Our instrument design and data reduction goals have been driven by a 0.1% error requirement.

In light of considerable natural variability from year to year and week to week in earths albedo we ask which limits on achievable accuracy the natural variability set?

We illuminate this problem by modelling natural albedo variability, based on Figure 3 in Bender et al (2006). The Figure gives an annual cycle and the 1 S.D. variability limits around this cycle.

We use that data to generate 10, 20 and 30 years of daily simulated albedo values. We fit a straight line to the entire dataset as well as to a dataset made up of 100 annual samples (like realistic observing conditions – one data point per night). We calculate the change in albedo over the period simulated and express this change in percent of the average albedo. We do that for the daily sample and for the 100-times-a-year sample. We next show the histograms of these accumulated errors. To the left in each row is the histogram of errors in albedo for the ‘all-nights period’ and to the right is the histogram for the ‘100 annual samples’.

The S.D. of these errors are:

years all days 100 days
10 0.102 0.184
20 0.077 0.132
33 0.060 0.105

The above suggest that natural variability will impose an uncertainty on estimates of global albedo at about 0.1% if just 10 years of nightly data are analysed, but that 30 years of data are needed if only 100 samples a year are made.

Any errors our measurements have above the 0.1% level will thus only make the problem worse.

The above analysis assumes global coverage of our data; all sampling effects due to lunar phases being are ignored by the random selection of 100 point per year.

Overall the analysis suggests that we may be able to achieve climate-change related data of interest in a decade as long as we do not allow errors in our observations or data-reduction much above the 0.1% level.
On shorter time-scales the high accuracy of our method will allow us to track such phenomena as the daily or weekly changes in clouds masses as well as more subtle changes such as seasonal changes. A volcanic eruption would be yum-yum!