FourCastNet, short for Fourier ForeCasting Neural Network, is a global data-driven weather forecasting model that provides accurate short to medium-range global predictions at 0.25 • resolution. FourCastNet accurately forecasts high-resolution, fast-timescale variables such as the surface wind speed, precipitation, and atmospheric water vapor. It has important implications for planning wind energy resources, predicting extreme weather events such as tropical cyclones, extra-tropical cyclones, and atmospheric rivers. FourCastNet matches the forecasting accuracy of the ECMWF Integrated Forecasting System (IFS), a state-of-the-art Numerical Weather Prediction (NWP) model, at short lead times for large-scale variables, while outperforming IFS for small-scale variables, including precipitation. FourCastNet generates a week-long forecast in less than 2 seconds, orders of magnitude faster than IFS. The speed of FourCastNet enables the creation of rapid and inexpensive large-ensemble forecasts with thousands of ensemble-members for improving probabilistic forecasting. We discuss how data-driven deep learning models such as FourCastNet are a valuable addition to the meteorology toolkit to aid and augment NWP models.

The formation of adduct ions consisting of uranium oxycations and water was studied using an ion trap-secondary ion mass spectrometer. The U(IV) and U(V) species [UO(OH)]+ and [UO2]+ were produced by bombarding the surface of UO3 using molecular primary ions, and the U(VI) species [UO2(OH)]+ was generated by O2 oxidation of [UO(OH)]+ in the gas phase. All three ions formed H2O adducts by termolecular association reactions:  [UO(OH)]+ (a U(IV) species) added three water molecules, for a total of five ligands; [UO2]+ (U(V)) added three or four water molecules, for a total of five or six ligands; and [UO2(OH)]+ (U(VI)) added four water molecules for a total of six ligands. Addition of a seventh ligand was not observed in any of the systems. These analyses showed that the optimum extent of ligation increased with increasing oxidation state of the uranium metal. Hard kinetic models were fit to the time-dependent mass spectral data using adaptive simulated annealing (ASA) to estimate reaction rates and rate constants from kinetic data sets. The values determined were validated using stochastic kinetic modeling and resulted in rate data for all forward and reverse reactions for the ensemble of reactive ions present in the ion trap. A comparison of the forward rate constants of the hydration steps showed that in general, formation of the monohydrates was slow, but that hydration efficiency increased upon addition of the second H2O. Addition of the third H2O was less efficient (except in the case of [UO2]+), and addition of the fourth H2O was even more inefficient and did not occur at all in the [UO2(OH)]+ system. Reverse rate constants also decreased with increasing ligation by H2O, except in the case of [UO(OH)(H2O)4]+, which prefers to quickly revert to the trihydrate. These findings indicate that stability of the hydrate complexes [UO y H z (H2O) n ]+ increases with increasing n, until the optimum number of ligands is achieved. The results enable correlation of uranium hydration behavior with oxidation state.

No abstract