Kalman filter

• Improvement to the forecast error covariances modeling to make them flow dependent

Xa(t0)

Pa(t0)

t

Xf(t0+T)

Pf(t0+T)

• Error growth depends on the dynamics which govern the evolution of the forecast

• Goal: obtain an evolution equation for the forecast error covariances

Diapositive suivante

Revenir à la première diapositive

Afficher la version graphique