Calibration of a water content reflectometer and soil water dynamics for an agroforestry practice

Download Calibration of a water content reflectometer and soil water dynamics for an agroforestry practice

Post on 14-Jul-2016

215 views

Category:

Documents

1 download

TRANSCRIPT

Calibration of a water content reflectometer and soil waterdynamics for an agroforestry practiceRanjith P. Udawatta Stephen H. Anderson Peter P. Motavalli Harold E. GarrettReceived: 9 March 2010 / Accepted: 19 November 2010 / Published online: 4 December 2010 Springer Science+Business Media B.V. 2010Abstract Water content reflectometers allow tem-poral and continuous assessment of spatial differ-ences in soil water dynamics. We hypothesized thatvolumetric soil water content estimated by the watercontent reflectometers (CS616 Campbell Sci. Inc.,Logan, UT) is influenced by clay content andtemperature and therefore site- and or soil-specificequations are required for accurate estimations of soilwater. Objectives of the study were to developcalibration equations and to evaluate soil waterdynamics for an agroforestry system using theimproved calibration equation. Putnam silt loam(fine, smectitic, mesic Vertic Albaqualfs) and Menfrosilt loam (fine-silty, mixed, superactive, mesic TypicHapludalfs) soils were selected with 2354% clay.Soils were packed in cylinders and sensors weremonitored at 5, 10, 15, 20, 25, 30, 35, and 40C.Calibration equations for volumetric water content(hv) as a function of sensor measured period,temperature, and clay content were developed. Coef-ficient of determination (r2) and root mean squareerror (RMSE) were used to compare goodness of fit.RMSE varied between 0.028 and 0.040 m3 m-3 forsoil specific and soil-combined linear and quadraticequations with period. Coefficients of determinationranged between 0.89 and 0.96 for these calibrations.RMSE decreased and r2 increased as temperature wasincluded. The effect of temperature varied with watercontent, with the strongest effect at high watercontents. Clay content did not contribute significantlyto improve predictability. Water content estimated bythe linear calibration equation with period andtemperature showed differences in hv influenced byvegetation and soil depth, and closely followedprecipitation events and water use by vegetation.The field study showed significant differencesbetween the two treatments. Also the importance oftemperature correction is emphasized during periodswith large diurnal fluctuations and site specificcalibration equations. Results of the study showedthat water content reflectometers can be used toestimate hv with less than 4% error and may needsite specific calibration and a temperature correctionto research more precise estimates.Keywords Cornsoybean CS616 CS615 Sensitivity analyses Soil water sensorsAbbreviationsCEC Cation exchange capacityEC Electrical conductivityK Dielectric constantR. P. Udawatta (&) H. E. GarrettCenter for Agroforestry, University of Missouri,203 Anheuser-Busch Natural Resources Building,Columbia, MO 65211, USAe-mail: UdawattaR@missouri.eduR. P. Udawatta S. H. Anderson P. P. MotavalliDepartment of Soil, Environmental and AtmosphericSciences, University of Missouri, 302 Anheuser-BuschNatural Resources Building, Columbia, MO 65211, USA123Agroforest Syst (2011) 82:6175DOI 10.1007/s10457-010-9362-3hv Volumetric soil water content m3 m-3RMSE Root mean square errorC PeriodTDR Time domain reflectometryWCR Water content reflectometerIntroductionAccurate and continuous estimation of soil watercontent is important in many plantsoilwater andhydrologic studies. Gravimetric, nuclear, electromag-netic, and tensiometer methods can be used toestimate soil water content (Zazueta and Xin 1994).Capacitance sensors (Dean et al. 1987; Kellenerset al. 2004a), impedance sensors (Hilhorst et al. 1993;Seyfried and Murdock 2004), and transmission lineoscillators (Campbell and Anderson 1998) are elec-tromagnetic approaches to measure soil water contentwhich are often preferred over neutron probe meth-odology. Relatively inexpensive CS616 water contentreflectometer (WCR) sensors (Campbell Sci. Inc.,Logan, UT), a type of a transmission line oscillator,uses a technique similar to time domain reflectom-eters (TDR) but does not require a separate pulse andsampling unit (Kelleners et al. 2005). WCR areincreasingly being used in field and laboratoryexperiments to research water balance, plant wateruse, irrigation, precision farming, and movement ofchemicals and ions (Seyfried and Murdock 2001,2004; Seobi et al. 2005; Anderson et al. 2009). Somepossible reasons for the preference for these units areease of installation, fewer regulatory and safetyconcerns, and cost effectiveness. Data can be col-lected continuously and either stored on-site ortransmitted to a remote computer via a telephone orradio line (Seyfield and Murdock 2001). Therefore,they are easier to use in an in-field monitoringsystem.In WCR sensors, two wave guides 30 cm long and0.32 cm diameter with a 3.2 cm spacing are attachedto a probe head with embedded circuitry; thusallowing an increase in the distance between thesensor and a data logger (Seyfried and Murdock2001; Chandler et al. 2004). Inside the probe head,voltage pulses are generated and the reflected pulsetriggers the next pulse. The output is proportional tothe number of reflections per second. Reflections aredivided by a scaling factor which can be read by adata logger as period. Sensors can be verticallyinstalled to estimate integrated soil profile watercontent or horizontally to measure water content bysoil depth.The wide disparity between dielectric permittivity(k) of air (1), soil (2.43.5), and water (80) is used tomeasure water content; thus it is an indirect mea-surement of soil water content. The WCR techniquemeasures equilibrium oscillation frequency or periodof an applied voltage, which is directly related tok. The travel time varies with the k of the medium inwhich the wave guide is inserted (Fellner-Feldegg1969). With an empirical calibration equation, themeasured wave period in microseconds is then relatedto volumetric soil water content (hv; Chandler et al.2004). Dielectric permittivity also varies with tem-perature. For example, dielectric constants are 87.9,78.4, and 55.6 for water at 0, 25, and 100C,respectively, and 1.0059 for 100C air.Manufacturer provided calibration estimates watercontent in sand reasonably well (Seyfried and Mur-dock 2001). In contrast, studies have shown thatfactory calibration overestimates soil water content inmany soils (Seyfried and Murdock 2001; Quinoneset al. 2003; Stangl et al. 2009). The WCR sensors use1545 MHz frequency range to estimate hv (Seyfriedand Murdock 2001) where TDR probes use up toabout 1 GHz (Or and Wraith 1999). The frequencyrange used in the WCR sensors is affected byvariations in clay content, clay type, and soilelectrical-conductivity (Campbell 1990; Seyfriedand Murdock 2004). However, the effect of claycontent can be corrected by using simple linear orquadratic functions (Chandler et al. 2004). Further-more, due