High frequency CO2 and wind speed measurements were used to examine the urban baseline eddy covariance CO2 flux and analyse the CO2 rich plume from a local power station. A reliable relationship between high frequency CO2 maxima and the rate of CO2 emission at the power station was established. This relationship was shown to be highly dependant on wind speed. The ensemble mean plume was found to be approximately Gaussian in horizontal profile with a width dependant on wind speed. The relationship between peak CO2 mixing ratio and averaging time was shown to be a simple power law with a time exponent of approximately 0.5. The large, short pulses in CO2 mixing ratio in the power plant plume were found to have an approximately Lorentzian shape. These pulses generated negative vertical eddy flux measurements so data from the plume sector were necessarily excluded from the flux baseline results. The plume-excluded flux had a similar magnitude and variability to those reported in other urban CO2 flux studies despite this site not being ideal due to the proximity of roughness elements to the measurement point. A linear-eddy model was used to relate the mean concentration plume profile to plume profiles for different percentiles of concentration. It showed that for high concentration percentiles and short dispersion times the plume width can be several times greater than the standard mean concentration plume width.