D21    Development of temporal error correlation model

Responsible partner
Due date
This is part of work-package 3300, the purpose of which is modelling and correction of GPS temporal error correlations.

ObjectivesTo develop and test a model for the temporal correlation of ground-based GPS observation errors in 4-dimensional variational data assimilation.

Methodology/Work Description: With the introduction of 4-dimensional variational data assimilation several observations from the assimilation window, for example a 6 hour period, and from the same station may be utilised. Experiences from the 4D-Var assimilation of surface observations have shown that the sensitivity of the assimilation to systematic observation errors may become critical and that models for the temporal correlation of observation errors need to be specified (Jaervinen et al, 2000). Furthermore, due to the coding design of variational data assimilation schemes, it may be easier to implement models for the temporal correlation of errors, rather than models for the spatial correlations of errors. Models for the temporal correlation will alternatively be developed from innovation vectors, i.e. differences between GPS observations and model data, or from differences between GPS observations and high quality radiosonde observations.

The efficiency of the developed temporal error correlation model  will be implemented and tested through data assimilation and forecast experiments with and without application of the error correlation model.

As in WP3200 we will also make use of the existing snad new time series of vertical position estimates. The temporal correlation - or decorrelation with time - will be studied. Studies indicate so far decorrelation times fo the order 2-4 days. We will based on results from more extensive studies try to separate the observed temporal variations in the ZTD, estimated using GPS data, into the decorrelation of the signal (ZTD) and the GPS error.

HIRLAM 4DVar, which is used at both SMHI and DMI, first became available for pratical experimentation late 2005, which prevented SMHI from finalising this deliverable as planned. Instead
DMI, in connection with their 4DVar simulations in WP4000, invented at model for correction of the temporal error correlations of the GPS ZTDs. The work of DMI on the temporal error model and
corrections is described in deliverable D23