Significant advancements have been made to the Community Multiscale Air Quality (CMAQ) forecast system in the recent years. Meanwhile, the atmospheric chemistry observing system has been improved in its coverage and accessibility. Integrating the observations into the air quality models through data assimilation has a great potential to reduce the forecasting errors; this methodology has been clearly demonstrated in weather forecasting. In this study, we implement the chemical data assimilation using CMAQ model and surface and aloft ozone observations. CMAQ model background error covariance is estimated using the observational (Hollingworth-Lonnberg) method with AIRNOW and ozonesonde observations along with CMAQ ozone forecasts. The error statistics is then utilized in the following data assimilation tests through truncated singular value decomposition (TSVD) regularization. Both statistical interpolation and variational data assimilation methods are implemented. Results from different schemes are compared and discussed in terms of their performance and application feasibility in future forecasting operations.