to this low frequency range, WCRestimates are often affected by temperature andrequires soil specific calibrations (Seyfried andMurdock 2001; Chandler et al. 2004).Clay content, especially soils containing smectiticclays found in subsurface horizons of poorly-drainedclaypan soils of Major Land Resource Area 113(USDA-NRCS 1998) or in other regions, may affectWCR readings. Smectitic clays have relatively highsurface charge which may attenuate the signal fromWCR sensors and affect their ability to estimateprofile hv. In addition, soils with high smectitic claycan undergo as much as 30% volume change due towetting and drying. Therefore, soils characterized by62 Agroforest Syst (2011) 82:6175123clay-rich subsurface horizons affect water movement,retention, and hv. These soils often retain water forextended periods of time and their shrinkage crackswill seal during wet periods, or channel flow throughthese cracks and infiltrate soil water during dryperiods. Therefore, soil profiles with relatively highwater infiltration in surface horizons and relativelylow infiltration in subsurface horizons may needindividual calibration equations to better understandwater dynamics. In support of this, Serrarens et al.(2000) observed that the measurement error doubledwhen a single calibration equation was used in a TDRcalibration study with six soil depths.A good and accurate understanding of plantwateruse, hydrologic relationships, and soil water dynam-ics are especially important in agroforestry alleycropping practices where grass, trees, and crops mayshare the same area. Roots of this mixed vegetationmay occupy the same soil volume but with differentdensities and distribution patterns. Trees and grass inagroforestry alley cropping practices use soil waterfrom different soil depths over a longer period ascompared to crop plants since annual crops haverelatively shallow root systems and shorter growingseasons than the grass, trees, and shrubs in agrofor-estry practices (Udawatta et al. 2005; Fernandez et al.2008). Lack of good calibration relationshipsbetween sensor output and hv may restrict accurateprediction of soil water dynamics in multi-speciespractices such as agroforestry practices. The objec-tives of this study were to: (1) examine clay contentand temperature effects on WCR hv readings, (2)develop calibration equations for hv, in soils havingmoderate to high clay content, and (3) evaluateseasonal soil water dynamics for an agroforestrypractice using improved calibration equations thattake into account soil and site factors.Materials and methodsSoil materialsTwo soil types were selected, Putnam soil from theclaypan region and Menfro soil from the Mississippivalley wooded slopes region, to represent a range inclay content. The selected two soils have clay contentvarying from 23 to 54% (Table 1). Bulk soil materialwas obtained from the A and Bt horizons of a Putnamsilt loam (fine, smectitic, mesic, Vertic Albaqualfs) atthe Paired-Watershed study at the Greenley MemorialResearch Center, in Novelty, MO (40010N,92110W). These soils have a montmorillonitic claymineralogy. Bulk soil was also obtained from the Ahorizon of a Menfro silt loam (fine-silty, mixed,superactive, mesic Typic Hapludalfs) at the Horti-culture and Agroforestry Research Center in NewFranklin, MO (HARC; 39010N, 92450W). The claymineralogy is mixed but dominated by montmoril-lonite (est. 6075%) with lesser amounts of illite(http://www2.ftw.nrcs.usda.gov/osd/dat/M/MENFRO.html). Soil texture, pH (1:1 soil:water), cationexchange capacity (CEC), and electrical conductivity(EC; 1:1 soil:water) were determined at the Universityof Missouri Soil Characterization Laboratory usingstandard methods for soil survey (Soil Survey Staff1984).The poorly drained Putnam silt loam soil occurs inthe northeast region of Missouri. Most areas with thissoil are used for cultivation of corn (Zea mays L.),soybean (Glycine max (L.) Merr.), and other graincrops. The deep, well-drained Menfro silt loam soiloccurs along loess bluffs near the Missouri andMississippi Rivers. Most areas with this soil are usedfor pasture and some areas for grain crops and nativehardwoods. Both Putnam and Menfro are developedTable 1 Selected soil properties for Putnam and Menfro soils used in the laboratory experimentSoil Depth, cm Particle size analysisClay, % Silt, % Sand, % Finesilt, %TexturalclassCEC,cmol kg-1pHCaCl2 pHH2O Electricalconductivity,dS m-1Putnam A 010 23.4 71.4 5.2 46.4 Silt loam 18.0 6.4 5.8 0.12Menfro A 010 32.6 63.8 3.6 28.1 Silty clay loam 22.7 6.4 5.1 0.23Putnam Bt 3050 53.9 43.4 2.7 27.0 Silty clay 37.9 5.3 4.8 0.31Agroforest Syst (2011) 82:6175 63123in loess material. Putnam is underlain by glacialmaterial.Laboratory calibrationTwo horizons from Putnam soil (010 and 3050 cmdepths) and one horizon from Menfro soil (010 cm)were used for the calibration (Table 1). Nine (somewith eight) volumetric soil water content values(0.05, 0.10, 0.15, 0.20, 0.25, 0.30, 0.35, 0.40, and0.45 m3 m-3) and eight temperature levels (5, 10, 15,20, 25, 30, 35 and 40C) with three replicates wereused in the laboratory study. Model CS616 WCRsensors (Campbell Sci. Inc., Logan, UT) were usedfor the study. The hv treatment values were targetvalues with actual hv verified after the laboratorystudy was completed. Initial soil water content wasdetermined and then measured amounts of water wereadded and thoroughly mixed to obtain the desired hv.Pre-determined weights for each 5 cm increment ofsoil were packed to a desired bulk density of1.25 Mg m-3 in 54 cm long and 10 cm diameterpolyvinyl chloride (PVC) cores with sealed bottoms.WCR sensors were inserted and the openings of thePVC tubes were sealed with four layers of saran wrapand duct tape. Sealed soil PVC columns were placedhorizontally on a laboratory cart to facilitate transportto a walk-in, temperature-controlled environmentalchamber. Gravimetric soil moisture percentage at thebeginning and end were evaluated to assure nomoisture loss during the study.Sensors were attached to a multiplexer (ModelAM16/32; Campbell Sci. Inc., Logan, UT) and themultiplexer was attached to a datalogger (ModelCR23X-4 m; Campbell Sci. Inc., Logan, UT) torecord data at 10-min intervals. The unit was poweredby a 12 V deep cycle marine battery. At eachtemperature, volumetric water content estimated bythe manufacturer provided equation and periodreadings were collected for two consecutive daysafter soil inside the core reached the specifiedtemperature before starting another horizon.hv 0:0663 0:0063 C 0:0007 C2 1where hv is volumetric water content estimated by themanufacturer-provided equation and C is period (ls).Period and hv estimated by the manufacturer-provided equation (Eq. 1; Campbell Sci. Inc. (2002)were downloaded to a laptop computer for analysis.At the end of each run, three soil samples from eachPVC core were oven-dried to determine gravimetricwater content. The gravimetric water contents weremultiplied by the bulk density to determine volumet-ric water content. This experimentally measured hvwas used to compare with