Precipitation Depth [mm, in]

Precipitation Intensity [ mm/hr, in/hr]

Through 1999, the network consisted of analog recording weighing rain gauges. In 2000, the digital gauges were placed adjacent (~1 m separation distance) to the analog gauges. Both the analog and digital network of gauges were in operation from 2000 to 2004 to enable a comparative analysis of the two systems. In regards to precipitation observations, they concluded that 1) two individual digital rain gauges recorded precipitation equivalently; 2) high errors in event intensities may be produced when analog charts are digitized at short time intervals; 3) for several different measures of precipitation, the analog and digital data were equivalent. [ Keefer et al. , 2008]

For the analog network record, different numbers of rain gauges were in operation during different periods of time. The most notable cases were from January 1980 to June 1991 and from October 1998 to October 2004 when the analog operational network was scaled back to nine gauges (4, 13, 42, 44, 46, 60, 68, 80, and 81) during the non-monsoon months because of financial considerations. Analysis by Osborn et al. [1979b] indicated that this number of gauges was more than adequate to characterize the variability of the winter frontal rainfall. The exact turn-on and turn-off dates for each of the gauges is part of the database and they are returned with any data query via the web database interface. Modified from Goodrich et al., 2008

Table 1: SWRC rain gauge locations and years of operation. Coordinates are not survey grade and are based on digital gauge locations

^AAnalog rain gauges operational year round. See DAP database for exact dates of operation.

^MAnalog rain gauges only operational during the monsoon season. See DAP database for exact dates of operation.

^DAnalog rain gauges using a daily (24-hour) clock/chart.

^SAnalog rain gauges using a 6-hour clock/chart.

^WAnalog rain gauges using a weekly clock/chart.

Analog Gauges (1953-1999)

The analog network consisted of Belfort 0.2032 m (8 inch) unshielded weighing-recording gauges (use of trade names in this report is for information purposes only and does not constitute an endorsement by USDA-ARS). The gauges were installed so that the top edge of the gauge orifice was nominally 0.91 m (36 inches) above the ground surface. The gauges record accumulated rainfall versus time as a pen trace on a paper chart attached to the side of a rotating, clock-driven drum.

Digital Gauges (2000-Present)

The digital instrumentation, developed in house, and conceptually similar to the design described by Hanson et al. [2001], was placed in the same gauge housing as the analog system to minimize the differences between analog and digital gauges caused by wind effects due to the gauge profile. The digital gauges employ an electronic load cell in which the weight of precipitation in a collection bucket is converted into a voltage. A linear relationship between voltage and observed precipitation depth was defined for each gauge using a set of calibration weights in the laboratory when the gauge was fabricated. In addition to the basic measurement device, considerable effort was directed to integrating data logging and telemetry electronics, in an attempt to maximize operational reliability and enhance protection from the environment and vandalism. Modified from Goodrich et al., 2008

Table 2: SWRC rain gauge instrument specifications for analog and digital eras.

Analog Gauges (1953-1999)

For data reduction and QA/QC procedures the following definition of an event was adopted for the network of gauges. An event begins when any one of the gauges in the network detects measurable rainfall. As long as there is not a hiatus of more then 60 min between breakpoints on any of the gauges in the network the event continues until the last breakpoint on any gauge prior to a 60 min hiatus of no measurable rainfall. If rainfall is measured in any of the gauges after a 60 min hiatus over the entire network, it is considered the start of a new event.

The digitizing resolution for analog gauge charts has always been 0.01 in for depth and whole minutes for time, with break points identified by visual inspection. Thus breakpoints consist of time, accumulated rain depth pairs, with nonuniform time intervals and depths that are multiples of 0.01 in. Modified from Goodrich et al., 2008

Digital Gauges (2000-Present)

The event definition for digital gauges differs from that of the analog gauges. An event begins for an individual gauge once 0.01 in of precipitation is detected in the course of an hour for that particular gauge. The event will continue until a 60 minute hiatus in precipitation is detected. The event definition for each gauge is independent of the other gauges in the network.

Voltages from a load cell are sampled every second and averaged over the course of a minute. A threshold of 0.01 in of precipitation accumulated in within 60 minutes is required for data to be recorded as a breakpoint on the datalogger.

Table 3: SWRC rain gauge error specifications and data recording resolutions for analog and digital eras.

*Error specifications are provided by the manufacturers of the weighing mechanisms of the rain gauges. These values do not include other sources of error from digitization, wind, etc.

The first SWRC network gauges were installed on WGEW in the early 1950s. These gauges provided sparse coverage of WGEW in a pseudo-grid format. Gauge installation followed recommended protocols as defined in Brackensiek et al. (1979). Gauges were installed on level concrete slabs in regions unobscured by tall vegetation or man-made structures. As the network expanded more complete and dense spatial coverage of WGEW was achieved.

