Walnut Gulch Experimental Watershed

The Walnut Gulch Experimental Watershed (WGEW) encompasses the 150 square kilometers in southeastern Arizona, U.S.A. that surrounds the historical western town of Tombstone (31o 42'N, 110o 03'W). The watershed is contained within the upper San Pedro River Basin which encompasses 7600 square kilometers in Sonora, Mexico and Arizona. The watershed is representative of approximately 60 million hectares of brush and grass covered rangeland found throughout the semi-arid southwest and is a transition zone between the Chihuahuan and Sonoran Deserts. Elevation of the watershed ranges from 1250 m to 1585 mMSL. Cattle grazing is the primary land use with mining, limited urbanization, and recreation making up the remaining uses.Walnut Gulch, being dry about 99% of the time, is an ephemeral tributary of the San Pedro River.


The Walnut Gulch Experimental Watershed was selected as a research facility by the United States Department of Agriculture (USDA) in the mid-1950’s.  Prior appropriation water laws resulted in conflicts between upstream land owner conservation programs and downstream water users.  Technology to quantify the influence of upland conservation on downstream water supply was not available.  Thus, scientists and engineers in USDA selected the Walnut Gulch watershed for a demonstration/research area which could be used to monitor and develop technology to address the problem.  In 1959, facilities needed for soil and water research in the USDA were identified in a United States Senate document.  This report created the national program of USDA-ARS research on soil and water processes. The Southwest Watershed Research Center in Tucson, Arizona was created in 1961 to administer and conduct research on the Walnut Gulch Experimental Watershed.  Subsequent legislation (Clean Water Legislation of the 1970's) added water quality thrusts to the research program.  

Research on the Walnut Gulch Experimental Watershed is currently conducted within the definitions of the Agricultural Research Service’s National Programs.  All SWRC research conducted at the watershed is within the Water Resource Management Program of the Natural Resources and Sustainable Agricultural Systems Category.  

Walnut Gulch Experimental Watershed is also a partner in the ARS Experimental Watersheds and Watershed Program.  Fourteen ARS research centers are operating over 100 long-term research watersheds.  These exceptional outdoor laboratories make major contributions to national scale projects including GEWEX, AMERIFLUX, ARS Rangeland Carbon Flux, USDA-NRCS Soil Climate Analysis Network (SCAN), and the Surface Radiation Network (SURFRAD).  

The long-term data bases and substantial infrastructure of the Walnut Gulch Experimental Watershed have attracted collaborative efforts with other federal and state agencies, universities, and foreign researchers. Collaborative research efforts have included ARS Hydrology Laboratory, ARS Water Conservation Laboratory, ARS Jornada Experimental Range, USDA Natural Conservation Research Service, US Geologic Survey, NASA, Arizona Department of Water Resources, Cochise County, University of Arizona, Arizona State University, and researchers from Mexico, Australia, Europe, Africa, and Asia.  

Walnut Gulch Experimental Watershed is one of two ARS experimental watersheds on western rangelands and the only one on southwest semi-arid rangelands.  Walnut Gulch Experimental Watershed has developed a reputation as the leading semi-arid research watershed in the world. The land comprising Walnut Gulch Experimental Watershed is under the ownership and control of Federal agencies, State of Arizona, private land owners or leaseholders.  The research activities and access to the field sites are arranged in cooperation with the appropriate federal and state agencies and the private landowners or leaseholders.    

SWRC has designed and developed instrumentation to specifically measure and monitor the hydrology of these semi-arid rangelands and has used this instrumentation to develop extensive, world renowned databases of the hydrology of semi-arid rangelands.  From these sources SWRC scientists have produced over 1500 manuscripts and several computer simulation models so that the hydrologic knowledge gained can be transferred to a variety of users.  Some of the hydrologic models developed, in whole or in part, include CREAMS, RUSLE, WEPP, KINEROS, EPIC, SPUR and CLIMATE.  A complete list of publications, models and databases is available from the SWRC.


