 |
NARSTO Workshop on Innovative Methods for Emission Inventory Development and Evaluation University of Texas, Austin October 14-17, 2003 |
 |
Tropospheric Chemical Data Assimilation and Inverse Modeling at the Meteorological Service of Canada
Richard Ménard1, Alain Robichaud1, Yan Yang2,
Saroja Polavarapu2, Jacek Kaminski3, and Emanuel Cosme4
1 Air Quality Research Branch, Meteorological Service of Canada
2 Data Assimilation and Satellite Meteorology Division, Meteorological Service of Canada
3 York University
4 McGill University
A tropospheric chemical data assimilation effort is developing at the meteorological service of Canada. The objectives are to implement in real-time an air quality forecasting and assimilation system and to develop in collaboration with universities the science of chemical data assimilation and inverse methodologies. While data assimilation has been used for operational weather prediction for decades its application to atmospheric chemistry is recent. Chemical data assimilation has specific challenges and objectives that are different from typical meteorological data assimilation problems.
One effort consist in assimilating surface ozone observations in real-time using the model CHRONOS with a domain covering North America. We will discuss the estimation of error statistics, the choice of covariance models, the quality control, the validation of objective analyses, and the predictability skill due to assimilation. Results from this and the past year will be presented. While developing an operational effort, research is also being pursued in a dual estimation of state and sources, i.e. a combined data assimilation and inverse modeling system. In particular we found that the estimation problem can be separated into two estimation problems one for the observed and the other for the unobserved variables, leading to a computationally effective method to perform multivariate (or multispecies) and dual statesource estimation. We will also discuss a method used obtain the cross error covariance statistics needed for the implementation of such systems, and some preliminary results will be presented. Finally and if time permits, we will also discuss our efforts to assimilate satellite observations using a global coupled meteorological-air quality model, GEM-AQ, in particular with CO observations from MOPITT and results from aerosol assimilation.