PDAFomi_obs_f Module

PDAF-OMI routines for full observations

This module contains subroutines to handle full observations. Further, it contains routines to restrict the global full vector of observations to those observations that are relevant for a process-local model subdomain. The routines are

• PDAFomi_gather_obs \n Gather full observation information
• PDAFomi_gather_obsstate \n Gather a full observed state vector (used in observation operators)
• PDAFomi_init_obs_f \n Initialize full vector of observations for adaptive forgetting factor
• PDAFomi_init_obsvar_f \n Compute mean observation error variance for adaptive forgetting factor
• PDAFomi_prodRinvA \n Multiply an intermediate matrix of the global filter analysis with the inverse of the observation error covariance matrix
• PDAFomi_likelihood \n Compute likelihood for an ensemble member
• PDAFomi_add_obs_err \n Add observation error to some matrix
• PDAFomi_init_obscovar \n Initialize global observation error covariance matrix
• PDAFomi_set_domain_limits \n Set min/max coordinate locations of decomposed grid
• PDAFomi_get_domain_limits_unstr \n Find min/max coordinate locations in unstructured grid
• PDAFomi_get_local_ids_obs_f \n Find observations inside or close to process domain
• PDAFomi_limit_obs_f \n Reduce full observation vector to part relevant for local process domain

The routine PDAFomi_get_domain_limits_unstr assumed geographic coordinates in radians and within the range -pi to +pi for longitude (- is westward) and -pi/2 to +pi/2 for latitude.

Revision history: * 2019-06 - Lars Nerger - Initial code * Later revisions - see repository log