designing and implementing a network for sensing water ... - pubs/jones_et_al-2017-jawr… ·...

DESIGNING AND IMPLEMENTING A NETWORK FOR SENSING WATER QUALITY AND

HYDROLOGY ACROSS MOUNTAIN TO URBAN TRANSITIONS1

Amber Spackman Jones, Zachary T. Aanderud, Jeffery S. Horsburgh, David P. Eiriksson, Dylan Dastrup,

Christopher Cox, Scott B. Jones, David R. Bowling, Jonathan Carlisle, Gregory T. Carling, and Michelle A. Baker2

ABSTRACT: Water resources are increasingly impacted by growing human populations, land use, and climatechanges, and complex interactions among biophysical processes. In an effort to better understand these factors insemiarid northern Utah, United States, we created a real-time observatory consisting of sensors deployed at aquaticand terrestrial stations to monitor water quality, water inputs, and outputs along mountain to urban gradients. TheGradients Along Mountain to Urban Transitions (GAMUT) monitoring network spans three watersheds with similarclimates and streams fed by mountain winter-derived precipitation, but that differ in urbanization level, land use,and biophysical characteristics. The aquatic monitoring stations in the GAMUT network include sensors to measurechemical (dissolved oxygen, specific conductance, pH, nitrate, and dissolved organic matter), physical (stage, temper-ature, and turbidity), and biological components (chlorophyll-a and phycocyanin). We present the logistics of design-ing, implementing, and maintaining the network; quality assurance and control of numerous, large datasets; anddata acquisition, dissemination, and visualization. Data from GAMUT reveal spatial differences in water quality dueto urbanization and built infrastructure; capture rapid temporal changes in water quality due to anthropogenic activ-ity; and identify changes in biological structure, each of which are demonstrated via case study datasets.

(KEY TERMS: monitoring; instrumentation; urbanization; sensor network; environmental observatory; qualityassurance/quality control.)

Jones, Amber Spackman, Zachary T. Aanderud, Jeffery S. Horsburgh, David P. Eiriksson, Dylan Dastrup, ChristopherCox, Scott B. Jones, David R. Bowling, Jonathan Carlisle, Gregory T. Carling, and Michelle A. Baker, 2017. Designingand Implementing a Network for Sensing Water Quality and Hydrology across Mountain to Urban Transitions.Journal of the American Water Resources Association (JAWRA) 1-26. https://doi.org/10.1111/1752-1688.12557

INTRODUCTION

Monitoring water systems with high temporal andspatial resolution for an extended duration provides

important insight into aquatic ecosystem processes(Parr et al., 2002; Kirchner et al., 2004; Rundel et al.,2009; Halliday et al., 2012; Rode et al., 2016). In thepast decade, the use of in situ sensors in environmen-tal monitoring has increased (Hart and Martinez,

1Paper No. JAWRA-16-0223-P of the Journal of the American Water Resources Association (JAWRA). Received November 24, 2016;accepted June 5, 2017. © 2017 The Authors. Journal of the American Water Resources Association published by Wiley Periodicals, Inc. onbehalf of American Water Resources Association. This is an open access article under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs License, which permits use and distribution in any medium, provided the original work is properly cited, the use isnon-commercial and no modifications or adaptations are made. Discussions are open until six months from issue publication.

2Research Engineer (Spackman Jones) and Assistant Professor (Horsburgh), Utah Water Research Laboratory, Research Technician (Cox),Professor (Jones), and Research Professional (Carlisle), Department of Plants, Soils, and Climate, and Professor (Baker), Department of Biol-ogy and the Ecology Center, Utah State University, 8200 Old Main Hill, Logan, Utah 84322; Associate Professor (Aanderud) and ResearchTechnician (Dastrup), Department of Plant and Wildlife Sciences, and Assistant Professor (Carling), Department of Geological Sciences, Brig-ham Young University, Provo, Utah 84602; and Hydrologist (Eiriksson), Global Change and Sustainability Center, and Professor (Bowling),Department of Biology, University of Utah, Salt Lake City, Utah 84112 (E-Mail/Spackman Jones: [email protected]).

JOURNAL OF THE AMERICAN WATER RESOURCES ASSOCIATION JAWRA1

JOURNAL OF THE AMERICAN WATER RESOURCES ASSOCIATION

AMERICAN WATER RESOURCES ASSOCIATION

https://doi.org/10.1111/1752-1688.12557

http://creativecommons.org/licenses/by-nc-nd/4.0/

http://creativecommons.org/licenses/by-nc-nd/4.0/

2006; Porter et al., 2012; Laney et al., 2015; Blaenet al., 2016; Pellerin et al., 2016); however, data gapsstill exist at scales ranging from watersheds to theglobe (Montgomery et al., 2007; Harding et al., 2014;Peters et al., 2014; Goodman et al., 2015), and guid-ance on sensor deployment, use, and data manage-ment remains limited (Rundel et al., 2009; Laneyet al., 2015; Lundquist et al., 2015; Pellerin et al.,2016). As Lundquist et al. (2015) observe, there is apaucity of literature regarding “. . . how instrumentsare actually installed, maintained, and quality-con-trolled, likely because technicians are paid to fixproblems rather than write about them.” Further-more, there is a lack of documentation and standard-ization of quality control (QC) in environmentalsensor networks (Strachan et al., 2016), casting doubton the reliability and comparability of resulting data(Campbell et al., 2013), even though quality-con-trolled and annotated datasets are of high value forreuse (Porter et al., 2012). Despite these concerns,high-frequency water quantity and water qualitymonitoring are essential to capture hydrologic andchemical patterns in aquatic systems, test hypotheses(Horsburgh et al., 2011; Rode et al., 2016), and facili-tate water resource management (Parr et al., 2002;Pellerin et al., 2016).

Globally, mountains play an important role in pro-viding water resources from snow and ice to down-stream urban population centers (e.g., Viviroli et al.,2007; Immerzeel et al., 2010; Buytaert and De Bi�evre,2012), but are underrepresented in environmentaldata collection networks (Strachan et al., 2016). TheIntermountain West of the United States (U.S.)encompasses high elevation landscapes from theSierra Nevada east to the Rocky Mountains, is char-acterized by arid to semiarid climate (Wise, 2012), andprovides water resources to well over 30 million peoplein urban centers in the U.S. and Mexico (Vano et al.,2014). In Utah, nearly 86% of the state’s populationresides in the rapidly growing urban corridor along theWasatch Front (Hale et al., 2015), a population that ishighly dependent on mountain water resources. Moni-toring of water storage and water quality fluxes isincreasingly important in this region because of highrates of population growth (Kotkin, 2013), long-termdroughts (Cook et al., 2004), and reduced snowpack(Gillies et al., 2012; Luce et al., 2013; Scalzitti et al.,2016). There is growing concern that current watersupplies will be inadequate for increased waterdemand (Montgomery et al., 2007; Bardsley et al.,2013), and dwindling water supplies increase the sig-nificance of water quality.

Within the urban context, flows in natural con-veyances are abstracted into drainage pipes, canals,and other man-made infrastructure that provide watersupply, flood control, and stormwater management

(Kaushal and Belt, 2012). Return flows from thesesystems significantly affect water quantity and qual-ity in urban streams (Groffman et al., 2003), hencetracking water as it passes through Utah’s urbanareas requires monitoring not only the streams, butalso significant inflows such as stormwater outfalls.Water quality in urban streams can be highlydynamic both spatially and temporally, driven notonly by signals from upstream watersheds (e.g.,spring snowmelt) but also by diversions, localstormwater inputs, and urban groundwater (Bhaskarand Welty, 2012; Kaushal et al., 2014; Hall et al.,2016c; Gabor et al., 2017). Quantifying the varyinghydrologic response from land uses that differ inurban infrastructure is challenging (Ryan et al.,2010), but it can be critically important for under-standing the function of urban streams, predictingpotential flooding, and assessing water qualityimpacts on urban streams and downstream receivingwaters (Paul and Meyer, 2001; Walsh et al., 2005,2016).

In this article, we describe a water quality sensornetwork for a mountain to urban environmentalobservatory that is part of the innovative UrbanTransitions and Aridregion Hydro-sustainability pro-ject (iUTAH: http://iutahepscor.org). This statewide,multi-university effort seeks to understand theimpacts of population increase, changing land use,and climate change on Utah’s water resources to pro-vide better information in planning for the sustain-ability of natural and urban systems. The seminalinfrastructure of iUTAH is a real-time observatorynetwork of terrestrial climate and aquatic stationscalled GAMUT (Gradients Along Mountain to UrbanTransitions) that collectively captures changes inwater resources along a gradient from Utah’s highelevation mountains through the state’s most denselypopulated urban areas. GAMUT is a cooperativeeffort between Utah’s three major research universi-ties (Utah State University, University of Utah, andBrigham Young University).

Our study combines the expertise of techniciansand scientists to describe the design, deployment, andoperation of the GAMUT network. We provide speci-fics on station selection and sensor deployment, main-tenance considerations, data integration andmanagement, and post-processing. We describeimportant lessons learned in network implementationand operation, information that we wish we had apriori and that we believe will be useful for a widecommunity of scientists who are now developing sen-sor networks for monitoring aquatic and terrestrialsystems (e.g., McDowell, 2015; Hinckley et al., 2016).We present our findings as follows: Gradients alongMountain to Urban Transitions Network outlines therequirements that drove our work; Network Design

JAWRA JOURNAL OF THE AMERICAN WATER RESOURCES ASSOCIATION2

JONES, AANDERUD, HORSBURGH, EIRIKSSON, DASTRUP, COX, JONES, BOWLING, CARLISLE, CARLING, AND BAKER

http://iutahepscor.org

details the methods used in designing a network tomeet those requirements; Network Implementationprovides the results of our specific implementation ofthe principles laid out in the design, including solu-tions to challenges we encountered, resourcesrequired to implement the network, and how GAMUThas catalyzed further research; and Case Studies pre-sents three brief data vignettes to illustrate the util-ity of GAMUT data for assessing the effects ofurbanization and anthropogenic activity on waterquality.

GRADIENTS ALONG MOUNTAIN TO URBANTRANSITIONS NETWORK

GAMUT was conceptualized as an in situ waterresearch facility to provide insights into biophysicalprocesses that impact water resources, facilitate newprojects by institutional researchers and educators,and improve existing monitoring and data infrastruc-ture to catalyze Utah’s competitiveness for researchfunding — as Hinckley et al. (2016) observe, moni-toring networks have potential to engage the scien-tific community to synergize scientific discovery. Theoverarching objective of GAMUT was to capture howwater quantity and quality change in multiplewatersheds along the gradient from the high moun-tains of Utah to the state’s population centers in thevalleys. In our selection of watersheds to instrument,we also wanted to represent gradients in the rateand types of urbanization and land-use change.These gradients are not specific to Utah, but arecommon in the Intermountain West region wherewater begins as mountain snowpack, flows throughrivers and streams, is stored in reservoirs, and iseventually used by populations living in the moun-tain valleys (Brown et al., 2005; Grimm et al., 2008).The following design requirements and principlesemerged from our original conceptualization of thenetwork:

1. Multiple watersheds were required to capturedifferent patterns of urbanization, differentstream sizes, and different mountain watersources.

2. Each watershed needed to be monitored along anelevation gradient and through urban areas.

3. Both aquatic and terrestrial climate stationswere required to capture water fluxes andinstream water processes.

4. An advanced suite of water quality observationsfor aquatic sites was necessary to capture biologi-cal and chemical parameters of interest.

5. The sensor network needed to be standardized sothat it could be managed and operated by multi-ple collaborating institutions and to ensure com-parability of data across sites and watersheds.

6. The network needed to capture the effects ofhuman water management infrastructure com-mon to urban Utah watersheds (e.g., dams andreservoirs, diversions, and stormwater returnflows).

7. The network needed to observe variables at highfrequencies and for extended durations to cap-ture seasonal variation (e.g., spring snowmeltrunoff, summer agricultural diversions) and dis-crete natural and anthropogenic events (e.g., pre-cipitation, agricultural returns, stormwaterflows).

8. The resulting data needed to be accessible to abroad audience (i.e., scientists across domains,educators and students, and stakeholders) periUTAH’s data policy (Horsburgh and Jones,2016).

9. Data generated by GAMUT needed to be pub-lished in standardized formats to be discoverableon a broader scale and to facilitate integrationwith other monitoring networks.

NETWORK DESIGN

Designing the GAMUT network required specifica-tion of monitoring hardware (e.g., sensors, datalog-gers, communication peripherals) as well as a planfor operating and maintaining the network and itsresulting datasets. In the following subsections, wedescribe the methods we used to design these aspectsof GAMUT. We follow with a section to describe inmore detail the specific implementation of thesedesigns.

Monitoring Station Design and Siting

To meet the requirements for the GAMUT net-work, we established standard designs for both aqua-tic and climate stations. Station design includedvariables to be measured, sensors to be used, andhow stations would be standardized in equipmentand programming, a crucial component to optimizeusability of monitoring network data (Thorpe et al.,2015; Hinckley et al., 2016). Based on our experience(e.g., Bowling et al., 2010; Horsburgh et al., 2010;Eiriksson et al., 2013), we needed sensors to be easilyserviceable, consistent across sites, and replaceable.


DESIGNING AND IMPLEMENTING A NETWORK FOR SENSING WATER QUALITY AND HYDROLOGY ACROSS MOUNTAIN TO URBAN TRANSITIONS

We sought robust and documented equipment fromestablished manufacturers to minimize time spenttroubleshooting and to ensure that technicians couldaccess support from vendors. Where possible, wesought to use sensor technology implemented byagencies and other observatories (e.g., U.S. GeologicalSurvey [USGS], National Ecological Observatory Net-work) to facilitate data comparability.

We designed all stations to include onsite datarecording and storage with real-time connectivity viaactive telemetry connections. We selected reliable andstandardized equipment for supplying power and pro-viding communications to ensure that the stationscould operate autonomously, that data were consis-tently collected across all sites, and that data weredependably streamed to a centralized base station(ESIP EnviroSensing Cluster, 2014). The power andcommunication equipments installed at each GAMUTstation are detailed in Table 1. We used manufac-turer estimates of sensor power consumption todevelop power budgets for the GAMUT stations andselected battery and solar panel sizes that exceededthe power needs of the sensor suite with the goal ofkeeping stations fully functional for 7-10 days on bat-tery power without a charge (Campbell Scientific,2011; Balam, 2013).

