We introduce SpLIIF to generate implicit neural representations and enable arbitrary downscaling of weather variables. We train a model from sparse weather stations and topography over Japan and evaluate in- and out-of-distribution accuracy predicting temperature and wind, comparing it to both an interpolation baseline and CorrDiff. We find the model to be up to 50% better than both CorrDiff and the baseline at downscaling temperature, and around 10-20% better for wind.

In recent years, data-driven, deep learning-based approaches for precipitation nowcasting have attracted significant attention, showing promising results. However, many existing models fail to fully exploit the extensive atmospheric information available, relying primarily on precipitation data alone. This study introduces two novel deep learning architectures, SmaAt-fUsion and SmaAt-Krige-GNet, specifically designed to enhance precipitation nowcasting by integrating multi-variable weather station data with radar datasets. By leveraging additional meteorological information, these models improve representation learning in the latent space, resulting in enhanced nowcasting performance. The SmaAt-fUsion model extends the SmaAt-UNet framework by incorporating weather station data through a convolutional layer, integrating it into the bottleneck of the network. Conversely, the SmaAt-Krige-GNet model combines precipitation maps with weather station data processed using Kriging, a geo-statistical interpolation method, to generate variable-specific maps. These maps are then utilized in a dual-encoder architecture based on SmaAt-GNet, allowing multi-level data integration. Experimental evaluations were conducted using four years (2016--2019) of weather station and precipitation radar data from the Netherlands. Results demonstrate that SmaAt-Krige-GNet outperforms the standard SmaAt-UNet, which relies solely on precipitation radar data, in low precipitation scenarios, while SmaAt-fUsion surpasses SmaAt-UNet in both low and high precipitation scenarios. This highlights the potential of incorporating discrete weather station data to enhance the performance of deep learning-based weather nowcasting models.

Data assimilation of observational data into full atmospheric states is essential for weather forecast model initialization. Recently, methods for deep generative data assimilation have been proposed which allow for using new input data without retraining the model. They could also dramatically accelerate the costly data assimilation process used in operational regional weather models. Here, in a central US testbed, we demonstrate the viability of score-based data assimilation in the context of realistically complex km-scale weather. We train an unconditional diffusion model to generate snapshots of a state-of-the-art km-scale analysis product, the High Resolution Rapid Refresh. Then, using score-based data assimilation to incorporate sparse weather station data, the model produces maps of precipitation and surface winds. The generated fields display physically plausible structures, such as gust fronts, and sensitivity tests confirm learnt physics through multivariate relationships. Preliminary skill analysis shows the approach already outperforms a naive baseline of the High-Resolution Rapid Refresh system itself. By incorporating observations from 40 weather stations, 10\% lower RMSEs on left-out stations are attained. Despite some lingering imperfections such as insufficiently disperse ensemble DA estimates, we find the results overall an encouraging proof of concept, and the first at km-scale. It is a ripe time to explore extensions that combine increasingly ambitious regional state generators with an increasing set of in situ, ground-based, and satellite remote sensing data streams.

A data-driven model for predicting the surface temperature using neural networks was proposed to alleviate the computational burden of numerical weather prediction (NWP). Our model, named TPTNet uses only 2m temperature measured at the weather stations of the South Korean Peninsula as input to predict the local temperature at finite forecast hours. The turbulent fluctuation component of the temperature was extracted from the station measurements by separating the climatology component accounting for the yearly and daily variations. The effect of station altitude was then compensated by introducing a potential temperature. The resulting turbulent potential temperature data at irregularly distributed stations were used as input for predicting the turbulent potential temperature at forecast hours through three trained networks based on convolutional neural network (CNN), Swin Transformer, and a graphic neural network (GNN). The prediction performance of our network was compared with that of persistence and NWP, confirming that our model outperformed NWP for up to 12 forecast hours.