The original rainfall and runoff instrumentation on Walnut Gulch Experimental Watershed was installed in 1954-55.  The initial network of 20 precipitation recording gauges was expanded in the early 1960's to the 88 gauge network currently in place on the watershed.  Five supercritical precalibrated flumes were constructed prior to 1955 to measure runoff from the heavily sediment laden ephemeral streams.  All five flumes failed or were badly damaged within two years.  They failed for hydrologic, hydraulic, and structural reasons.  Following extensive hydraulic model research at the Agricultural Research Service (ARS) Outdoor Hydraulic Structures Laboratory in Stillwater, Oklahoma, the original five flumes were rebuilt using a design known as the Walnut Gulch Supercritical flume.  Six additional flumes were added later.  

Hydro-meteorologic and soil erosion/sedimentation data are collected from 125 instrumented installations on WGEW.  Precipitation is measured with a network of 88 weighing-type recording raingauges arranged in a grid throughout the watershed.  Various runoff measuring structures are used to monitor small watershed (< 40 ha) runoff. These structures include a broad-crested V-notch weir, H-flumes, and Santa Rita supercritical flow flumes.  Currently there are 8 small watersheds being monitored.  Runoff from watersheds greater than 40 ha is measured using either livestock watering ponds or large supercritical flow flumes.  The largest flume, at the outlet of the Walnut Gulch Experimental Watershed has a flow capacity of 650 cubic meters/sec.  There are 10 stock pond watersheds and 11 large flume watersheds currently being monitored.  Sediment from the small watersheds monitored with the V-notch weir or H-flumes is sampled with automatic pump samplers.  Sediment from watersheds equipped with the Santa Rita supercritical flow flumes is sampled with a total-load automatic traversing slot sampler.  Meteorological, soil moisture and temperature and energy flux measurements are made at two vegetation/soil complexes. Permanent vegetation plots and transects have been established to evaluate the impacts of management practices and global change on vegetation.  

Because of the growing obsolescence of existing rainfall and runoff mechanical sensors with analog data-recording, impending reduction in staff and the labor intensive requirements to collect and process the charts, SWRC began a multi-year effort in 1996 to fully reinstrument Walnut Gulch Experimental Watershed with electronic sensors and digital data-logging capability combined with radio telemetry to allow remote data transmission and monitoring.  This reinstrumentation greatly enhances our research and cooperative capabilities as well as maintaining the viability of hydrologic data collection and long term continuous record.  

A high resolution, self contained, simple raingauge was designed by SWRC field technicians that has been laboratory and field tested under simulated and natural rainfall.  The gauge consists of a precision, temperature compensated load cell, which measures the weight of a platform-mounted container that collects water during a precipitation event. As water accumulates in the container, the voltage output from the load cell changes.  The programmed datalogger samples voltage every second and averages at 1 minute interval.   To minimize data storage requirements and transmission time, only time stamps and voltages commensurate with precipitation detectable to 0.25mm precision are recorded.  The capacity of the raingauge is 200mm (8 in) before it must be serviced.  A very unique feature of the raingauge design is that all electronics, data logger, and radio/modem components are housed in a metal below-ground cylinder, thus reducing vandalism, lightning interference, and temperature effects.  

The conversion from analog to digital output of the runoff measuring instruments was done by attaching a precision linear potentiometer to the output gear shaft of the currently used water-level recorders.  The voltage output from the potentiometer is collected by a data logger which averages 1-second samples at 1-minute intervals and records flow data (time stamp and voltage) only when a minimum depth threshold has been exceeded (0.003m at small flumes, 0.015m at large flumes and stock tanks).  At sites where automatic sediment sampling is done, the data logger controls the operation of the sampler and records each sample’s begin time and total time to collect the sample.  Samples are collected when flow depth is greater than 0.06m.  

On a daily basis, all locations are automatically and sequentially queried and data are transmitted to a dedicated computer at the Tombstone field office.  Data are archived, used to generate daily reports and written to the Tucson SWRC network server.  Daily data radio transmission time and size can range from a minimum of 1.5 hour and 300 KB for non-event days to over 4 hours and 1MB for a day with rainfall/runoff. 


