In order to monitor global climate change on a decade-to-decade basis in support of national and foreign policy decisions, it will be necessary to better quantify and to substantially reduce the measurement errors inherent in estimates of global-mean temperature, as well as to develop an improved understanding of the processes that contribute to short term variability of global-mean temperature. To achieve these goals, the panel recommends the following actions:
(1) The nations of the world should implement a substantially improved temperature monitoring systems10 that ensures the continuity and quality of critically important data sets. Needed measurements include not only the conventional climatic variables (temperature and precipitation), but also the time-varying, three-dimensional spatial fields of ozone, water vapor, clouds, and aerosols, all of which have the potential to cause surface and lower to mid-tropospheric temperatures to change relative to one another. Management of climate data sets also needs additional attention and support. Raw and processed measurements and follow-on products need to be accessible in a form that enables a number of different research groups to replicate the processing of the more widely disseminated datacontinue
10 The NRC report Adequacy of Climate Observing Systems (NRC, 1999) describes characteristics that should be incorporated into the design of climate monitoring systems to facilitate the detection of climate change.
sets and to develop new and improved temperature algorithms. To ensure such access, the ongoing documentation of instrumentation and observing practices, the archiving of data sets, and the provision of raw and processed data sets in electronic form to the scientific community should be regarded as integral parts of the climate monitoring effort and afforded high priority in terms of funding.
(2) The scientific community should perform a more comprehensive analysis of the uncertainties inherent in the surface, radiosonde, and satellite data sets. Such an assessment should involve a detailed analysis of the sensitivity of global-mean temperatures derived from these three different measurement systems to the various choices made in the processing of the raw datae.g., corrections for instrument changes, adjustments for orbital decay effects in the satellite measurements, and procedures for interpolating station data onto grids. Such studies should also address the comparison of data sets with different sampling characteristics.
(3) Natural as well as human-induced changes should be taken into account in climate model simulations of atmospheric temperature variability on the decade-to-decade time scale. In particular, the studies described in Finding #4 need to be repeated with improved models and with an experimental design that reflects the uncertainties in natural and human-induced forcings.
(4) The scientific community should explore the possibility of exploiting the sophisticated protocols that are now routinely used to ensure the quality control and consistency of the data ingested into operational numerical weather prediction models, to improve the reliability of the data sets used to monitor global climate change.break