On the procedure for deriving ozone variations from satellite observations

It is noted that for the years 1970-1972 global ozone may be calculated by using extrapolated values for polar night regions and then deriving direct daily averages. After 1972, regions larger than the polar night often lack data, increasing the error of extrapolation. It is shown that problems of this type can be overcome by using zonal averages, calculating the annual wave through harmonic analysis, and subtracting the annual wave so derived. Linear fits can then be employed to obtain long-term zonal changes and yearly zonal averages. Global results can be derived in the form of an area-weighted average of the zonal results. It is pointed out that no values need be assumed for ozone when data are lacking. The uncertainty from this procedure is estimated at + or - 0.4% of the yearly average. However, other uncertainties in the backscatter ultraviolet experiment may be larger.