Irrigation Hydrology: Crossing ScalesWesley W. Wallender1 and Mark E. Grismer2
Abstract: Hydrology is the science concerned with distribution, circulation, and properties of water of the earth and its atmoacross the full range of time and space scales. Subject matter ranges widely from chemical and physical properties to the relatioto living things. Irrigation hydrology is constrained to analysis of irrigated ecosystems in which water storage, applications, or dvolumes are artificially controlled in the landscape and the spatial domain of processes varies from micrometers to tens of kwhile the temporal domain spans from seconds to centuries. The continuum science of irrigation hydrology includes the surfaceface~unsaturated and groundwater systems!, atmospheric, and plant subsystems. How do we scale up highly nonlinear physical, cheand biological processes understood at natural scales to macro- and mega-scales at which we measure and manage irrigatedtems? How do we measure, characterize, and include natural heterogeneity in scaling nonlinear processes? In this paper,scaling issues and related research opportunities in irrigation hydrology with the hope of helping the irrigation-drainage engscience profession better address scaling problems in formulating designs affecting irrigated ecosystems.
While irrigation has dramatically increased agricultural produtivity in the short term, subsequent water resource depletionwell as soil and water contamination threaten long-term sustability of irrigated ecosystems. Improved understanding of pcesses crossing space and time scales sets the foundatioquantification of both positive and negative impacts neededmanage irrigation-drainage projects for long-term sustainabiHydrology is the science of choice to better understand compecosystems because water, a ubiquitous and life-supportingstance, links physical and biological systems at multiple tempand spatial scales.
Hydrology is the science concerned with distribution, circution, and properties of water of the earth and its atmosphere,the full range of time and space scales. Subject matter rawidely from chemical and physical properties to the relationwater to living things. Irrigation hydrology is constrained to twater-related science of irrigated ecosystems in which the spdomain of processes varies from micrometers to hundreds olometers while the temporal domain spans from seconds to
1Professor, Depts. of Land, Air and Water Resources~Hydrology! andBiological and Agricultural Engineering, Univ. of California–DaviDavis, CA 95616.
2Professor, Depts. Of Land, Air and Water Resources~Hydrology! andBiological and Agricultural Engineering, Univ. of California–DaviDavis, CA 95616.
turies. The continuum science of irrigation hydrology includessurface, subsurface~unsaturated and groundwater systems!, atmo-spheric, and plant subsystems in which at least one part ofwater storage, application, or drainage systems is artificially ctrolled in the landscape. In this paper, we discuss scaling issand related research opportunities in irrigation hydrology withhope of helping the irrigation-drainage engineering/science pfession better address scaling problems in formulating desaffecting irrigated ecosystems.
Continuum, Scales, and Scaling
Viewing irrigation hydrology within the broader context of hydrologic sciences is in the spirit of the National Research Coun~NRC 1990! report emphasizing analysis of ‘‘cycling of water aall scales from microprocesses of soil water to the global pcesses of hydroclimatology.’’ It remains a challenge to precisdefine appropriate space and time scales of seemingly simplecesses many of which are in fact nonlinear and occur acrosserogeneous systems. Fig. 1 schematically illustrates how winfluences and is influenced by transport and transformationair, solutes, soil, and biota~plants and animals! that are vital forlife and spans the continuum from groundwater through thedose zone to the atmosphere.
Irrigation hydrology falls within the three-dimensional domashown in Fig. 1 and is studied at several time and space sclisted in Table 1. The microscale is the smallest scale at whicsystem can be considered a continuum. Below this scale, maals are viewed as discrete particles and properties such as wcontent and flow velocity are not continuous functions. At thscale gas, liquid, and solid phases are considered as indepeyet interacting continua. The macroscale is larger than the miscale but smaller than the scale of the entire system of inteGas, liquid, and solid phases are commonly considered as o
s
.
IGATION AND DRAINAGE ENGINEERING / JULY/AUGUST 2002 / 203
dall
i-ofra-ssen, aen
-thenoco
or a
leage
turaony-laryolu
ecue-diso-ry
ticudle
fullx-t inrainlk
es,besis ager
stegth
re,er-
cap-being
2,ure.
thea-
ed’’rve.the
aryshedred
ts ofthe
es-e isme-ap-ar,ould
aulicn-ven
ob-
ofneityusetem.toneale.fpply
t
lapping ~spatially averaged! continua. At the megascale anabove, spatial variations are ignored and the system is usudescribed using temporal and spatial averages.
The ‘‘natural’’ scale may be defined as a ‘‘characteristic’’ dmension in time and space associated with an observationvariable or a process while ‘‘scaling’’ refers to concepts, or padigms used to adjust relationships between variables or proceas the scale of observation differs from the natural scale. Oftemore intuitive characteristic measure is the representative elemtary volume~REV! used by Bear~1972!. For homogeneous systems REV is the minimum volume of observation such thatvolume average of a property, such as soil porosity, doeschange as the averaging volume increases. In this case, theresponding natural length scale is several pore diameters,least 3–5 grain diameters~Clausnitzer and Hopmans 1999! forglass beads. For nonstationary properties~e.g., trend in variablemean! and inhomogeneous~heterogeneous! systems the lowerlimit of the REV for porosity is related to the particle size whithe upper limit is related to the distance across which averporosity trends.
Even in homogeneous systems, differences between naand measurement length scales may compromise validity of nlinear constitutive relationships commonly used in irrigation hdrology such as that between soil-water content and capilpressure, or matric suction potential. This occurs despite restion of conflicts ~Corey and Klute 1985! between definition ofpotentials based on the aqueous phase and the water mol~Nitao and Bear 1996!. Measurement of soil-water content rquires a sample large enough to include the entire pore-sizetribution, or approximately equivalent to that required for the prosity REV. However, neglecting adsorptive force fields, capillapressure is defined only across an air-water interface in a parlar interconnected pore space. As an analogy for soil, if a bunof capillary tubes having a range of diameters that are initiallyof water ~saturated! is subjected to a particular suction that eceeds the capillary rise value of some of the tubes, it will resulonly those tubes having a diameter greater than that value ding; the smaller tubes remain ‘‘saturated.’’ The bundle, or ‘‘buwater content’’ includes both the ‘‘saturated’’ and empty tubwhile the applied capillary pressure applies to only those tuthat drained. That is, the natural scale of capillary pressurethe pore scale, while that for water content is several times larAs the relationship between capillary pressure (Pc) and pore hy-draulic radius~R! is hyperbolic~i.e., Pc}1/R), the relationshipbetween water content and capillary pressure is a nonlinearfunction. Similarly, for a homogeneous soil, the natural lenscale of capillary pressure is microscopic~Nitao and Bear, 1996!and the constitutive relationship is highly nonlinear. Pressuwhich varies within the soil-water content-based REV, is av
Fig. 1. The water continuum in hydrology
204 / JOURNAL OF IRRIGATION AND DRAINAGE ENGINEERING / JULY/A
y
a
s
-
tr-t
l-
-
le
-
-
-
t.
