Thursday, 27 January 2011: 3:30 PM
607 (Washington State Convention Center)
One of the most important considerations in scientific research is the quality of the data used for new and important findings. However, a problem with the data, even if it is undenounced to the researcher can completely nullify the results, especially after it is released by a data center. Therefore a citation reference, version number or another way of identifying the data is needed for all data sets used in research. Especially in creating a climate data record, the quality of the input data also needs to the best available data at that time with metadata. In the world of satellite data, the need for a citation is increasing important since sensors change very quickly and products are always being re-processed over time with new releases super seceding older versions. An example of this can be made with the SSM/I passive microwave daily gridded brightness temperature products archived and available from the National Snow and Ice Data Center (NSIDC). Analysis of the gridded brightness temperature data showed a large number of bad scans in the data set. These bad scans should have been flagged and removed by the quality control software, but were not. A count of all bad scans that were partially or fully detected was completed from 1 March through 30 September 1989. Although the effect of bad scans on individual days may be small, the cumulative effect of bad scans over multiple days or in small subset regions may be significant, especially when the problem occurs over many years of data. Therefore the SMM/I gridded data were reprocessed for 1988-2001. However, there is no way to determine whether individuals have updated the data, which in this case was released over 10 years later unless data citations are required for all publications, one does not know if bad data were used in the analysis. Users of the SSM/I gridded data should update their holdings with the newly gridded data. However, notifying users of the data is a difficult task and the use of a citation at least allows one to identify which data set is being used.
![](images/flag.png)
![](images/blue_ribbon.png)