The Walnut Gulch Experimental Watershed is located primarily in a high foothill alluvial fan portion of the larger San Pedro River watershed. The surface geology is dominated by fan deposits, but in southern and southeastern parts of WGEW a complex history of tectonism has resulted in igneous-intrusive and volcanic rocks, and highly disturbed Paleozoic and Mesozoic rocks in the Tombstone Hills. Cenozoic alluvium is very deep and is composed of coarse-grained fragmentary material, the origin of which is readily traceable to present-day mountain flanks on the watershed. The alluvium consists of clastic materials ranging from clays and silts to well-cemented boulder conglomerates with little continuity of bedding. This alluvial fill material is more than 400 m deep in places and serves as a huge ground water reservoir. Depth to ground water varies greatly in the watershed ranging from 50 m at the lower end to 145 m in the central parts of the watershed.

Topographic expression of the alluvium is that of low undulating hills dissected by present stream channels whose routes are controlled by geologic structures. Upland slopes can be as great as 65% while slopes in the lower lying areas can be as small as 2 to 3%. Major channel slopes average about 1% with smaller tributary channels averaging 2 to 3%. The remaining mountainous portion of the watershed consists of rock types ranging in age from pre-Cambrian to Quaternary, with rather complete geologic sections. Rock types range from ridge-forming limestone to weathered granite intrusions. The geologic structural picture of the mountainous area is complex, with much folding and faulting. This folding and faulting, along with igneous intrusions has resulted in large areas of shattered rock, which influence the watershed hydrology.

The watershed hydrology is, in places, controlled by past geologic events and structures. Intrusive igneous dikes in the Tombstone Hills influence ground water movement and change the surface drainage. The Schieffelin granodiorite alters the course of the Walnut Gulch main stream, acts as a probable ground water barrier between the ground water in the Tombstone Hills and the deep alluvial basin, and has caused numerous small perched water tables along its perimeter. Highly compacted conglomerate beds greatly alter the path of stream channels and, in places, divert streams at more than right angles. High angle faults form new paths for streamflow, making channels arrow-straight in some places and causing diversions in others.



Soils of the Walnut Gulch Experimental Watershed are dominantly sandy, gravely loams that vary from deep, relatively mature, and well drained soils to thin, immature soils. All soils are strongly reflective of a semiarid climate and the parent material upon which they have formed, but vary in texture and composition with landform and the length of time that the surface has been exposed to biochemical weathering. The soil profile can contain up to 60% gravel in the uppermost 10 cm and less than 40% in the underlying horizons. Soil surface rock fragment cover (erosion pavement) can range from nearly 0% on shallow slopes to over 70% on the very steep slopes. 

Geology exerts a major control on soil distribution, maturity, thickness, and permeability. Most soils of the watershed are unconsolidated, but locally near-surface soil horizons may be moderately to well consolidated owing to the deposition of calcrete. Where accelerated erosion of the last century has not stripped the upper horizons, soils tend to be thick, mature loams rich in sand and gravel and of high carbonate (calcrete) content. The warm, semiarid climate results in relatively slow biochemical reduction of bedrock. Soils of Holocene age are typically coarse, permeable, and poorly developed. Surfaces that were first exposed to weathering processes prior to Holocene time are deeper, more mature, and generally more argillaceous than the younger soils. Time has been inconsequential relative to soilforming processes in areas of bare rock. In contrast, where surfaces of fan terraces remain and have been exposed to weathering processes throughout the late-Cenozoic and Quaternary periods, time has been sufficient to yield deep, argillic soils, even where climatic conditions have been generally arid to semiarid. Nowhere in the watershed has time been adequate to yield clayey soils, rich in iron and aluminum oxides and hydroxides, that are indicative of long-term warm, moist conditions. 