sensor-measured periodand estimated hv. Relationships between independentvariables (period, temperature, clay content) andexperimentally-measured hv were developed usinglinear and quadratic regressions for each soil horizonand all three horizons combined (Eqs. 27; SAS Inst.1989). Initially period was used in a linear form andthen in a quadratic form. Subsequently temperaturewas also incorporated in linear and quadratic forms.The accuracy of the manufacturer-provided calibra-tion equation was compared with the measured hvvalues. Coefficients of determination and RMSEwere used to evaluate calibration equations for eachsoil and all three soils combined to determine themost suitable equations for the studied soils. Thefollowing relationships were evaluated between theexperimentally measured hv versus period, tempera-ture, and clay% in various forms.hv b0 b1 C 2hv b0 b1 C b2 C2 3hv b0 b1 C b2 Temp 4hv b0 b1 C b2 C2 b3 Temp 5hv b0 b1 C b2 C2 b3 Temp b4 Temp2 6hv b0 b1 C b2 Temp b3 clay% 7where; hv is experimentally measured volumetricwater content, C is period, Temp temperature, andclay% clay content (g/g * 100%).Comparison of vegetation, depth, and temperatureeffects on water contentThis research utilized an on-going long-term pairedwatershed study located at the Greenley ResearchCenter in Novelty, MO. This study is examining theeffects of agroforestry and grass vegetative bufferstrips on water quality in three adjacent watershedswith row crop agriculture (Udawatta et al. 2002,2004, 2006). CS616 WCR sensors and 107B soiltemperature probes (Campbell Sci. Inc., Logan, UT)64 Agroforest Syst (2011) 82:6175123were installed at four sites within the vegetativebuffers and row crop areas and were placed horizon-tally at four depths (5, 10, 20, and 40 cm) at each site.The predominant soil for the watershed study wasPutnam silt loam (fine, smectitic, mesic VerticAlbaqualfs) which contains a distinct argillic horizonat depths varying from 10 to 85 cm across thewatersheds. Additional details on soil, site, andexperimental details can be found in Udawatta et al.(2002). Sensors were attached to a multiplexer andthe multiplexer was attached to a data logger torecord period, hv estimated by the manufacturerprovided equation, and soil temperature at 10-minintervals.Soil samples were collected from the field siteduring dry periods and wet periods to examine thefitness of the equations developed in the currentstudy. Gravimetric water content was measured andthese values were converted to volumetric watercontents by multiplying the bulk density. The periodvalues and water contents estimated by the manufac-turer provided equation were recorded at the time ofsoil sampling. The period readings were converted tovolumetric water contents by the Eq. 8 developed inthis study.hv 0:311 0:0193 C 8To understand vegetation, depth, and temperatureeffects on soil water dynamics, data were collectedfrom March 12 to November 19, 2007. Period, hv, andtemperature data were downloaded to a laptop com-puter for subsequent analysis. Weekly period valueswere extracted at 12:00 noon on each seventh daystarting from March 12. The period was converted tohv using the linear Eq. 9 developed in the laboratorystudy (Eq. 4) to compare soil water dynamics betweenthe two management treatments and depths.hv 0:283 0:0199 C 0:00198 Temp 9where, hv is volumetric water content estimated bythe linear equation developed in this study, C isperiod (ls), and Temp soil temperature (C).The effect of diurnal temperature fluctuations onestimated water content was compared in the rowcrop area at the 5 cm depth at 30 min intervals forJuly 24, 2007 (date was selected due to the highestdifference in temperature readings during a day forthe year). CS616 sensor-measured period values wereconverted to hv with the following four equations:1. Manufacturer-provided quadratic Eq. 1,2. Manufacturer-provided quadratic Eq. 1 using themanufacturer-provided temperature correctedperiod (9),Cc Cunc 20 Temp 0:526 0:052 Cunc 0:00136 C2unc 10where, Cc = corrected period, Cunc = uncorrectedperiod3. Calibration Eq. 8 developed in this study withperiod only and,4. Calibration Eq. 9 developed in this study withperiod and temperature.Differences in hv between row crop and agrofor-estry treatments and by soil depth were declaredsignificant at the a = 0.05 level using least signifi-cance difference (LSD) at each measured date.Results and discussionSoil propertiesProperties for the collected bulk soils for the labo-ratory study differed in clay and silt content and otherchemical properties (Table 1). The textural classes ofthe Putnam soil horizons from the Greenley Centerwere silt loam and silty clay, respectively, while theMenfro soil from the HARC in New Franklin was asilty clay loam. Soil clay content, fine silt content,EC, CEC, and pH measured in 0.01 M CaCl2 and d-H2O varied among horizons. The surface horizon ofthe Putnam soil had 23.4% clay. The Putnam Bthorizon (3050 cm) had 2.3 times more clay (53.9%)than the Putnam surface A horizon (010 cm). CECand EC of the Putnam Bt horizon were higher thanthe Putnam surface horizon but soil pHCaCl2 andpHH2O were lower.The 32.6% clay content of the Menfro A horizon(010 cm) was in-between the clay contents of thetwo Putnam soil horizons. CEC followed a similarpattern. Values of pHCaCl2 for the Putnam surface soiland the Menfro soil were similar. However, thePutnam surface soil had a higher pHH2O than that ofthe Menfro soil. Soil organic carbon content of thesesoils were \3%. Organic matter can affect theAgroforest Syst (2011) 82:6175 65123dielectric response of soil (Campbell Sci. Inc. 2002).A study by Khele et al. (2008) observed a linearincrease of dielectric constant from 3.1 to 7.6 withinthe 811 GHz range when organic matter content wasincreased from 0 to 20%.Measured and sensor-estimated water contentThe manufacturer-provided equation estimated sig-nificantly higher water content as compared tomeasured values, especially for higher hv (Fig. 1).The difference between measured- and estimated-hvincreased with increasing water content for all threesoils irrespective of clay content or other physicalproperties. At the highest water content levels withineach soil, the manufacturer-provided equation esti-mated 0.65, 0.70, and 0.77 m3 m-3 hv for Putnam A,Menfro A, and Putnam Bt, respectively, whilemeasured values were 0.46, 0.41, and 0.45 m3 m-3hv, respectively. The bulk density of these soil coreswas 1.25 Mg m-3 and therefore the porosity wasapproximately 0.52 m3 m-3. The manufacturer-pro-vided equation estimated 25, 35, and 48% morewater-filled pore volumes than the estimated soilporosity for Putnam A, Menfro A, and Putnam Btrespective soils. Therefore, the manufacturer-providedequation was less suited to estimate soil water in thesesoils. Similar to the results of this study, in a soilmoisture sensor comparison research, Walker et al.