p
aged across the fluid phase and the resulting function~Fig. 2!depends on the averaging method. Conceptually, an averageillary pressure versus soil-water content relationship couldscaled up. For example, combining two adjacent REVs havwater contentsu1 and u2 , yields an average water contentuave
5@u11u2#/2. Entering the nonlinear function shown in Fig.uave for the larger sample volume leads to an ‘‘average’’ pressPc,ave5@Pc,11Pc,2#/2 which may fall on an upscaled functionThe difference betweenPc,est and Pc,ave is the error if the un-scaled function is used with the data from a device measuringlarger volume of soil. Furthermore, there are multiple combintions of adjacent water contents each giving a unique ‘‘averagpressure and these pairs do not fall on a single scaled-up cuHence, the original and the upscaled functions depend onmethod of averaging. While conceptually an average capillpressure versus soil-water content relationship can be establifor the REV, microscopic capillary pressure cannot be measunondestructively. Furthermore, most laboratory measuremensoil-water content are at sample sizes considerably larger thansoil-water content REV. Instead, some ‘‘averaged’’ capillary prsure determined from a wick, ceramic candle, or ceramic platassigned to the measured soil-water content to give voludependent, constitutive relations. If the relationship between cillary pressure and soil-water content were fundamentally lineno scaling issue between the different measurement scales wresult. Measured relationships between unsaturated soil hydrconductivity ~K! and capillary pressure, or soil-water content ecounter the same difference in scale problem combined with egreater nonlinearity~i.e., K}R2).
For heterogeneous porous media, hydraulic function~s! varyby location within the medium and this also causes scaling prlems. When adjacent volumes are averaged, the functions~linearor nonlinear! are averaged and, as above, there is a cloudpoints rather than a single upscaled curve. Together, heterogeof the medium and nonlinearity of processes, or relations caproblems in scaling processes across the irrigated ecosysBeven and Fisher~1996! describe the ‘‘scaling problem’’ as a seof concepts that enable information or models developed atparticular scale to be used in making predictions at another sc
Harvey ~1997! and Buggman~1997! suggest six causes oscale and scaling problems for hydrologic systems that also ato the irrigated ecosystem.
Fig. 2. Conceptual nonlinear function between soil-water contenuand average soil-water capillary pressurePc
UGUST 2002
Table 1. Space and Time Scales for Water Properties and Processes in Irrigation Hydrology
rent
tua
orandsesenthae to. Ontedp-
ot t
profromtriccticof inale
ndiketpothelvecale-tionar
ha
erenchernsoil-r
tsthe
m att
ofent
near
ationn-tiales
td forw,
soil
andor
ornotbio-odelsth
n is
theltsestnceub-theher-
1. Spatial heterogeneity in surface/subsurface processes;2. Nonlinearity in system responses at different scales;3. Processes may require threshold scales to occur~related to
nonlinearity!;4. Dominant processes of concern may change with diffe
scales;5. Development of emerging processes resulting from mu
on natural systems.Fundamentally, the latter four causes are special cases of
combination of those associated with process nonlinearitysystem heterogeneity. Focusing also on the last four cauWagenet~1998! noted that scale translation efforts have oftfailed due to either a key controlling process or characteristichas been overlooked, or when multiple factors interact/combincreate unique phenomena heretofore not considered/studiedpossible opportunity in analysis of scaling problems associawith irrigation hydrology is measurement of integrative water aplication or drainage processes at several scales from the plthe region.
Scaling in Irrigation and Drainage Systems
Irrigation-drainage engineering has focused on analyses ofcesses occurring at several spatial/temporal scales rangingmicroscopic pore-scale phenomena to regional, or water-disscale issues associated with water use and delivery. In pralength, area- or volume-dependent measurements are takendependent and dependent variables resulting in length scdependent constitutive models. For example, Wallender~1987!developed a model to predict yield from soil-water content afound a model dependence on water content sample size. Lwise, the scale ‘‘jump’’ in agricultural hydrology from that awhich individual processes such as infiltration, drainage, evatranspiration, soil loss, or crop yield have been studied, toglobal scale at which climate change impacts manifest themsehas presented conceptual as well as practical problems of sand scaling~Schulze 2000!. Thus, in order to predict human impacts on water resources, air quality, and agricultural producirrigation hydrologists need to consider ‘‘upscaling’’ of nonlineprocesses in the presence of heterogeneity.
Pore to Laboratory Column Scales
In recent hydrologic research, several schemes and methodsbeen developed to upscale, or volume average~e.g., Boe 1994;
JOURNAL OF IRR
l
a
,
t
e
o
-
te--
-
-
ss
ve
Whitaker 1999; Ekrann and Aasen 2000! physical laws,~e.g., themomentum equation! applicable to pore-scale processes in ordto describe processes, or phenomena at the column, or bscale. In irrigation hydrology, the obvious processes of concthat are upscaled from the pore scale include assessment ofwater hydraulic properties~e.g., Desbarats 1995; Or and Tulle1999; Held and Celia 2001! plant-water uptake~Cermak and Na-dezhdina 1998! and water-quality related processes~e.g., Mac-donald et al. 1999!. At microscales, water and its constituenflow and transform as well as interact with soil surfaces inrhizosphere~i.e., root-, soil-, and soil-water interfaces!, the unsat-urated zone below the root zone and the groundwater systetime scales from seconds to millenia~Table 1!. Both measuremensupport size limitations as well as developing realistic solutionscontinuity-type equations at microscales have led to measuremand modeling at macroscales. For example, the microscale limomentum ~Navier-Stokes! equations are simplified to theStokes’ equations and space-averaged to derive Darcy’s equ~Whitaker, 1999!. Combining the Darcy equation with mass coservation to model spatially averaged velocity or water potenresults in the Richard’s equation. Similarly, the Navier-Stokequations are simplified for the case of free-surface~e.g., open-channel! flow in border or furrow irrigation to yield the St. Venanequations. These cross-section averaged equations are solveone- or two-dimensional flow velocities and depths in furroborder, and basin irrigation systems.
When considering micro-scale heterogeneities acrosspores, or from soil cores to plot scales, Hopmans et al.~2002!summarize application of ‘‘aggregation’’ methods~Wheatcraftand Tyler 1986! to scale up pore-scale measurements of largesmall hydraulic conductivity ‘‘textures’’ to an average value fthe core as a whole.
When considering microbial degradation of contaminantsnutrients in soils, models of bacterial populations often doresolve pore-scale variability problems, rather, substrate andmass concentrations are taken as bulk averages. These munrealistically predict monotonically increasing biomass groweverywhere except where the limiting substrate concentratiovery small. Macdonald et al.~1999! examines the possibility ofbiofilm mass-transfer limitations at the pore scale using bothtraditional biofilm model as well as previously published resufrom an upscaling model. Results from the biofilm model suggthat limitations on biofilm growth due to mass-transfer resistacould be significant in coarse-grained soils with adequate sstrate availability. The upscaling approach also indicated thatdegree of mass transfer resistance is reduced at higgroundwater velocities.
