A study on assimilating potential vorticity data

The correlation that exists between the potential vorticity (PV) field and the distribution ofchemical tracers such as ozone suggests the possibility of using tracer observations as proxyPV data in atmospheric data assimilation systems. Especially in the stratosphere, there areplentiful tracer observations but a general lack of reliable wind observations, and the correlationis most pronounced. The issue investigated in this study is how model dynamics would respondto the assimilation of PV data. First, numerical experiments of identical-twin type were conductedwith a simple univariate nuding algorithm and a global shallow water model based onPV and divergence (PV-D model). All model fields are successfully reconstructed through theinsertion of complete PV data alone if an appropriate value for the nudging coefficient is used. A simple linear analysis suggests that slow modes are recovered rapidly, at a rate nearly independentof spatial scale. In a more realistic experiment, appropriately scaled total ozone datafrom the NIMBUS-7 TOMS instrument were assimilated as proxy PV data into the PV-Dmodel over a 10-day period. The resulting model PV field matches the observed total ozonefield relatively well on large spatial scales, and the PV, geopotential and divergence fields aredynamically consistent. These results indicate the potential usefulness that tracer observations, as proxy PV data, may offer in a data assimilation system.