The use of high-precision satellite measurements to derive regional source distributions of methane and carbon dioxide is explored. Three key areas to be considered are first, validation of the satellite data with ground-based FTIR observations, second, modelling the trace gas distribution (the 'forward model'), and third, retrieving the source distribution.

Observations of CO2 and CH4 spectral lines are used to derive a total gas column or partial column below a specified pressure level; the column may then be compared to the satellite column measurement taking into account the different averaging kernels of the two measurements. The scatter in the comparisons to suitable satellite instruments may be further reduced by combining the measurements with similar observations of N2 or O2 to derive column-weighted mean mixing ratio. Techniques for such comparisons are discussed, and time series of mean mixing ratios of CO2 and CH4 at the NDSC site in Lauder, New Zealand are examined.

A preliminary retrieval algorithm has been developed for deriving methane sources from satellite measurements of mean mixing ratio. A detailed regional meterological model (the New Zealand Limited Area Model) has been used to simulate the methane field for assumed source distributions, and is discussed in detail by Struthers et al (these proceedings). This 'forward model' has been coupled with an optimal estimation inverse model to form the source retrieval algorithm. We calculate, first, the sensitivity of the mixing ratio to the emission (measurement weighting functions), second, we simulate retrievals for assumed methane sources, and third, we calculate the sensitivity of the retrieved emission to the assumed emission (measurement averaging kernels). The last two require assumptions about the uncertainty in the satellite measurements and the a priori used in their retrieval of the CH4 column. Finally, we assess the practicality of characterizing regional methane emission from satellites.