IGATION AND DRAINAGE ENGINEERING / JULY/AUGUST 2002 / 205
or-aluatesapdif-sapdelyorelmo
aof
sesfromogin
edho
i-the
rvinenthe-eener,cs athi
n otrainticaup
porfo
owthew are
ths thfobe
calth
ysandre-
ratechsur
nten
dinr ims i
o-
x-
tesl toss a
ulicorn toricalog-re-
andy in
nstion.ntialphy,ribu-rcymsuch
on
ay,
forimu-
forthe
iallyiesrop-
tys arevia a
re
insoil
del-96;
and98;998;id-
tion
When considering crop consumptive-use processes, the wof Cermak and Nadezhdina~1998! in upscaling plant-water uptake measurements from individual trees to stands may be vable. They estimated sapwood depth according to xylem wcontent and more precisely according to radial patterns offlow rate in five coniferous and four broad-leaved species offerent diameter, age, and site conditions and compared it towood cross-sectional area, a simple biometric parameter wiused for scaling up the transpiration data between trees and fstands. Sapwood area estimated by the two methods was aequal in some species~e.g., Cupressus arizonica!, but differedsignificantly in other species~e.g., Olea europaea, Pinus pine!.Radial pattern of sap flow rate was a more reliable indicatorsapwood then xylem water content for sap flow scaling purpoSapwood area could be used for upscaling transpiration ratesmeasuring points~e.g., stems! to whole trees and from trees tstands only for the same measurement periods. That is, averasapwood cross-sectional areas within the tree stand providreasonable estimate of the stand transpiration rate as a willustrating how heterogeneity effects~i.e., variable soils, nutri-ents, tree growth, shading, etc.! within the tree stand on transpration rates appear to diminish when transitioning betweentree and stand scales.
There are unexplored processes at the microscale deseupscaling research that will lead to improved water managemFor example, a common, though not yet well-understood pnomenon is soil shrink/swelling and cracking during and betwirrigation events~Waller and Wallender 1993; Lima and Grism1994; Nichols and Grismer 1997!. Solution of mass, momentumenergy, and entropy balance equations of continuum mechanithe microscale is one opportunity to understand and simulatebehavior and build upon these early attempts. The predictiowater and solute transport is complicated by simultaneous sand its causal stress in the solid phase. With improved theoreunderstanding at the microscopic scale leading to superiorscaled models which are consistent with measurement supdevelopment of more efficient salt management strategiescommonly irrigated and drained cracking soils will evolve.
In addition to saturated and multiphase~unsaturated! flow inirrigated porous media, free-surface flow in canals as well as fland erosion in borders and furrows is governed bymicroscopic-scale Navier-Stokes equations. Imagine water flothe leading edge of the advancing front in furrow irrigation wheflow is fully three dimensional. The flow domain evolves witime as soil aggregates disintegrate. Fluid density changes asoil particles are entrained in the water. Recommendationserosion control as well as pesticide and herbicide runoff canguided via the understanding of these processes at the microsThus, there are numerous research opportunities to combinesolution of the balance equations with process theory from phics, chemistry, and biology at microscales to better understand predict sediment/pollutant flow and transport but, again,search on upscaling to measurement support is needed.
Appropriate measurement technology is required to caliband validate microscale models. Remote sensing methods sunuclear magnetic resonance are currently available to meazero-, first-, and second-order tensor fields such as water co~zero order or scalar!, flow velocity ~first order or vector! andvelocity gradient~second order or informally called tensor! at thepore scale and may assist in developing a greater understanof microscopic processes and space averaging methods foproved management and design of irrigation-drainage systemthe future.
206 / JOURNAL OF IRRIGATION AND DRAINAGE ENGINEERING / JULY/A
k
-r
-
stst
.
gale
g.
tsf
l-t,r
t
er
e.e-
aset
g-n
Laboratory to Plot Scales
Modifying the characterization of soil-water properties from labratory columns~order of 1021 m! to plots ~order of 10 m! in-volves scaling of hydraulic functions, or aggregation of soil tetures ~Hopmans et al. 2002!. For example, Tidwell and Wilson~1997! describe procedures for upscaling permeability estimato the plot and field scales. Such an analysis is also criticaevaluation of soil moisture variability and recharge rates acrolandscape due to small changes in topography. Green~1995! de-veloped a method of upscaling both water content and hydraconductivity on the assumption of uniform matric potential fupscaling water content, and use of van Genuchten’s functiorepresent the upscaled water- retention curve. He used numeexperiments to address how the combination of undulating topraphy and moisture-dependent anisotropy affected soil-waterdistribution and distribution of recharge rates in both spacetime. He found that small-scale soil layers created anisotropupscaled~spatially averaged over many small layers! hydraulicconductivity. Values of anisotropy at moderately low saturatiowere generally much higher than at saturation or near saturaMoreover, zones of soil-water accumulation acted as prefererecharge areas. Different combinations of landscape topograanisotropy, and steady recharge rates affected the spatial disttions of both Darcy flux and upscaled soil-water contents. Daflux varied by three or four orders of magnitude, from a minimubeneath topographic crests to a maximum beneath swalesthat vertical Darcy flux at a depth of 2 m was strongly correlated(R250.96) with topographic curvature. This latter observatiagain suggests that larger-scale measurements~e.g., topography,tree stands! that incorporate smaller-scale heterogeneities mprovide better insight into processes of concern~e.g., seepagetranspiration! than when modeled at the smaller scales.
At this scale range, again there are ample opportunitiesupscaling research. For example, as cited above, upscaled slated flow patterns have potentially significant ramificationsroot zone leaching of pesticides and solutes to groundwater. Inlatter case, Buyuktas and Wallender~2001! modeled three-dimensional flow and transport to subsurface drains for a spatvariable soil and found that upscaled soil hydraulic propertcould be used for this case to represent heterogeneous soil perties.
Plot to Field Scales
Upscaling from plot ~order 101 m! to the field scale~order102– 103 m! is again complicated by nonlinearity and variabiliof process parameters such that some process descriptionrevised, or recharacterized so as to include heterogeneitiesrange of statistical methods~e.g., kriging, renormalization, andaggregation!. For the irrigation hydrologist, such analyses ausually applied to soil-water transport processes~i.e., soil-waterretention and permeability! that have been studied extensivelythe past decade by engineers, geologists, hydrologists, andscientists and is available in their respective journals~e.g., Rubinand Gomez-Hernandez 1990; Russo 1992; Indelman 1993; Inman and Dagan 1993a,b,c; Sanchezvila et al. 1995; King 19Wen and Gomez-Hernandez 1996; Ewing 1997; HristopulosChristakos 1997; Bierkens and vanderGaast 1998; Hunt 19Tegnander and Gimse 1998; Wen and Gomez-Hernandez 1Hristopulos and Christakos 1999; Sahimi and Mehrabi 1999; Twell and Wilson 1999; Sanchez-Vila et al. 1999!. On the otherhand, in ecological engineering studies, Ahn and Mitsch~2002!found that taking a scaling perspective based on wetland func
UGUST 2002
/
-thepit
pte
dales
angt a
lea
to-odencetivesynu-
g oput-ure
-meiat
at-
1-nt.et
thont
pri-atlua
ng
aningne
ea-sseed.ap
e Segin
c-. Fua-ted
, threa
nsti-rest,ons.ansderandForl-eter
-y tothemandp-ace-
ise,rs oruchthe
am-ign
-en-nles ofam-
sedsity,
-d
-in-
is dif-arge
fluxe to-k ofnedpti-
on toata/orkdatam-ms
andthatIn
theon-
o-
and ‘‘performance’’ ~e.g., water use, nutrient uptaketransformation, plant coverage, hydraulic residence times! to beuseful towards comparison of 1 m2 versus 1 ha systems. In comparing the small and large wetland systems, they found nearlysame performance during time periods of one to two years desobvious heterogeneities~e.g., preferential flow! in the 1 ha sys-tem.