(2004) observed that CS615 sensors predicted greatersoil water content near saturation than the soilporosity. In a laboratory calibration study, Stengeret al. (2005) observed 1519% overestimation inAustralia with manufacturer-provided equation.VWC = -0.31 + 0.019*period r = 0.92Period (microsec)VWC = -0.31 + 0.019*period r = 0.920.00.10.20.30.40.50.60.70.8Volumetric Water Content (m3m-3 )VWC = -0.34 + 0.019*period r = 0.890.00.10.20.30.40.50.60.70.815 20 25 30 35 40Volumetric Water Content (m3m-3 )Period (microsec)VWC = -0.28 + 0.018*period r = 0.96Menfro APutnam A Putnam BtCombined15 20 25 30 35 40Fig. 1 Relationshipsbetween sensor-measuredperiod and volumetric soilwater content (VWC) forPutnam A, Menfro A, andPutnam Bt and all threehorizons combined. Filledand empty circles denotemeasured andmanufacturer-providedcalibration estimated watercontent, respectively. Thedistribution of sensor-measured period for eachmeasured volumetric watercontent indicates thetemperature effect66 Agroforest Syst (2011) 82:6175123Stangl et al. (2009) also noticed overestimationanomalies as high as 67%. In contrast, soil watercontent was underestimated for volcanic soils (Stengeret al. 2005). Capacitance sensors behave the sameway, water content was underestimated at lower hv andoverestimated at higher hv; as high as 80% (Kellenerset al. 2004b).Individual sensor precision appears to be high asindicated by the narrow range in measured values. Allsensors responded to differences in hv similarlyamong the three horizons. The three replicatedsensors for each hv had a very small standard errorfor period (\0.02), for values between 17 and 40 ls.Similar to our results Stangl et al. (2009) alsoobserved low sensor variability for CS 615. Whensensors were removed it was also noticed that none ofthe rods were either bent or tips were touching tocause larger differences in reading.Sensor measured period had a significant relation-ship with measured soil water for the three soilsindividually and the three soils combined (Table 2;Fig. 1). Period alone accounted for 8996% of thevariation in measured-hv of individual soils in linearand quadratic calibration equations (Table 2). Amongthe three soils, the Putnam Bt soil had the bestrelationship (r2 = 0.96 and RMSE = 0.028 m3 m-3)between period and measured-hv. A quadratic regres-sion equation improved the r2 of Menfro soil, but therewas no change for Putnam soils as compared to thelinear calibration. This corroborates with previousstudies which showed that differences between linearversus quadratic and cubic calibrations were eithersmall or did not improve estimated water contents(Chandler et al. 2004; Stenger et al. 2005; Stangl et al.2009). Table 2 shows that linear and quadratic cali-brations estimate hv within 0.0280.040 m3 m-3 forthese three soils.Slopes and intercepts among the three soils of thelinear calibration were not significantly different andtherefore sensor-measured period and hv were com-bined to develop a single calibration for the threesoils (Table 2; Fig. 1). Sensor period in linear andquadratic calibrations explained 92% of the variationin hv for the three soils combined. This 92%predictability could be attributed to several factors.The relationship between k and hv is directly propor-tional to the free water content such as with sand(Zazueta and Xin 1994; Kelleners et al. 2004a). Thesoils used in this study contained up to 54% clay. Infine textured soils, the presence of bound watercauses high dielectric loss affecting measured hv(Jones et al. 2005; Kelleners et al. 2004b). Accordingto Chandler et al. (2004) and Seyfried and Murdock(2001), variation between sensors due to scatter alsoaffects the reading. In their studies, calibration ofindividual sensors reduced the scatter and improvedthe predictability. In the current study packing alsomay have contributed to sensor readings. Compac-tion, air gaps, porosity, and spatial variability affecttravel time and thereby the measured hv (Serrarenset al. 2000; Vaz and Hopmans 2001; Stangl et al.2009).Table 2 Relationshipsbetween CS616 sensor-measured period (C; inmicrosec) and measuredvolumetric soil watercontent (m3 m-3) forPutnam A, Menfro A, andPutnam Bt and all threehorizons combinedRegression Relationship Coefficient ofdetermination, r2Significancelevel,P [ FRoot meansquareerror, m3 m-3Putnam Ahv = -0.309 ? 0.0197 * C 0.92 0.001 0.039hv = -0.387 ? 0.0259 * C - 0.0011 * C2 0.92 0.001 0.039Menfro Ahv = -0.339 ? 0.0198 * C 0.89 0.001 0.040hv = -0.0673 - 0.0115 * C ? 0.0058 * C2 0.91 0.001 0.038Putnam Bthv = -0.283 ? 0.0183 * C 0.96 0.001 0.028hv = -0.282 ? 0.0182 * C ? 0.00002 * C2 0.96 0.001 0.029Combinedhv = -0.311 ? 0.0193 * C 0.92 0.001 0.038hv = -0.283 ? 0.0182 * C ? 0.00002 * C2 0.92 0.001 0.039Agroforest Syst (2011) 82:6175 67123Root mean square error (RMSE) for linear andquadratic equations with period was 0.038 and0.039 m3 m-3, respectively, for the three soils com-bined as compared to 0.15 m3 m-3 for the manufac-turer-provided equation. RMSE is preferred as anadditional measure of quality of models (Stengeret al. 2005). It is a measure of goodness of fitcompared to correlation and regression coefficients,and these values indicated that period alone can beused to estimate hv with less than 0.04 m3 m-3difference in measured- and estimated-hv.Coefficients of determination and RMSEs for bothlinear and quadratic calibrations presented in Table 2were similar and, therefore, the linear equation maybe preferable because it is simpler to use. Resultsstrongly suggest the importance of development ofsite or soil specific equations as compared to themanufacturer-provided equation for more preciseestimates of hv. Equations presented in Table 2 maybe useful to estimate hv for soils with similar claytypes as found in northern and central Missouri. Itshould be noted that clay mineralogy affects dielec-tric properties and therefore, further research isrequired to understand the effects of clay mineralogyon sensor performance.Temperature effectFigure 2 shows a representative example for thePutnam A horizon of relationships between hv andperiod for selected temperature values. The other twosoils examined in this study also exhibited thispattern (data not presented). As hv increases, theperiod increased up to *0.36 m3 m-3 hv for alltemperature values. Slope steepness was higher forhigher temperatures and lower for lower temperaturesfor hv \ 0.36 m3 m-3. Slope steepness was signifi-cantly reduced beyond this water content and theincreasing water content had a small effect on themeasured period.The effect of temperature on measured period andhv was positive and linear (Fig. 3). The measuredperiod was always higher with higher temperaturesfor the same hv irrespective of soil horizon. At lowerhv values, the slope steepness with increasingtemperature was low and concomitant differences inestimated hv were smaller as compared to higher hv.For example, at the 0.03 m3 m-3 hv, Putnam Bt soilsindicated a 0.725 ls increment in measured period,from an increase in temperature from 5 to 40C. Thisrepresents 0.013 m3 m-3 change in estimated-hvusing the manufacturer-provided quadratic equationbetween these two temperature values. Similar to theresults for soils in this study, Campbell