Search

Global dynamical reanalyses of the atmosphere and ocean fundamentally rely on observations, not just for the assimilation (i.e. for the definition of the state of the Earth-system components), but also in many other steps along the production chain. Observations are used to constrain the model boundary conditions, for the calibration or uncertainty determination of other observations, and for the evaluation of data products. This requires major efforts, including data rescue (for historical observations), data management (including metadatabases), compilation and quality control, and error estimation. The work on observations ideally occurs one cycle ahead of the generation cycle of reanalyses, allowing them to make full use of it. In this paper we describe the activities within ERA-CLIM2, which range from surface, upper-air and Southern Ocean data rescue to satellite data recalibration and from the generation of snow cover products to the development of a global station data metadatabase. The project has not produced new data collections. Rather, the data generated has fed into global repositories and will serve future reanalysis projects. The continuation of this effort is first contingent upon the organization of data rescue, but also upon a series of targeted research activities to address newly-identified in-situ and satellite records.

Snow cover variability has significant effects on local and global climate evolution. By changing surface energy fluxes and hydrological conditions, changes in snow cover can alter atmospheric circulation and lead to remote climate effects. To document such multi-scale climate effects, atmospheric reanalysis and derived products offer the opportunity to analyze snow variability in great detail far back to the early 20th century. So far only little is know about their quality. Comparing snow depth in four long-term reanalysis datasets with Russian in situ snow depth data, we find a moderately high daily correlation (around 0.6–0.7), which is comparable to correlations for the recent era (1981–2010), and a good representation of sub-decadal variability. However, the representation of pre-1950 inter-decadal snow variability is questionable, since reanalysis products divert towards different base states. Limited availability of independent long-term snow data makes it difficult to assess the exact cause for this bifurcation in snow states, but initial investigations point towards representation of the atmosphere rather than differences in assimilated data or snow schemes. This study demonstrates the ability of long-term reanalysis to reproduce snow variability accordingly.