There is also some question as to whether commonly accemethods for estimating evapotranspiration~e.g., Penman-Monteith equation! by irrigation hydrologists that are validatefrom lysimeter-scale analyses apply at larger field/region scdespite their mathematical elegance~Schulze 2000!. Perhaps in aneffort to better upscale from the plant to canopy gaseous exchprocesses in a manner that may be useful to us, Wohlfahrt e~2000! develop a model that quantifies CO2 and H2O gas ex-change of whole plants and the canopy. The model combines~gas exchange, energy balance! and canopy~radiative transfer,wind attenuation! scale simulations. Net photosynthesis and smatal conductance are modeled using a nitrogen-sensitive mof leaf gas exchange. An analytical solution to the energy balaequation is adopted to calculate leaf temperatures. Radiatransfer, considered separately for the wave bands of photothetically active, near-infrared and long-wave radiation, is simlated by means of a model that accounts for multiple scatterinradiation using detailed information on canopy structure as indata. Partial pressures of CO2 and H2O, as well as air temperatures within the canopy are not modeled, but instead, measvalues are used as input data to the model.
Also of potential significant importance to irrigation hydrologists is the upscaling of nonlinear surface erosion processessured at the plot scale up to field and watershed scales assocwith hillslope [email protected]., vineyards and orchards; see Btany and Grismer~2000!#. Le Bissonnais et al.~1998! measuredcrust formation, runoff, and erosion during two seasons fromm2, 20-m2, and 500-m2 plots and up to a small 70 ha catchmeRunoff fraction of total rainfall, sediment concentration and nsoil loss from the 20-m2 and 500-m2 plots were practicallyequivalent and several times greater than that obtained from1-m2 plots and 70 ha catchment. Interestingly, peak sediment ccentration and soil loss from the 1-m2 plots and 70 ha catchmenwere quite similar, despite different processes~e.g., preferentialflow! occurring at each scale. Overall, ground cover was themary factor controlling erosion, though the variation in resultsthe different scales underscored the need for additional evation. King et al.~1998! incorporates this research into upscaliof an erosion model from small areas to regions.
Next consider upscaling processes and measurements,their implications for irrigation and drainage system design useconomic sampling principles. First we discuss processes. Geally, irrigation design is based on results of time and araveraged mass balance equations. For overland flow procesolution of the momentum equation is approximated or ignorIn the case of surface irrigation modeling, a predetermined shis assumed for the surface flow profile as a replacement for thVenant equation. Likewise, infiltration, an input to the volumbalance equation of surface flow, is measured over areas ranfrom 1–100 m2 and infiltration rate and cumulative depth funtions are generally based on area-averaged measurementssprinkle irrigation performance evaluation, the momentum eqtion is ignored and application volume depends on the projecarea of the catch can and on the duration of the test. Thusaverage and spatial variation of water application depths are aand time-averaged measurements.
JOURNAL OF IRR
e
d
el.
f
l
-
f
d
a-ed
e-
-
d
r-
s,
et.
g
or
e-
Rather than using scales of measurement that optimize cotutive relations and are consistent with the processes of intethey have been largely determined by practical consideratiFor sprinkle irrigation assessments, readily available oil cdrove their use in measuring uniformity. Fischer and Wallen~1988! recommended larger collectors to enhance accuracyprecision and to reduce the duration and number of tests.surface irrigation system evaluation, although variability of infitration measurements decreases as the area of the infiltromincreases~Wallender 1987!, the area of the infiltrometer is constrained by practical considerations such as water, the abilitconstruct and install large infiltrometers, or the ability to solveinverse problem of determining the infiltration function fromeasurements of flow geometry, slope, hydraulic roughness,water advance. Similarly, the construction and function of gysum blocks, tensiometers, neutron probes determines the spand time-averaging of soil-moisture measurements. Likewdrainage and groundwater measurements, using piezometewells, are area or volume as well as duration dependent. Smeasurement scales may be incongruent with that optimal forirrigation-drainage process of interest~Wallender 1987!.
Last, economic principles are used to determine optimal spling for design. Optimal sampling minimizes the sum of deserror cost and sampling cost~Ito et al. 1999!. They calculate thecost of a design error~or cost information! as the difference be-tween optimal returns to water in which infiltration function information is complete and incomplete. The optimal sampling dsity minimizes the sum of unit information and infiltratiomeasurement cost. The use of suboptimal space and time scaprocesses and constitutive relations carry into suboptimal spling strategies and system designs.
A water budget analysis serves to illustrate this compromiprocess. To control sampling cost by decreasing sampling denseveral fields are aggregated and all but one [email protected].,evapotranspiration~ET!, surface water delivery, drainage, tailwater, change in vadose zone storage# is measured or assumed anthe remaining term is selected as the closure term~e.g., deeppercolation!. To further control sampling cost, in addition to decreasing sampling density, the duration between samples iscreased. Less frequent measurements are taken because itficult ~i.e., numerous simultaneous measurements over a larea! and expensive to measure changes in storage as well asof water and solutes using small sample areas spaced closgether over large areas. However, rarely is the cost of the lacinformation included in the analysis. Sampling cost is constraiby arbitrary budgets instead of the sum of sampling and subomal management.
Field to Region-Watershed Scale
Upscaling field-scale measurements, processes and simulatithe regional or watershed scale involves aggregation of dprocesses collected at multiple levels into a coherent framewfor subsequent analyses of impacts at the larger scales. Thismanagement is crucial to later application and is often accoplished with the assistance of Geographic Information Syste~GIS!. Multispectral remote sensing data are often analyzedincorporated to produce a vegetative cover thematic layer,may be used to provide data for ET and infiltration estimation.irrigation hydrology, agricultural fields become a part, perhapslargest fraction, of the overall landscape or watershed to be csidered. For example, Purkey and Wallender~2001! modeled theeffect of removing land parcels from agricultural production l
IGATION AND DRAINAGE ENGINEERING / JULY/AUGUST 2002 / 207
surpo
hanent
ly inionredul-de
ewi
ataface
andtionaterobom
landet-
rony agutioina
st in
nizith
ive,sicaedtheromsca
enuatetisHisimuuiredi-ta.s toile
eldsaap-t atsoiobre-
d tomu-
of
opedthe
ngactster-
ldcts ofnishis ofes isingpy
logy,e of
and-ere
gasmea-
andof
tem/t inntdelsrentho-ing
onte-
tondea-
s ofeter-
on
fieldt and
-
od-allyhed.Mscellsbothand-ortenrib-a-oil
ound
ngse-3-i-
cated up- and down-gradient of shallow groundwater on subface drainage. For single parcels, the 31% drainage reductiontential of downgradient land retirement was more attractive tthe 16% reduction associated with the upgradient land retiremRegulatory and water management agencies are increasingterested in the overall impact of irrigated agricultural producton the landscape~including groundwater! as the health of watesupplies for local communities and environment are intertwinwith agricultural practices in the region. Regional scale agrictural economic-type models and evaluations have also beenveloped~e.g., Lee and Howitt 1996!, but we retain our focus heron physically based models and the processes associatedupscaling~e.g., Davy and Crave 2000!. We consider upscalingfrom the field to region and underscore the importance of dmanagement and inclusion of impacts on ground and surwater systems within the agricultural production region.
