Runoff Volume Over Watershed Area [mm, in, cf]

Runoff Rate [ mm/hr, in/hr, cfs]

During 1953-1954, five large trapezoidal critical depth runoff measuring flumes were constructed with design capacities ranging from 43 to 225 m³ s^-1 (1500 to 8000 ft ³ s^-1 ). A sparse rain gauge network of 30 gauges was established. By the end of the season only one of the five flumes remained intact and operational. The structures failed because they were (1) hydrologically undersized, (2) hydraulically inadequate, and (3) structurally incapable to withstand the loads involved. These failures stimulated a joint project with personnel of the ARS Hydraulic Structures Laboratory in Stillwater, OK [ Gwinn, 1964; Gwinn, 1970; Brakensiek et al., 1979; Smith et al., 1982] to develop, test, and calibrate new runoff measuring flumes for high-velocity, sediment laden flow conditions. The resulting Walnut Gulch Supercritical Flow Flume became an accepted standard for basic flow-measuring structures at the ARS semiarid watersheds [Renard et al., 2008].

Runoff measurements on the small watersheds began in 1962 with the installation of broad crested V notch weirs at 63.101, a shrub dominated site (Lucky Hills) and 63.112, a grass dominated site (Kendall). At Lucky Hills, three additional weirs and two H flumes were installed during the period 1963-1965. Four weirs were installed near the city of Tombstone beginning in 1972. However, the weirs on watersheds with well-defined channels had measurement problems associated with the sediment load in that sediment would deposit in the ponds behind the weirs invalidating the weir rating curve. On the basis of the scale model testing of the large flumes, Smith et al. [1982] designed a metal supercritical flume (Smith flume or Santa Rita flume) to be used on small watersheds. These flumes were installed in the SRER in 1975. Currently, runoff at all of the small WGEW watersheds with the exception of 63.105, 63.106, and 63.112 is measured with Santa Rita flumes [Stone et al., 2008].

Table 1: SWRC flume/weir locations and years of operation. Coordinates are not survey grade.

The measurement of runoff at Walnut Gulch is affected by the ephemeral nature of the runoff, high flow velocities, high sediment concentrations in the flow, and the initial upstream channel geometry created by the previous flow(s) sediment transport/deposition. The high sediment loads present a problem for structures that measure runoff stage at critical depth. Structures such as weirs and zero slope flumes retard the flow velocity to tranquil or subcritical conditions that, for sediment-laden flows, cause sediment deposition in the pond above or within the measurement section and invalidate the stage-discharge relationship of the measurement structure. Because of this, runoff measurement at Walnut Gulch has relied on super- critical flumes that channel the flow through the structure at a velocity high enough to minimize sediment deposition within the measurement section. A full discussion of the hydraulic factors involved and the evolution of flume design at Walnut Gulch is given by Smith et al. [1982].

Analog Gauges (1953-1999) / Digital Gauges (2000-Present)

Runoff was originally measured using a stilling well, float, and analog stage recorders (Stevens A-35, Friez FD-4, Friez FW-1) [see Brakensiek et al. , 1979] with mechanical clocks to record the timing of the event (Note: use of trade names in this report is for information purposes only and does not constitute an endorsement by the USDA-ARS). As of 2000, digital recorders consisting of potentiometers attached to the stilling well gear mechanism and a Campbell Scientific data logger were added to all of the runoff measurement stations.

Original Critical Flow Flume (OCF)

The first five critical flow measuring stations were constructed in WGEW by 1954. These flumes were simply smooth flow constrictions that contracted the flow sufficiently to cause critical flow at a smooth over-fall, but created some backwater. By the end of 1954 all of these structures were severely undermined or overtopped. The failures occurred because the structures were inadequate to carry the weight of the water involved, hydrologically too small, and hydraulically inadequate with resulting downstream scour undermining the concrete. [Smith et al., 1982]

Walnut Gulch Super Critical Flume (WGSF)

WGSF's replaced the OCF's and are currently used today for flow measurement on the large watersheds at WGEW. These flumes were designed to accelerate the sediment laden flow to supercritical velocity through the throat of the structure to minimize deposition of sediment on the flume and in the stilling well. The acceleration of the flow is caused by the geometry of the structure which features a curved entrance approach, a shallow, sloped v-notch floor, and sidewalls with a one-to-one slope. Stage is measured though an intake at the center of the v-notch which is routed to a stilling well and discharge is calculated from rating curves unique to each flume.

Porous Dikes (PD)

After the WGSF's were constructed, it was observed that flows were frequently asymmetrical through the measurement section of the WGSF. The asymmetry was due to the short-approach section of the flume and the variable channel geometry of the alluvial channel directly upstream. On the basis of measurements of flow velocity and design testing using scale models, porous dikes were installed upstream from all of the large flumes with the exception of 63.015 to guide the flow along the centerline of the flume.