Over two dozen soil series have been identified on the watershed. The major soil series presently defined on this area are Blacktail (fine, mixed, thermic, Aridic Argistolls), McAllister (fine-loamy, mixed, thermic, Ustollic Haplargids), Elgin (fine, mixed, thermic, Ustollic Paleargids), Sutherland (loamy-skeletal, carbonatic, thermic, shallow Ustollic Paleorthids), Monterosa (loamy-skeletal, mixed, thermic, shallow Ustollic Paleorthids), Stronghold (coarse-loamy, mixed, thermic, Ustollic Calciorthids), Luckyhills(coarse-loamy, mixed, thermic, Ustochreptic Calciorthids). 

The soil series can be combined into soil groups with similar geologic parent material and/or geomorphic surfaces. The Baboquivari-Combate-Bodecker Group consists of permeable, immature soils formed on late-Holocene channel, flood-plain, and alluvial-terrace deposits in all parts of the watershed. Mature, poorly transmissive soils of the Forrest-Bonita Group are derived principally from early- to mid-Holocene cienega and inset deposits of alluvium. Deep sandy gravel loams of the Blacktail- Elgin-Stronghold-McAllister-Bernardino Group occur on beds of conglomerate. The soils of the Luckyhills-McNeal Group tend to be sandy and gravelly loams that are immature compared with soils where rilling and gully erosion have been less extensive. The Sutherland-Mule-Tombstone soils are very gravelly, mature loams that typically contain well developed pedogenic calcrete. Volcanic-terrain soils of the Epitaph- Graham-Grizzle Group are mostly thin, clay-rich loams containing abundant gravel and cobble clasts of basalt or andesite and tuff. Most soils of igneous and carbonate rocks in the Tombstone Hills, the Mabray-Chiricahua-Rock-Schieffelin-Lampshire- Monterosa Group, are very immature, shallow gravel and cobble loams. In headwater areas of the watershed are shallow clay-, sand-, and gravel-loam soils of the Budlamp- Woodcutter Group.


Major watershed vegetation includes the grass species of black grama (Bouteloua eriopoda), blue grama (B. gracilis), sideoats grama (B. curtipendula), bush muhly (Muhlenbergia porteri), and Lehmann lovegrass (Eragrostis lehmanniana); and shrub species of creosote bush (Larrea tridentata), white-thorn (Acacia constricta), tarbush (Flourensia cernua), snakeweed (Gutierrezia sarothrae), and burroweed (Aplopappus tenuisectus). These represent the two main vegetation structural types, “shrubdominated” and “grass-dominated” on WGEW. Shrub-dominated indicates 20% or more of site vegetation cover contributed by whitethorn acacia, creosote bush and tarbush, which together constitute the bulk of total vegetation cover at such sites. Grass-dominated refers to open sites with widespread, appreciable grass cover and indicates less than 15% of site vegetation cover contributed by shrubs, primarily species other than white-thorn, creosote and tarbush such as mormon tea (Ephedra trifurca) and soaptree yucca (Yucca elata). 

Native and exotic grasses are found at all elevations throughout WGEW. Important native forage species include black grama, sideoats grama, slim tridens and bush muhly. The most widespread invasive species is Lehmann lovegrass appearing between 1967 and 1994 and establishing in disturbed areas such as road right-of-ways. Woody vegetation at WGEW consists of shrubs, subshrubs and trees. Prominent shrub species included whitethorn acacia, creosote bush and tarbush. Common subshrubs include desert zinnia (Zinnia acerosa), mariola (Parthenium incanum) and fairy-duster (Calliandra eriophylla). Trees are found along limited riparian zones and in open woodland at higher elevations including mesquite, oak and juniper species. Other vegetation life forms at WGEW include annual and perennial forbs, cacti, and other xerophytes including yucca and agave species. These typically form a small portion of total vegetative cover, although forbs can be transiently abundant following sufficient precipitation. 