Integrated modeling of interdependent surface, soil-watergroundwater processes requires multiple scales of investigaor analysis, to achieve objectives associated with improved wmanagement, or environmental protection. There have been plems in some efforts to upscale land-use cover information frfield to regional scales. For example, Yang~1997! investigated theeffects of changing spatial scale on the representation ofcover and found that significant errors may be introduced in herogeneous areas where different land cover types exhibit stspectral contrasts. The extent of agreement between spatiallgregated coarse resolution land cover datasets and full resoldatasets changed depending on the properties of the origdatasets, including the pixel size and class definition. To assiidentification and resolution of such problems, Geurink~1999!designed a database structure that includes elements for orgation and characterization of field and laboratory observations wutilities for both data management and derivation of extensintegrated model parameters. Selection and resolution of phyproperties, stored and maintained in the database, are governtemporal and spatial sensitivity of hydrologic processes ofregion. The broadest range of applications for data acquired fthe database can be attained when data are independent ofstructure, and process algorithm of a model. Data managemutilities have been outlined to query and acquire data, evaland correct data errors and omissions, characterize data stacally, and develop data sets for integrated model calibration.results suggest that uncertainty and sensitivity analyses of slation results can be used to refine observation resolution reqments for physical properties and to guide identification of adtional locations or physical properties for which to acquire daMoreover, centralized, hierarchal accessibility to data appearpromote multidisciplinary application of common data whmaintaining protection protocols.
At the surface, where crop growth and water use~or ET! are ofprimary importance practically, Olesen et al.~2000! developed acrop simulation model~CLIMCROP! for investigating differentmethods of aggregating simulated county and national crop yifor winter wheat in Denmark with and without irrigation forrange of soil types and climatic conditions. While the model ctured most of the spatial variation in observed yields, excepthe coarsest resolutions of soil data, the finest resolution ofand climate data gave the best fits between simulated andserved spatial autocorrelation in yield. The model better repsented interannual yield variability on loamy soils as comparethat for sandy soils. The results indicated that upscaling of silated yields for Danish conditions required a spatial resolutionsoil data of 10 km2 or finer and that more climate~i.e., ET!
208 / JOURNAL OF IRRIGATION AND DRAINAGE ENGINEERING / JULY/A
--
.-
-
th
,r-
g-
nl
a-
lby
le,t
ti-
--
l-
stations are required if yields at larger scales~i.e., regional! are tobe estimated. Crop simulation models have also been develfor rice production and associated evolution of methane frompaddies~e.g., Matthews et al. 2000; van der Gon et al. 2000!.
Irrigation hydrologists may take some direction in developiagricultural landscape models that incorporate landscape impfrom those developed for ecological systems or forested washeds. Cernusca et al.~1998! used a combined approach of fiemeasurements and process-based modeling to assess effeland-use changes in the Eastern Alps, the Swiss Alps, the SpaPyrenees, and the Scottish Highlands ecosystems. An analysstructures and processes in the context of land-use changperformed, scaling from the leaf to the landscape level includanalysis of vegetation and soil-type spatial distribution, canostructure, ecosystem water relations, catchment area hydroecosystem microclimate and energy budgets, gas exchangsingle plants to canopies, and potential risks associated with luse changes. At each level of integration, modeling results wvalidated by direct field measurements. For example, withexchange processes, the up-scaling began with experimentalsurements of gas exchange by single leaves using CO2/H2O po-rometers, proceeds to the canopy level using Bowen-ratioEddy-correlation techniques, and finally to the gas exchangecomposite landscapes using a scintillation anemometer sysdifferential optical absorption spectroscopy system equipmencombination with a instrumented aircraft flying at a constaheight above the landscape. A hierarchy of process-based mowas used to link the gas exchange processes at the diffescales: a single leaf model based on a biochemical model of ptosynthesis’ a newly developed multispecies canopy model takinto account multiple light scattering and nonuniform distributiof leaves, and a landscape model providing a pixel-by-pixel ingration. In an analogous fashion, Llorens and Gallart~2000! up-scaled interception by individual pine needles and bunchesdetermine stem flow, throughfall, and eventually canopy acatchment scale forest water storage through combined field msurements and modeling. They found that modeled estimateforest canopy water storage were 30% greater than that dmined from more commonly used indirect estimates basedrainfall-throughfall regressions.
Lumped process and distributed models used to upscaleobservations of processes have also been evaluated in foreswatershed hydrology. For example, Yu~2000! evaluated threetypes of forest hydrologic models, ‘‘black-box,’’ lumped parameter, and physically based distributed models~PBDMs!. Black-box models are essentially statistical input-output watershed mels, while lumped-parameter and PBDM models are empiricderived models based on hydrological processes in the watersRelative to black-box and lumped-parameter models, PBDconsidered watershed spatial heterogeneity through differentand flow pathways and can express hydrological processes inspatial and temporal dimensions. The writer suggested that lscape ecology knowledge combined with GIS could suppPBDMs to solve the upscaling problem. Birkinshaw and Ew~2000! also successfully use a physically based, spatially distuted river catchment model combined with ‘‘point-based’’ equtions for transformation of carbon and nitrogen pools in the sand subsequent advection dispersion of these nutrients to grand surface waters in the watershed. Soutter and Loague~2000!also found that application of a ‘‘point-scale’’ pesticide leachimodel coupled with variation in soil-type data and land/water-uconditions was adequate to evaluate regional dibromochloropropane~DBCP! leaching in the Fresno, Calif. area. Prev
ously, but and in a similar fashion, Kersebaum and Wenkel~1998!evaluated the effect of different data qualities and aggregalevels on results of an integrated soil-water and nitrogen dynics model for an area of three districts that were part of a formlarge-scale study on nitrogen losses to groundwater. Measments and simulations of groundwater recharge and nitroleaching indicated that aggregation of soil map units had a rtive small effect as compared to averaging of weather dataerrors of land-use characterization. Use of long-term averweather data lead to underestimation of groundwater rechrates. Nitrogen uptake by plants in the regional study was oless than that simulated with actual weather data and obsecrop rotations. Refsgaard et al.~1999! uses a combined data aggregation and distributed modeling approach to evaluate regioscale nitrate leaching and groundwater contamination. Reaavailable European level data from standard databasessupplemented by selected data from national sources. Moderameters were obtained from these data by use of ‘‘transfer futions’’ and upscaled from point to field to catchment scales useffective parameters with a statistically based data aggregaprocedures that preserved areal distribution of soil types, vegtion types, and agricultural practices on a catchment basiscomparing simulation results with measurements for a pairsmall Danish watersheds, the upscaling/aggregation procedurpeared to be applicable in many areas with regard to root zprocesses such as runoff generation and nitrate leaching, buimportant limitations with regard to stream hydrograph shapeto failure to account for scale effects in relation to stream-aquinteractions. Groundwater contamination issues also pose huhealth risks that require combined physical molding and hearisk assessments at the regional scale~e.g., Pelmulder 1994; Pelmulder et al. 1996! and such analyses may need to be includwith regional-scale economic assessments~e.g., Lee and Howitt1996! of irrigated agricultural impacts to formulate regional watquality policies and management of agricultural practices.
