data assimilation and extract severe weather information beyond raw model outputs. Most recent efforts are highlighted as follows:
| 1.Cost-Effective Data Assimilation Using Machine Learning | | | 2.Random Forest (RF)-based convective outlook guidance | | |
* ======================================================================== *
Highlights of Recent and Ongoing Research
[Cost-Effective Data Assimilation Using Machine Learning]
From Wang and Wang (2023; In Preparation)
[Random Forest (RF)-based convective outlook guidance]
