Development of a Long-Term (1884-2006) Serially Complete Dataset of U.S. Temperatures and Precipitation for Climate Services

by: Jinsheng You, Kenneth G Hubbard, Martha Shulski

Volume 2015, No. 1, 31 Jan 2015

Serially complete climate datasets with no missing data are necessary for a diverse group of users working in many economic sectors. In this article we describe the procedures used to create a Serially Complete Data set (SCD) for the U.S. We include the selection criterion applied to potential SCD stations, the various procedural steps and the details applied to each step. A few observations that were not previously digitized were obtained from observers official paper reports. The methods used to estimate missing data are the Spatial Regression Test and the Inverse Distance Weighting technique. Using the criterion for selecting stations we were able to include 2144 stations for the SCD that had at least 1 element (maximum/minimum temperature and/or precipitation) for a continuous period of at least 40 years. In addition, the quality control procedure assigned confidence intervals to all observations and many of the estimates. We continue to explore the options for estimating any missing data that remain after our 3 step approach and we look forward to changing the base data set form TD 3200 to GHCN.

<<View Full Article>>