Ultimately, watershed contamination, or landscape water-models developed to assess effects of changing land use, oragement conditions need to incorporate contamination effectthe overall ecological system. Habersack~2000! discusses theriver-scaling concept as a basis for ecological assessmentsnotes that since river morphology results from transport of waand sediments, the size of project areas and the analysis pdures are critical factors. Restricting the assessment of abioticbiotic river components and its variability to a certain scale mneglect the fact that ecological integrity depends on the streprocess scale. He uses a two-step procedure for assessinecological integrity at various temporal and spatial scales. Duthe so-called downscaling phase, abiotic and biotic componare analyzed at the regional, catchment, sectional, local, and pscales and illustrate the importance of mass balance analysesapplication of fractals and self-similarity studies for channel dvelopment. As the scale varies dramatically, results obtained fvarious analysis tools are significantly different but interdepdent. Biotic analyses are also performed at these scales, sothe interrelations between channel morphodynamics and haquality can be derived at the end of this first phase. In the secphase, upscaling integrates and aggregates the results fromfirst step in order to yield overall conclusions on ecological interity.
Finally, assessment of farm and regional water use efficiein agriculture is essential towards rational allocation of limitwater resources~Grismer 2001a,b!. For example, Tuong andBhuiyan ~1999! evaluate the water losses associated with r
JOURNAL OF IRR
-
-
-
e
d
-
e-
-
n-
p-
d
-
n-n
d
e-d
he
st
nd
att
e
production and identify that water used during land preparatand seepage and percolation during crop growth are the primlimitations and discuss strategies for increasing farm-level wause efficiency and upscaling from on-farm to system-level wasavings.
Region-Watershed to Global Scale
Upscaling from regional to global conditions requires remosensing measurements and incorporation of global circulamodels coupled with changing land-use and anthropomorphicpacts on climate change and variability. This scale of researcapplication is typically beyond the scope normally consideredthe irrigation hydrologist, but may soon come to be an issueconcern as land-use impacts and agricultural policies and ptices in one country often affect neighboring nations as wellmore distant nations across the ocean~e.g., dust storms in ChinaMongolia appearing in California in spring 2001!. In addition toglobal climate change considerations are aspects of populagrowth and adequate nutrition~Renault and Wallender 2000!.Harvey~2000! discusses general aspects and problems assocwith upscaling to the global scale, while De Grandi et al.~2000!illustrate use of satellite imagery and data to evaluate ecosyschanges in Africa.
OpportunitiesAs a profession, irrigation engineers/hydrologists are developcreative measurement and sampling technology as well as anacal and statistical theory to improve monitoring and assessmewater quantity and quality impact from irrigated agriculturThough a daunting task, in gaining a better understanding ofgation hydrology processes through measurement and modacross broad space and time scales, we hope to discover linkwhich transform results from one time and space scale to anowith no or limited additional measurement.
This approach to field and laboratory research combined wmodeling leading to optimal irrigation and drainage managemis recognized as an important strength in irrigation hydrology.such, several research/application opportunities remain includ~1! scaling up fundamental physical and biological processetime and space to better represent the agricultural productiontem as part of the broader landscape;~2! developing indicators toidentify system function~malfunction! at all scales; and~3! man-age production systems, at appropriate time and space scfrom input ~e.g., water and nutrient application! as well as output~e.g., productivity and environmental impact! perspectives.Clearly, new scale-appropriate measurement methods are neto facilitate such studies, an opportunity in its own right.
In reviewing the ASCE articles database, it was interestingnote that in the past three decades there have been only a haof articles published in symposia or journals that consider in thkeywords, or titles the combination of irrigation or drainage whydrology. Irrigation-drainage engineering scaling issues havebeen overtly considered though they are apparent in manysearch articles, perhaps as a result of combined empiricism‘‘judgement’’ often required of the engineering discipline. In cotrast, scale and scaling issues have been a dominant restheme in our sister fields of hydrology, soil science, and agrictural economics during the past decade. In hydrology, forample, books by Stewart et al.~1996! and Sposito~1998! provideexcellent reviews of the subject, while in soil science, entire jonal issues~e.g., see Finke et al. 1998! have been devoted to th
IGATION AND DRAINAGE ENGINEERING / JULY/AUGUST 2002 / 209
r the ii.e.blee,datavel
havury tolicy
tech
ace
-
ter,
s.’’
ion
.’’cifir,
sys
-of
R.,
ses
-
tchis
cs.’’
es
cio-rs.’’
t forr,
dge.’’
b-r,
g
cale-eo-
s.,
ta-
-
opic
per-
rent
-
, V.
problems of upscaling processes. Perhaps we should consideour role as engineers/hydrologists of the agricultural landscapless to develop new techniques of information management,new software, but more to better apply what is already availato solve the ‘‘real world’’ problems we face. For examplGeurink ~1999! in working with the multiple scales associatewith an integrated surface/ground water model developed a dbase structure that facilitates analyses and scaling between leHowever, Dumanski et al.~1998! noted ‘‘Innovations in the in-formational sciences are so rapid and constant that we oftento run hard just to avoid being left behind.’’ Consequently, ochallenge is to develop the information paradigms necessardevelop a better understanding of the major issues and pochanges related to our science, and then apply those newniques that suit these requirements.
References
Ahn, C., and Mitsch, W. J.~2002!. ‘‘Scaling considerations of mesocosmwetlands in simulating large created freshwater marshes.’’Ecol. Eng.,18~3!, 327–342.
Battany, M. C., and Grismer, M. E.~2000!. ‘‘Rainfall runoff, infiltrationand erosion in hillside vineyards: Effects of slope, cover and surfroughness.’’Hydrolog. Process.,14, 1289–1304.
Bear, J.~1972!. Dynamics of fluids in porous media, Dover, New York.Beven, K. J. and Fisher, J.~1996!. ‘‘Remote sensing and scaling in hy
drology.’’ Scaling up hydrology using remote sensing, Stewart, J. B.,Engman, E. T., Fedes, R. A. and Kerr, Y., eds., Wiley, ChichesU.K., 1–18.
Bierkens, M. F. P., and vanderGaast, J. W. J.~1998!. ‘‘Upscaling hydrau-lic conductivity: Theory and examples from geohydrological studieNutrient Cycl. Agroecosyst.,50~1–3!, 193–207.
Birkinshaw, S. J., and Ewen, J.~2000!. ‘‘Nitrogen transformation compo-nent for SHETRAN catchment nitrate transport modeling.’’J. Hy-drol., 230~1-2!, 1–17.
Boe, O.~1994!. ‘‘Analysis of an upscaling method based on conservatof dissipation.’’Transp. Porous Media,17~1!, 77–86.
Bugmann, H.~1997!. ‘‘Scaling issues in forest succession modelingElements of Change 1997 - Session One: Scaling from site-speobservations to global model grids, S. J. Hassol and J. Katzenbergeeds., Aspen Global Change Institute, Aspen, Colo. 45–57.
Buyuktas, D., and Wallender, W. W.~2001!. ‘‘Numerical simulation ofwater flow and solute transport to tile drains.’’J. Irrig. Drain. Eng., inpress.
Clausnitzer, V., and Hopmans, J. W.~1999!. ‘‘Determination of phase-volume fractions from tomographic measurements in two-phasetems.’’ Adv. Water Resour.,22~6!, 577–584.
Cermak, J., and Nadezhdina, N.~1998!. ‘‘Sapwood as the scaling parameter: Defining according to xylem water content or radial patternsap flow?’’Ann. Sci. Forest. (Paris),55~5!, 509–521.
Cernusca, A., Bahn, M., Chemini, C., Graber, W., Siegwolf,Tappeiner, U., and Tenhunen, J.~1998!. ‘‘Ecomont: A combined ap-proach of field measurements and process-based modeling for asing effects of land-use changes in mountain landscapes.’’Ecol. Mod-ell., 113~1–3!, 167–178.
