Machine learning (ML) is actively being explored as a potential way of representing unresolved sub-grid processes with more realism than traditional parameterizations. A typical approach has been to generate training data with cloud-resolving simulations, and concatenate the vertical profiles of several atmospheric variables into input and output vectors of a feed-forward neural network. However, these networks lack the connections to directly propagate information through the vertical column.  Here we examine if predictions can be improved by instead traversing the vertical column using a recurrent neural network (RNN), which also respects the fact that physical laws do not change by height. By using datasets published in other studies, we test ”vertical RNNs” on three different problems (non-orographic gravity waves, moist physics and non-local unified parameterization). In each case, we find that bidirectional RNNs have similar or higher offline accuracy as feed-forward models while using fewer trainable parameters. While prognostic climate simulations were not performed in this exploratory work, the RNNs are more stable in offline autoregressive tests than ResNets, the previous state-of-the-art.

Accurate wildfire forecasting can inform regional management and mitigation strategies in advance of fire occurrence. Existing systems typically use fire danger indices to predict landscape flammability, based on meteorological forecasts alone, often using little or no direct information on land surface or vegetation state. Here, we use a vegetation characteristic model, weather forecasts and a data-driven machine learning approach to construct a global daily ~9 km resolution Probability of Fire (PoF) model operating at multiple lead times. The PoF model outperforms existing indices, providing accurate forecasts of fire activity up to 10 days in advance, and in some cases up to 30 days. The model can also be used to investigate historical shifts in regional fire patterns. Furthermore, the underlying data driven approach allows PoF to be used for fire attribution, isolating key variables for specific fire events or for looking at the relationships between variables and fire occurrence.