UMBC ebiquity
DDDAS (Dynamic Data Driven Applications Systems) 2024 to Appear in Springer
Towards a Dynamic Data Driven AI Regional Weather Forecast Model
November 8, 2024
The advent of long-term reanalysis datasets such as ECMWF ERA 4/5 has enabled the development of AI-driven machine learning models for weather forecasting. The major benefit of AI as an approach is its ability to reduce computational forecast time from tens of hours to tens of seconds, thereby enabling a variety of new applications rang- ing from extreme regional weather event forecasting to first responder aid for wildfires, severe storms, floods, oil spills, tornadoes, and other extreme events in real time. Today, several operational weather forecast centers are evaluating these models as complements or alternatives to their existing models. However, similar efforts in applying AI/ML ap- proaches to mesoscale weather forecasting have lagged behind due to the lack of a reanalysis of current operational regional weather forecast models. Recently, the ECMWF made publicly available Copernicus Eu- ropean Regional ReAnalysis (CERRA) at spatial resolutions of 11km (0.10) and 5.5km (0.050) from 1984 to the present. We present the first demonstration of a successful AI regional forecast at 5.5 km spatial reso- lution employing the Nvidia FourCastNet (FCN) model with its Adaptive Fourier Neural Operator (AFNO) and transformer self-attention model- ing approach. We describe the training of a regional FourCastNet model in the NASA Center for Climate Studies (NCCS) Adapt cluster at the Goddard Space Flight Center using five years of CERRA reanalysis data at 3-hour intervals for five variables at four pressure levels. We show the RMSE forecast errors of a 5.5km implementation trained on five years of data improved for all variables but one over a forecast trained on three. We also devise a nesting scheme wherein our regional model is boundary forced by a global forecast. We find that our model improves on the per- formance
InProceedings
Downloads: 56 downloads