Topic: Non-Gaussian Data Assimilation

Friday, January 20, 2023 from 7-9 UTC

Organisers and Conveners: Steven Fletcher (CIRA/CSU, US), James Taylor (RIKEN, Japan)

One of the underlying assumptions underpinning many data assimilation schemes (including variational, Kalman filter, or ensemble-based) is that the background, observational and model errors are Gaussian in distribution. However, this assumption is often false, with errors taking a non-Gaussian distribution. In this session, we invite all contributions to the development of non-Gaussian DA.

Program / Abstracts (PDF Download)

Questions asked during the event (Yopad link)

Session recordings (RIKEN website)


  • A Quantile Conserving Ensemble Filtering Framework: Regressing Probit-Transformed Quantile Increments to Update Unobserved Variables
  • Revisiting and Repurposing the Gaussian Anamorphosis EnKF
    Hristo G. CHIPILSKI, Ian Grooms, Mohamad El Gharamti, Jeffrey Anderson, Ricardo Baptista
  • A two-step nonlinear non-Gaussian framework for data assimilation applied to assimilation of wind direction observations
    Ian GROOMS
  • Improving vortex position accuracy with a new multiscale alignment ensemble filter
    Yue (Michael) YING, Jeffrey Anderson, Laurent Bertino
  • "Ensemblized" linear least squares (LLS)
    Patrick N. RAANES

Time Zones:
09 UTC
Europe:            07 – 09 am GMT (London)      | 08 – 10 am CET (Berlin)
Asia/Australia: 03 – 05 pm CST (Shanghai)   | 04 – 06 pm JST (Tokyo)      | 06 – 08 pm AEDT (Sydney)
Americas:        11pm – 01 am PST (San Fran.)  | 00 – 02 am MST (Denver)   | 02 – 04 am EST (New York)

How to attend to ISDA-Online events?
Simply subscribe to the ISDA-Online mailing list