Although historical records indicate that most of the Walnut Gulch Experimental Watershed was grassland approximately 100 years ago, shrubs now dominate the lower two-thirds of the watershed. However, since 1967 there is no evidence of widespread shift from grass dominated to shrub dominated conditions at WGEW. In fact, vegetation has changed very little overall in total cover and composition. Any ongoing vegetation change appears to be incremental rather than wholesale. 

Total absolute vegetation cover is spatially variable and typically low in this semiarid area. Shrub canopy ranges between 30 to 40% on shrub-dominated areas and grass canopy cover ranges from 10 to 80% on grass-dominated sites. Vegetation spatial distribution is closely linked to soil type and variations in annual and August precipitation. Average annual herbaceous forage production is approximately 1200 kg/ha.

Water Balance

The Walnut Gulch Experimental Watershed water balance, although variable from year to year as well as across the area, is obviously controlled by precipitation. The annual water balance is illustrated for average conditions. Given the average 350 mm input precipitation, approximately 327 mm is detained on the surface for subsequent infiltration. Essentially all of the infiltrated moisture is either evaporated or transpired by vegetation back to the atmosphere. Based on data collected from small watersheds, less than 1.5 hectare, approximately 23 mm of the incoming precipitation is in excess of that which is intercepted and/or infiltrates. We refer to this as "onsite runoff". As the runoff moves over the land surface and into dry alluvial channels, transmission losses begin. Approximately 20 mm of transmission losses occur and about 2 mm of surface runoff is measured at the watershed outlet. The 20 mm of transmission losses result in some ground water recharge and some evaporation and transpiration from vegetation along the stream channels. Quantities for ground water recharge and evaporation and transpiration of channel losses are not shown because their quantification is difficult and very site specific. This is an area of active research. The geology along and beneath the stream channels create some reaches that are underlain by impervious material, whereas in other locations, the channels extend to regional ground water and permit appreciable recharge. Where the channels are underlain by impermeable material, riparian aquifers connected to the channels support phreatophytes. Runoff from the entire watershed is about 2 mm.


Walnut Gulch Experimental Watershed lies in the transition zone between the Sonoran and the Chihuahuan Deserts. The climate is classified as semi-arid, with mean annual temperature at Tombstone of 17.7°C and mean annual precipitation of 350 mm. On average there are 53 days of precipitation per year and most accumulation is as rainfall. The precipitation regime is dominated by the North American Monsoon with slightly more than 60% of the annual total coming during July, August and September; about 1/3 coming during the six months October through March. Summer events are localized short-duration, high-intensity convective thunderstorms driven by the intense solar heating of the land surface and moisture inputs from the Gulf of Mexico and Gulf of California. Winter storms are generally slower moving, frontal systems from the Pacific Ocean. These frontal systems generate longer duration and lower intensity precipitation that covers larger areas. The two opposite phases of the oceanatmosphere phenomenon El Nino-Southern Oscillation (ENSO), referred to as El Nino and La Nina, affect winter precipitation with greater than normal precipitation during El Nino periods and less than normal precipitation during La Nina episodes. Virtually all runoff is generated by summer thunderstorm precipitation and runoff volumes and peak flow rates vary greatly with area and on an annual basis. Potential evaporation (Class A USWB pan) is approximately 260 cm per year which is nearly 7.5 times the annual precipitation.


Precipitation varies considerably from season to season and from year to year on the Walnut Gulch Experimental Watershed. Annual precipitation varied from 170 mm in 1956 to 541 mm in 1983; summer rainfall (July, August and September) varied from 93 mm in 1960 to 325 mm in 1999; and winter precipitation (January, February and March) varied from 0 mm in 1972 to 175 mm in 1993. Nearly two-thirds of the annual precipitation on the Walnut Gulch Experimental Watershed occurs during the North American Monsoon as high intensity, convective thunderstorms of limited areal extent. The moisture source for these thunderstorms is primarily the Gulf of Mexico and the Gulf of California. 

Winter rains (and occasional snow) are generally low-intensity events associated with slow-moving cold fronts, and are generally of greater areal extent than summer rains. Convective storms can occur during the winter as well. Runoff on the Walnut Gulch Experimental Watershed results almost exclusively from convective storms during the summer season. 

