Automated detection of weather fronts using a deep learning neural network

Biard, James C.; Kunkel, Kenneth E.

doi:https://doi.org/10.5194/ascmo-5-147-2019

Articles | Volume 5, issue 2

https://doi.org/10.5194/ascmo-5-147-2019

© Author(s) 2019. This work is distributed under
the Creative Commons Attribution 4.0 License.

https://doi.org/10.5194/ascmo-5-147-2019

© Author(s) 2019. This work is distributed under
the Creative Commons Attribution 4.0 License.

Articles | Volume 5, issue 2

21 Nov 2019

| 21 Nov 2019

Automated detection of weather fronts using a deep learning neural network

James C. Biard and Kenneth E. Kunkel

Data sets

• National Weather Surface Coded Surface Bulletins 2003- National Weather Service https://doi.org/10.5281/zenodo.2642801

National Weather Surface Coded Surface Bulletins 2003- (JSON format) J. C. Biard https://doi.org/10.5281/zenodo.2646544

National Weather Surface Coded Surface Bulletins 2003- (netCDF format) J. C. Biard https://doi.org/10.5281/zenodo.2641072

DL-FRONT MERRA-2 weather front probability maps over North America 1980- J. C. Biard and K. E. Kunkel https://doi.org/10.5281/zenodo.2641072

DL-FRONT MERRA-2 vectorized weather fronts over North America 1980-2018 (JSON format) J. C. Biard and K. E. Kunkel https://doi.org/10.5281/zenodo.2669180

DL-FRONT MERRA-2 vectorized weather fronts over North America 1980-2018 (netCDF format) J. C. Biard and K. E. Kunkel https://doi.org/10.5281/zenodo.2669505

DL-FRONT MERRA-2 weather front processing and analysis artifacts J. C. Biard and K. E. Kunkel https://doi.org/10.5281/zenodo.2712481

Short summary

A deep learning convolutional neural network (DL-FRONT) was around 90 % successful in determining the locations of weather fronts over North America when compared against front locations determined manually by forecasters. DL-FRONT detects fronts using maps of air pressure, temperature, humidity, and wind from historical observations or climate models. DL-FRONT makes it possible to do science that was previously impractical because manual front identification would take too much time.