The investigation of regional climate anomalies with a linear barotropic model and an adjoint technique

Abstract

The adjoint of a steady-state linear barotropic model is presented as a tool for diagnosing regional upper-level vorticity anomalies occurring in general circulation model (GCM) experiments. As a starting point, a simulation of a chosen GCM vorticity anomaly is made using a standard linear barotropic model forced by the corresponding anomalies in the GCM's upper-level divergence and transient-forcing fields. the associated adjoint model is then used to interpret this linear simulation over a chosen geographical region, in terms of the linear model's global forcing field. to do this, the adjoint model is forced by a local information source situated over the chosen region. the resulting adjoint solution consists of a global spatial weighting function which can then be used to identify regions of the forcing—local or remote—associated with the regional anomaly. the validity of barotropic linear theory can also be quantified as a by-product. As an illustration, the model is applied to a large interdecadal anomaly over Europe, arising in a 100-year doubled CO2 integration of the Hamburg coupled ocean-atmosphere GCM. Linear barotropic theory is found to account for over 60% of the European interdecadal anomaly amplitude in terms of anomalous divergence and the effects of transients. Although a large part of the effects of anomalous divergence are spatially concurrent with the European anomaly and just upstream of it, remote forcing regions over the east Pacific and tropical Atlantic are also implicated. Anomalous transients are found to play a secondary role. the linear results are compared with some one-point correlation maps of the GCM's interdecadal variability.