We developed a plan for locating stations withineach watershed. In designing watershed observato-ries, placement of monitoring sites is dependent onthe scientific goals of the study, the topographic andland-use characteristics of the watershed(s), as wellas logistical aspects such as access, telemetry options,and physical infrastructure for installation (Strobland Robillard, 2008; ESIP EnviroSensing Cluster,2014). In order to span elevations and mountain tourban environments in each watershed, we decided toplace aquatic monitoring stations: (1) in a high

elevation first- or second-order stream; (2) in a mid-elevation second- or third-order stream, which maycorrespond to immediately below a significantimpoundment to capture the effects of a dam andreservoir; (3) at a low elevation valley site; and (4)near the terminus of each stream within or below theurban area of interest. For climate and terrestrialmonitoring, we attempted to locate stations in: (1)high elevation mountain headwater areas; (2) mid-elevation areas near reservoirs; and (3) low elevationin the valley/urban areas. Where possible, weplanned to co-locate aquatic stations with existingdischarge gaging stations to take advantage of his-toric and ongoing data collection efforts by federalagencies and local water districts. Furthermore, weattempted to approximately co-locate climate andaquatic stations where possible. Ideally, the locationof each station provides measurements that are rep-resentative of a relatively large area (valley scale forclimate sites, reach scale for aquatic sites). To thisend, climate stations were positioned in open areasand aquatic stations were sited within the mainchannel flow. This enables more accurate interpola-tion between sites and minimizes bias caused bylocalized climatic and aquatic features (World Meteo-rological Organization, 2008).

Fundamental and Enhanced Water QualityStations. We designed aquatic monitoring stationsto collect data for a set of “fundamental” water qual-ity variables. These include dissolved oxygen (DO),specific conductance (SC), pH, water temperature,turbidity, and stream stage. We determined thatobservations for these sensors could help in answer-ing many, but not all of our driving research ques-tions, particularly in urban areas. Therefore, weadded a set of “enhanced” variables to measure ataquatic stations bracketing sites up- and downstreamof urban areas. Enhanced variables included biologi-cal constituents (i.e., chlorophyll-a and phycocyaninpigments), nutrients (i.e., nitrate), and fluorescentdissolved organic matter (fDOM). Many of the aquaticvariables are measured by sensors attached to a mul-tiparameter sonde. Table 2 lists the variables mea-sured at GAMUT sites and provides a justificationand basis for why we chose to measure each variable.Specific sensors used to measure these variables arealso included in Table 2, and details of their deploy-ment are described in more detail in the NetworkImplementation section.

Climate Stations. Climate stations were designedto complement aquatic stations and provide infrastruc-ture for research activities related to water supply,soil moisture, evapotranspiration, and biogeochem-istry. The core suite of sensors acquired for climate

TABLE 1. Power, Communications, and Peripheral Components atGradients Along Mountain to Urban Transitions Network Sites.The battery, radio or modem, and datalogger are housed in anenclosure attached to a mast or tower along with the solar paneland antenna.

Component Manufacturer and Model

Battery PowersonicCharge controller Morningstar SunSaverSpread-spectrum radio Campbell Scientific RF450Cell phone modem RAVENDatalogger Campbell Scientific CR3000-RC

(climate), CR800 (aquatic)Solar panel SolartechAntenna Campbell Scientific 14201 YagiEnclosure Campbell Scientific ENC16/18Mast Campbell Scientific UT20Tower ROHN 25SS020



stations measures air temperature, relative humidity,barometric pressure, wind speed and direction, radia-tion, precipitation, snow depth, soil moisture, and soiltemperature. See Table 2 for the complete list of vari-ables, sensors, and rationale.

Operational Design

The operational design for GAMUT includes theplans and procedures for how the network would beoperated across multiple watersheds. Settling on a

TABLE 2. Site Type, Variables Measured, Rationale for Inclusion, and Sensor Manufacturer and Model.

Site Type Variables Rationale Sensor Model

Fundamental andenhanced aquatic

Dissolved oxygen Important for aquatic organisms and the health ofaquatic ecosystems. Used by State of Utah as an overallindicator of water quality

YSI EXO2 599100-01

Specific conductance, watertemperature

Temperature influences biological activity and growth.Specific conductance measures the concentration ofdissolved constituents. Both are used by the State ofUtah as water quality indicators

YSI EXO2 599870-01

pH Determines the solubility and biological availability ofchemical constituents in water

YSI EXO2 599795-02

Stage Measure of stream water level needed to calculatedischarge

Campbell ScientificCS451

Turbidity Optical measure of water clarity that is related toconcentrations of total suspended solids (e.g., Joneset al., 2011)

Forest TechnologySystems DTS-12

Enhanced aquatic Fluorescent dissolved organicmatter (fDOM)

DOM includes important components of the carbon cycle,is important in aquatic food webs, and can indicateaquatic-terrestrial linkages (e.g., Gabor et al., 2015)

YSI EXO2 599101-01

Phycocyanin, chlorophyll-a Indicators for the concentration of photosyntheticpigments present in cyanobacteria and algae

YSI EXO2 599102-01

Nitrate Important biological macro-nutrient Satlantic SUNA V.2Terrestrial climate Air temperature, relative humidity Air temperature can control rates of biological growth,

chemical reactions, and affects nearly all other weatherparameters. Relative humidity is a measure of thewater vapor content of air

Campbell ScientificHC2S3

Air temperature Redundant measure of air temperature Apogee T110Barometric pressure The weight of the atmosphere. Indicates changes in

weather patternsCampbell ScientificCS106

Wind speed, wind direction Important for monitoring and predicting weatherpatterns. Affects rates of evaporation, aeration, andmixing in surface waters

RM Young 5303

Precipitation Measure of the delivery of atmospheric water to thesurface of the earth. Amount and duration ofprecipitation affects water availability for humans andecosystems

Geonor T-200B

Snow depth Indicator of the amount of water stored in solid form onthe surface of the earth relating to water availability

Judd CommunicationsUltrasonic DepthSensor

Incoming and outgoing shortwaveand longwave radiation

Indicators of the amount of energy from the sun reachingthe earth’s surface and the amount of radiation emittedby the earth’s surface and lower atmosphere. Importantin estimating an energy budget

Hukseflux NR01

Incoming shortwave radiation Redundant measure of incoming shortwave radiation Apogee SP-230Incoming and outgoingphotosynthetically activeradiation

Indicators of the amount of light available forphotosynthesis

Apogee SQ-110

Infrared surface temperature Influences physical, chemical, and biological processes atthe soil surface

Apogee SI-111

Soil moisture, soil temperature,soil conductivity

Important in estimating the exchange of water and heatbetween the atmosphere and soil

Acclima ACC-SEN-SDI

Enclosure humidity Quality assurance/control variable indicating moistureintrusion into the datalogger enclosure

Campbell ScientificCS210

Enclosure open door sensor Quality assurance/control variable indicating whenmaintenance actions were performed at a station

Campbell Scientific18166



design for the operational aspects of GAMUT wasimportant up front given that we planned to deploysites in three watersheds managed by different tech-nicians employed by separate organizations.GAMUT’s operational design needed to include thefollowing: plans for quality assurance (QA) and QC,site and sensor maintenance, rating curve develop-ment for aquatic stations, and data collection andmanagement.

Quality Assurance/Quality Control Plan. Toensure procedural consistency across the watersheds,we developed and implemented standard protocols fordata QA and QC. Campbell et al. (2013) differentiatebetween QA and QC of sensor data: quality assurancerefers to a “set of processes or steps taken to ensurethat the sensor network and protocols are developedand adhered to in a way that minimizes inaccuraciesin the data produced,” whereas quality control “occursafter the data are generated and tests whether theymeet the necessary requirements for quality outlinedby the end users.”

Protocols to ensure that the data are reliable areimportant given the geographic scope of the GAMUTnetwork, the distribution of technicians across insti-tutions, the potential turnover of personnel, and thebroad audience for which the data are intended. Reg-ular maintenance of stations and sensors, includingcleaning and calibration, is essential to QA (Parret al., 2002; Campbell et al., 2013). Manufacturersprovide guidelines for sensor maintenance; however,the recommended periodicity is typically unspecified.Our plan was to implement a minimum frequency ofmonthly site visits and to increase the frequency ifthe monitoring of data or site conditions revealedissues (Wagner et al., 2006). An aquatic site visitinvolves cleaning sensors to minimize the effects offouling and performing calibration for sensors thatare subject to drift. We initially adopted calibrationcriteria from the USGS (Wagner et al., 2006) andfrom sensor manufacturers (e.g., Xylem, 2012).

A detailed record of field activities is essential todocument environmental conditions and site and sta-tion maintenance actions such as calibrations, sensordeployments, and retrievals (World MeteorologicalOrganization, 2008; ESIP EnviroSensing Cluster,2014) and is important for post-processing as datacorrections should not be made unless the source oferror can be explained by field notes or data fromother stations or other variables (Wagner et al.,2006). We planned an online equipment managementsystem, currently under development, to ensure thatimportant information about what activities wereperformed where, when, and by whom would berecorded in standardized formats and be accessibledigitally.

Post-processing of raw environmental sensor data,which consists of adjustments to data along with theapplication of flags, or data qualifiers, to annotatedata points, is usually required before those data canbe reliably used in scientific analyses (Mourad andBertrand-Krajewski, 2002; Horsburgh et al., 2011;Campbell et al., 2013). To perform these functions,GAMUT technicians use Observations Data Model(ODM) Tools (Horsburgh et al., 2015), a software pro-gram designed for post-processing of time series data.Our project data policy gave GAMUT a goal of per-forming QC post-processing within six months of orig-inal data collection.

We adopted a series of standardized post-proces-sing steps for all variables across GAMUT sites, con-sistent with practices and recommendationsdescribed in the literature (Campbell et al., 2013;Horsburgh et al., 2015). These steps are designed toadvance the raw time series data from GAMUT sen-sors to a quality-controlled product suitable for scien-tific analysis (subject to any limitations of the datanoted in data qualifiers) and include addressing outof range values and erroneous data due to sensormalfunction or environmental conditions, correctionfor sensor drift and calibration, filling data gaps, con-ducting a final data review, and applying data flags.More details are provided on the implementation ofthese QC steps in the Supplemental Materials (FileS1) for this article.

Discharge Rating Curve Development. Streamdischarge is an essential quantity for aquatic monitor-ing, allowing the comparison of flow rate between sitesand time periods as well as the quantification of con-stituent transport. For GAMUT aquatic sites, opera-tion of the station and development of a continuousrecord of discharge required establishing rating curvesto translate stream stage measurements to discharge(Kennedy, 1982; Schmadel et al., 2010). Our designwas to use standard methods to manually measure dis-charge (Rantz, 1982; Turnipseed and Sauer, 2010;Mueller et al., 2013), associate those measurementswith concurrent stage readings, and fit relationships toresulting data to develop a rating curve (Herschy,2009). The rating curve can then be used with high-fre-quency water level data to derive discharge (Hors-burgh et al., 2010).

For GAMUT sites co-located with existing gagingstations, we adopted the discharge measurementsfrom those gages. For all other aquatic sites, periodicdischarge measurements were made using severalflow gaging methods to capture the wide range offlows observed at GAMUT sites. Instead of includingthese techniques in the NETWORK IMPLEMENTA-TION section, details are in the Supplemental Mate-rials (File S2), along with the steps we undertook to



develop stage-discharge rating curves and generatehigh-frequency estimates of discharge. Althoughthese methods are understood to be standard, theyare typically documented in disparate sources.

Design for Data Collection and Dissemination

Early on, participants in the iUTAH project com-mitted to openly publish data to a broad audience.This was codified in a data policy (Horsburgh andJones, 2016) that outlines timelines and proceduresfor data sharing designed to maximize the impactand use of datasets collected within iUTAH facili-ties and by iUTAH research teams. For GAMUT,Jones et al. (2015) provide a complete descriptionof the data management cyberinfrastructure thatsupports the network. In short, raw data arestreamed directly into operational databases andmade available online in near real time. Wedesigned the GAMUT cyberinfrastructure so thattime series data are stored using the Consortium ofUniversities for the Advancement of HydrologicScience, Inc. (CUAHSI) ODM (Horsburgh et al.,2008) and published in Water Markup Language(WaterML) format using WaterOneFlow web ser-vices (Zaslavsky et al., 2007). This makes theGAMUT data available in a national context, per-mitting discovery and download along with datafrom any other networks registered with theCUAHSI Water Data Center. Using standardizedformats also permitted us to integrate visualizationof agency data (e.g., USGS) with GAMUT data(http://data.iutahepscor.org/tsa).

NETWORK IMPLEMENTATION

At the time of writing, GAMUT includes 40instrumented climate, aquatic, or storm drain moni-toring sites, each collecting a subset of 141 vari-ables, depending on the site type, resulting in thegeneration of 2,012 individual time series, consist-ing of all of the observations for a variable mea-sured using a specific method at a particular site.Currently, the GAMUT time series comprise over174 million individual data values after approxi-mately 3.5 years of network operation. In the fol-lowing subsections, we illustrate how we applied thedesign procedures and principles described in theprevious section to create a monitoring network thatmet our requirements. We also address our specificfindings and discuss considerations for networkimplementation.

Watershed Selection

We selected the Logan River, Red Butte Creek,and the Provo River watersheds as the bases forGAMUT (Figure 1). These watersheds met ourcriteria of mountain snow water sources in differentranges, varying levels and patterns of urbanization,and water bodies of differing sizes. The threewatersheds were also strategically viable giventheir proximity to the three participating institu-tions.

The Logan River originates high in the Bear RiverMountains with headwaters near the Utah-Idaho bor-der (2,900 m), flows through forest and rangeland, isimpounded to create several small reservoirs inLogan Canyon, and then flows through lower eleva-tions in Cache Valley (1,380 m), which is slowly tran-sitioning from agricultural to urban land use, beforeterminating at Cutler Reservoir on the Bear River.The average daily discharge (1971-2015) at the USGSgage near the outlet of Logan Canyon (USGS10109000 Logan River Above State Dam, near Logan,Utah) is 6.51 m3/s from a catchment area of 554 km2,and the mean elevation is 2,300 m (U.S. GeologicalSurvey, Surface Water Data for U.S.: USGS AnnualStatistics. Accessed September 23, 2016, http://waterdata.usgs.gov/nwis/; all streamflow and catchmentareas are derived from this source). Deployment andmaintenance of GAMUT in the Logan River water-shed are managed by personnel at Utah StateUniversity.

Red Butte Creek originates in the Wasatch Moun-tains in Salt Lake County (2,300 m) in a forested,protected research natural area (Ehleringer et al.,1992), is impounded by a dam in Red Butte Canyon,and then flows through the University of Utah cam-pus and highly urbanized portions of Salt Lake City(1,300 m) where the creek joins the subsurface andstorm drain system and eventually terminates inthe Jordan River. Red Butte Creek has a catchmentarea of 20.8 km2, mean elevation of 2,012 m, and anaverage daily discharge (1964-2015) near the mouthof the canyon (USGS 10172200 Red Butte Creek atFort Douglas, near SLC, Utah) of 0.114 m3/s.Scientists and technicians from the University ofUtah manage GAMUT installations in Red ButteCreek.