V-Notch Weir (VNW)

V-notched weirs measure low discharge more accurately than horizontal weirs. The V-notch is usually a 90 degree opening with the sides of the notch inclined 45 degrees with the vertical. The approach velocity can be neglected if the minimum distance from the weir to the channel banks is at least twice the head and if the minimum distance from the channel bottom to the crest is at least twice the head.

Smith Concrete Flume (SCF)

The Smith Concrete Flume is a modified version of the WGSF designed for use in small channel flow measurement. The flume design incorporates two improvements over the WGSF. The slope of the floor breaks at the entrance to the throat defined by the walls, rather than at the entrance to the curved approach. Also, the curvature of the approach wall was reduced to decrease the tendency of waves occurring from rapid changes in flow direction upon entrance.

Smith Extension Flume (SEF)

The Smith Extension Flume was installed at watershed 63.102 as a modification to the original SCF. The throat of the flume was extended downstream to further constrict flow and prevent inundation of bedload sediment.

Santa Rita/Smith Flume (SRF)

The Santa Rita or Smith Flume was designed in the same manner as the SCF, but constructed from metal and intended to measure smaller flows. These flumes used at WGEW and the SRER have been equipped with automated slot-samplers for measuring suspended sediment during runoff events.

H Flume (HF)

The H flume was designed to measure discharge rates up to approximately 0.85 m³ s^-1. The H flume has a V-shaped throat for low-flow sensitivity and a flat floor approach to allow sediments and debris to pass through relatively easily. H flumes installed at small watersheds in WGEW are equipped with automated pump samplers to measured suspended sediment during flow events.

Replogle Flume (RF)

The Replogle flume or broad-crested weir is a long-throated flume in which flow is contracted by raising the elevation of the flume floor. The contraction results in critical flow from which discharge can be measured using a stage-discharge relationship unique to the flume.

Venturi Flume (VF)

The Venturi flume is characterized by a relatively long section or throat and a horizontal floor. These flumes measure discharge in an open channel by contracting flow by either tapering the sidewalls or changing the elevation of the floor, or both. Assuming critical flow, discharge can be calculated from flow depth measured at the outlet.

Table 2: SWRC flume/weir instrument specifications. Coordinates are not survey grade.

^aOCF , original critical flow flume; WGSF, Walnut Gulch supercritical flume; PD, porous dike installed; VNW, V notch weir; SCF, Smith concrete flume; SEF, Smith extension flume; SRF, Santa Rita or Smith flume; HF, H flume; RF, Replogle flume; and VF, Venturi flume.

^bFor the period of record after the installation of the WGSF on large flumes. For the entire period of record for 63.101, 63.105, 63.106, and 63.112. The period of record for the remainder of the small WGEW watersheds is after the installation of the SRF or in the case of 63.102, after the installation of the SCF. For the period 1976-Present for the SRER flumes.

Data Quality

The quality of the runoff records of WGEW has been primarily impacted by the type of measuring structure, the sediment characteristics affecting the alluvial channels upstream from the measuring structures, the sediment load in the flow, and the mechanism used for recording the event. A qualitative assessment of the runoff records based on the first criteria above was done by the unit scientists in 1989 for the large flumes.

Large Watersheds

The quality of the runoff data collected before the installation of the WGSF were considered poor because of the inadequacy of the original structures, data collected after the installation but before the porous dikes were installed were considered fair to good because of the need to compensate for asymmetrical flows, and data collected after the dikes were installed were considered good to excellent.

Small Watersheds

The period of record before the installation of the SRF is considered poor because of invalid rating tables due to sedimentation while the period after is considered good to excellent. For the small watersheds that do not have SRF, the sediment load does not affect the runoff measurement and entire period of record is considered good to excellent.

Sources of Error

The high sediment loads can cause sediment deposition in the intakes of the stilling wells during the recession of the hydrograph, which effectively slows the rate of water exiting the stilling well. The result is a recession curve that slowly approaches zero. Because of this, most of the recessions on the large watersheds have been estimated manually on the basis of observation of flow recession rates during runoff events.

The analog stage recording system (prior to 2000) consisted of a mechanical clock, which rotated a drum chart via a gear and a pen activated by changes in water level in the stilling well via a float, gear, and cam mechanism. In the early records, the pens could run out of ink, thus missing the event. The pens were eventually replaced with felt tip pens in the 1980s, which minimized that problem. Problems with the mechanical clocks included failure to rewind the clock, mechanical failure, and inconsistent clock speed. In addition, there was no central synchronization of time throughout the flume and rain gauge network. For the digital data (2000 - present), the recording times for the rain gages and runoff measurement stations are synchronized on a daily basis.

