earth-comparator-set-lives-outside-the-supplied-archive

# Earth: the supplied archive is an input, never the comparison set Budget acquisition time at design, before you write analysis code. In Earth science a number for a physical quantity is not a result until it stands beside independent estimates of the same quantity, and those never live in the folder you were handed. Plan to fetch four things: the other measurement techniques or instruments that estimate this quantity; the prior published assessments or synthesis values for it; the operational or reference baseline the field scores against (persistence, climatology, the incumbent product); and, where the claim concerns the future, a projection ensemble spanning the low and high forcing pathways the archive supports. Plot every independent estimate's absolute value with its uncertainty interval, side by side, one panel per unit of analysis (region, basin, epoch, variable, lead). Only then add the pairwise differences. A plot of differences from your chosen reference is not an intercomparison, and it is the commonest way a correct set of numbers fails to register as one. Express the same result in each normalisation the field uses, per-unit-area intensity, absolute total, and the impact-relevant equivalent, and state whether the headline is area- or mass-weighted. Repeat the entire analysis under every scenario the projection data supports, state the ordering across scenarios, report the share falling in each severity class, and carry it to the end-of-century horizon. Where...