The Provo River originates high in the UintaMountains in Summit County (3,600 m), flowsthrough relatively remote mountains and forestbefore being impounded to create a large reservoir(Jordanelle), and then flows through the mid-eleva-tion Heber Valley, which is currently transitioningfrom agriculture to ex-urban land use with rapid pop-ulation growth (25% for Heber City in the past fiveyears) (U.S. Census, 2015 Quick Facts. Accessed



http://data.iutahepscor.org/tsa

http://waterdata.usgs.gov/nwis/

http://waterdata.usgs.gov/nwis/

October 12, 2016, http://www.census.gov/quickfacts/table/PST045215/4934200,49043). After leaving theHeber Valley (1,660 m), the Provo River flowsthrough a second large reservoir (Deer Creek), downProvo Canyon and into the city of Provo, Utah, andultimately discharges to Utah Lake. A gage beforethe river enters Jordanelle Reservoir (USGS10155000 Provo River near Hailstone, Utah) recordsan average daily discharge (1950-2015) of 7.76 m3/sfrom a catchment area of approximately 596 km2,and a gage in the Heber Valley (USGS 10155500Provo River near Charleston, Utah) records an aver-age daily discharge (1992-2015) of 7.22 m3/s from acatchment area of approximately 930 km2. The meanelevation of the watershed is 2,450 m. GAMUT in theProvo River is managed and maintained by staff fromBrigham Young University.

Monitoring Site Selection

We sited aquatic water quality stations at five loca-tions and terrestrial climate stations at four locationsin each watershed. Stations and locations are detailedin Table 3, and Figure 2 provides a representation ofeach site within the watershed. All monitoring sites,regardless of type, were subject to several siting con-siderations. First, reliably communicating data innear real time were a challenge given remote moni-toring sites and mountainous topography. We createda mixed telemetry system using both cellular andspread-spectrum radio technologies to overcome thesechallenges. Second, we considered the likelihood ofvandalism and theft at potential monitoring locations(also described by Campbell et al., 2013; ESIPEnviroSensing Cluster, 2014). Third, since the

FIGURE 1. Location of Watersheds Selected for the Gradients Along Mountain to Urban Transitions Network. Adapted from Baskin et al. (2002).



http://www.census.gov/quickfacts/table/PST045215/4934200,49043

http://www.census.gov/quickfacts/table/PST045215/4934200,49043

iUTAH project has a strong education and outreachcomponent, we considered sites that were visible tothe public and accessible by student and othergroups. Furthermore, for all GAMUT sites, partner-ships with landowners (the U.S. Forest Service formany GAMUT sites), local city and county govern-ments, conservation districts, and universities werecritical during the initial permitting phase, and manysites required legal access agreements between theuniversity and the landowner.

Ultimately, we worked to balance scientific needswith physical site constraints, communication

constraints, public engagement goals, site security,and partnership potential. Our planning and siteselection process was iterative and took well over ayear to complete. Iteratively revising network designallows for practitioners to incorporate important les-sons learned through their experience (Strobl andRobillard, 2008). Some legal access agreements tookmonths to negotiate, which was a limiting step andcan be a major constraint and timing considerationfor implementing new networks. In addition, despiteour best efforts to secure stations and sensors, wehave experienced damage, including theft of cable

TABLE 3. Gradients Along Mountain to Urban Transitions Network Station Locations and Land-Use Types Organized by Watershed andOrdered by Elevation. Elevations are in meters. Site classifications are determined in part by Woods et al. (2001).

Watershed Site Name Site Type Elevations Land Use Latitude Longitude

Red Butte Knowlton Fork Climate Climate 2,010 Wasatch montane 40.810122 �111.76695Knowlton Fork Aquatic Fundamental aquatic 1,990 Wasatch montane 40.809522 �111.765472Todd’s Meadow Climate 1,763 Semiarid foothills 40.789054 �111.796416Above Red Butte ReservoirAquatic

Enhanced aquatic 1,674 Semiarid foothills 40.779602 �111.806669

Above Red Butte ReservoirClimate

Climate 1,655 Semiarid foothills 40.780567 �111.807222

Red Butte Gate Fundamental aquatic 1,579 Urban transition 40.774228 �111.817025Cottam’s Grove Fundamental aquatic 1,505 Urban transition 40.763958 �111.828286Conner Road Storm drain 1,499 Urban 40.762522 �111.828439Green Infrastructure ResearchFacility Climate

Climate 1,488 Urban 40.7608 �111.830474

Green Infrastructure ResearchFacility Storm Drain

Storm drain 1,486 Urban 40.760912 �111.829696

Fort Douglas Storm drain 1,473 Urban 40.759012 �111.831446Dentistry Building Storm drain 1,463 Urban 40.757989 �111.832084Foothill Drive Enhanced aquatic 1,459 Urban 40.757225 �111.8337221300 East Enhanced aquatic 1,353 Urban 40.744995 �111.854441900 West Fundamental aquatic 1,291 Urban 40.7416 �111.9176

Provo River Trial Lake Climate 3,040 Uinta subalpine forest 40.678111 �110.948339Beaver Divide Climate 2,508 Uinta subalpine forest 40.612508 �111.098289Soapstone Climate Climate 2,388 Uinta subalpine forest 40.573928 �111.043503Soapstone Aquatic Fundamental aquatic 2,367 Uinta subalpine forest 40.579503 �111.047669Woodland Fundamental aquatic 2,136 Exurban 40.5578613 �111.168625Below Jordanelle Reservoir Enhanced aquatic 1,790 Exurban 40.59507 �111.42864Sage Creek Canal 1,690 Exurban 40.488245 �111.440195Sage Creek Flood Canal 1,690 Exurban 40.488245 �111.440195Lower Midway Fundamental aquatic 1,676 Exurban 40.50707 �111.44991Charleston Climate Climate 1,659 Exurban 40.484717 �111.462558Charleston Aquatic Enhanced aquatic 1,658 Exurban 40.48498 �111.46245

Logan River TW Daniels ExperimentalForest

Climate 2,629 Wasatch montane 41.864805 �111.507494

Franklin Basin Climate Climate 2,109.52 Semiarid foothills 41.949815 �111.581352Franklin Basin Aquatic Fundamental aquatic 2,110.3 Semiarid foothills 41.9502 �111.580553Tony Grove Climate Climate 1,927.86 Semiarid foothills 41.885493 �111.568767Tony Grove Aquatic Fundamental aquatic 1,886.1 Semiarid foothills 41.875846 �111.564533Utah Water ResearchLaboratory west bridge

Enhanced aquatic 1,414 Urban transition 41.739034 �111.795742

Spring Creek Storm drain 1,386 Urban 41.710961 �111.833736Main Street (Highway 89/91) Fundamental aquatic 1,377 Urban 41.721091 �111.835096River Heights Bridge Storm drain 1,373 Urban 41.725147 �111.825917Blacksmith Fork aboveconfluence with Logan River

Fundamental aquatic 1,366 Urban 41.704431 �111.8508

Logan River Golf Course Climate 1,364 Urban 41.705643 �111.854268Mendon Road (600 South) Enhanced aquatic 1,353 Agricultural 41.720533 �111.886928



and solar panels and cable damage by wildlife.Suggestions of best practices for avoiding site damageare included in the Supplemental Materials (File S1).In particular, purchasing an extra set of all equip-ment permits quick replacement of sensors or othercomponents to minimize potential data loss.

Factors we considered for locating aquatic stationsinclude anchoring the sensors and datalogger enclo-sure, accessing the sensors for maintenance at highand low water levels, ensuring that the sensorsremain submerged at low water levels, protecting thesensors from debris and shear stress at high flowlevels, preventing sedimentation in the sensor hous-ings, and assessing the likelihood for water freezing

and resulting sensor damage. Despite our attentionto these considerations, flows in Utah’s rivers arehighly variable, and we have had cases of sensorsexposed to air due to low water levels as well as sen-sor housings shearing at high flow levels. Some ofour aquatic sites are also prone to sedimentation insensor housings during snowmelt or significant stormevents. We have yet to experience sensor damage dueto freezing, likely because we made efforts to ensurethat even when the surface of the water is frozen, thesensors remain submerged in flowing water below theice. Our experience has been that varying environ-mental conditions require that we adopt an adaptivestrategy for managing stations, adjust site visit

FIGURE 2. Conceptual Diagram of Gradients Along Mountain to Urban Transitions Network Site Locations Relative to Each Other andMajor Features within Each Watershed. Not to scale.



frequency as necessary, and utilize housings thatenable modifying sensor positions in response to flowconditions.

We also considered the development of ratingcurves when selecting locations for aquatic stations.Important factors include a suitable stream gagingcross section (i.e., straight river reach and uniformflows across stream) nearby and a natural hydrologiccontrol and streambed that are not prone to shifting.These qualities minimize the likelihood that ratingcurves will need to be re-created after high flowevents (Rantz, 1982).

For climate stations, we made similar considera-tions in determining locations. We consistentlydeployed stations in open areas with low vegetationto prevent obstruction of radiation sensors, provide alevel area for valid snow depth readings, and avoidpotential interference with wind and precipitationmeters from nearby trees, buildings, or other tallobjects. Furthermore, we sought un-irrigated loca-tions with natural vegetation, even in urban areas, toprevent interference with precipitation gaging and toprovide representative radiation readings. Weacknowledge that siting climate monitoring in moun-tain topography requires a balance for selecting idealsettings for sensing different variables. For example,precipitation is most accurately gaged in protectedzones, whereas air temperature, humidity, and windshould be measured in open areas to be generallyrepresentative (Strachan et al., 2016).

Another factor for site selection was co-location withexisting monitoring sites to augment data collection byother entities, reduce redundancy, and facilitate inte-gration. All three watersheds include USGS gages withlong discharge records, and we deployed our waterquality monitoring equipment adjacent to these gageswhere possible. In the Provo River, we co-locatedGAMUT aquatic sites with additional gages maintainedby the Central Utah Water Conservancy District, andtwo GAMUT climate sites were co-located with existingSnow Telemetry (SNOTEL) sites operated by the U.S.Department of Agriculture’s National Resources Con-servation Service (NRCS). The Logan River watershedcontains an experimental forest with a long record ofmeteorological and soil observations (Mahat and Tar-boton, 2014), and we integrated our monitoring withexisting infrastructure at that site. In Red Butte Creek,climate stations were located to complement and/orreplace a previous sensor network maintained by theUniversity of Utah (Ehleringer et al., 1992).

Sensor Deployment and Station Installation

Our network design specified the sensors that wewould use for monitoring (Table 1), but we needed to

implement the physical installation of the stationsand the deployment of sensors. In general, implemen-tations were standardized to each site type, althoughin some cases, effective installations involved address-ing site-specific challenges.

Aquatic Station Implementation. Across allaquatic sites, sensors are housed in acrylonitrilebutadiene styrene (ABS) pipes extending into theriver with a mast to which the instrumentation enclo-sure and solar panel are attached (Figure 3). Sondesand turbidity sensors are housed in 10.16 cm (4 in)ABS pipe, pressure transducers are housed in5.08 cm (2 in) ABS pipe, and nitrate sensors arehoused in 15.24 cm (6 in) polyvinyl chloride pipe.Sensor housings terminate in pump screens or pipecaps with holes drilled into the bottom to allow ade-quate water flow for accurate measurements whileprotecting the sensors from debris during high flows.At some sites, existing structures (e.g., bridges, con-crete walls) were used to mount these housings. Atsites with no structures present, a sensor mountingframe was designed, fabricated, and deployed (Fig-ure 3b) consisting of two vertical fence posts cemen-ted into the ground with horizontal sensor mountingposts affixed to the vertical posts using structural fit-tings. This platform allows for flexibility of installa-tion in a variety of streambank situations. Eachaquatic site was also equipped with a graduatedstage plate, with locations surveyed to local bench-marks to provide a permanent reference for observa-tions of water surface elevations.

Climate Station Implementation. All climatestations were deployed by erecting a ~6 m towerbased in concrete to which cross arms were connectedfor mounting sensors (Figure 3a). Manufacturerguidelines were generally followed in sensor installa-tion. Sensor arms were typically mounted 2 m abovethe ground, although deep snowpack required thatsensors be mounted higher at some high elevationsites. This was an important consideration as sensorscan be buried by deep snow, and any snow “creep”can shear cross arms, instrument enclosures, andsensors from their mountings. Radiation sensors weremounted to a mast arm on the south side of the towerto eliminate the risk of shading from the tower andsolar panel, though reflection from the solar panelmay occur. We found that at high elevation sites, pre-cipitation gages needed to be mounted to 2.5 m ped-estals to reduce the possibility of snow interferencewith gage orifices, whereas lower elevation precipita-tion gages could be mounted to 1 m pedestals. Tominimize variability in wind data caused by localmicro terrain and vegetation, anemometers weremounted near the top of towers (World Meteorological



Organization, 2008). Soil moisture/temperature sen-sors were installed in pits adjacent to the sensortower, and cables were protected from rodents withflexible conduit.

Storm Drain Station Implementation. Forstorm drain sites, acoustic Doppler velocity meters(ADVM: Teledyne ISCO 2150 flow module) weremounted to adjustable scissor rings that expand to fit

pipe diameters ranging between 40.64 and 203.2 cm.The ADVM sensors were positioned in the bottom ofstorm drain pipes and measure both water depth andvelocity to instantaneously determine discharge usingthe pipe geometry. In some storm drains, hydraulicconditions (e.g., pipes with slopes great enough tocause “rooster tail” flows from low depth, high veloc-ity water impacting the face of the flow module)invalidated the methods used by the ADVM to

FIGURE 3. Examples of Gradients Along Mountain to Urban Transitions Network Station Installations with Schematics: (a, d) typical climatesite, (b, e) aquatic site mounted to a bridge, (c, f) aquatic site with custom sensor mounting framework. Green text indicates sensors whilebrown text signifies infrastructure and peripherals. DO, dissolved oxygen; SC, specific conductance; ABS, acrylonitrile butadiene styrene.



measure discharge. In these cases, we mounteddownward-facing sonic sensors, designed for measur-ing snow depth, to the top of the scissor ring to gen-erate an additional water depth measurement forflow calculations. We use the depth measurementswith site-specific constants to generate discharge esti-mates using Manning’s equation.