Summing individual storm events to generate monthly and seasonal values for precipitation illustrates some water supply and forage management problems. The ensemble of individual storm events such as that shown below for August 27, 1982 resulted in the following August isohyetal map. The ratio of maximum point precipitation of 100 mm to the minimum of 40 mm (a ratio of more than 2:1) has been measured with considerable regularity. But more importantly, although these extremes were only 4 km apart, they occurred in the same pasture of one ranch. The maximum rainfall value produced good forage whereas the minimum rainfall produced less than normal forage. 

The precipitation variability during the summer season when most forage production occurs in the Walnut Gulch Experimental Watershed is indicated. Again, the variability is appreciable with the amounts of 240 mm and 170 mm being less than 5 km distant. Spatial precipitation variability is proportionally ameliorated by nonsummer rains in either the early year (January-March) or late in the calendar year (October-December). Both seasons’ precipitation can provide antecedent moisture for early season forage grasses.


Runoff at the Walnut Gulch Experimental Watershed is typical of many semi-arid regions in that the channels are dry for most of the year. Runoff only occurs as the result of rainfall and the hydrographs are "flashy" meaning that the flood peak arrives very quickly after the start of runoff and the duration of runoff is short. Almost all of the annual runoff and all of the largest events occur between July and September as a result of high intensity, short duration, and limited areal extent thunderstorms. Runoff occurs very infrequently in the early fall as a result of tropical cyclones and in the winter as a result of slow moving frontal systems both of which cover large areas and have rainfall of low intensities and long durations. 

Although these fall and winter rainfall events generate little runoff at the Walnut Gulch Experimental Watershed, this is not the case for the San Pedro River just downstream from where Walnut Gulch enters the river. For the same period of record (1963-1996), the top six annual maximum peak flow events at the outlet of the Walnut Gulch Experimental Watershed occurred in the summer months, while for the San Pedro, two of the top six occurred in the fall and two occurred in the winter. 

Watershed size or scale plays an important role on the dominant processes determining runoff characteristics. At the hillslope scale, the rates and amounts of runoff are influenced by rainfall intensity and soil-vegetation characteristics. Runoff occurs when the rainfall intensity is greater than the infiltration capacity of the soil, a process referred to by hydrologists as rainfall excess or "Hortonian Flow". The importance of rainfall intensity in the generation of runoff can be illustrated by plotting the frequency of the maximum 30 minute rainfall intensity for rainfall events in the non-summer months, for events in the summer months that do not produce runoff, and for events in the summer months that do produce runoff. As can be seen, the average 30 minute intensity for the summer runoff producing rainfall is twice and three times as large than for the non runoff producing summer and winter rainfall events respectively. The influence of the interaction between rainfall intensity and soils and vegetation can be illustrated by comparing the frequency of runoff producing summer events between the Lucky Hills shrub dominated watershed 102 and the Kendall's grass dominated watershed 112. In this case the average 30 minute intensity is 10 mm/hr greater for the Kendall's watershed meaning that it takes higher rainfall intensities to produce runoff on the grassed watershed. In contrast to runoff at the hillslope scale, runoff at the watershed scale is controlled more by infiltration of water into the alluvial channels (transmission losses) and the spatial distribution of thunderstorm rainfall. The result of these two factors leads to a decrease in unit runoff depth and peak discharge with increasing area. 

The runoff data from the Walnut Gulch Experimental Watershed have been used for flood frequency analysis, water yield estimations, and validation of hydrologic and sediment yield models. Current uses of the data include small and large scale water balance estimates, runoff and sediment yield linkages with the Upper San Pedro River Basin, and validation of remote sensing algorithms and simulation models integrated with Geographic Information Systems.