Inc. (2002)reported -0.8 to 1.8% water content error for0.12 m3 m-3 hv between 10 and 40C. Studying thetemperature effects on Lolalita sandy loam, Searlaloam, and Larimer loam soils at 0.10 m3 m-3, Seyfriedand Murdock (2001) observed a 0.09 m3 m-3 differ-ence in estimated hv over a temperature range of 5 and45C when soil specified equations were used. Claycontents for the three soils in their study ranged from 5to 29% whereas the Menfro and Putnam soils in thisresearch ranged from 23 to 54% clay. The variation inthe observed effects of temperature between hv in thisstudy, Campbell Inc. (2002), and Seyfried and Mur-dock (2001) could be possibly due to differences inclay contents and clay mineralogy of the soils used ineach study.In nearly saturated soils, the hv was greatlyinfluenced by the temperature. Period values forPutnam Bt were 35.54 and 39.15 ls at 5 and 40C,respectively, for the 0.45 m3 m-3 hv. The correspond-ing difference in estimated hv was 0.17 m3 m-3 hv,with the manufacturer-provided quadratic equation.Campbell Inc. (2002) observed a 9% change in water1520253035400.0 0.1 0.2 0.3 0.4 0.5Campbell CS 616 Period (microsec)Volumetric Water Content (m3 m-3)5C 10C15C 20C25C 30C35C 40CFig. 2 Sensor-measured period versus measured volumetricsoil water content for the eight selected temperature values forPutnam A horizon material68 Agroforest Syst (2011) 82:6175123content error for a soil with 0.30 m3 m-3 hv between10 and 40C. In another study with 0.30 m3 m-3 hv,the difference in hv was 0.155 m3 m-3 between 5 and40C (Seyfried and Murdock 2001).The manufacturer-provided temperature correction(Eq. 3) was evaluated with estimated and measured hv.This quadratic equation corrects the period, and thecorrected period is used to estimate hv using themanufacturer-provided quadratic equation. The mea-sured period was less than 23 ls for the entiretemperature range for the three soils withhv \ 0.15 m3 m-3 (Fig. 3). For the 0.10 m3 m-3 hvat 5C, both Menfro and Putnam Bt period values were*20 ls and the corrected period was 20.45 ls. Theestimated hv, using the 20 and 20.45 ls period in thequadratic equation, were 0.088 and 0.098 m3 m-3,respectively, whereas the measured hv was 0.10m3 m-3. The temperature corrected period valueswere 41.48 and 27.65 ls for Putnam Bt at 5 and 40Cfor the 0.45 m3 m-3 hv. The manufacturer-providedequation estimated 0.877 and 0.295 m3 m-3 watercontents for these periods, respectively. Irrespective ofthe clay content, the manufacturer-provided quadraticequation estimated similar hv values with corrected anduncorrected periods at 20C.Since the temperature effect appears to be linearand uniform at low volumetric water contents, thetemperature response can be easily incorporated intocalibration equations to estimate volumetric watercontent using period (Figs. 2 and 3). Although periodvalues leveled off at higher water contents, whentemperature was included in the calibration, RMSEdecreased for the linear and quadratic equations(Tables 2 and 3). Equations developed using theperiod and temperature explained 9397% of thevariability in hv for soils with 2354% clay (Table 3).When the data for all three soils were combined, 95%of the variation in hv was accounted for by period andtemperature in a linear calibration. The RMSE was0.13 m3 m-3 for the three soils combined with themanufacturer-provided equation. The laboratory hvdata were reasonably and well represented by linearand quadratic calibration equations for these soildata; the predictability was better than with themanufacturer-provided equation.The temperature effects on dielectric properties arecomplex (Seyfried and Murdock 2004) and may bedue to the interactive effects of temperature and theamount of bound water, clay mineralogy, and ionvalence. It should be noted that the dielectric constantis directly proportional to the free water of the media(Zazueta and Xin 1994). Therefore, these sensorsmay cause errors in measurement, particularly forareas with large diurnal fluctuations. Results alsoshow that the temperature effect was small forsmaller hv and higher for higher hv. However,inclusion of a temperature correction might beimpractical until temperature-moisture combo sen-sors become available since it would require instal-lation of temperature sensors at each depth andlocation where hv is being measured. According toSeyfried and Murdock (2004), the temperatureeffect should be acknowledged and included whenusing the sensors.Effect of clay contentThe clay content among the three soils variedbetween 23 and 54% and accounted for only 13%of the variation in hv while period alone explained92% of the variation in hv (Table 4; Fig. 1). The bestcalibration equation with clay, temperature, andperiod accounted for 95% of the variation (Table 4).Adding squared terms each for period, temperature,or clay content did not improve r2 or reduce RMSE.Although clay content influences the dielectric prop-erties due to variations in charge, period accountedfor most of the variation in hv. Research shows thatclay content affects period and requires soil specificcalibrations to improve the estimate for a given hv(Seyfried and Murdock 2001; Chandler et al. 2004).1520253035400 5 10 15 20 25 30 35 40 45 50Campbell CS616 Period (microsec)Temperature (C)0.450.360.310.250.200.140.100.03Fig. 3 Campbell CS616 sensor-measured period versus tem-perature for Putnam Bt soil with soil water content valuesbetween 0.03 and 0.45 m3 m-3Agroforest Syst (2011) 82:6175 69123For example, comparing three calibration methodswith TDR sensors, Quinones et al. (2003) stated thatnon-continuous wetting, continuous wetting, andsensors at known soil water levels had consistentrelationships. In contrast, Seyfield and Murdock(1996) found that a single equation can be used todescribe differences in soil water for the same soilsused in their study. Another factor that may haveaffected the lower contribution by the clay contentcould be the small (2354%) range in clay content forsoils used in this study; i.e., soils with very low andvery high clay contents were not included. Seyfriedand Murdock (2001) had clay contents as low as 5and 10% compared to 23% in this study. In Australia,Stangl et al. (2009) used 6489% clay soils with sixCS615 sensors to develop calibration relationshipsand found slopes and intercepts were different amongsoils and horizons. However, these equations cannotbe compared directly as those were developed forCS615 sensors. Period reading for the Stenger et al.(2005) and Stangl et al. (2009) were between 0.6 and2.2 ms as compared to 1540 ms in the current study.Period and volumetric soil moisture data from theStenger et al. (2005) were used to compare thequadratic equation developed in this study. Volumet-ric water contents was converted to period usinghv = -0.283 ? 0.0182 * C ? 