Datalogging, Telemetry, Power, and Data Publication

Our initial design included industry standard data-loggers and power and communications peripherals,but we needed to program measurement intervals andaveraging procedures as well as determine the fre-quency of communication and mechanisms for eventualdata publication. We selected 15 min as the frequencyfor recording data in an attempt to observe actual tem-poral fluctuations in variables of interest and estimateprocess rates while avoiding capturing sensor noise,generating unnecessarily large datasets, and strainingpower resources. Several sensors include internal pro-cessing for value reporting, and we incorporated aver-aging in the datalogger programs to minimize spuriousdata points, reduce sensor noise, and capture condi-tions over the measurement recording interval. Ataquatic stations, factory settings were used for vari-ables reported by the sondes. Given a single measure-ment command from the datalogger, the sonde’sonboard processing performs burst sampling, outlierexclusion, and averaging algorithms with stabilizationcriteria specific to each variable, returning processedresults. For the turbidity sensor, a single measurementcommand triggers a burst of 100 instantaneous mea-surements made over five seconds, and a suite of statis-tics are returned. For the pressure transducer, weimplemented burst sampling by calling for the sensorto make 25 instantaneous measurements (requiringabout 20 s) and report the mean. For most climate vari-ables, we programmed the datalogger to scan at 10-sec-ond intervals and average values over 15 min. Forsensors that measure variables that are prone to noise(i.e., snow depth, soil moisture, and precipitation), weimplemented burst sampling to better capture instan-taneous values. For these variables, measurements aremade every 10 s during the final minute of the 15-mininterval and the average is reported. Generic datalog-ger programs implemented for GAMUT aquatic and cli-mate sites are provided as Supplemental Materials(File S3— aquatic, File S4— climate).

Station dataloggers store data in local memory,and GAMUT uses a variety of telemetry connectionsto transmit data, including spread-spectrum radioswhere line-of-sight is available and commercial cellu-lar band modems where spread-spectrum radios areimpractical. One or more base stations in each

watershed retrieves the data from all sites and is con-nected to the Internet, permitting data to be trans-mitted to a centralized location, uploaded tooperational databases, and made accessible. Our ini-tial design was to communicate with sites hourly toprovide data in near real time; however, in somecases, this frequency contributed to power losses. Thepower budgets we developed for GAMUT suggest thatstations should be fully functional for 7-10 days onbattery power alone; however, this assumes new bat-teries and the original suite of sensors. At some sites,sensors have been added, which, along with agingbatteries, reduce the longevity of the battery’s effec-tive charge. At the time of writing, we have experi-enced a number of cases of battery failure,particularly at high elevation climate sites in thewinter where cold temperature, snow accumulation,and rime on solar panels are common and whereadditional peripheral sensors have been deployed, allof which may strain power resources. Based on ourestimates for GAMUT sites, communications canaccount for 20-30% of the power budget, and onestrategy for reducing power consumption is to reducethe frequency of communications, which we havedone for select sites. Battery failure may also be pre-vented by establishing a voltage threshold for cuttingpower to the system, but we have not yet imple-mented this practice.

The workflow for data streaming from field sensorsto operational databases to dissemination via theInternet is described by Jones et al. (2015). We alsouse HydroShare, a community repository for hetero-geneous resource types (http://www.hydroshare.org),to provide long-term archival, publication, and simpli-fied access for the following GAMUT data resources:(1) raw data in a flat CSV file for each monitoringsite; (2) quality-controlled data for each variable ateach site with the script of editing steps; and (3)stage-discharge relationships as a package consistingof individual discharge measurements, the resultingrelationship, and pertinent metadata (e.g., iUTAHGAMUT Working Group 2016, 2017a, b). Raw dataare updated in HydroShare on a daily basis. There issome lag in the publication of quality-controlled andderived data products due to the time needed fortechnicians to review and generate these datasets,but quality-controlled data are generally publishedwithin six months. Resources containing stage-dis-charge relationships are updated as needed.

Quality Assurance Implementation

We implemented QA in GAMUT by employing con-sistent procedures for cleaning, calibration, andmaintenance of sensors, by recording those activities,



http://www.hydroshare.org

and by regularly monitoring data. The following sub-sections describe sensor maintenance, including casesthat prompted us to modify our maintenance protocolwhen our experience revealed deficiencies in ourpractices. In general, technicians record field andmaintenance activities on uniform field sheets as wellas digitally while we develop an online equipmentmanagement system (Jones et al., 2015).

To monitor data, a technician in each watershedperforms regular visual inspections (2-3 times perweek) of raw data to identify and document potentialproblems and to prioritize field activities. We alsoimplemented automated alerts to identify possibleissues in data streams occurring between regularvisual checks of the data and to reduce the requiredfrequency of visual checks. The alerts are pro-grammed as stored procedures in our operationaldatabases, which run daily and send email notifica-tions when data screening criteria are not met. Therules currently implemented for GAMUT includechecks of battery voltage range, checks for “no data”values, checks for data persistence (e.g., flat line),and checks to ensure data are current (Jones et al.,2015) and are consistent with community recommen-dations for sensor data QA/QC (Campbell et al., 2013;ESIP EnviroSensing Cluster, 2014; Integrated OceanObserving System, 2015).

Sensor Maintenance at Aquatic Stations. Ouroriginal QA plan called for monthly site visits forcleaning, calibration, and other maintenance. Afterobserving large shifts in the fDOM, phycocyanin, andchlorophyll-a data associated with calibration events,we discovered that calibration coefficients variedmore than expected. We determined that the prepara-tion of calibration solutions and field calibration pro-cedures were introducing more error than if theoriginal calibrations had been retained. We changedour protocols to only calibrate these sensors in thelaboratory under constant temperature, with suffi-cient time for equilibration, and under controlled con-ditions for calibration solution preparation andstorage. We concluded that optical sensors (i.e.,fDOM, DO, phycocyanin, chlorophyll-a, and nitrate)are generally stable enough to require calibrationchecks only every three to four months or more andshould only be calibrated if needed. Our experience issimilar to that of other users of these instrumentsand informal guidance provided by sensor manufac-turers (YSI). For DO, calibrations should still be per-formed in the field at the elevation at which thesensor is measuring. We have continued monthly cal-ibration checks for pH and SC sensors, which aremore prone to drift.

After observing large diurnal fluctuations in stagedata that were not independently corroborated and

were correlated with water temperature, we deter-mined that the pressure transducer temperature com-pensation was invalidated at some sites.Communication with the manufacturer (Campbell Sci-entific) verified that this is due to scale buildup thatmay occur in systems with significant calcium carbon-ate content, which is all of the aquatic sites in theLogan River and Red Butte Creek. The one pressuretransducer deployed in the upper Provo River has notexhibited this behavior, which we conclude is becausethe upper Provo River is more pH-neutral than Loganand Red Butte Creek. To prevent scale buildup, we nowregularly (every two to three months) rinse the pres-sure transducers in a vinegar solution for 5-10 min,depending on the visible condition of the sensor.

While automated wipers that clean sensor facesminimize the effects of fouling on the aquatic sensors,we needed to clean sensors at least monthly toremove sediment from the sonde measurement cup,to ensure that wipers on all sensors are functioning,and to remove biofilms and scale from sensor bodieswith a cloth or soft-bristled brush. For some sitesduring some seasons, more frequent cleaning is nec-essary (e.g., sediment accumulation during springsnowmelt runoff necessitates weekly visits at someaquatic sites). Several times each year, the probehousings and pump screens need to be removed andcleaned of biological growth. Additional proceduresfor aquatic site maintenance include checking sensorwipers, which may need periodic replacement, andchecking the pressure transducer desiccant andreplacing when expired. The technicians’ regularvisual monitoring of data and automated alerts alsohelp identify environmental conditions that mayrequire additional attention. Manufacturer recom-mendations for regular maintenance of sensors andequipment are outlined in the Supplemental Materi-als (File S1).

Sensor Maintenance at Climate Stations. TheGAMUT climate stations are mostly autonomous andrequire relatively little maintenance, as problemswith sensors are typically identifiable with data mon-itoring procedures. Regular maintenance includesmonthly inspections to check that sensors are notcontaminated by dirt, insect activity, etc.; adjustmentto verify that sensors remain level; and general clean-ing to ensure optimal operation. A few seasonal cir-cumstances necessitate additional maintenance.During the winter, solar panels and radiometersmust periodically be cleared of snow, as solar radia-tion data can be impacted by snow accumulation onthe sensors, and snow, ice, or rime on solar panelscan prevent station batteries from recharging. Wedetect snow accumulation by monitoring precipita-tion, station power, and incoming shortwave



radiation measurements. The precipitation gagesinstalled in GAMUT also require routine (at leasttwice per year) replacement of antifreeze and oil tothe measurement bucket. Manufacturer recom-mended maintenance for sensors deployed at climatesites is included in the Supplemental Materials(File S1).

As climate sensors are not easily calibrated, wesought other methods to verify sensor readings. Somevariables (i.e., incoming shortwave radiation and airtemperature) are measured by two independent sen-sors at each climate station, which facilitates datacomparison and validation. For variables that are notmeasured in pairs, we verify by comparing readingsbetween sites. As of this writing, we have plans tomaintain a spare set of equipment to be used as rov-ing reference sensors (Campbell et al., 2013) to spotcheck readings of deployed sensors.

Quality Control Implementation

Within GAMUT, QC consists of regular review ofdata series and post-processing to apply flags andadjust data to generate an approved, reviewed dataseries, which is performed by technicians in each ofthe three GAMUT watersheds using the ODM Toolssoftware. ODM semantics use QC levels to designatethe level of post-processing associated with a dataset.For GAMUT, we determined to use QC level 0 (QC0)for raw data streaming from sensors, and QC level 1(QC1) to designate data series that have beenreviewed, have corrections applied, and are approvedby technicians. GAMUT also uses QC level 2 to repre-sent derived products (e.g., discharge derived fromstage). These levels are consistent with thosedescribed by Porter et al. (2012).

As they performed QC, GAMUT techniciansobserved that, although we collectively set up a frame-work to guide post-processing, application of edits andcorrections was often subjective. For example, twotechnicians performing QC post-processing on thesame raw dataset could arrive at two separate results.To promote consistency in the transformation fromQC0 to QC1 in GAMUT, we developed and imple-mented more specific guidelines for QC, describedbelow, including a general QC workflow, priorities forQC (e.g., which time series would be processed), andvariable-specific post-processing steps. We also discov-ered several unusual cases requiring innovative QCsolutions. This was particularly important as thenumber of personnel conducting post-processing grewbeyond the three watershed technicians.

Quality Control Workflow. To perform QC, atechnician reviews the QC0 data, performs the

necessary edits using ODM Tools, and saves theresulting Python script wherein each edit is capturedas a line of code in a text file. The data and script arethen reviewed by a supervising technician, revised ifneeded, and the processed data are committed to theoperational database as QC1. The script serves as therecord of the transfer from QC0 to QC1, and techni-cians make comments in the script to annotate therationale for corrections. For GAMUT, the process forcreating new QC1 data series or updating existingseries with ODM Tools is used as described by Hors-burgh et al. (2015).

We made several decisions to specifically imple-ment the QA/QC framework for GAMUT. First, weneeded to determine how to handle data gaps, anoma-lies, and periods of erroneous data. Figure 4 showsexamples from GAMUT of these common QC cases.We concluded that for periods of two hours or less, lin-ear interpolation could reasonably be used to fill gapsor periods of verified erroneous data. For longer peri-ods, or if the technician judges that linear interpola-tion is inappropriate, we assign data to values of�9,999 to represent “No Data.” Including a “No Data”value (rather than leaving the period blank) indicatesproper data collection did not occur and permits theassignment of a qualifier to provide an explanation.We settled on a standard set of flags from which alltechnicians could select the appropriate qualifier toexplain periods of questionable data (e.g., sediment,ice, snow) (Table 4). In all cases of interpolation orassignment to �9,999, a flag is applied to alert datausers and provide relevant details.

We determined that linear drift correction shouldbe used for all cases of aquatic sensor calibration andcleaning (Figure 4c) unless there is no perceptibleshift in the data. The value by which to shift data fordrift corrections is determined by visual estimation,by calculation based on the slope of the data at thetime of calibration, or by the difference in the pre-and post-calibration readings reported by the sensor.We use ODM Tools to apply filters to identify anoma-lies and data gaps, interpolate, fill data gaps, assignvalues as �9,999, apply qualifiers, and perform lineardrift correction (Horsburgh et al., 2015).

Variable-Specific Quality Control. We devel-oped specific recommendations for data review andpost-processing for each of the GAMUT variables thatundergo QC, and these details are included in theSupplemental Materials for this article (File S2). Sev-eral variables exhibited behavior that was outside ofour (and the sensor manufacturers’) expectations,requiring unconventional QC solutions. Figure 5illustrates six of these conditions, which are docu-mented in detail as technical notes in the Supplemen-tal Materials (File S4).



Our solutions might be applicable to other monitor-ing networks using similar sensors, but more broadly,observatory developers should expect to face similarQC issues that may require unfamiliar or uniquesolutions. Sensors are tools for which the proper usemay depend on the particular situation to which theyare applied. Although we are inclined to rely on theresults of sensors from trusted manufacturers, thesecases show that scientists and researchers should be

skeptical of sensor outputs and seek independentmethods of data verification.

Resources to Create and Maintain GAMUT

The initial proposal for GAMUT anticipated thatoperating the network would require the support ofseveral personnel at each institution and some fundsfor sensor servicing and repair. The original plancalled for three full-time field leads/technicians, eachemployed by one of the supporting institutions andassigned to the associated GAMUT watershed; onepart-time data manager; and several university fac-ulty as project leads, all of which are roles outlined inSutter et al. (2015).

The installation and maintenance of GAMUT siteswas labor intensive. While the time required varieddepending on ease of access and technician experi-ence, generally speaking, after preparation and plan-ning, the physical installation of each aquatic sitetook approximately one day for two to three individu-als to complete and each climate site took approxi-mately two to three days for three to five individualsto complete. Significant time was also invested in sit-ing and installing repeater telemetry stations. As the

FIGURE 4. Examples of Raw Data Requiring Quality Control Post-Processing: (a) data outliers in soil temperature (degrees Celsius), (b) periodof sensor malfunction in turbidity (nephelometric turbidity units [NTU]) data, (c) for specific conductance (microSiemens per centimeter), sensor

calibration in raw data and drift correction in quality-controlled data, (d) gaps in air temperature (degrees Celsius) measurements.

TABLE 4. Standardized Qualifiers/Flags Used in Gradients AlongMountain to Urban Transitions Network.

Code Description

LI Linear interpolationSM Sensor malfunctionPF Power failureS Suspicious valuesICE Ice interference with sensorSNOW Snow interference with sensorMNT Erroneous or missing data due to maintenanceSED Sediment interference with sensorLWT Data suspicious due to low water. Sensor likely dryCAL Improper or erroneous calibrationCOR_PT Pressure transducer data corrected to

remove erroneous data signalZERO Value set to zero



technicians and data manager developed and imple-mented the GAMUT QA/QC Plan, it became apparentthat the regular site visits and post-processingrequired support from additional personnel. ForGAMUT, we have found the ideal arrangement to bea technician in each watershed with two to threewell-trained assistants, along with a data managerwith one assistant. A more mature network mayrequire fewer personnel.

