A Demonstration of an Improved Filtering Technique for Analyzing Climate
Records via Comparisons of Satellite MSU/AMSU Instrument Temperature
Products from Three Research Groups
Abstract
Climate data records typically exhibit considerable variation over short
time scales both from natural variability and from instrumentation
issues. The use of linear least squares regression can provide overall
trend information from noisy data, however assessing intermediate time
periods can also provide useful information unavailable from basic trend
calculations. Extracting the short term information in these data for
assessing changes to climate or for comparison of data series from
different sources requires the application of filters to separate short
period variations from longer period trends. A common method used to
smooth data is the moving average, which is a simple digital filter that
can distort the resulting series due to the aliasing of the sampling
period into the output series. We utilized Hamming filters to compare
MSU/AMSU satellite time series developed by three research groups (UAH,
RSS and NOAA STAR), the results published in January 2017 (Swanson
2017).