Forecasts of atmospheric chemical concentrations are becoming of increased importance. At the moment a variety of methods (statistical, expert systems, and CTM-based) are being used to forecast air quality and support air pollution abatement strategies in certain regions (over 120 cities in the US are issuing air pollution forecasts). An increased ability to forecast air pollution has obvious important societal consequences. Furthermore the use of chemical forecasting in support of comprehensive atmospheric chemistry and air pollution studies is becoming the standard mode of operation. In addition, forecasting provides one of the best model evaluations, as the model cannot be tuned in advance. In this paper we report on some recent advances in chemical data assimilation using 4-dimensional variational techniques. Computational tools will be presented along with results from studies aimed at implementing data assimilation techniques. The applications include: a systematic evaluation of the impact of the assimilation of individual chemical species and set of species on model prediction skills and to provide guidance on what species to focus assimilation efforts on to improve operational forecast of air quality; and the design of optimal network designs and adaptive measurement strategies, and analysis of measurements needed to support long term forecasting efforts at a fixed location (such as urban air quality), and also at designing strategies for intensive field operations (e.g., identification of critical quantities to be measured, additional spatial locations to conduct measurements, and when/where to critically deploy resources.