0.00002 * C2. Periodvalues were regressed to examine whether an equa-tion developed for high clay is comparable to thequadratic equation developed in the current study.Regression coefficient was 0.98 between the periodvalues of Stenger et al. and the current study.However, further studies may be required to validatethe accuracy of the equation when VMC is predictedfor clay percentages higher than values in the currentstudy.Table 3 Relationships between CS616 sensor-measured period (C; in microsec), soil temperature (C), and measured volumetricsoil water content (m3 m-3) for Putnam A, Menfro A, and Putnam Bt and all three horizons combinedRegression relationship Coefficient ofdetermination, r2Significancelevel, P [ FRoot mean squareerror, m3 m-3Putnam Ahv = -0.278 ? 0.0206 * C - 0.0024 * Temp 0.96 0.001 0.029hv = -0.193 ? 0.0141 * C ? 0.00012 * C2 - 0.0026 * Temp 0.96 0.001 0.029Menfro Ahv = -0.315 ? 0.0206 * C - 0.0020 * Temp 0.93 0.001 0.034hv = 0.283 - 0.0249 * C ? 0.00084 * C2 - 0.0025 * Temp 0.96 0.001 0.026Putnam Bthv = -0.258 ? 0.0186 * C - 0.00157 * Temp 0.97 0.001 0.022hv = -0.211 ? 0.0151 * C ? 0.00006 * C2 - 0.00159 * Temp 0.97 0.001 0.022Combinedhv = -0.283 ? 0.0199 * C - 0.00198 * Temp 0.95 0.001 0.031hv = -0.129 ? 0.0083 * C ? 0.00021 * C2 - 0.00213 * Temp 0.95 0.001 0.030Table 4 Relationships between CS616 sensor-measured period (C; in microsec), soil temperature (C), and clay (%) with measuredvolumetric soil water content (m3 m-3) for all three horizons combinedRegression relationship Coefficient ofdetermination, r2Significancelevel, P [ FRoot mean squareerror, m3 m-3hv = 0.279 - 0.0011 * clay 0.01 0.172 0.133hv = 0.544 - 0.0162 * clay ? 0.000193 * clay2 0.03 0.074 0.132hv = -0.297 ? 0.0192 * C - 0.0003 * clay 0.92 0.001 0.038hv = -0.270 ? 0.0198 * C - 0.00198 * Temp - 0.00031 * clay 0.95 0.001 0.03170 Agroforest Syst (2011) 82:6175123Although sensors could provide hv for compari-sons, a soil specific calibration is required to obtain ahigh degree of accuracy in hv (Leib et al. 2003). Thisis especially true when the same soil volume isutilized by multi-species vegetation with differentlengths of growing season, root distribution patterns,and moisture requirements, such as in agroforestrysystems. Accurate information on parameters such aswater use, profile moisture patterns, and peak demandare important for developing multi-species manage-ment practices for environmental and economicbenefits. In spite of proper calibrations and frequentmaintenance and checking, Zazueta and Xin (1994)questioned the long-term stability of the calibration.Sensor-estimated field water contentfor agroforestry and crop treatments by depthThe equation developed in this study estimated fieldsoil volumetric water content better than the watercontents estimated by the manufacturer providedequation (Fig. 4). Slopes were 0.96 and 1.90 for theequations developed in this study and the manufac-turer provided relationships with the field measuredwater content values. The slopes of the two equationswere significantly different. The manufacturer pro-vided equation estimated much higher water contentsespecially for higher water content values. Theequation developed in the study estimated watercontents similar to soil porosity.The annual rainfall at the Greenley Center in 2007was 893 mm which corresponds to 97% of the long-term mean annual rainfall. Rain occurred on 77 daysof the 253-day (March 12 and November 19) studyperiod with amounts ranging from 0.25 to 58 mm(Fig. 5). Precipitation amount received on day 2, 3,and 6 during this period were between, 2530, 3035,and 2025 mm, respectively. About 42% of therainfall occurred during the March to May periodwhen the evapotranspiration demand was relativelylow. Approximately 33% of the rainfall occurredduring the period between June 8 and October 26.This period corresponds with the cropping period ofthe watershed.The experimental design of a long-term study thatevaluates soil water dynamics of an agroforestrysystem was used to examine management and deptheffects on hv. Soil water content was higher due towinter recharge and reduced evapotranspiration at thebeginning of the analysis for both treatments and fourdepths (Fig. 5). Soil water content was slightly higherfor all four depths in the agroforestry treatment ascompared to the crop treatment until mid-April. Thiscould possibly be due to the beneficial effects ofperennial vegetation on soil physical properties, suchas increased porosity and carbon accumulation andreduced bulk density (Seobi et al. 2005; Andersonet al. 2009; Udawatta and Anderson 2008). Porosityvalues for agroforestry soil were 0.53, 0.49, 0.52, and0.57 m3 m-3 for the 10 cm depth increments ascompared to 0.49, 0.46, 0.48, and 0.59 m3 m-3 forthe respective depths of the crop treatment (Seobiet al. 2005).y = 0.9659x + 0.0368R = 0.8860.100.150.200.250.300.350.400.450.50VWC estimated by the Equation developed in this study (m3 m-3 )Measured Volumetric Water Content (m3 m-3)y = 1.9038x -0.0577R = 0.86770.100.200.300.400.500.600.700.800.900.10 0.20 0.30 0.40 0.50VWC estimated by Campbell Equation(m3 m-3 )0.10 0.20 0.30 0.40 0.50Fig. 4 Field measured volumetric water content and volumet-ric water contents (VWC) estimated by the manufacturerprovided equation (a) and equation developed in this study (b).Please note that the scale of Y axis is different for the twofiguresAgroforest Syst (2011) 82:6175 71123As the vegetation became active and began totranspire, soil water content decreased. The agrofor-estry treatment lost more hv compared to the croptreatment until the crop was established. However,these differences were not significant. Statisticallylower (P \ 0.05) hv persisted in the agroforestrytreatment compared to the row crop treatment withineach depth during the crop period. Among themeasured 37 sampling dates during the study period,significant differences were found between crop andagroforestry treatments for 13, 17, 18, and 18sampling dates for 5, 10, 20, and 40 cm depths,respectively. This was attributed to the greatertranspiration from the trees in the agroforestry buffertreatment compared to the soybeans in the row croptreatment. In Missouri, bud break for oaks occursduring the MarchApril period and trees start to usesoil water. Thus, the trees with higher leaf areas inPPT (mm)02040605 cm depth0.10.20.30.40.510 cm depth0.10.20.30.40.520 cm depth0.10.20.30.40.540 cm depthVWC (m3 m-3 )0.10.20.30.40.5CropAgroforestryJulian Date March April May June July August Sept. Oct. Nov. Precipitation 71 99 127 155 183 211 239 267 295 3235 cm depth10 cm depth20 cm depth40 cm depthPrecipitation VWC (m3 m-3 )VWC (m3 m-3 )VWC (m3 m-3 )Fig. 5 Daily precipitation and volumetric soil water contentestimated with the linear calibration at 12:00 noon (n = 4) forcrop and agroforestry treatments at the paired watershed studyfor 5, 10, 20, and 40 cm depth during 2007. The gray areashows the crop period for soybeans. Bars on the 40-cm depthgraph indicate LSD values for significant differences in watercontent between crop and agroforestry treatments at thea = 0.05 level72 Agroforest Syst (2011) 82:6175123early spring would have transpired more waterrelative to row crops. Although precipitation replen-ished soil water resulting in small losses from the soilprofile, soil water depletion occurred