Significant monetary resources were required toinitially purchase sensors and infrastructure. At thetime of acquisition, the sensors for a GAMUT climatestation cost ~$25,000, a fundamental aquatic stationcost ~$14,000, and an enhanced aquatic station cost~$41,000. Each station required ~$1,700 in power andcommunications equipment, and each telecommunica-tions repeater station cost ~$2,800. In addition to theobvious upfront costs, funds must be allocated to sup-port network longevity. Many manufacturers recom-mend sensor replacement, factory recalibration, and/or cable replacement after one to five years (see

Table S2 in Supplementary Materials). Damagecaused by vandalism, wildlife, and environmentalevents may require more frequent equipment replace-ment.

After the initial hiring of technicians and procure-ment of sensors, it was six months before the firstGAMUT stations were deployed and nine monthsbefore the first GAMUT data were available online.After 15 months, the majority of the existing GAMUTstations were operational. Documenting and begin-ning full implementation of the QA/QC Plan occurredafter 18 months. To acquire enough data to derivesufficiently robust stage-discharge relationships andto catch up on the backlog of post-processing datatook approximately three years. It was also only aftermultiple years of data collection that variable-specificQC issues became apparent. Finally, to document thisprocess and present the results in a scientific journaltook four years. Although we are aware that individ-ual circumstances will vary, we present these timeframes as a point of reference for others who may be

FIGURE 5. Variable-Specific Cases of Data Requiring Quality Control Post-Processing Showing Raw Data and the Quality-Controlled Data. (a)Phycocyanin (relative fluorescence units [RFU]) with calibration events for which calibration coefficients shifted. Data were corrected by retroac-tively applying corrected calibration coefficients. (b) Negative precipitation (centimeters) resulting from high power voltages corrected by eitherinterpolation or reassignment to �9,999 to represent “No Data.” (c) Negative precipitation (centimeters) caused by sensor noise and evaporationof accumulated precipitation. Data were corrected using an algorithm that compares each point to the previous to eliminate decreases. (d) Waterlevel (centimeters) prior to and after sensor cleaning. Diurnal fluctuations in level are erroneously associated with water temperature (degreesCelsius), so the slope of the temperature-level relationship was used to remove the incorrect water temperature compensation. (e) Elevatedwater level (centimeters) due to stream ice damming. Data were corrected by interpolation or reassignment to �9,999. (f) Inverted soil tempera-ture (degrees Celsius) due to a sensor firmware issue. Data were corrected by reversing the sign during these periods.



considering building observatories similar toGAMUT, and we anticipate that the findings we pre-sent in this article may help expedite the establish-ment of other monitoring networks.

Extensibility and Synergy

As mentioned, an intended outcome of GAMUTwas its capacity to catalyze further science, whichHinckley et al. (2016) assert to be the greatest infor-mation contribution of long-term monitoring. Weanticipated that research efforts would benefit fromthe foundation provided by GAMUT watersheds andsites, the GAMUT physical infrastructure and cyber-infrastructure, and the GAMUT data. Indeed, wewanted the GAMUT network to serve as a “backbone”onto which researchers could add complementarymonitoring infrastructure. Furthermore, we antici-pated that we would need to address situations suchas additional or revised monitoring locations as welearned more about important processes in the threewatersheds. The design of GAMUT is scalable andmodular, permitting the addition and removal of sen-sors at stations, the transferring of entire stations tonew locations, and the addition of new stations to thenetwork. Our upfront decisions to standardize equip-ment across watersheds and stations facilitates flexi-bility and expansion.

After deploying the first set of GAMUT stationsand the initial phase of monitoring, we recognizedthat we were not capturing important components ofthe hydrologic system in urban and urbanizing areas.As a result, we deployed additional sites to bettercapture stormwater inputs and downstream urbanareas in Red Butte Creek, stormwater outfalls, andimportant tributaries to the Logan River, and anagricultural canal on the Provo River. We also re-located aquatic sites downstream to better capturewater quality conditions of the Provo River beforereaching its terminus at Utah Lake, an impairedwater body (UDWQ, 2016). These changes were madewithout major modifications to our telemetry networkor underlying data management cyberinfrastructure.

Climate stations were envisioned as data collectionplatforms that could be added to for studies beyondinitial iUTAH funding. To accomplish this, weacquired dataloggers and multiplexors with capacityfor expansion beyond the initial sensor suitedescribed in our design. Most stations have beenupgraded to include sensors that measure soil oxy-gen, soil carbon dioxide, and soil heat flux. At severalhigher elevation sites, monitoring rates of sapflux forparticular tree species have been undertaken in con-junction with GAMUT (Chan and Bowling, 2016a, b,c, 2017).

Furthermore, because of the scale of GAMUT dataand the other available resources, researchers arechoosing to build on GAMUT by using GAMUT dataand working in GAMUT watersheds. Examplesinclude efforts to better understand nitrogen dynam-ics in snow, soil, and water in Red Butte Creek (Hall,2016a, b; Hall et al., 2016a, b); analyses for trace ele-ments and isotopes in precipitation, snowpack, sur-face water and groundwater, and plants, algae, andmoss in all three GAMUT watersheds (Carling et al.,2015, 2016; Hall, 2016c; Hall et al., 2016b); andexperiments to study effects of nutrients and pharma-ceuticals and the structure of bacterial communitiesat GAMUT aquatic sites (Ogata and Baker, 2016).Researchers have conducted intensive synoptic sam-pling in each of the watersheds to help validate sen-sor readings, to understand relationships withunmeasured variables, and in an attempt to bettercapture groundwater interactions.

Other facilities have used GAMUT as a spring-board. In the Logan River watershed, a sister net-work monitoring urban stormwater has beendeployed in an adjacent canal (Melcher and Hors-burgh, 2017), which adopts components of GAMUT’sdesign and implementation, including the GAMUTtelemetry network. In the Provo River watershed, arecent toxic algal bloom on Utah Lake prompted thedeployment of buoyed monitoring platforms by theState of Utah. For this effort, the State has used theexpertise of GAMUT technicians and the GAMUTcyberinfrastructure.

In addition to biophysical studies, GAMUT is ahub for education and outreach efforts and socialscience research. The GAMUT watersheds and aqua-tic stations are the central venue for summer insti-tutes that train K-12 teachers, undergraduates, andhigh school students. Several faculty members areincorporating GAMUT station visits and data intotheir university courses, and an outreach effort at alocal museum features GAMUT data in an interactivedisplay. Social science researchers have broadenedthe idea of environmental monitoring into a socio-eco-logical observatory as they collect social water sciencedata (Flint et al., 2017) within and adjacent toGAMUT watersheds.

All of these additional efforts have built fromGAMUT’s base infrastructure and baseline datasets,and all were facilitated by forethought in planning,locating, and instrumenting the GAMUT networkstations. This is an important consideration in build-ing networks like GAMUT, as the original fundingunder which the infrastructure is built rarely extendsbeyond the three- to five-year period of the originalresearch grant (Thorpe et al., 2015). Designing exten-sibility into the network from the beginning wasimportant in catalyzing these types of new efforts



that can bring new sources of funding to supportlonger term sustainability.

DATA CASE STUDIES

We have described in detail the practices that wefollowed and our insights and findings in the installa-tion and operation of GAMUT. Aside from the focus onthe logistics of the network, data from GAMUT revealinsights into water quality, especially as it is impactedalong the mountain to urban gradient. In this section,we provide three brief examples of applications ofGAMUT data. We use GAMUT data to show simplyhow urbanization affects fDOM pulses and availabilityas well as the frequency of algal blooms indicated bypeaks in chlorophyll-a and phycocyanin. We also useGAMUT data to provide insight on the effects of builtinfrastructure on aquatic systems and to provideevidence for a specific anthropogenic activity.

Urbanization Increases Pulses of Organic Matter andAlgae Blooms

Runoff events in urbanized areas carry pulses ofnutrients and organic matter to streams, which het-erotrophic bacteria may exploit. We compared fDOMbetween the three most downstream sites in eachwatershed and identified cases of “pulses,” which wedefined as an increase in fDOM of at least 100%within an hour. In Red Butte Creek, the most urban-ized of the GAMUT watersheds, fDOM pulsesoccurred 22 times over a three-month period, some-times lasting up to three days (Figure 6b). By com-parison, levels of fDOM remained relatively constantin the Provo River (~30 quinine sulfate units [QSU])and Logan River (1.5 QSU) over the same time per-iod. The pulses of fDOM in Red Butte Creek mostlycoincide with weather driven episodes that transportsediment and other materials to streams (Wilsonet al., 2013), a process that is expedited by urbaniza-tion. The transport of fDOM with sediment is corrob-orated by concurrent spikes in turbidity measured atthe same site in Red Butte Creek and the generalpaucity of turbidity spikes in the Logan and Provowatersheds (Figure 6c). High percentages of impervi-ous surfaces and multiple storm drain outfalls lead toflashy flow regimes characteristic of urban streams(Hong et al., 2012), which, for Red Butte Creek, alsocorrespond to higher frequency of fDOM pulses.

We also compared the patterns of photosyntheticpigments chlorophyll-a and phycocyanin between thethree watersheds to give an indication of the

frequency of potential algae blooms. If either of thesevariables increased over 100% within an hour, weidentified that period as a bloom. Generally, thesespikes were more common in Red Butte Creek thanthe other two less-urbanized rivers. Phycocyaninpeaks, which may represent cyanobacteria blooms(Figure 6e) were more frequent than chlorophyll-apeaks, which may represent green algal blooms (Fig-ure 6d). In Red Butte Creek, 236 cyanobacteriablooms occurred over three months. On days withpigment spikes, the average increase of phycocyaninconcentrations was 200% compared to days withoutelevated levels. Over the same time period, 75 greenalgal blooms occurred, which increased chlorophyll-aconcentrations an average of 313% per day comparedto days without elevated levels. We also observedphotosynthetic pigment spikes in the Provo Riverfrom mid-November to the end of December (33 algaland 11 cyanobacterial). In both cases, we cannotdemonstrate whether these elevated photosyntheticpigments resulted from blooms in the river or reser-voirs upstream of the monitoring station and/or weregenerated from sloughing of benthic periphyton closerto the sensor locations. Regardless, these patternsare similar to those observed by Reed et al. (2016),who found algal biomass and growth rates and harm-ful cyanobacterial blooms more common in urbanizedcreeks and stormwater ponds compared to forestedand agricultural tidal creeks. Our findings supportclaims that developed lands may aggravate waterquality issues (Paul and Meyer, 2001; Walsh et al.,2005, 2016; Kaushal et al., 2014). Because thesepeaks appeared and disappeared within single days,high-frequency data were essential in capturing theflashiness of algal pigments in these systems.

Reservoir Size Structures Water Quality

Built infrastructure, specifically the size and char-acteristics of a reservoir system, impacts water chem-istry (Ward and Stanford, 1983; Stanford and Ward,2001). As mentioned, all of the GAMUT watershedsinclude reservoirs for water resource management,but the size of the catchments and reservoir systemsvary. Jordanelle Reservoir on the Provo River is large(maximum surface area of ~13.5 km2), while thereservoirs on the Logan River (1st, 2nd, and 3rddams, combined total surface area of 0.088 km2) andRed Butte Creek (Red Butte Reservoir, surface areaof 0.038 km2) are relatively small (Figure 2). Weexamined the effects of these reservoirs on waterquality by comparing data from the aquatic sitesabove and below the reservoirs in each watershed.

The longer retention times associated with the lar-ger reservoir on the Provo River translated into the



most dramatic changes in water quality. For exam-ple, DO is 80-90% of saturation directly below thedam in contrast to the site 29 km above the dam,which is consistently near saturation (Figure 7a).This observation could be due to oxygen consumptionin reservoir sediments and reservoir stratification(Friedl and W€uest, 2002), or caused by differences instream geomorphology above and below the reservoirthat influence physical reaeration and the capacityfor benthic algae to influence the diurnal DO signal(e.g., Erwin et al., 2016; Hall, 2016d). In Red ButteCreek and the Logan River, the overall DO levels aresimilar above and below the dams. Below JordanelleReservoir on the Provo River, the pH was substan-tially higher than above the reservoir, with greaterdiurnal variability and divergent seasonal patterns

(Figure 7b). In the Logan River and Red Butte Creek,pH stayed relatively constant through this time per-iod, and the values above and below the reservoirsdid not deviate from each other to the degree occur-ring in the Provo. SC provides an indicator of thelevel of dissolved constituents and can be used to dis-tinguish surface runoff from groundwater or base-flow. In the Provo River, SC (Figure 7c) belowJordanelle Reservoir was consistently higher thanabove the reservoir by 300%, primarily due tochanges in lithology between the sample sites, butalso potentially reflecting greater evaporative waterlosses that concentrate dissolved solutes. Carlinget al. (2015) also found greater temporal variabilityin trace element and ion concentrations above Jor-danelle Reservoir than below. However, SC between

FIGURE 6. Data Are Water Chemistry and Biology Variables from Urban Aquatic Sites in Three Watersheds Showing: (a) precipitationintensity (millimeters per 15 min) and pulses of (b) fDOM (quinine sulfide units), (c) turbidity (nephelometric turbidity units), (d)

chlorophyll-a (relative fluorescence units), and (e) phycocyanin (relative fluorescence units).



the sites above and below Red Butte Reservoir waslower by 17%, which may reflect groundwater inputsto lower monitoring sites (Hall et al., 2016c). Therewas no notable difference above and below the damson the Logan River.

Signs of Construction Are Visible as Turbidity

The late winter/spring of 2011 was wet with deepsnowpack and an extended melt period that resultedin high river flows throughout the IntermountainWest (Alexander et al., 2015). During this time, theLogan River experienced extended flooding thatresulted in damage to city infrastructure on the orderof $100,000, with county-wide per capita impact of$4.04 (FEMA, 2011). In response to this flooding, theNRCS, Cache County, and City of Logan applied forfederal funding through the Emergency WatershedProtection Program. Those funds were used in part toimplement a plan for optimal flood hazard protection.City and county engineers chose a hard engineeringapproach to increase hydraulic efficiency of the river.Construction began in late winter/early spring of2014, wherein the channel was deepened throughexcavation, and the banks were stabilized. Channelroughness was also reduced by removing coarsewoody debris and installing a liner topped with

native and imported cobbles and boulders (McMillenLLC, 2012a, b).

These weekday construction activities within thechannel created tractable increases in turbidity alongthe Logan River. On Monday to Friday during work-ing hours (08:00 AM-17:00 PM), excavation and bankstabilization efforts caused daily turbidity increasesof almost 67-fold at the Logan Main Street aquaticstation (Figure 8). Conversely, during the weekends,turbidity held relatively constant. Elevated turbidityis an established indicator of soil and land distur-bance (i.e., construction work) with higher potentialfor erosion near or within waterways in both moun-tain and urban areas (Wolman and Schick, 1967;Anderson and Potts, 1987). The high-resolution datameasured by GAMUT allowed the rapid assessmentof water quality and the potential to more preciselyidentify and quantify human-induced changes along ariver’s reach.