Runoff Transmission Losses

In semi-arid areas such as the Walnut Gulch Experimental Watershed, ranching, wildlife, increasing populations, urbanization, expanding industry, and needs of downstream water users all compete for limited water resources. The increased demand for water resources creates pressure to develop new sources of water and requires better methods of quantifying the water budget; assessing streamflow; and assessing the interaction between streamflow, flooding, infiltration losses in channel beds and banks, evapotranspiration, soil moisture, and ground water recharge. 

An important component of the Walnut Gulch Experimental Watershed water budget is streamflow abstraction from infiltration in the channel beds and banks, called transmission losses. Transmission losses are important because water infiltrates when flood waves move through the normally dry stream channels, reducing runoff volumes and flood peaks, and affecting components of the hydrologic cycle, such as soil moisture and ground water recharge. The importance of recharge through the Walnut Gulch Experimental Watershed ephemeral stream channels has been confirmed by ground water mounding (increases in water levels in wells in and adjacent to the main channels) after flood events. Owing to the small diameter of the runoff producing storms, most flows traverse dry channels and large reductions in runoff occur. The entire watershed is highly dissected by a dense channel network providing significant opportunity for transmission losses.

An example of transmission losses is presented. The August 27, 1982 storm, was isolated in subwatershed 6 on the upper 95 km2 of the watershed (and not all of that produced runoff). The runoff measured at Flume 6 amounted to 2.46x105 m3 with a peak discharge of 107 m3s-1. Runoff traversing 4.2 km of dry streambed between Flume 6 and Flume 2 resulted in significant infiltration losses. For example, in the 4.2 km reach the peak discharge was reduced to 72 m3s-1 and 48,870 m3 of water were absorbed in the channel alluvium. During the course of the 6.66 km from Flume 2 to Flume 1, the peak discharge was further reduced, and 41,930 m3 of runoff was infiltrated in the channel alluvium.

Erosion and Sedimentation

Erosion and sediment transport are highly variable across the Walnut Gulch Experimental Watershed, largely in response to variability in precipitation and runoff. The processes of erosion, sediment transport, and deposition on uplands and in ephemeral stream channels are being studied at several locations on the Walnut Gulch Experimental Watershed across a range of spatial scales. 

At the plot scale, rainfall simulator experiments are being conducted to quantify the relationships between, rainfall, runoff, and sediment yield. Experiments are being designed to use rare earth elements as tracers to quantify the spatial variability of erosion at the plot scale. The patterns and rates of soil erosion and redistribution within small watersheds are being determined by analyzing the distribution of fallout 137Cesium. Results are being coupled with sediment data collected as part of the longterm monitoring program. Traversing slot sediment samplers located at the outlet of Santa Rita critical depth runoff measuring flumes collect sediment samples. The traversing slot sampler was designed to measure sediment concentrations under high velocity, sediment laden flow conditions. Currently, sediment concentration samples are collected at seven small (0.18 to 5.42 ha) watersheds. Sediment concentrations and total event sediment yields are related to storm-runoff characteristics, and statistical relationships have been developed to estimate sediment yields for events with missing data. Sediment yields 0.07 to 5.7 t ha-1 yr-1, with an areal average of 2.2 t ha-1 yr-1. For six of the seven watersheds between 6 and 10 events produced 50% of the total sediment yields over the eleven year period. On the seventh watershed two storms produced 66% of the sediment because of differences in the geomorphology and vegetation characteristics of that area. Differences between sediment yields from all watersheds were attributable to instrumentation, watershed morphology, degree of channel incision, and vegetation. 

Sediment yield is monitored at stock ponds located at the outlets of 10 small watersheds. These data are used to quantify long-term sediment yield rates and provide data critical to developing sediment budgets for semi-arid rangeland watersheds. 

Collected field data and the Walnut Gulch Experimental Watershed sedimentmonitoring network provide critical data for developing simulation models and rangeland assessment methods. These data have been used to develop equations to predict hydraulic geometry and erosion rates in small channels as functions of discharge, shear stress distribution, and soil properties. Collected data have been used in conjunction with current erosion prediction technologies, such as CREAMS, WEPP, and RUSLE, to improve the scientific understanding of small watershed erosion processes.