Over the last decade, a number of groups have begun producing decadal-time-scale, near-global, ocean state estimates (the OceanObs'09 Community White Papers (CWP) by Lee and 26 others [2009] and Stammer and 26 others [2009] provide an overview and preliminary evaluation, respectively). Their primary purpose is the understanding of the ocean climate and its variability. Such efforts put demanding requirements on the observing system. As a bridge between the modeling and observation communities, the ocean state estimation (OCEST) groups necessarily assess the skill and deficiencies in both data sets and models. Estimation tools developed by these groups can also be used for defining observational requirements for a climate monitoring and prediction system, although focus has been mainly on production of state estimates. This white paper takes as its major premises that: (A) the climate system is truly global and must be observed globally if it is to be understood; (B) the problem of observing the ocean in its climate role involves some very long time scales; (C) all elements of the oceanic state are dynamically connected, such that measuring any specific element may contain at least some information on a much larger part of the state, and which dynamically consistent interpolators may be able to explore efficiently both forward and backward in time; (D) the ocean is a noisy fluid, full of eddies and variability across many space and time scales; a significant fraction of oceanic energy resides in fast motions (less than a month) and small space scales (less than 100 km). These premises have immediate implications for observational systems design which underlie the remainder of the document.