CONCLUSIONS

GAMUT is the product of a community effort ofscientists to create an environmental observatory tomonitor fluxes of water quantity and quality along

FIGURE 7. Water Chemistry Variables from Aquatic Stations above and below Reservoirs in Each of the Gradients Along Mountain toUrban Transitions Watersheds: (a) dissolved oxygen as percent saturation, (b) pH, (c) and specific conductance (microSiemens per

centimeter).



mountain to urban gradients in three watersheds innorthern Utah, U.S. The contrasts in watershed char-acteristics, including degrees and patterns of urban-ization, can be effectively compared using GAMUTdata given standard practices for station design,installation, operation, and data management.Though the network was designed and planned byexperienced scientists with what we consider to beadequate time and funds, we encountered unexpectedsetbacks. Overall, network design required an itera-tive approach to overcome these challenges.

For effective QA/QC for GAMUT, documentedstandard practices and coordination between person-nel were essential. Cleaning and calibration of sen-sors must be performed consistently and regularlyand must also be well documented to enable post-pro-cessing. We found that actively monitoring data wasessential for identifying and addressing problems tominimize potential data loss. Although subjectivity inperforming post-processing may not be completelyovercome, we implemented a standard QC workflowthat includes recording post-processing steps, madedecisions about consistently handling inaccuratedata, and developed variable-specific guidelines toaddress this challenge.

GAMUT revealed nuances with data that haveresulted in changes in field procedures, novel solu-tions for post-processing data, and even adjustmentsto hardware by sensor manufacturers. We emphasizethe importance of experience in these cases and con-clude that scientists should remain skeptical andseek independent verification of sensor data, even forsensors from trusted manufacturers. We suggest thatthere is room for manufacturers to further clarify rec-ommended procedures and frequencies for mainte-nance as well as documentation of algorithms anddata processing.

Our three case studies demonstrate the utility ofthe high-frequency data generated by GAMUT to ini-tially assess impacts of urbanization, built infrastruc-ture, and anthropogenic activity. The water qualitydynamics were only evident with high-frequency data

over the spatial extent of GAMUT. Some of theseresearch questions were conceptualized at GAMUT’soutset as we developed hypotheses to deductively testvia water quality monitoring, but others have onlybeen revealed through the analysis of GAMUT data.

GAMUT is underlying infrastructure that servesas a vehicle for other research endeavors. Research-ers have confidence in the GAMUT network and datagiven its consistency of data collection and commit-ment to standardized operation, maintenance, andpost-processing. Although we attempted to collect abroad suite of variables at representative sites in thestudy watersheds and apply standardized QA/QC toaddress data consistency and usability, ultimatelysecondary users of the data should be familiar withthe provenance of the GAMUT data so that they canmake their own assessment of potential bias andassumptions in determining whether the data meettheir specific needs.

SUPPORTING INFORMATION

Additional supporting information may be foundonline under the Supporting Information tab for thisarticle: (S1) Supplemental Guidelines for GAMUTQuality Assurance and Quality Control; (S2) Generationof Rating Curves and Discharge Data for GAMUT; (S3)Generic Datalogger Programs for GAMUT AquaticSites; (S4) Generic Datalogger Programs for GAMUTClimate Sites; and (S5) Post-Processing Solutions toUnexpected Sensor and Data Issues for GAMUT.

ACKNOWLEDGMENTS

This work was supported by the U.S. National Science Founda-tion under EPSCoR grant IIA-1208732 awarded to Utah StateUniversity, as part of the State of Utah EPSCoR Research Infras-tructure Improvement Award. Any opinions, findings, and conclu-sions or recommendations expressed in this material are those of

FIGURE 8. Turbidity (nephelometric turbidity units) Measured in the Logan River in March 2014 Showing Spikes Corresponding toInstream Construction and Bank Stabilization.



the authors and do not necessarily reflect the views of the NationalScience Foundation. The authors gratefully acknowledge the workof many field, lab, and data technicians, in particular, Joe Craw-ford, now with the Central Utah Water Conservancy District. Wealso appreciate the input and suggestions by three anonymousreviewers, which helped improve the manuscript.

LITERATURE CITED

Alexander, M.A., J.D. Scott, D. Swales, M. Hughes, K. Mahoney,and C.A. Smith, 2015. Moisture Pathways into the U.S. Inter-mountain West Associated with Heavy Winter PrecipitationEvents. Journal of Hydrometeorology 16:1184-1206. https://doi.org/10.1175/JHM-D-14-0139.1.

Anderson, B. and D. Potts, 1987. Suspended Sediment and Turbid-ity Following Road Construction and Logging in Western Mon-tana. Water Resources Bulletin 23:681-690. https://doi.org/10.1111/j.1752-1688.1987.tb00842.x.

Balam, H., 2013. Tower of Power — Calculating Power Budgets forRemote iUTAH EPSCoR Weather Stations. USU Spring RunoffConference, Logan, Utah, April 9-10.

Bardsley, T., A. Wood, M. Hobbins, T. Kirkham, L. Briefer, J. Nier-meyer, and S. Burian, 2013. Planning for an Uncertain Future:Climate Change Sensitivity Assessment toward AdaptationPlanning for Public Water Supply. Earth Interactions 17:1-26.https://doi.org/10.1175/2012EI000501.1.

Baskin, R.L., K.M. Waddell, S.A. Thiros, E.M. Giddings, H.K. Had-ley, D.W. Stephens, and S.J. Gerner, 2002. Water-QualityAssessment of the Great Salt Lake Basins, Utah, Idaho, andWyoming — Environmental Setting and Study Design. U.S.Geological Survey Water Resources Investigations Report 02-4115, 45 pp. http://pubs.usgs.gov/wri/wri024115/.

Bhaskar, A.S. and C. Welty, 2012. Water Balances along an Urban-to-Rural Gradient of Metropolitan Baltimore, 2001–2009. Envi-ronmental & Engineering Geoscience XVIII:37-50. https://doi.org/10.2113/gseegeosci.18.1.37.

Blaen, P.J., K. Khamis, C.E.M. Lloyd, C. Bradley, D. Hannah, andS. Krause, 2016. Real-Time Monitoring of Nutrients and Dis-solved Organic Matter in Rivers: Capturing Event Dynamics,Technological Opportunities and Future Directions. Science ofthe Total Environment 569:647-660. https://doi.org/10.1016/j.scitotenv.2016.06.116.

Bowling, D.R., S.B. Marchetti, C.K. Lunch, E.E. Grote, and J. Bel-nap, 2010. Carbon, Water, and Energy Fluxes in a SemiaridCold Desert Grassland during and following Multiyear Drought.Journal of Geophysical Research 115:1-16. https://doi.org/10.1029/2010JG001322.

Brown, D.G., K.M. Johnson, T.R. Loveland, and D.M. Theobald,2005. Rural Land-Use Trends in the Conterminous UnitedStates, 1950–2000. Ecological Applications 15:1851-1863.https://doi.org/10.1890/03-5220.

Buytaert, W. and B. De Bi�evre, 2012. Water for Cities: The Impactof Climate Change and Demographic Growth in the TropicalAndes. Water Resources Research 48:1-13. https://doi.org/10.1029/2011WR011755.

Campbell, J.L., L.E. Rustad, J.H. Porter, J.R. Taylor, E.W.Dereszynski, J.B. Shanley, C. Gries, D.L. Henshaw, M.E. Mar-tin, W.M. Sheldon, and E.R. Boose, 2013. Quantity Is Nothingwithout Quality. BioScience 63:574-585. https://doi.org/10.1525/bio.2013.63.7.10.

Campbell Scientific, 2011. Power Supplies. Logan, Utah. https://s.campbellsci.com/documents/af/technical-papers/pow-sup.pdf, accessedSeptember 2016.

Carling, G.T., D. Dastrup, and T.H. Goodsell, 2016. GAMUT WaterChemistry Data. HydroShare. https://www.hydroshare.org/

resource/07f508a29bb043d29084d18d4981be13/, accessed Octo-ber 2016.

Carling, G.T., D.G. Tingey, D.P. Fernandez, S.T. Nelson, Z.T. Aan-derud, T.H. Goodsell, and T.R. Chapman, 2015. Evaluating Natu-ral and Anthropogenic Trace Element Inputs along an Alpine toUrban Gradient in the Provo River, Utah, USA. Applied Geochem-istry 63:398-412. https://doi.org/10.1016/j.apgeochem.2015.10.005.

Chan, A. and D.R. Bowling, 2016a. T.W. Daniels Sapflux Aspen.HydroShare. https://www.hydroshare.org/resource/39226ebf17294b6cb2f3ff8bbc3cc25b/, accessed October 2016.

Chan, A. and D. Bowling, 2016b. T.W. Daniels Sapflux SubalpineFir. HydroShare. https://www.hydroshare.org/resource/6a1f646b8fa54f1e995cb19d3d30c457/, accessed October 2016.

Chan, A. and D. Bowling, 2016c. Knowlton Fork Sapflux. Hydro-Share. https://www.hydroshare.org/resource/9f7a391dd2c14acba509d335e2a6d6d6/, accessed October 2016.

Chan, A.M. and D.R. Bowling, 2017. Assessing the Thermal Dissi-pation Sap Flux Density Method for Monitoring Cold SeasonWater Transport in Seasonally Snow-Covered Forests. TreePhysiology 2017:1-12. https://doi.org/10.1093/treephys/tpx049.

Cook, E.R., C. Woodhouse, C. Eakin, D. Meko, and D. Stahle, 2004.Long-Term Aridity Changes in the Western United States.Science 306:1015-1018. https://doi.org/10.1126/science.1102586.

Ehleringer, J.R., L. Arnow, T. Arnow, I. McNulty, and N. Negus,1992. Red Butte Canyon Research Natural Area: History, Flora,Geology, Climate, and Ecology. Great Basin Naturalist52:950121.

Eiriksson, D., M. Whitson, C.H. Luce, H.P. Marshall, J. Bradford,S.G. Benner, T. Black, H. Hetrick, and J.P. McNamara, 2013.An Evaluation of the Hydrologic Relevance of Lateral Flow inSnow at Hillslope and Catchment Scales. Hydrological Processes654:640-654. https://doi.org/10.1002/hyp.9666.

Erwin, S.O., J.C. Schmidt, and T.M. Allred, 2016. Post-Project Geo-morphic Assessment of a Large Process-Based River RestorationProject. Geomorphology 270:145-158. https://doi.org/10.1016/j.geomorph.2016.07.018.

ESIP EnviroSensing Cluster, 2014. Community Wiki Document onBest Practices for Sensor Networks and Sensor Data Manage-ment. Federation of Earth Science Information Partners. http://wiki.esipfed.org/index.php/EnviroSensing_Cluster, accessed Jan-uary 2016.

FEMA (Federal Emergency Management Agency), 2011. Utah —Flooding. FEMA-4011-DR. https://www.fema.gov/pdf/news/pda/4011.pdf, accessed September 2016.

Flint, C.G., A.S. Jones, and J.S. Horsburgh, 2017. Data ManagementDimensions of Social Water Science. Journal of the AmericanWater Resources Association. https://doi.org/10.1111/1752-1688.12568.

Friedl, G. and A. W€uest, 2002. Disrupting Biogeochemical Cycles-Consequences of Damming. Aquatic Sciences 64:55-65. https://doi.org/1015-1621/02/010055-11.

Gabor, R., S. Hall, D. Eiriksson, Y. Jameel, M. Millington, T. Stout,M. Barnes, A. Gelderloos, H. Tennant, G. Bowen, B. Neilson, andP. Brooks, 2017. Persistent Urban Influence on Surface WaterQuality via Impacted Groundwater. Environmental Science &Technology 49:4425-4432. https://doi.org/10.1021/es504881t.

Gabor, R.S., M.A. Burns, R.H. Lee, J.B. Elg, C.J. Kemper, H.R.Barnard, and D.M. McKnight, 2015. Influence of Leaching Solu-tion and Catchment Location on the Fluorescence of Water-Soluble Organic Matter. Environmental Science and Technology49:4425-4432.

Gillies, R., S.-Y. Wang, and M. Booth, 2012. Observational andSynoptic Analyses of the Winter Precipitation Regime Changeover Utah. Journal of Climate 25:4679-4698. https://doi.org/10.1175/JCLI-D-11-00084.1.

Goodman, K., S. Parker, J. Edmonds, and L. Zeglin, 2015. Expand-ing the Scale of Aquatic Sciences: The Role of the National



https://doi.org/10.1175/JHM-D-14-0139.1

https://doi.org/10.1175/JHM-D-14-0139.1

https://doi.org/10.1111/j.1752-1688.1987.tb00842.x

https://doi.org/10.1111/j.1752-1688.1987.tb00842.x

https://doi.org/10.1175/2012EI000501.1

http://pubs.usgs.gov/wri/wri024115/

https://doi.org/10.2113/gseegeosci.18.1.37

https://doi.org/10.2113/gseegeosci.18.1.37

https://doi.org/10.1016/j.scitotenv.2016.06.116


https://doi.org/10.1029/2010JG001322

https://doi.org/10.1029/2010JG001322

https://doi.org/10.1890/03-5220

https://doi.org/10.1029/2011WR011755

https://doi.org/10.1029/2011WR011755

https://doi.org/10.1525/bio.2013.63.7.10

https://doi.org/10.1525/bio.2013.63.7.10

https://s.campbellsci.com/documents/af/technical-papers/pow-sup.pdf

https://s.campbellsci.com/documents/af/technical-papers/pow-sup.pdf

https://www.hydroshare.org/resource/07f508a29bb043d29084d18d4981be13/

https://www.hydroshare.org/resource/07f508a29bb043d29084d18d4981be13/

https://doi.org/10.1016/j.apgeochem.2015.10.005

https://www.hydroshare.org/resource/39226ebf17294b6cb2f3ff8bbc3cc25b/

https://www.hydroshare.org/resource/39226ebf17294b6cb2f3ff8bbc3cc25b/

https://www.hydroshare.org/resource/6a1f646b8fa54f1e995cb19d3d30c457/

https://www.hydroshare.org/resource/6a1f646b8fa54f1e995cb19d3d30c457/

https://www.hydroshare.org/resource/9f7a391dd2c14acba509d335e2a6d6d6/

https://www.hydroshare.org/resource/9f7a391dd2c14acba509d335e2a6d6d6/

https://doi.org/10.1093/treephys/tpx049

https://doi.org/10.1126/science.1102586

https://doi.org/10.1002/hyp.9666

https://doi.org/10.1016/j.geomorph.2016.07.018

https://doi.org/10.1016/j.geomorph.2016.07.018

http://wiki.esipfed.org/index.php/EnviroSensing_Cluster

http://wiki.esipfed.org/index.php/EnviroSensing_Cluster

https://www.fema.gov/pdf/news/pda/4011.pdf

https://www.fema.gov/pdf/news/pda/4011.pdf

https://doi.org/10.1111/1752-1688.12568

https://doi.org/10.1111/1752-1688.12568

https://doi.org/1015-1621/02/010055-11

https://doi.org/1015-1621/02/010055-11

https://doi.org/10.1021/es504881t

https://doi.org/10.1175/JCLI-D-11-00084.1

https://doi.org/10.1175/JCLI-D-11-00084.1

Ecological Observatory Network (NEON). Freshwater Science34:42-52. https://doi.org/10.1086/679446.

