Bjerknessenterets mål er å forstå klima
til nytte for samfunnet.

Seminar: Prior and Posterior Inflation for Ensemble Filters: Theoretical Formulation and Application to Community Atmosphere Model

Mohamad E. Gharamti from NCAR will give a seminar talk on May 6


Short biography:

Mohamad (a.k.a Moha) has been an active Data Assimilation (DA) researcher since 2010. Moha has worked on various topics in DA including reduced adjoint 3/4D-VARs, Kalman filtering, extended Kalman filtering and ensemble Kalman smoothers/filters. His work focuses on investigating novel DA techniques that tackle mainly state-parameter estimation problems and non-Gaussianity. Currently, Moha is a member of the Data Assimilation Research Section (DAReS) at NCAR and his research encompasses formulating smart inflation/localization procedures in addition to their application in DART. To date his active areas of application are: subsurface hydrology, marine ecosystems and atmosphere.

Mohamed E Gharamti
Mohamed E Gharamti


In this talk, temporally and spatially varying adaptive inflation algorithms for ensemble filters are presented. The algorithms provide an efficient variance correction strategy for the prior and posterior ensemble statistics. The derivation of the adaptive prior scheme is first presented following Bayes Theorem. Then, I will introduce a new posterior inflation scheme featuring an observation-impact removal strategy as a way to sequentially compute the inflation distribution. The usefulness of posterior inflation is investigated and compared to prior inflation with an 80-member ensemble in the Community Atmosphere Model. 6-hour forecasts of the atmospheric state, in the Troposphere and lower Stratosphere, are generated over the month of September 2010. GPS Radio Occultation refractivity observations in addition to wind and temperature data from aircraft, ACARS and satellites are assimilated. The following questions are addressed: What inflation scheme is more effective at handling sampling errors? When model bias dominates other error sources, which inflation strategy yields a better fit to the observations? Does inflating both the prior and the posterior ensemble perturbations help mitigate for different error sources? The evaluation of the inflation algorithms is assessed using observation space diagnostics. The inflation patterns are studied and correlations to observation network densities are examined.


Arranged date for the seminar talk: May 06, 2019

BCCR Seminar room 4020 at 14:15