Corey, A. T., and Klute, A.~1985!. ‘‘Application of the potential conceptto soil water equilibrium and transport.’’Soil Sci. Soc. Am. J.,49~1!,3–11.
Davy, P., and Crave, A.~2000!. ‘‘Upscaling local-scale transport processes in large-scale relief dynamics.’’Phys. Chem. Earth PartA—Sold Earth Geodesy,25~6–7!, 533–541.
De Grandi, G. F., Mayaux, P., Malingreau, J. P., Rosenqvist, A., SaaS., and Simard, M.~2000!. ‘‘New perspectives on global ecosystemfrom wide-area radar mosaics: Flooded forest mapping in the tropiInt. J. Remote Sens.,21~6–7!, 1235–1249.
210 / JOURNAL OF IRRIGATION AND DRAINAGE ENGINEERING / JULY/A
ats,
-s.
e
-
c
-
s-
,
Dumanski, J., Pettapiece, W. W., and McGregor, R. J.~1998!. ‘‘Relevanceof scale dependent approaches for integrating biophysical and soeconomic information and development of agroecological indicatoNutrient Cycl. Agroecosyst.,50, 13–22.
Ekrann, S., and Aasen, J. O.~2000!. ‘‘Steady-state upscaling.’’Transp.Porous Media,41~3!, 245–262.
Ewing, R. E. ~1997!. ‘‘Aspects of upscaling in simulation of flow inporous media.’’Adv. Water Resour.,20~5–6!, 349–458.
Finke, P. A., Bouma, J., and Hoosbeek, M. R.~1998!. ‘‘Preface.’’NutrientCycl. Agroecosyst.,50, 1.
Fischer, G. R., and Wallender, W. W.~1988!. ‘‘Collector size and testduration effects on sprinkler water distribution measurement.’’Trans.ASAE,31~2!, 538–542.
Geurink, J. S.~1999!. ‘‘Concepts of data management and assessmenmulti-scale integrated hydrologic modeling.’’ PhD Civil EngineeUniv. of South Florida, Tampa, Fla., 355.
Green, T. R.~1995!. ‘‘The roles of moisture-dependent anisotropy anlandscape topography in soil-water flow and groundwater recharPhD Civil Engineering, Stanford Univ., Palo Alto, Calif., 327.
Grismer, M. E.~2001!. ‘‘Regional alfalfa yield, ETc and water value inwestern states.’’J. Irrig. Drain. Eng.,127~3!, 131–139.
Grismer, M. E.~2002!. ‘‘Regional cotton lint yield, ETc and water valuein Arizona and California.’’Agric. Water Manage.,54~3!, 227–242.
Habersack, H. M.~2000!. ‘‘The river-scaling concept~RSC!: A basis forecological assessments.’’Hydrobiologia,422, 49–60.
Harvey, L. D. D. ~1997!. ‘‘Upscaling in global change research.’’Ele-ments of Change 1997– Session One: Scaling from site-specific oservations to global model grids, S. J. Hassol and J. Katzenbergeeds., Aspen Global Change Institute, Aspen, Colo. 14–33.
Harvey, L. D. D.~2000!. ‘‘Upscaling in global change research.’’Clim.Change,44~3!, 223.
Held, R. J., and Celia, M. A.~2001!. ‘‘Pore-scale modeling and upscalinof nonaqueous phase liquid mass transfer.’’Water Resour. Res.,37~3!,539–549.
Hopmans, J. W., Nielsen, D. R., and Bristow, K. L.~2002!. ‘‘How usefulare small-scale soil hydraulic property measurements for large-svadose zone modeling?’’Heat and Mass Transfer in Natural Environment, D. Smiles, P. A. C. Raats, and A. Warrick, eds., American Gphysical Union Geophysical Monograph Series, in press.
Hristopulos, D. T., and Christakos, G.~1997!. ‘‘An analysis of hydraulicconductivity upscaling.’’Nonlin. Anal.-Theory Methods Appl.,30~8!,4979–4984.
Hristopulos, D. T., and Christakos, G.~1999!. ‘‘Renormalization groupanalysis of permeability upscaling.’’Stoch. Environ. Res. Risk Asses13~1–2!, 131–160.
Hunt, A. G. ~1998!. ‘‘Upscaling in subsurface transport using cluster stistics of percolation.’’Transp. Porous Media,30~2!, 177–198.
Indelman, P.~1993!. ‘‘Upscaling of permeability of anisotropic heterogeneous formations 3. Applications.’’Water Resour. Res.,29~4!, 935–943.
Indelman, P., and Dagan, G.~1993a!. ‘‘Upscaling of conductivity of het-erogeneous formations - General approach and application to isotrmedia.’’ Transp. Porous Media,12~2!, 161–183.
Indelman, P., and Dagan, G.~1993b!. ‘‘Upscaling of permeability of an-isotropic heterogeneous formations 1. The general framework.’’WaterResour. Res.,24~4!, 917–923.
Indelman, P., and Dagan, G.~1993c!. ‘‘Upscaling of permeability of an-isotropic heterogeneous formations 2. General structure and smallturbation analysis.’’Water Resour. Res.,29~4!, 925–933.
Ito, H., Raghuwanshi, N. S., and Wallender, W. W.~1999!. ‘‘Economicsof furrow irrigation under partial infiltration information.’’J. Irrig.Drain. Eng.,125~3!, 105–111.
Kersebaum, K. C., and Wenkel, K.-O.~1998!. ‘‘Modeling water and ni-trogen dynamics at three different spatial scales: Influence of diffedata aggregation levels on simulation results.’’Nutrient Cycl. Agro-ecosyst.,50~1–3!, 313–319.
King, P. R. ~1996!. ‘‘Upscaling permeability - Error analysis for renormalization.’’ Transp. Porous Media,23~3!, 337–354.
King, D., Fox, D. M., Daroussin, J., LeBissonnais, Y., and Danneels
UGUST 2002
rge
ndy
cy
V.on
essnal
i-
ter
to
fcasniv.
risk
nsem
—,
-
pic
a.’’
ctre-
gy
on.’’
rg,
and
s.’’
r
ts of
s
from
onln,
n
~1998!. ‘‘Upscaling a simple erosion model from small areas to a laregion.’’ Nutrient Cycl. Agroecosyst.,50~1–3!, 143–149.
Le Bissonnais, Y., Benkhadra, H., Chaplot, V., Fox, D., King, D., aDaroussin, J.~1998!. ‘‘Crusting, runoff and sheet erosion on siltloamy soils at various scales and upscaling from m2 to small catch-ments.’’Soil Tillage Res.,46~1–2!, 69–80.
Lee, D. J., and Howitt, R. E.~1996!. ‘‘Modeling regional agriculturalproduction and salinity control alternatives for water quality polianalysis.’’Am. J. Agric. Econom.78~1!, 41–53.
Lima, L. A., and Grismer, M. E.~1994!. ‘‘Application of fracture me-chanics to cracking of saline soils.’’Soil Sci.,158~2!, 86–96.
Llorens, P., and Gallart, F. A.~2000!. ‘‘A simplified method for forestwater storage capacity measurement.’’J. Hydrol.,240~1–2!, 131–144.
