Stochastic Rainfall Modeling at Sub‐kilometer Scale

New measurement devices allow observing rainfall with unprecedented resolution. Such observations often reveal new features of rainfall occurring at the local scale (areas of about 1–25 km2). In particular, the joint effects of the advection of rain storms over the ground, and the deformation of spatial rain patterns along time, generate a complex rain field dependence structure characterized by strong space‐time interactions. When a high‐resolution is desired, stochastic rainfall models must therefore be upgraded to account for these new features of rain fields. In this paper, we propose to improve the meta‐Gaussian framework, which is typically used to model space‐time rain fields, to the specific case of sub‐kilometer rainfall. Particular attention is paid to the reproduction of the main features of local scale rainfall, namely: (1) a skewed distribution of rain intensities with the presence of intraevent intermittency and (2) a space‐time dependency structure with strong and complex space‐time interactions. The resulting model, able to generate high‐resolution, continuous and space‐time rain fields at the local scale, is validated and applied to a real data set collected by a network of drop‐counting rain gauges recording rainfall at a 1 min frequency. The combination of these data with the proposed model results in a complete framework that allows resolving the features of high‐resolution rainfall (1 min temporal resolution, 100 m spatial resolution) over a small alpine catchment in Switzerland. Key Points A stochastic rainfall model appropriate for very high‐resolution rainfall features is proposed An efficient Bayesian estimation is used to infer model parameters The proposed framework is applied to a very dense rain gauge network