15-Minute Precipitation (TD3260)

General Information

The data are taken at principal (primary) and secondary stations, and those cooperative observer stations operated by the National Weather Service (NWS) and the Federal Aviation Administration (FAA). The files cover the United States, Pacific protectorates, Puerto Rico, and Virgin Islands. Initially a number of gage types were used; these were gradually superseded by the Fischer Porter gage, introduced in 1963. At present some 2000 of these are used, with the remaining gages being the Universal Rain Gage. Both types have automated readout, the Fischer Porter on tape and the Universal on chart.

Up to 2700 15-minute precipitation stations have taken data since the start of the period of record. Approximately 2300 were active in 1988. For the 15- minute precipitation database, the period of record is 1971 to present.

For complete information on this dataset, write the NCDC and ask for their TD-3260 Precipitation Documentation Manual.

Hourly Stations in the 15-Minute Database

Over the period from January 1984 through October 1993, the NCDC included some hourly data in their TD-3260 15-minute data set. In comparisons with nearby stations with true 15 minute data, the hourly data indicates more intense runoff since the whole hour of precipitation is assumed to occur during the first 15 minute timestep as opposed to being distributed amongst the four 15 minute timesteps for the hour. Anybody using these data should be wary of this potential inaccuracy.

We have been aware of this problem for a little over a year and have contemplated ways of fixing it. The station attributes are no different between the 15-minute stations and the "pseudo" stations so there is not an easy way to distinguish one from another other than by analyzing the timing of the events.

Starting with our 1997 discs (Volume 5.0), we have modified how we process the 15-Minute dataset. We postprocess the data, searching for stations that fit an ‘hourly’ profile. These stations have their data record statistic set to 60Min instead of 15Min. We make no other changes to the data. Thus, the Hourly stations will still show up in the data, but you can filter them out if you so desire.

We decide that a station is an hourly one by looking at all of the events reported by that station. If less than 1% of the events fall on non-hourly times, we call it an hourly station. Events that are flagged or have a 0 value (starting of the month usually) are skipped. Interestingly enough, this testing has turned up four types of stations: true 15-Minute stations, true Hourly stations, stations that started as 15-Minute and changed to Hourly, and stations that started as Hourly and changed to 15-Minute. For a list of the stations that seem to have changed, or that seem to be Hourly stations, see the text file ANOMALY.TXT in the \REPORTS directory on the disc.

To speed up the identification of the Hourly and the 15-Minute stations, we have created two files that can be use to mark a set of stations: HPD.DBF and QPD.DBF. They are in the \DBASE directory. The HPD.DBF file contains all the Hourly stations, while the QPD.DBF file contains the 15-Minute stations. You can load either file by using the File-Open Marked Stations command.


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