JEDI-MPAS 1.0.0, a new data assimilation (DA) system for the MPAS model, was publicly released for community use. This article describes JEDI-MPAS's implementation of the ensemble–variational DA technique and demonstrates its robustness and credible performance by incrementally adding three types of microwave radiances (clear-sky AMSU-A, all-sky AMSU-A, clear-sky MHS) to a non-radiance DA experiment. We intend to periodically release new and improved versions of JEDI-MPAS in upcoming years.

In the mountains, the combination of large model error and observation sparseness is a challenge for data assimilation. Here, we develop two variants of the particle filter (PF) in order to propagate the information content of observations into unobserved areas. By adjusting observation errors or exploiting background correlation patterns, we demonstrate the potential for partial observations of snow depth and surface reflectance to improve model accuracy with the PF in an idealised setting.

To better exploit the observational information available for all scales in data assimilation systems, we investigate a new method to introduce scale separation in the algorithm. It consists in carrying out the analysis with spectral localisation for the large scales and spatial localisation for the residual scales. The performance is then checked explicitly and separately for all scales. Results show that accuracy can be improved for the large scales while preserving reliability at all scales.

A data assimilation scheme was developed to improve snow water equivalent (SWE) simulations by updating meteorological forcings and snowpack states using passive microwave satellite observations. A chain of models was first calibrated to simulate satellite observations over northeastern Canada. The assimilation was then validated over 12 stations where daily SWE measurements were acquired during 4 winters (2012–2016). The overall SWE bias is reduced by 68 % compared to original SWE simulations.

This study proposed a new cloud retrieval method based on the particle filter (PF). The PF cloud retrieval method is compared with the Multivariate and Minimum Residual (MMR) method that was previously established and verified. Cloud retrieval experiments involving a variety of cloudy types are conducted with the PF and MMR methods with measurements of Infrared radiances on multi-sensors onboard both GOES and MODIS, respectively.

This study investigates the assimilation of optical reflectances, snowdepth data and both combined into a multilayer snowpack model. Data assimilation is performed with an ensemble-based method, the Sequential Importance Resampling Particle filter. Experiments assimilating only synthetic data are conducted at one point in the French Alps, the Col du Lautaret, over five hydrological years. Results of the assimilation experiments show improvements of the snowpack bulk variables estimates.

A realistic ocean circulation model is adapted to explicitly simulate model uncertainties and an ensemble data assimilation -stochastic perturbations, altimetric data and 4-D observation operator- is developed in order to control the Gulf Stream dynamic. The performance of the ensemble system is measured through probabilistic approach; the update then adjusts the bias and the dispersion of the ensemble (reliability) and reduces the uncertainty by 30% (resolution) for the SSH variable.

In this paper, a simple and generic implementation approach is presented, with the aim of transforming a deterministic ocean model (like NEMO) into a probabilistic model. With this approach, several kinds of stochastic parameterizations are implemented to simulate the non-deterministic effect of unresolved processes, unresolved scales, and unresolved diversity. The method is illustrated with three applications, showing that uncertainties can produce a major effect in the model simulations.