Grimm, N.B., D. Foster, P. Groffman, J.M. Grove, C.S. Hopkinson,K.J. Nadelhoffer, D.E. Pataki, and D.P.C. Peters, 2008. TheChanging Landscape: Ecosystem Responses to Urbanization andPollution across Climatic and Societal Gradients. Frontiers inEcology and the Environment 6:264-272. https://doi.org/10.1890/070147.

Groffman, P.M., D.J. Bain, L.E. Band, K.T. Belt, G.S. Brush, andJ.M. Grove, 2003. Down by the Riverside: Urban Riparian Ecol-ogy. Frontiers in Ecology and the Environment 1:315-321.https://doi.org/10.1890/1540-9295(2003) 001[0315:DBTRUR]2.0.CO;2.

Hale, R.L., A. Armstrong, M.A. Baker, S. Bedingfield, D. Betts, M.Buchert, T. Crowl, R.R. Dupont, J.R. Ehleringer, C. Flint, J.Grant, S. Hinners, J.S. Horsburgh, A.S. Jones, C. Licon, S.E.Null, A. Odame, D.E. Pataki, D. Rosenberg, M. Runburg, P. Sto-ker, and C. Strong, 2015. iSAW: Integrating Structure, Actors,and Water to Study Socio-Hydro-Ecological Systems. Earth’sFuture 3:110-132. https://doi.org/10.1002/2014EF000295.

Hall, R.O., 2016d. Metabolism of Streams and Rivers: Estimation,Controls, and Application. Streams in a Changing Environment.Academic Press, San Diego, California.

Hall, S., 2016a. iUTAH Precipitation Ions and Nitrogen Stable Iso-tope Composition. HydroShare. https://doi.org/10.4211/hs.7ce47150ee344d4e95f2fd7bb4f660ac, accessed October 2016.

Hall, S., 2016b. Red Butte Canyon Meadow and Lawn Soil Biogeo-chemical Data. HydroShare. https://www.hydroshare.org/resource/5f9ec9cf259f454a8ab964e3e8d59131/, accessed October2016.

Hall, S., 2016c. Riparian Plant, Algae, and Moss Nitrogen and Car-bon Isotopes across the iUTAH Watersheds. HydroShare.https://www.hydroshare.org/resource/0e00b7cd6e11438d9f5b491dc3b6185c/, accessed October 2016.

Hall, S.J., M.A. Baker, S.B. Jones, J.M. Stark, and D.R. Bowling,2016a. Contrasting Soil Nitrogen Dynamics across a MontaneMeadow and Urban Lawn in a Semi-Arid Watershed. UrbanEcosystems 19:1083-1101. https://doi.org/10.1007/s11252-016-0538-0.

Hall, S.J., E.M. Ogata, S.R. Weintraub, M.A. Baker, J.R. Ehlerin-ger, C.I. Czimczik, and D.R. Bowling, 2016b. Convergence inNitrogen Deposition and Cryptic Isotopic Variation acrossUrban and Agricultural Valleys in Northern Utah. Journal ofGeophysical Research: Biogeosciences 121:2340-2355. https://doi.org/10.1002/2016JG003354.

Hall, S.J., S.R. Weintraub, D. Eiriksson, P.D. Brooks, M.A. Baker,G.J. Bowen, and D.R. Bowling, 2016c. Stream Nitrogen InputsReflect Groundwater across a Snowmelt-Dominated Montane toUrban Watershed. Environmental Science and Technology50:1137-1146. https://doi.org/10.1021/acs.est.5b04805.

Halliday, S.J., A.J. Wade, R.A. Skef, C. Neal, B. Reynolds, P. Row-land, M. Neal, and D. Norris, 2012. Science of the Total Envi-ronment an Analysis of Long-Term Trends, Seasonality andShort-Term Dynamics in Water Quality Data from Plynlimon,Wales. Science of the Total Environment 434:186-200. https://doi.org/10.1016/j.scitotenv.2011.10.052.

Harding, R., G. Weedon, H. van Lanen, and D. Clark, 2014. TheFuture for Global Water Assessment. Journal of Hydrology518:1860193.

Hart, J.K. and K.Martinez, 2006. Environmental Sensor Networks: ARevolution in the Earth System Science? Earth-Science Reviews78:177-191. https://doi.org/10.1016/j.earscirev.2006.05.001.

Herschy, R., 2009. Streamflow Measurement. Routledge, Taylor andFrancis Group, London, United Kingdom. ISBN: 978-0-415-4132-8.

Hinckley, E.L.S., S.P. Anderson, J.S. Baron, P.D. Blanken, G.B.Bonan, W.D. Bowman, S.C. Elmendorf, N. Fierer, A.M. Fox,K.J. Goodman, K.D. Jones, D.L. Lombardozzi, C.K. Lunch, J.C.

Neff, M.D. SanClements, K.N. Suding, and W.R. Wieder, 2016.Optimizing Available Network Resources to Address Questionsin Environmental Biogeochemistry. BioScience 66:317-326.https://doi.org/10.1093/biosci/biw005.

Hong, B., K.E. Limburg, M.H. Hall, G. Mountrakis, P.M. Groffman,K. Hyde, L. Luo, V.R. Kelly, and S.J. Myers, 2012. An Inte-grated Monitoring/Modeling Framework for Assessing Human-Nature Interactions in Urbanizing Watersheds: Wappinger andOnondaga Creek Watersheds, New York, USA. EnvironmentalModelling and Software 32:1-15. https://doi.org/10.1016/j.envsoft.2011.08.006.

Horsburgh, J.S. and A.S. Jones, 2016. iUTAH Research Data Pol-icy. HydroShare. https://doi.org/10.4211/hs.61666076104f4306b4ddbf22a3fc52ca.

Horsburgh, J.S., S.L. Reeder, A.S. Jones, and J. Meline, 2015. OpenSource Software for Visualization and Quality Control of Con-tinuous Hydrologic and Water Quality Sensor Data. Environ-mental Modelling & Software 70:32-44. https://doi.org/10.1016/j.envsoft.2015.04.002.

Horsburgh, J.S., A. Spackman Jones, D.K. Stevens, D.G. Tarboton,and N.O. Mesner, 2010. A Sensor Network for High FrequencyEstimation of Water Quality Constituent Fluxes Using Surro-gates. Environmental Modelling & Software 25:1031-1044.https://doi.org/10.1016/j.envsoft.2009.10.012.

Horsburgh, J.S., D.G. Tarboton, D.R. Maidment, and I. Zaslavsky,2008. A Relational Model for Environmental and WaterResources Data. Water Resources Research 44:W05406. https://doi.org/10.1029/2007WR006392.

Horsburgh, J.S., D.G. Tarboton, D.R. Maidment, and I. Zaslavsky,2011. Components of an Environmental Observatory Informa-tion System. Computers & Geosciences 37:207-218. https://doi.org/10.1016/j.cageo.2010.07.003.

Immerzeel, W.W., L. van Beek, and M. Bierkens, 2010. ClimateChange Will Affect the Asian Water Towers. Science 328:1382-1385. https://doi.org/10.1126/science.1183188.

Integrated Ocean Observing System, 2015. Manual for Real-TimeQuality Control of Dissolved Oxygen Observations.

iUTAH GAMUT Working Group, 2016. Discharge Rating Curve atLogan River near Mendon Road Advanced Aquatic Site(LR_MR_AA). HydroShare. https://www.hydroshare.org/resource/1c41a65ee28a47b0b319070290a30846/, accessed Octo-ber 2016.

iUTAH GAMUT Working Group, 2017a. iUTAH GAMUT NetworkQuality Control Level 1 Data at Provo River near SoapstoneBasic Aquatic (PR_ST_BA). HydroShare. https://www.hydroshare.org/resource/e640b21cb9b04d20a952cf156459a729, accessedOctober 2016.

iUTAH GAMUT Working Group, 2017b. iUTAH GAMUT NetworkRaw Data at Red Butte Creek near Foothill Drive AdvancedAquatic Site (RB_FD_AA). HydroShare. https://www.hydroshare.org/resource/7999192b91bf489c981a1059fc95ebc2/, accessedOctober 2016.

Jones, A.S., J.S. Horsburgh, S.L. Reeder, M. Ram�ırez, and J. Cara-ballo, 2015. A Data Management and Publication Workflow fora Large-Scale, Heterogeneous Sensor Network. EnvironmentalMonitoring and Assessment 187:348. https://doi.org/10.1007/s10661-015-4594-3.

Jones, A.S., D.K. Stevens, J.S. Horsburgh, and N.O. Mesner, 2011.Surrogate Measures for Providing High Frequency Estimates ofTotal Suspended Solids and Total Phosphorus Concentrations.Journal of the American Water Resources Association 47:239-253. https://doi.org/10.1111/j.1752-1688.2010.00505.x.

Kaushal, S.S. and K.T. Belt, 2012. The Urban Watershed Contin-uum: Evolving Spatial and Temporal Dimensions. Urban Ecosys-tems 15:409-435. https://doi.org/10.1007/s11252-012-0226-7.

Kaushal, S.S., P.M. Mayer, P.G. Vidon, R.M. Smith, M.J. Pennino,T.A. Newcomer, S. Duan, C. Welty, and K.T. Belt, 2014. Land



https://doi.org/10.1086/679446

https://doi.org/10.1890/070147

https://doi.org/10.1890/070147

https://doi.org/10.1890/1540-9295(2003) 001%5b0315:DBTRUR%5d2.0.CO;2

https://doi.org/10.1890/1540-9295(2003) 001%5b0315:DBTRUR%5d2.0.CO;2

https://doi.org/10.1002/2014EF000295

https://doi.org/10.4211/hs.7ce47150ee344d4e95f2fd7bb4f660ac

https://doi.org/10.4211/hs.7ce47150ee344d4e95f2fd7bb4f660ac

https://www.hydroshare.org/resource/5f9ec9cf259f454a8ab964e3e8d59131/

https://www.hydroshare.org/resource/5f9ec9cf259f454a8ab964e3e8d59131/

https://www.hydroshare.org/resource/0e00b7cd6e11438d9f5b491dc3b6185c/

https://www.hydroshare.org/resource/0e00b7cd6e11438d9f5b491dc3b6185c/

https://doi.org/10.1007/s11252-016-0538-0

https://doi.org/10.1007/s11252-016-0538-0

https://doi.org/10.1002/2016JG003354

https://doi.org/10.1002/2016JG003354

https://doi.org/10.1021/acs.est.5b04805



https://doi.org/10.1016/j.earscirev.2006.05.001

https://doi.org/10.1093/biosci/biw005

https://doi.org/10.1016/j.envsoft.2011.08.006


https://doi.org/10.4211/hs.61666076104f4306b4ddbf22a3fc52ca

https://doi.org/10.4211/hs.61666076104f4306b4ddbf22a3fc52ca




https://doi.org/10.1029/2007WR006392

https://doi.org/10.1029/2007WR006392

https://doi.org/10.1016/j.cageo.2010.07.003

https://doi.org/10.1016/j.cageo.2010.07.003


https://www.hydroshare.org/resource/1c41a65ee28a47b0b319070290a30846/

https://www.hydroshare.org/resource/1c41a65ee28a47b0b319070290a30846/

https://www.hydroshare.org/resource/e640b21cb9b04d20a952cf156459a729

https://www.hydroshare.org/resource/e640b21cb9b04d20a952cf156459a729

https://www.hydroshare.org/resource/7999192b91bf489c981a1059fc95ebc2/

https://www.hydroshare.org/resource/7999192b91bf489c981a1059fc95ebc2/

https://doi.org/10.1007/s10661-015-4594-3

https://doi.org/10.1007/s10661-015-4594-3

https://doi.org/10.1111/j.1752-1688.2010.00505.x

https://doi.org/10.1007/s11252-012-0226-7

Use and Climate Variability Amplify Carbon, Nutrient, andContaminant Pulses: A Review with Management Implications.Journal of the American Water Resources Association 50:585-614. https://doi.org/10.1111/jawr.12204.

Kennedy, E., 1982. Computation of Continuous Records of Stream-flow. Applications of Hydraulics. U.S. Geological Survey, Tech-niques of Water-Resources Investigations, Book 3, Chapter A13,59 pp. https://pubs.usgs.gov/twri/twri3-a13/.

Kirchner, J.W., X. Feng, C. Neal, and A.J. Robson, 2004. The FineStructure of Water-Quality Dynamics: The (High-Frequency)Wave of the Future. Hydrological Processes 18:1353-1359.https://doi.org/10.1002/hyp.5537.

Kotkin, J., 2013. America’s Growth Corridors: The Key to NationalRevival. New York City, New York. https://www.manhattan-institute.org/html/americas-growth-corridors-key-national-revival-5857.html, accessed September 2016.

Laney, C.M., D.D. Pennington, and C.E. Tweedie, 2015. Filling theGaps: Sensor Network Use and Data-Sharing Practices in Eco-logical Research. Frontiers in Ecology and the Environment 13(7):363-368. https://doi.org/10.1890/140341.

Luce, C.H., J. Abatzoglou, and Z. Holden, 2013. The Missing Moun-tain Water: Slower Westerlies Decrease Orographic Enhance-ment in the Pacific Northwest USA. Science 342:1360-1364.https://doi.org/10.1126/science.1242335.

Lundquist, J.D., N.E. Wayand, A. Massmann, M.P. Clark, F. Lott,and N.C. Cristea, 2015. Diagnosis of Insidious Data Disasters.Water Resources Research 51:3815-3827. https://doi.org/10.1002/2014WR016585.

Mahat, V. and D.G. Tarboton, 2014. Representation of CanopySnow Interception, Unloading and Melt in a ParsimoniousSnowmelt Model. Hydrological Processes 28:6320-6336. https://doi.org/10.1002/hyp.10116.

McDowell, W.H., 2015. NEON and STREON: Opportunities andChallenges for the Aquatic Sciences. Freshwater Science 34:386-391. https://doi.org/10.1086/679489.

McMillen LLC, 2012a. Final Damage Survey Report Cache County,Utah, Logan River — Bank Protection. https://www.cachecounty.org/public-works/current-projects/riverprojects/current-projects/loganriver.html, accessed September 2016.

McMillen LLC, 2012b. Damage Survey Report Cache County, Utah.Logan River — Debris Removal. https://www.cachecounty.org/public-works/current-projects/riverprojects/current-projects/loganriver.html, accessed September 2016.

