Summary of the Day (TD3200)

General Information

It was during the mid-1800s that the beginning of a network of climatological observers began to form in the United States. There was not then, as there is not now, funding to provide paid observers at more than a handful of stations; the decision to provide uniform equipment and permit people to make the observations, either as volunteers or on an "other duties as assigned" basis, allowed the network to spread even as it achieved a reasonable geographic saturation.

The most basic stations comprise a Min/Max thermometer, a rain gage, and a calibrated staff for snow measurements; others have equipment for pan evaporation.

The file is called the Summary of the Day because it is based on a single reading during each 24-hour period. More sophisticated stations, of course, have evolved over the years, with the ability to measure dozens of parameters and send signals via satellite on a sub-second cycle.

The cooperative observer system still strongly relies on its non-professional staff -- housewives, farmers, foresters, park rangers, some thousands of people with an interest in weather, in record-keeping, in precision, and in routine.

Some stations in Climatedata show a WBAN number. WBAN stations were originally those that reported on an hourly basis, using the Weather Bureau Army Navy forms, and received a WBAN identifier. Now, all current stations with WBAN numbers either are, or are located with, National Weather Service First Order stations which report on an hourly basis. It is not known whether all the non-current WBAN-numbered stations on Climatedata were also located with First Order Stations.

In 1948, data began to be stored on punch cards, with some states' data being retroactively entered on card. In 1969, then and future data storage was converted to magnetic tape in card image format. In 1982, the historic files were transmuted to an element file structure, to which records since that date have been added. The NCDC has passed the data received since 1982 through extensive computer editing; the backfiles are gradually being examined and edited as well. In 1993, a comprehensive check was done to identify the most obviously erroneous values and to assign a more valid value for those cases. See the discussion on the Error Corrections below.

For complete information on this dataset, write the NCDC and ask for their TD-3200 Precipitation Documentation Manual.

VALHIDD Error Corrections

In 1993, the NCDC issued an error-corrections file. This file was created by the NCDC software application called ValHiDD (Validation of Historical Daily Data). It is run against the NCDC TD3200 data (the summary of the day data), and creates a listing of individual observations which are considered “erroneous”. This listing shows the old (wrong) value, the new (corrected) value, and a reason for deciding the old value was wrong. Reasons for being wrong include failing out of range checks, having a minimum value greater than a maximum value, entering value in the wrong decimal format, and others. For a comprehensive discussion of the ValHiDD program and its use, see A Deterministic Approach to the Validation of Historical Daily Temperature and Precipitation Data from the Cooperative Network in the Bulletin of the American Meteorological Society, Vol. 73.

We have taken this list of corrections and applied it to our data. We have also created a text file containing a list of the old values, the new values, the dates for the value, and the type of error. The file has a header section which describes the error codes and how to interpret the file. This file is located on the CD disk in a directory called errfile.

Station History

The National Climatic Data Center maintains an historical record of the stations in the Cooperative Network; this is the basis of the Station History view. Station numbers are assigned by state, based on alphabetization. As stations have changed over the years, the match between and numeric order has shifted slightly. Station numbers themselves are 4 digits, but are coupled with the 2-digit state code. Some station IDs also have 2 extra, final digits -- the division identifiers. A division is a region within a state with climatologically similar behavior. Thus, a station number would usually be simply the center 4 digits shown in the example below.

StateStation NumberDivision ID
06148407
Divisional boundaries shifted in the early 1950s, but have been largely stable since, allowing the accumulation of divisional statistics; these have become the basis for quality control checking by the NCDC. Data are compared to divisional records to determine whether they fit reasonably within the envelope of these historic statistics. A glance at the station history view reveals how frequently stations have changed. The decision as to whether a station move triggers a number change or not is made by the Cooperative Program Managers of the National Weather Service, who decide whether the location of the shifted station produces data compatible with the old; if not, a new name/ID is assigned. Beyond a certain distance (approximately 5 miles) a new number is assigned no matter the climatological congruence of the stations old and new. In some cases a new name will arise for an area; it gets a new number though the site may not have changed at all. Definitive answers to questions about station history require examining the original forms at NCDC, though the computerized history file is undergoing editing and rectification at present. The station history files have passed through many transmutations, and are passing through one even now. Climatedata’s presentation of the information required merging three files. Our general impression is that the files recording the succession of stations have only recently begun to receive the care that the main database has received. We advise circumspection in the use of that view of the data.


Return to Hydrosphere Home Page
Send suggestions to hdpweb@hydrosphere.com
Last Updated March 10, 1998
Copyright © 1997 Hydrosphere Data Products, Inc.