Analog Gauges (1953-1999)

Prior to 1968 there was not a regular schedule of in-field rain gauge calibration. Occasional field checks were made by measuring the amount of rainfall accumulated in the collection bucket with a standard volumetric tube measurement. If differences existed between the measurements, a correction factor was developed from the tube measurement and applied. As of 1968, each gauge was checked and adjusted annually, usually prior to the summer monsoon, with a set of standard weights through the full range of a pen sweep on the analog gauges "with the following sequence of calibrated weights - 0.05, 0.10, 0.15, 0.20, 0.25, 0.30, 0.40, 0.50, 0.75, 1.00, 1.25, 1.50, 2.00, 2.50, 3.00, 3.50, etc., to the maximum" [ Chery and Osborn , 1971, p. 5].

Chart on and off times were set and compared respectively, to the wristwatches of field technicians whose watches were set to a broadcast standard time. If clocks stopped or gained/lost more than 15 min over the course of a 7-d period (weekly gauge visits), the clocks were replaced with spares and the malfunctioning clocks were serviced in Tombstone. In these cases the event start times would be adjusted to account for the fast/slow clock using a linear correction. In the case where a clock stopped, start times from the nearest operating gauge were used. However, because of the daily time resolution of the analog charts and the use of a wrist watch, "time at one gauge at any instance will be, at best, within ±5 minutes with any other gauge in the network" [Chery and Kagan, 1975, p. 49]. Modified from Goodrich et al., 2008

Digital Gauges (2000-Present)

The voltage to depth relationship of each digital gauge is also verified annually using the calibration weights in the field. At this time the gauge is cleaned and inspected by a hydrological technician. Each data logger clock time is checked daily via telemetry and periodically reset to National Institute of Standards and Technology (NIST) standard time. The time kept by the base station computer is manually set to NIST standard time once per month and then all data logger clocks are updated via telemetry to match the base. Though the base station computer may deviate from NIST time by about ±2 min per month, the network of 88 data logger clocks stay within less than one minute of each other and standard time [Keefer et al., 2008]. Modified from Goodrich et al., 2008

Daily maintenance procedures include checking battery voltages and water depths in buckets from daily maintenance reports. Daily precipitation depths are evaluated for each gauge and compared with surrounding gauges and meteorological data to screen for potential problems. Collection buckets are emptied when water depth reached a threshold of 3-4 inches. During these maintenance visits buckets are inspected for leaks and gauges are inspected for damage and repaired as needed. Additionally, gauges are checked if data communication via telemetry is not received at the Tombstone field office. All maintenance procedures, problems, and repairs are logged by technicians for future reference.

Analog Gauges (1953-1999)

The analog gauges produced an ink line on a paper chart to record accumulated rainfall versus time. The vast majority of analog gauges were fitted with clock gears and charts such that one revolution of the drum is equivalent to a 24-h period. A smaller number of gauges were set up with weekly clocks and charts, to aid in interpreting the storm event timing of the daily gauges. Several gauges were also equipped with 6-h clocks and charts to provide finer time resolution for several nested WGEW source area catchments. Modified from Goodrich et al., 2008

Digital Gauges (2000-Present)

The digital rain gauge network was outfitted with Campbell scientific dataloggers to record voltages from the load cell. The network is currently comprised mainly of CR10X and some CR1000 dataloggers. Data is only recorded if a precipitation event is initiated, i.e., 0.01 in of rain is detected. Data continues to be recorded until a 60 min hiatus in rainfall is reached. In addition to recording accumulated precipitation, various diagnostic measures, as well as hourly and daily summaries, are recorded for maintenance purposes.

Dataloggers are programmed using Campbell Scientific's LoggerNet. CR10X dataloggers are programmed using EdLog and CR1000 dataloggers use CRBasic. While the programming languages differ in syntax, the logic and recording functions where shown to be interchangeable through thorough testing of the programs.

Analog Gauges (1953-1999)

Accumulated precipitation charts were collected from the gauges weekly by technicians. The charts were logged in at the Tombstone field office shortly after they are retrieved from the gauges and notes were compiled from the charts to aid in data processing. The charts are then sent to the SWRC in Tucson and inspected for continuity and completeness and queued for coding. Modified from Goodrich et al., 2008

Digital Gauges (2000-Present)

Shortly after midnight on a daily basis, data from the WGEW sites are downloaded automatically via radio and are transmitted to a computer at the SWRC Tombstone field office. VHF, spread spectrum, and cellular modems are used depending on the geographic location of the gauge. Raw data are archived in the Tombstone office and a series of batch processes are executed using LoggerNet to parse out relevant data to be processed in Tucson and to generate daily maintenance reports. The daily data are then transfered to a SWRC server residing in Tucson. Modified from Nichols and Anson., 2008

Analog Gauges (1953-1999)

