One of the recurring points made in discussions here has been the need to include climate-related data of many types, from a wide range of partners potentially scattered across the globe.  Ideally this would be the case.  But I wish to raise a warning flag about such partnerships.  For several decades I have listened to colleagues who actually perform climate research bemoan the problems involved in dealing with long term data sets collected from a changing variety of instruments under protocols that have evolved in time.  Many of the resulting shifts in calibration, accuracy, and overall data quality were due to changes in equipment and/or data collection methods (timing, observer type, averaging methods, etc.).  While it is often technically possible to correct for the effects of such shifts, the information needed is frequently missing or is not detailed enough.  Assumptions must then be made, and additional uncertainties are introduced into the data. These then provide ammunition for people who wish to dispute or even disparage the findings of climate researchers.  I believe it is absolutely critical that any data sets ingested into the planned Climate Service data base must pass several tests.  (1) Point observation climate data must be collected at well-chosen locations that are not strongly influenced by local effects, or likely to be so influenced over the next several decades or even centuries.  (2) Instruments used for point measurements must be mounted in well-documented ways following best current practice (e.g., attention to local heating or moisture sources, representativeness of local surface conditions, redundant sensors, fan-aspiration of temperature sensors, etc.).  "Best current practice" will of course change with time, so I think it is critical that such changes should not be implemented abruptly, but should be phased in over a multi-year period so that the correspondence between data acquired under the old and new methods can be firmly established.  (3) Remote-sensing instruments, whether earth or space-based, must provide frequent comparisons to high-quality ground-based sensors, and their spatial and temporal averaging methods must be well-defined and documented.  Any changes to these methods must provide for abundant overlap with or inclusions of the older methods, so that any shifts in the resulting measurements can be unambiguously determined.  (4) All instruments must be rigorously calibrated and periodically re-calibrated, with traceability to national and global standards.  The calibration procedure must be laid out in great detail so a knowledgeable reader would be able to replicate the process.  The personnel entrusted with these calibrations must be trained and must follow the agreed-upon methods exactly.  (5) Frequent maintenance of sensors and/or sensor self-checks is essential if the data are to be trusted.  (6) The devil is often in the details -- this is especially the case for climate-related data where climate variable changes may be small and difficult to detect.  Extensive metadata must be provided for each instrument and each measurement site, so that potential doubters can examine the information and be assured that the data are the best that can be expected, and that plausible uncertainty values have been developed and provided.  Given such considerations, I think it is imperative that all data collected into and made available by the new Climate Service should pass these tests.  Casual partners that cannot live up to these admittedly high standards simply should not be included.  This may cause difficulties with regard to private-public partnerships and developing countries, but credibility of the information to be provided by the Climate Service depends critically on such high standards.  If we are to eventually understand the world's climate and the reasons behind its variability, the data that our theories rest on must be as trustworthy as possible.  I think it should be a major (perhaps the principal) goal of the NOAA Climate Service to provide world leadership in collecting and distributing climate data that is of indisputable quality, so that future emphasis can be directed toward understanding what it all means, instead of arguing about the worth of the data.

Climate-related data of questionable quality will cause huge problems.  They will mislead researchers, will require large expenditures of time and money to improve to a usable state, and will provide a rationale for people and organizations who wish for any reason to question or deny the conclusions that may be drawn.  Inclusion of questionable data from any source will undermine the entire effort of the proposed Climate Service.  Data from the NOAA Climate Service must be regarded by all as highly trustworthy, and demonstrated to be so by means of great transparency and voluminous metadata.  Anything less will badly impact the organization's scientific utility.