from all fourdepths during the growing season.The pattern of changes in hv closely followed therainfall distribution (Fig. 5). Rain events rechargedthe soil profile on both treatments; the effect wasmore dominant on the two surface depths. Soils at 20and 40 cm depths started to lose water after mid-August for the crop treatment while the differenceswere smaller in the agroforestry treatment. Rainevents did not completely recharge the profile untilthe 58-mm rain event in October. Although soil watercontent remained high in the crop treatment while thesoil water content in the agroforestry treatmentcontinued to decrease, differences between the twotreatments and depths were not significant after thisrecharge. This pattern could be attributed to rainfalland reduced evapotranspiration demand. Soil waterdynamics were parallel to rainfall distribution andevapotranspiration demand. The analysis indicatedthat perennial vegetation with deeper roots used morewater and maintained lower soil water profile than theannual crops. In this study, with lower soil watercontent for the agroforestry treatment and an associ-ated increased soil water storage potential, theagroforestry buffer may reduce runoff during precip-itation events.Figure 6 shows an example of the differences in hvat the 5 cm depth on July 24 which had the highestrecorded diurnal temperature difference in 2007 in thefield study. Soil temperature values were 19.5 and32.2C at 6:00 a.m. and 2:50 p.m., respectively.Volumetric water contents estimated by manufac-turer-provided equations showed the largest fluctua-tions. The difference in volumetric water contentbetween the maximum and minimum for July 24 were0.023 and 0.043 m3 m-3 for the two manufacturer-provided equations without temperature correctionand with temperature correction, respectively (Fig. 6).Without temperature correction, the manufacturer-provided equation estimated larger values during theday while the temperature corrected manufacturer-provided equation estimated smaller values. Thelargest (0.335 m3 m-3; without temperature correc-tion) and the smallest (0.272 m3 m-3; with temper-ature correction) estimated hv were recorded between2:30 and 3:30 p.m. for the respective manufacturer-provided curves. The higher temperature appeared tohave significantly influenced the manufacturer-pro-vided equation and its estimate of hv.The linear calibration developed (period only) inthis study had a 0.013 m3 m-3 difference between themaximum and minimum hv. The highest (0.246 m3m-3) hv was at 3:00 p.m. and the smallest (0.233 m3m-3) hv was between 5:30 and 7:00 a.m. It closelyfollowed the maximum and minimum temperaturevalues. The calibration with period and temperaturehad the smallest difference (0.011 m3 m-3) betweenthe highest (0.239 m3 m-3) and the smallest (0.228m3 m-3) hv. Equations developed using period aloneand period with temperature showed smaller fluctua-tions as compared to the manufacturer-provided equa-tions. The results of the analysis also suggest thatsensor-estimated period values should be used toestimate volumetric water contents during the timeswhen temperature fluctuation was relatively small.Potential application and summaryThe study was designed to examine clay content andtemperature effects on WCR sensor-measured soilwater content since accurate values are required tounderstand potential water use by a combination oftrees, crops, and grass to evaluate soil and waterconservation practices. Within these systems, non-15202530350.200.250.300.350 3 6 9 12 15 18 21 24Temperature (C)Volumetric Water Content (m3m-3 )Time (Hour)Calibration 1 Calibration 2 Calibration 3Calibration 4 TemperatureFig. 6 Soil temperature and volumetric water content at 0.5 hintervals for the 5 cm soil depth in the row crop treatment onJuly 24, 2007 at the Greenley Center Paired Watershed. Linesdenote water content estimated using manufacturer-providedrelationship (Calibration 1), manufacturer-provided relation-ship with temperature correction (Calibration 2), linearcalibration developed in this study with period (Calibration3), and linear calibration developed in this study with periodand temperature (Calibration 4)Agroforest Syst (2011) 82:6175 73123uniform water use and highly dynamic changes in soilwater content could occur in the root zone andtherefore, shallow rooted plants may not receive therequired amount of soil water leading to yieldreductions.WCR sensors can be used for continuous monitor-ing of volumetric water content. Development of sitespecific calibrations ensures greater accuracy of soilmoisture measurements over general calibration equa-tions. The equations developed from this studyestimated soil water content with less than 4% errorthus providing an estimation of soil water dynamics.These levels of errors may be tolerable for manyapplications. The analysis also indicated that sites withhigher temperature fluctuations require a temperaturecorrection, especially during times with hv above0.15 m3 m-3. Calibration equations developed in thisstudy may be used for similar soils and temperatureconditions to estimate changes in soil water content forirrigation scheduling; to study root zone water dynam-ics and water balance; and to evaluate soil water andsolute movement within the profile. These calibrationequations may result in reduced error in hv estimatesand significant differences in volumetric water contentamong studies and/or management attributes.Frequent checking with a TDR system and/orgravimetric sampling can be employed to correctWCR estimated values. WCR soil water monitoringwith these procedures reduces the cost and allowsestimation of continuous soil water dynamics inseveral locations and depths. The required precision,frequency, and spatial density of data collection for aspecific cropping system and soil type and the costassociated with instrumentation are among severalfactors to be considered when designing an appro-priate calibration procedure or a soil water monitor-ing system. Future studies may be needed to examinewhether the initial calibration equation for the WCRsensor holds for extended periods of time and todetermine the effects of clay mineralogy.ReferencesAnderson SH, Udawatta RP, Seobi T, Garrett HE (2009) Soilwater content and infiltration in agroforestry buffer strips.Agrofor Syst 75:516Campbell JE (1990) Dielectric properties and influence ofconductivity in soils at one to fifty megahertz. Soil Sci SocAm J 54:332341Campbell GS, Anderson RY (1998) Evaluation of simpletransmission line oscillators for soil moisture measure-ment. Comput Electron Agric 20:3144Campbell Science Inc (2002) CS616 and CS625 water contentreflectometers-revision 9/04. Campbell Science Inc.,Logan, UTChandler DG, Seyfried M, Murdock M, McNamara JP (2004)Field calibration of water content reflectometers. Soil SciSoc Am J 68:15011507Dean TJ, Bell JP, Batty AJB (1987) Soil moisture measurementby an improved capacitance technique. Part 1, sensordesign and performance. J Hydrol 93:6778Fellner-Feldegg H (1969) The measurement of dielectrics inthe time