Melcher, A.A. and J.S. Horsburgh, 2017. An Urban Observatory forQuantifying Phosphorus and Suspended Solid Loads in Com-bined Natural and Stormwater Conveyances. EnvironmentalMonitoring and Assessment 189:285. https://doi.org/10.1007/s10661-017-5974-7.

Montgomery, J., T. Harmon, W. Kaiser, A. Sanderson, C. Haas, R.Hooper, B. Minsker, J. Schnoor, N. Clesceri, W. Graham, and P.Brezonik, 2007. The WATERS Network: An Integrated Environ-mental Observatory Network for Water Resources. EnvironmentalScience & Technology 41(19):6642-6647. https://doi.org/10.1021/es072618f.

Mourad, M. and J.-L. Bertrand-Krajewski, 2002. A Methodfor Automatic Validation of Long Time Series of Data inUrban Hydrology. Water Science and Technology 45:263.LP-270.

Mueller, D., C. Wagner, M. Rehmel, K. Oberg, and F. Rainville,2013. Measuring Discharge with Acoustic Doppler Current Pro-filers from a Moving Boat. Applications of Hydraulics. U.S. Geo-logical Survey, Techniques and Methods 3-A22 version 2, 116pp. http://pubs.usgs.gov/tm/3a22/.

Ogata, E.M. and M.A. Baker, 2016. Biofilm Response to NutrientEnrichment in Urbanizing Streams: Nutrient Diffusing Sub-strate Bioassay. HydroShare. https://www.hydroshare.org/resource/6612b5beff044810b83d56c4db451a85/, accessedOctober 2016.

Parr, T.W., M. Ferretti, I.C. Simpson, M. Forsius, and E. Kov�acs-L�ang, 2002. Towards a Long-Term Integrated Monitoring Pro-gramme in Europe: Network Design in Theory and Practice.Environmental Monitoring and Assessment 78:253-290. https://doi.org/10.1023/A:1019934919140.

Paul, M.J. and J.L. Meyer, 2001. Streams in the Urban Landscape.Annual Reviews in Ecology and Systematics 32:333-365.https://doi.org/10.1146/annurev.ecolsys.32.081501.114040.

Pellerin, B.A., B.A. Stauffer, D.A. Young, D.J. Sullivan, S.B.Bricker, M.R. Walbridge, G.A. Clyde, and D.M. Shaw, 2016.Emerging Tools for Continuous Nutrient Monitoring Networks:Sensors Advancing Science and Water Resources Protection.Journal of the American Water Resources Association 20460:1-16. https://doi.org/10.1111/1752-1688.12386.

Peters, D., H. Loescher, M. SanClements, and K. Havstad, 2014.Taking the Pulse of a Continent: Expanding Site-BasedResearch Infrastructure for Regional- to Continental-Scale Ecol-ogy. Ecosphere 5:1-23. https://doi.org/10.1890/ES13-00295.1.

Porter, J.H., P.C. Hanson, and C. Lin, 2012. Staying Afloat in theSensor Data Deluge. Trends in Ecology & Evolution 27:121-129.https://doi.org/10.1016/j.tree.2011.11.009.

Rantz, S.E., 1982. Measurement and Computation of Streamflow.U.S. Geological Survey Water Supply Paper 2175, 373 pp.http://pubs.usgs.gov/wsp/wsp2175/.

Reed, M.L., J.L. Pinckney, C.J. Keppler, L.M. Brock, S.B. Hogan,and D.I. Greenfield, 2016. The Influence of Nitrogen and Phos-phorus on Phytoplankton Growth and Assemblage Compositionin Four Coastal, Southeastern USA Systems. Estuarine, Coastaland Shelf Science 177:71-82. https://doi.org/10.1016/j.ecss.2016.05.002.

Rode, M., A.J. Wade, M.J. Cohen, R.T. Hensley, M.J. Bowes, J.W.Kirchner, G.B. Arhonditsis, P. Jordan, B. Kronvang, S.J. Halli-day, R.A. Skeffington, J.C. Rozemeijer, A.H. Aubert, K. Rinke,and S. Jomaa, 2016. Sensors in the Stream: The High-Fre-quency Wave of the Present. Environmental Science & Technol-ogy 50(19):10297-10307. https://doi.org/10.1021/acs.est.6b02155.

Rundel, P.W., E.A. Graham, M.F. Allen, J.C. Fisher, and T.C. Har-mon, 2009. Environmental Sensor Networks in EcologicalResearch. New Phytologist 182:589-607. https://doi.org/10.1111/j.1469-8137.2009.02811.x.

Ryan, R.J., C. Welty, and P.C. Larson, 2010. Variation in SurfaceWater — Groundwater Exchange with Land Use in an UrbanStream. Journal of Hydrology 392:1-11. https://doi.org/10.1016/j.jhydrol.2010.06.004.

Scalzitti, J., C. Strong, and A.K. Kochanski, 2016. A 26 Year High-Resolution Dynamical Downscaling over the Wasatch Moun-tains: Synoptic Effects on Winter Precipitation Performance.Journal of Geophysical Research 121:3224-3240. https://doi.org/10.1002/2015JD024497.A.

Schmadel, N.M., B.T. Neilson, and D.K. Stevens, 2010. Approachesto Estimate Uncertainty in Longitudinal Channel Water Bal-ances. Journal of Hydrology 394:357-369. https://doi.org/10.1016/j.jhydrol.2010.09.011.

Stanford, J.A. and J.V. Ward, 2001. Revisiting the Serial Disconti-nuity Concept. Regulated Rivers: Research & Management17:303-310. https://doi.org/10.1002/Rrr.659.Abs.

Strachan, S., E.P. Kelsey, R.F. Brown, S. Dascalu, F. Harris, G.Kent, B. Lyles, G. McCurdy, D. Slater, and K. Smith, 2016.Filling the Data Gaps in Mountain Climate Observatoriesthrough Advanced Technology, Refined Instrument Siting,and a Focus on Gradients. Mountain Research and Develop-ment 36:518-527. https://doi.org/10.1659/MRD-JOURNAL-D-16-00028.1.

Strobl, R.O. and P.D. Robillard, 2008. Network Design for WaterQuality Monitoring of Surface Freshwaters: A Review. Journalof Environmental Management 87:639-648. https://doi.org/10.1016/j.jenvman.2007.03.001.



https://doi.org/10.1111/jawr.12204

https://pubs.usgs.gov/twri/twri3-a13/


https://www.manhattan-institute.org/html/americas-growth-corridors-key-national-revival-5857.html



https://doi.org/10.1890/140341


https://doi.org/10.1002/2014WR016585

https://doi.org/10.1002/2014WR016585



https://doi.org/10.1086/679489

https://www.cachecounty.org/public-works/current-projects/riverprojects/current-projects/loganriver.html






https://doi.org/10.1007/s10661-017-5974-7

https://doi.org/10.1007/s10661-017-5974-7

https://doi.org/10.1021/es072618f

https://doi.org/10.1021/es072618f

http://pubs.usgs.gov/tm/3a22/

https://www.hydroshare.org/resource/6612b5beff044810b83d56c4db451a85/

https://www.hydroshare.org/resource/6612b5beff044810b83d56c4db451a85/

https://doi.org/10.1023/A:1019934919140

https://doi.org/10.1023/A:1019934919140

https://doi.org/10.1146/annurev.ecolsys.32.081501.114040

https://doi.org/10.1111/1752-1688.12386

https://doi.org/10.1890/ES13-00295.1

https://doi.org/10.1016/j.tree.2011.11.009

http://pubs.usgs.gov/wsp/wsp2175/

https://doi.org/10.1016/j.ecss.2016.05.002

https://doi.org/10.1016/j.ecss.2016.05.002

https://doi.org/10.1021/acs.est.6b02155

https://doi.org/10.1111/j.1469-8137.2009.02811.x

https://doi.org/10.1111/j.1469-8137.2009.02811.x

https://doi.org/10.1016/j.jhydrol.2010.06.004


https://doi.org/10.1002/2015JD024497.A

https://doi.org/10.1002/2015JD024497.A



https://doi.org/10.1002/Rrr.659.Abs

https://doi.org/10.1659/MRD-JOURNAL-D-16-00028.1

https://doi.org/10.1659/MRD-JOURNAL-D-16-00028.1

https://doi.org/10.1016/j.jenvman.2007.03.001

https://doi.org/10.1016/j.jenvman.2007.03.001

Sutter, R.D., S.B. Wainscott, J. Boetsch, C. Palmer, and D. Rugg,2015. Practical Guidance for Integrating Data Management intoLong-Term Ecological Monitoring Projects. Wildlife Society Bul-letin 39:451-463. https://doi.org/10.1002/wsb.548.

Thorpe, A., D. Barnett, S. Elmendorf, E. Hinckley, D. Hoekman, K.Jones, C. Meier, J. Parnell, L. Stanish, and K. Thibault, 2015.Introduction to the Sampling Designs of the National EcologicalObservatory Network (NEON) Terrestrial Observation System.Ecosphere 7:1-11. https://doi.org/10.1002/ecs2.1627.

Turnipseed, D. and V. Sauer, 2010. Discharge Measurements atGaging Stations. U.S. Geological Survey Techniques and Meth-ods 3-A8, 87 pp. http://pubs.usgs.gov/tm/tm3-a8.

UDWQ, 2016. Utah’s 2016 Draft Integrated Report, 298 pp. UtahDepartment of Environmental Quality, Salt Lake City, Utah.http://www.deq.utah.gov/ProgramsServices/programs/water/wqmanagement/assessment/currentIR2016.htm, accessed Septem-ber 2016.

Vano, J., B. Udall, D. Cayan, J. Overpeck, L. Brekke, T. Das, H.Hartmann, H. Hidalgo, M. Hoerling, G. McCabe, K. Morino, R.Webb, K. Werner, and D. Lettenmaier, 2014. UnderstandingUncertainties in Future Colorado River Streamflow. Bulletin ofthe American Meteorological Society 95:59-78. https://doi.org/10.1175/BAMS-D-12-00228.1.

Viviroli, D., H.H. Du, B. Messerli, M. Meybeck, and R. Weingart-ner, 2007. Mountains of the World, Water Towers for Humanity:Typology, Mapping, and Global Significance. Water ResourcesResearch 43:1-13. https://doi.org/10.1029/2006WR005653.

Wagner, R.J., R.W. Boulger, C.J. Oblinger, and B.A. Smith, 2006.Guidelines and Standard Procedures for Continuous Water-Quality Monitors: Station Operation, Record Computation, andData Reporting. U.S. Geological Survey Techniques and Meth-ods 1-D3. http://pubs.usgs.gov/tm/2006/tm1D3/.

Walsh, C.J., D.B. Booth, M.J. Burns, T.D. Fletcher, R.L. Hale, L.N.Hoang, G. Livingston, M.A. Rippy, A.H. Roy, M. Scoggins, andA. Wallace, 2016. Principles for Urban Stormwater Managementto Protect Stream Ecosystems. Freshwater Science 35:398-411.https://doi.org/10.1086/685284.

Walsh, C.J., A.H. Roy, J. Feminella, P. Cottingham, P.M. Groff-man, and R.P. Morgan, 2005. The Urban Stream Syndrome:Current Knowledge and the Search for a Cure. Journal of NorthAmerican Benthological Society 24:706-723. https://doi.org/10.1899/04-028.1.

Ward, J. and J. Stanford, 1983. The Serial Discontinuity Conceptof Lotic Ecosystems. In: Dynamics of Lotic Ecosystems, T. Fon-taine and S. Bartell (Editors). Ann Arbor Scientific Publishers,Ann Arbor, Michigan, pp. 29-42.

Wilson, H.F., J.E. Saiers, P.A. Raymond, and W.V. Sobczak, 2013.Hydrologic Drivers and Seasonality of Dissolved Organic CarbonConcentration, Nitrogen Content, Bioavailability, and Export ina Forested New England Stream. Ecosystems 16:604-616.https://doi.org/10.1007/s10021-013-9635-6.

Wise, E.K., 2012. Hydroclimatology of the US Intermountain West.Progress in Physical Geography 36:458-479. https://doi.org/10.1177/0309133312446538.

Wolman, M.G. and A.P. Schick, 1967. Effects of Construction onFluvial Sediment, Urban and Suburban Areas of Maryland.Water Resources Research 3:451-464. https://doi.org/10.1029/WR003i002p00451.

Woods, A., D. Lammers, S. Bryce, J. Omernik, R. Denton, M. Dome-ier, and J. Comstock, 2001. Ecoregions of Utah (Color Poster withMap, Descriptive Text, Summary Tables, and Photographs). U.S.Geological Survey, Reston, Virginia. https://www.epa.gov/eco-research/ecoregion-download-files-state-region-8#pane-42.

World Meteorological Organization, 2008. Guide to MeteorologicalInstruments and Methods of Observation. WMO-No 8. Geneva,Switzerland. http://www.wmo.int/pages/prog/www/IMOP/CIMO-Guide.html, accessed January 2015.

Xylem, 2012. EXO User Manual, Item# 603789REF, Revision F.YSI Incorporated, Yellow Springs, Ohio. https://www.ysi.com/File Library/Documents/Manuals/EXO-User-Manual-Web.pdf,accessed January 2015.

Zaslavsky, I., D. Valentine, and T. Whiteaker, 2007. CUAHSIWaterML. Open Geospatial Consortium Discussion Paper OGC07-041r1, 88 pp.



https://doi.org/10.1002/wsb.548

https://doi.org/10.1002/ecs2.1627

http://pubs.usgs.gov/tm/tm3-a8

http://www.deq.utah.gov/ProgramsServices/programs/water/wqmanagement/assessment/currentIR2016.htm

http://www.deq.utah.gov/ProgramsServices/programs/water/wqmanagement/assessment/currentIR2016.htm

https://doi.org/10.1175/BAMS-D-12-00228.1

https://doi.org/10.1175/BAMS-D-12-00228.1

https://doi.org/10.1029/2006WR005653

http://pubs.usgs.gov/tm/2006/tm1D3/

https://doi.org/10.1086/685284

https://doi.org/10.1899/04-028.1

https://doi.org/10.1899/04-028.1

https://doi.org/10.1007/s10021-013-9635-6

https://doi.org/10.1177/0309133312446538

https://doi.org/10.1177/0309133312446538

https://doi.org/10.1029/WR003i002p00451

https://doi.org/10.1029/WR003i002p00451

https://www.epa.gov/eco-research/ecoregion-download-files-state-region-8#pane-42

https://www.epa.gov/eco-research/ecoregion-download-files-state-region-8#pane-42

http://www.wmo.int/pages/prog/www/IMOP/CIMO-Guide.html

http://www.wmo.int/pages/prog/www/IMOP/CIMO-Guide.html

https://www.ysi.com/File

https://www.ysi.com/File