Skip to content

jvhs0706/regional-forecast-new

Repository files navigation

Regional air pollutant forecasting

This repository introduces a deep-learning model that makes REGIONAL air pollution forecasting for the future 48 hours. It combines the ground observation of the past 72 hours and the WRF-CMAQ results for the future 48 hours. The proposed constitues a variety of structures: LSTM encoder-decoders, bidirectional LSTMs, time-distributed dense layers, and a newly proposed broadcasting layer.

model

The main contributions of the project is:

  • Breaking the spatial confinement of the ground-observation-based deep-learning air pollution forecast to the ground monitor stations.
  • Superceding the widely-used spatial correction (interpolation methods).
  • Proposing the novel broadcasting layer that may be applied to deep-learning tasks of similar nature.

The publication related to the project is under composition and review.

About

The new version for jvhs0706/regional-forecast

Resources

License

Stars

Watchers

Forks

Packages

No packages published