domain. J Phys Chem 73:616623Fernandez ME, Gyenge J, Licata J, Schlichter T, Bond BJ(2008) Belowground interactions for water between treesand grasses in a temperate semiarid agroforestry system.Agrofor Syst 74:185197Hilhorst MA, Balendonck J, Kampers FWH (1993) A broad-bandwidth mixed analog/digital integrated circuit formeasurement of complex impedances. IEEE J Solid-StatePhys 28:764769Jones SB, Wraith JM, Or D (2005) Time domain reflectometrymeasurement principles and applications. Hydrol Process16:141153Kelleners TJ, Soppe RWO, Robinson DA, Schaap MG, AyarsJE, Skaggs TH (2004a) Calibration of capacitance probesensors using electric circuit theory. Soil Sci Soc Am J68:430439Kelleners TJ, Soppe RWO, Ayars JE, Skaggs TH (2004b)Calibration of capacitance probe sensors in a saline siltyclay soil. Soil Sci Soc Am J 68:770778Kelleners TJ, Seyfried MS, Blonquist JM Jr, Bilskie J, Chan-dler DG (2005) Improved interpretation of water contentreflectrometer measurement in soils. Soil Sci Soc Am J69:16841690Khele VVN, Shaikh AA, Ramshetti RS (2008) Dielectricproperties of black soil with organic matters at microwavefrequency. Indian J Radio Space Phys 38:112115Leib BG, Jabro JD, Matthews GR (2003) Field evaluation andperformance comparison of soil moisture sensors. Soil Sci168:396408Or D, Wraith JM (1999) Temperature effects on soil bulkdielectric permittivity measure by time domain reflect-rometry: a physical mode. Water Resour Res 35:371383Quinones H, Rielle P, Nemeth I (2003) Comparison of threecalibration procedures for TDR soil moisture sensors. IrrigDrain 52:203217SAS Institute (1989) SAS/STAT users guide. Version 6, 4thedn. SAS Institute Inc, Cary, NCSeobi T, Anderson SH, Udawatta RP, Gantzer CJ (2005)Influences of grass and agroforestry buffer strips on soilhydraulic properties for an Albaqualf. Soil Sci Soc Am J69:893901Serrarens D, MacIntyre JL, Hopmans JW, Bassoi LH (2000)Soil moisture calibration of TDR multilevel probes. SciAgric 57:349354Seyfield MS, Murdock MD (1996) Calibration of the timedomain reflectrometry for measurement of liquid water infrozen soils. Soil Sci 161:879874 Agroforest Syst (2011) 82:6175123Seyfield MS, Murdock MD (2001) Response of a new soilwater sensor to variable soil, water content, and temper-ature. Soil Sci Soc Am J 65:2834Seyfried MS, Murdock MD (2004) Measurement of soil watercontent with a 50-MHz soil dielectric sensor. Soil Sci SocAm J 68:394403Soil SurveyStaff (1984) Soil survey laboratory methods man-ual. USDA Soil Conservation Service, National SoilSurvey Center, U.S. Govt. Printing Office, Washington,DCStangl R, Buchan GD, Loiskandl W (2009) Field use andcalibration of a TDR-based probe for monitoring watercontent in a high-clay landslide soil in Australia. Geo-derma 150:2331Stenger R, Barkle G, Burgess C (2005) Laboratory calibrationof water content reflectometers in their in situ verification.Aust J Soil Res 43:607615Udawatta RP, Anderson SH (2008) CT-measured pore char-acteristics of surface and subsurface soils as influenced byagroforestry and grass buffers. Geoderma 145:381389Udawatta RP, Krstansky JJ, Henderson GS, Garrett HE (2002)Agroforestry practices, runoff, and nutrient loss: a pairedwatershed comparison. J Environ Qual 31:12141225Udawatta RP, Motavalli PP, Garrett HE (2004) Phosphorusloss and runoff characteristics in three adjacent agricul-tural watersheds with claypan soils. J Environ Qual 33:17091719Udawatta RP, Anderson SH, Garrett HE (2005) Tree, grass,and crop root length densities and soil water contentwithin an agroforestry buffer system. In: Brooks KN,Ffolliott PF (eds) Moving agroforestry into the main-stream. The 9th North American agroforestry conferenceproceedings, pp 116, 1215 June 2005. Department ofForest Resources, University of Minnesota, St. Paul, MN(non-paginated CD-ROM)Udawatta RP, Motavalli PP, Garrett HE, Krstansky JJ (2006)Nitrogen and nitrate losses in runoff from three adjacentcorn-soybean watersheds. Agric Ecosyst Environ 117:3948United States Department of Agriculture-Natural ResourcesConservation Service (USDA-NRCS) (1998) World soilresources: global major land resource stresses map. SoilSurvey Division, USDA-NRCS, Washington, DCVaz CMP, Hopmans JW (2001) Simultaneous measurement ofsoil penetration resistance and water content combinedpenetrometer-TDR moisture probe. Soil Sci Soc Am J45:412Walker JP, Willgoose GR, Kalma JD (2004) In situ measure-ment of soil moisture: a comparison of techniques.J Hydrol 293:8599Zazueta FS, Xin J (1994) Soil moisture sensors. Bulletin 292.Florida Cooperative Extension Service, Inst. Food andAgricultural Sciences, University of Florida, Gainesville,FL, p 12Agroforest Syst (2011) 82:6175 75123Calibration of a water content reflectometer and soil water dynamics for an agroforestry practiceAbstractIntroductionMaterials and methodsSoil materialsLaboratory calibrationComparison of vegetation, depth, and temperature effects on water contentResults and discussionSoil propertiesMeasured and sensor-estimated water contentTemperature effectEffect of clay contentSensor-estimated field water content for agroforestry and crop treatments by depthPotential application and summaryReferences /ColorImageDict > /JPEG2000ColorACSImageDict > /JPEG2000ColorImageDict > /AntiAliasGrayImages false /CropGrayImages true /GrayImageMinResolution 149 /GrayImageMinResolutionPolicy /Warning /DownsampleGrayImages true /GrayImageDownsampleType /Bicubic /GrayImageResolution 150 /GrayImageDepth -1 /GrayImageMinDownsampleDepth 2 /GrayImageDownsampleThreshold 1.50000 /EncodeGrayImages true /GrayImageFilter /DCTEncode /AutoFilterGrayImages true /GrayImageAutoFilterStrategy /JPEG /GrayACSImageDict > /GrayImageDict > /JPEG2000GrayACSImageDict > /JPEG2000GrayImageDict > /AntiAliasMonoImages false /CropMonoImages true /MonoImageMinResolution 599 /MonoImageMinResolutionPolicy /Warning /DownsampleMonoImages true /MonoImageDownsampleType /Bicubic /MonoImageResolution 600 /MonoImageDepth -1 /MonoImageDownsampleThreshold 1.50000 /EncodeMonoImages true /MonoImageFilter /CCITTFaxEncode /MonoImageDict > /AllowPSXObjects false /CheckCompliance [ /None ] /PDFX1aCheck false /PDFX3Check false /PDFXCompliantPDFOnly false /PDFXNoTrimBoxError true /PDFXTrimBoxToMediaBoxOffset [ 0.00000 0.00000 0.00000 0.00000 ] /PDFXSetBleedBoxToMediaBox true /PDFXBleedBoxToTrimBoxOffset [ 0.00000 0.00000 0.00000 0.00000 ] /PDFXOutputIntentProfile (None) /PDFXOutputConditionIdentifier () /PDFXOutputCondition () /PDFXRegistryName () /PDFXTrapped /False /CreateJDFFile false /Description > /Namespace [ (Adobe) (Common) (1.0) ] /OtherNamespaces [ > /FormElements false /GenerateStructure false /IncludeBookmarks false /IncludeHyperlinks false /IncludeInteractive false /IncludeLayers false /IncludeProfiles false /MultimediaHandling /UseObjectSettings /Namespace [ (Adobe) (CreativeSuite) (2.0) ] /PDFXOutputIntentProfileSelector /DocumentCMYK /PreserveEditing true /UntaggedCMYKHandling /LeaveUntagged /UntaggedRGBHandling /UseDocumentProfile /UseDocumentBleed false >> ]>> setdistillerparams> setpagedevice

Recommended

View more >