Consideration of the condition and topography of the contributing watersheds is important as WGEW is comprised of mixed use land. Watersheds 63.102 and 63.106 were altered by vegetation management in the early 1980s. Watershed boundaries have been affected by urban development and the construction of new roads (e.g., 63.125). Several stock ponds exist at WGEW and storage in these structures should be considered when evaluating flume data. Ambiguity of watershed boundaries exists at SRER and has been improved at flumes 7 and 8 by constructing berms, but uncertainty remains.

Accuracy Specifications

Analog Recorders (1953-1999)

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

Digital Recorders (2000-Present)

A runoff event begins for a large flume once 0.05 ft of flow is detected. A threshold of 0.01 ft of flow is needed to trigger a runoff event at the small flume, and a flow depth of 0.2 ft will initiate automated sediment sampling. Continuous 1-minute data at a resolution 0.01 ft will continue to be collected until flow depth drops below these thresholds. Voltages from the potentiometer are sampled every second and averaged over the course of a minute and converted to stage using a calibration coefficient. The potentiometers used for the small and large flumes are 10 and 15 turn respectively.

The first concrete critical depth runoff measuring flumes were installed at WGEW in 1953 and 1954. Following the structural failure of the OCFs in the early 1950's, design and construction of the existing WGSFs began. Construction of the WGSFs was largely completed in the 1960s and continue to operate today. The installation of SRFs currently used at the small watershed locations began in the 1970s at WGEW and SRER.

The staff of hydrologic technicians located in Tombstone, AZ conduct annual maintenance and repairs prior to the monsoon season and as needed. During these visits the stilling well intakes are cleaned and inspected. The float pulley system is manually calibrated and a zero level is set. Maintenance is often required after runoff events, as sediment can deposit in the stilling well intake. Post-runoff maintenance includes, cleaning the stilling well intake, checking the float, verifying the calibration and zero level, and verifying the high water marks on the flume floor/walls.

Daily maintenance procedures include checking battery voltages and flow depths from daily maintenance reports. Daily runoff depths are evaluated for each flume and compared with other flumes and precipitation data to screen for potential problems. Additionally, flumes 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 Recorders (1953-1999)

The analog recorders produced an ink line on a paper chart to record flow depth versus time. At the large flumes an A-35 chart recorder recorded data continuously onto a chart until the relevent section was collected by a technician, after a runoff event. The recorders were fitted with clock gears and charts such that one revolution of the drum is equivalent to a 24-h period for FW-1 recorders at the small flumes/weirs.

Digital Recorders (2000-Present)

The digital flume network was outfitted with Campbell scientific dataloggers to record voltages from a potentiometer fitted to the existing float and pulley system. The network is currently comprised mainly of CR10X and some CR1000 dataloggers. Data is only recorded if a runoff event is initiated, i.e., 0.05 ft of flow for the large flumes and 0.01 ft of flow for the small flumes. Data continues to be recorded until the stage drops below these thresholds. In addition to recording runoff depth, various diagnostic measures, as well as hourly and daily summaries, are recorded for maintenance purposes.

Dataloggers are programmed using Campbell Scientific 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)

Runoff 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 runoff event. The coded runoff charts were digitized by an analog-to-digital converter coupled with a card punch. The operator entered coded information 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]. Once the charts were digitized, stage data was converted to discharge based on rating curves of stage-discharge relationship calibrated for each flume (Woolhiser and Saxton, 1965).

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 runoff data from the previous day and calculates volume and discharge based on calibrated rating curves. This processed runoff 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)

Quality control of runoff events consisted mainly of removing minor fluctuations occurring prior to the initiation of the hydrograph due to rain on the flume and estimation of the recession of the hydrograph due to sediment blockage of the intakes. The estimation of the receding limb of the hydrograph was done by fitting a French curve. In both cases the event is flagged as "estimated" in the data base.

Digital Gauges (2000-Present)

The runoff 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 runoff across 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. Additionally, an algorithim is used to fit a theoretical recession to the hydrograph if the stilling well intake was clogged with sediment. Modified from Nichols and Anson., 2008

The runoff data is archived daily in an SQLServer database once received at the SWRC Tucson office. The runoff data is then queued for the QA/QC application. Once the 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 runoff data can be accessed at the following websites:

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

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

Use of unit-source watersheds for hydrologic investigations in the semiarid Southwest

Prediction-runoff relation for very small semiarid rangeland watersheds

Transmission losses in ephemeral stream beds

Thunderstorm runoff in southeastern Arizona

Spatial characteristics of thunderstorm rainfall fields and their relation to runoff

A distributed model for small semiarid watersheds

Effects of watershed representation on runoff and sediment-yield modeling

Differentiating the impact of parameter lumping from the impact of geometric simplification in modeling runoff and erosion

Linearity of basin response as a function of scale in a semiarid watershed