Charts were coded by a technician in Tucson. In coding the charts, a technician ascertains the date, beginning time and classification codes of each precipitation event. Precipitation events are classified as significant or nonsignificant. A significant event causes runoff through any of the flow-measuring structures or has total rainfall of 0.25 inch with intensities greater than 0.50 inch per hour at any gage. This determination requires that all the records of a single event be examined as a group and the runoff records be consulted. The coded precipitation charts were digitized by an analog-to- digital converter coupled with a card punch. The operator entered coded information (date, begin time of event, type of precipitation, significance of event) and then separated the pen traces into appropriate line segments that accurately describe the event [Chery and Kagan , 1975]. Estimation of the digitization error can be found in work by Chery and Beaver [1976], Freimund [1992], and Keefer et al. [2008]. It should be noted that analog to digital conversion has evolved as technology has advanced (from manual reading, done prior to 1960, to an electromechanical analog to digital converter coupled with a card punch until the mid-1980s, through several solid state electronic digitizing tablets [ Osborn, 1963; Chery and Kagan, 1975; Keefer et al. , 2008]. Modified from Goodrich et al., 2008

Digital Gauges (2000-Present)

A scheduled Visual Basic script is executed every morning at the Tucson SWRC office after the raw data has been received. This program gathers the raw rain gauge data from the previous day and calculates precipitation depths, in breakpoint format, based on calibration records. These calibration coefficients allow voltages to be converted to precipitation depths. This processed precipitation data is then archived into an SQL database where it is queued for QA/QC procedure and made available for visual inspection on an internal maintenance website.

Analog Gauges (1953-1999)

Once chart digitizing is completed, event rainfall totals are plotted by gauge location and isohyetal maps are visually examined for missing data or spurious totals. On the basis of this examination, charts may be reexamined, recoded, and/or redigitized. For spurious or missing data, nearby gauges may be used to provide estimates, which are tagged as such in the database. Chery and Kagan [1975] reported on the amount of estimated analog data for the 6-year period from 1967 to 1972. On an annual basis, the percentage of rainfall duration that was estimated ranged from 3.4 to 12.6 percent. For the same period, the percentage of total rainfall event depth that was estimated ranged from 3.1 to 5.6 percent. It was also found that because of rain gauge accuracy and the processing resolution of the analog gauges, many small events (typically less than 1.27 mm or 0.05 inches) are not measured. Modified from Goodrich et al., 2008

Digital Gauges (2000-Present)

The precipitation events are then quality checked by visual inspection. A Windows-based visualization tool was developed using Borland Delphi. This program queries the database for "unchecked" events and displays them graphically for the user. In addition to displaying a graph of the time series of an event for a particular instrument, it also displays a color coded map which represents daily summary precipitation over the watershed. If the time series graph looks typical and the magnitude and duration of the event are judged to be within the range of expected values on the basis of the daily summaries across the watershed, the event is marked as "verified." Otherwise the event is marked as "not good." The program also provides methods for correcting common problems. Modified from Nichols and Anson., 2008

The raw precipitation data is archived daily in an SQLServer database once received at the SWRC Tucson office. This provisional data for rain gauges co-located with meteorological stations is pushed to an SWRC FTP site, from which the data is ingested and served by the Long Term Agroecosystem Research Network (LTAR) daily. The precipitation data is then queued for the QA/QC application. Once the precipitation data has been QA/QC'd (yearly) it is archived locally and served from the SWRC FTP site, accessible from the SWRC Online Data Access Project Site (DAP) in csv format.

SWRC precipitation data can be accessed at the following websites:

SWRC Online Data Access (DAP) - QA/QC'd data updated approximately every 90 days

USDA Long-Term Agroecosystem Research Network - 15-min data updated daily (provisional data at select gauges)

USDA National Agricultural Library Ag Data Commons - DAP mirror

Data Use Agreement

All data available through the SWRC data access website are in the public domain, and are not restricted by copyright.

The SWRC will review the research results to ensure sound scientific data interpretation in the context of our historical results and our in situ experience with these data. We expect that our support will be acknowledged through co-authorship and formal acknowledgment of field and/or data support in the manuscripts (see example below).

Datasets were provided by the USDA-ARS Southwest Watershed Research Center. Funding for these datasets was provided by the United States Department of Agriculture, Agricultural Research Service.

Please send 1 copy of the published manuscript to:

Southwest Watershed Research Center

2000 E. Allen Rd.

Tucson, AZ 85719

Rainfall point intensities in an air mass thunderstorm environment: Walnut Gulch, Arizona

Precipitation changes from 1956-1996 on the Walnut Gulch Experimental Watershed

Event to multidecadal persistence in rainfall and runoff in southeast Arizona

Analysis of spatial and temporal precipitation data over a densely gaged experimental watershed

Spatial characteristics of thunderstorm rainfall fields and their relation to runoff