MacDonald, T. R., Kitanidis, P. K., McCarty, P. L., and Roberts, P.~1999!. ‘‘Mass-transfer limitations for macroscale bioremediatimodeling and implications on aquifer clogging.’’Ground Water,37~4!, 523–531.
Matthews, R. B., Wassmann, R., Knox, J. W., and Buendia, L. V.~2000!.‘‘Using a crop/soil simulation model and GIS techniques to assmethane emissions from rice fields in Asia. IV. Upscaling to natiolevels.’’ Nutrient Cycl. Agroecosyst.,58~1–3!, 201–217.
National Research Council~NRC!. ~1990!. ‘‘Opportunities in the Hydro-logic Sciences.’’Committee on Opportunities in the Hydrologic Scences, Water Science and Technology Board, National AcademyPress, Washington, D.C., 368.
Nichols, J. R., and Grismer, M. E.~1997!. ‘‘Measurement of fracturemechanics parameters in silty-clay soils.’’Soil Sci.,162~5!, 309–322.
Nitao, J. J., and Bear, J.~1996!. ‘‘Potentials and their role in transport inporous media.’’Water Resour. Res.,32~2!, 225–250.
Olesen, J. E., Bocher, P. K., and Jensen, T.~2000!. ‘‘Comparison of scalesof climate and soil data for aggregating simulated yields of winwheat in Denmark.’’Agr. Ecosyst. Environ.,82~1–3!, 213–228.
Or, D., and Tuller, M.~1999!. ‘‘Liquid retention and interfacial area invariably saturated porous media: Upscaling from single-poresample-scale model.’’Water Resour. Res.,35~12!, 3591–3605.
Pelmulder, S. D.~1994!. ‘‘ A framework for regional scale modeling owater resources management and health risk problems with astudy of exposure sensitivity to aquifer parameters.’’ PhD thesis, Uof California, Los Angeles.
Pelmulder, S. D., Yeh, W. W. G., and Kastenberg, W. E.~1996!. ‘‘Re-gional scale framework for modeling water resources and healthproblems.’’Water Resour. Res.,32~6!, 1851–1861.
Purkey, D. R., and Wallender, W. W.~2001!. ‘‘Drainage reduction underland retirement over shallow water table.’’J. Irrig. Drain. Eng.,127~1!, 1–7.
Refsgaard, J. C., Thorsen, M., Jensen, J. B., Kleeschulte, S., and HaS. ~1999!. ‘‘Large scale modeling of groundwater contamination fronitrate leaching.’’J. Hydrol.,221~3–4!, 117–140.
Renault, D., and Wallender, W. W.~2000!. ‘‘Nutritional water productiv-ity and diets.’’Agric. Water Manage.,45~2000!, 275–296.
Rubin, Y., and Gomez-Hernandez, J. J.~1990!. ‘‘A stochastic approach tothe problem of upscaling of conductivity in disordered mediatheory and unconditional numerical simulations.’’Water Resour. Res.26~4!, 691–701.
Russo, D.~1992!. ‘‘Upscaling of hydraulic conductivity in partially saturated heterogeneous porous formation.’’Water Resour. Res.,28~2!,397–409.
Sahimi, M., and Mehrabi, A. R.~1999!. ‘‘Percolation and flow in geo-logical formations: upscaling from microscopic to megascoscales.’’Physica A,266~1–4!, 136–152.
JOURNAL OF IRR
e
n,
Sanchez-Vila, X., Axness, C. L., and Carrera, J.~1999!. ‘‘Upscaling trans-missivity under radially convergent flow in heterogeneous mediWater Resour. Res.,35~3!, 613–621.
Sanchez-Vila, X., Girardi, J. P., and Carrera, J.~1995!. ‘‘A synthesis ofapproaches to upscaling of hydraulic conductivities.’’Water Resour.Res.,31~4!, 867–882.
Schulze, R.~2000!. ‘‘Transcending scales of space and time in impastudies of climate and climate change on agrohydrologicalsponses.’’Agric. Ecosyst. Environ.,82~1–3!, 185–212.
Soutter, L. A., and Loague, K.~2000!. ‘‘Revisiting the Fresno DBCP casestudy simulations: The effect of upscaling.’’J. Environ. Qual.,29~6!,1794–1805.
Sposito, G.~1998!. Scale dependence and scale invariance in hydrolo,Cambridge Univ. Press, Cambridge, U.K.
Stewart, J. B., Engman, E. T., Fedes, R. A., and Kerr, Y.~1996!. ‘‘Scalingup hydrology using remote sensing.’’ Wiley, Chichester, U.K.
Tegnander, C., and Gimse, T.~1998!. ‘‘Flow simulations to evaluate up-scaling of permeability.’’Math. Geol.,30~6!, 717–731.
Tidwell, V. C., and Wilson, J. L.~1997!. ‘‘Laboratory method for inves-tigating permeability upscaling.’’Water Resour. Res.,33~7!, 1607–1616.
Tidwell, V. C., and Wilson, J. L.~1999!. ‘‘Permeability upscaling mea-sured on a block of Berea Sandstone: Results and interpretatiMath. Geol.,31~7!, 749–769.
Tuong, T. P., and Bhuiyan, S. I.~1999!. ‘‘Increasing water-use efficiencyin rice production: Farm-level perspectives.’’Agric. Water Manage.,40~1!, 117–122.
van der Gon, H. A. C. D., van Bodegom, P. M., Houweling, S., VerbuP. H., and van Breemen, N.~2000!. ‘‘Combining upscaling and down-scaling of methane emissions from rice fields: Methodologiespreliminary results.’’Nutrient Cycl. Agroecosyst.,58~1–3!, 285–301.
Wagenet, R. J.~1998!. ‘‘Scale issues in agroecological research chainNutrient Cycl. Agroecosyst.,50, 23–34.
Wallender, W. W.~1987!. ‘‘Sample volume statistical relations for watecontent, infiltration and yield.’’Trans. ASAE,30~4!, 1043–1050.
Waller, P. M., and Wallender, W. W.~1993!. ‘‘Changes in cracking, watercontent and bulk density of salinized swelling clay field soils.’’SoilSci.,156~6!, 414–423.
Wen, X. H., and Gomez-Hernandez, J. J.~1996!. ‘‘Upscaling hydraulicconductivities in heterogeneous media - An overview.’’J. Hydrol.,183~1–2!, R9–R32.
Wen, X. H., and Gomez-Hernandez, J. J.~1998!. ‘‘Upscaling hydraulicconductivities in cross-bedded formations.’’Math. Geol.,30~2!, 181–211.
Wheatcraft, S. W., and Tyler, S. W.~1988!. ‘‘An explanation of scale-dependent dispersivity in heterogeneous aquifers using concepfractal geometry.’’Water Resour. Res.,24, 566–578.
Whitaker, S. ~1999!. Theory and applications of transport in poroumedia: The method of volume averaging, Kluwer-Academic, Dor-drecht, The Netherlands, 219.
Wohlfahrt, G., Bahn, M., Tappeiner, U., and Cernusca, A.~2000!. ‘‘Amodel of whole plant gas exchange for herbaceous speciesmountain grassland sites differing in land use.’’Ecol. Modell.,125~2–3!, 173–201.
Yang, W.~1997!. ‘‘Effects of spatial resolution and landscape structureland cover characterization.’’ PhD thesis, Univ. of Nebraska–LincoLincoln, Neb.
