REVffiWS OF GEOPHYSICS, SUPPLEMENT, PAGES 202-209, APRIL 1991 U.S. NATIONAL REPORT TO INTERNATIONAL UNION OF GEODESY AND GEOPHYSICS 1987-1990

Catchment Hydrology

David C. Goodrich and David A.Woolhiser

U.S. Department ofAgriculture, AgriculturalResearch Service, Aridland WatershedManagement Unit Tucson, Arizona

Introduction runoff and receiving water models with integration procedures. The paper describes an integration procedure for the EPA Although important progress has been made in catchment Hydrologic Simulation Program-FORTRAN (HSPF) to link sur hydrology research, a detailed, process based, understanding of facerunoff, subsurfaceflow(and quality)and receivingwaters. hydrologic response over a range of catchment scales (0.01-500 km ) still eludes the hydrologic community. Selected research Hammer and Kadlec [1986] developed a mathematical model efforts, primarily from refereed journals, conducted by U.S. (or for overland flow through wetland vegetated areas. Long term U.S. based) investigators for the period of late 1986 to 1990 are simulations were conducted without significant volume balance reviewed. The review is restricted to examination ofentire catch errors with a one dimensional numerical algorithm. The algorithm ment response ratherthan that ofa single process component such was validated using data from a peatland located near Houghton as routing or infiltration. Therefore, point or plot scale studies, Lake, Michigan. modeling ofa single process or analyses more strictly focused on Sivapalan etal. [1987], under a number ofsimplifyingassump erosion and water quality issues are not addressed unless they tions,presenteda modelfor storm runoffproductionwithspatially provide insights into catchment hydrologic behavior. Research that variable, temporally constant, rainfall on heterogeneous catch will be addressed that does not fit neatly into the above criteria ments whichaccounts for the effectsof catchment topography on includes automation of catchment modeling efforts, as well as runoff production via the topographic index method of Beven aspects of similarity and scale in catchment hydrology. Every [1986].In a subsequentpaper Sivapalan et al. [1990]generalized attempt was made to conduct a thorough review ofthe literature. the geomorphic unit hydrograph approach (GUH) to account for Our apologies to authors whose contributions have been over spatially variable initial soil moisture conditions and partial area looked. storm runoff response from both Hortonian and Dunne runoff generation mechanisms. Physically-based flood frequency es Distributedcatchment modeling timates are then numerically generated from spatially variable, Newmodel developments, model comparisons and assessment, temporally constant storm realizations, for assumed catchment treatment ofvariability, and automation are examined here. In the characteristics. Both approaches were used to investigate catch case where a new model is developed and evaluated, review ment scaleand similarity issues and arereferred to in a latersection. comments relating to both aspects are included in the new model Hirschi and Barfield [1988a, 1988b] described a physically development section.Treatment of uncertaintyis a centralissue in basederosionmodelwhich utilizes a two soil layer Green-Ampt both model development and evaluation and could be the basis of model for infiltration. Kinematic runoff routing was used for rill an entire review topic in its own right. The issue of uncertaintyis routingand a dynamicsurface storage routinetreated depression not dealt with in detail here but two reviews on the topic by Beck storage. Sensitivity analysis and testing of the erosion model [1987] and Haan [1989] are brought to the reader's attention. components rather than the hydrologic components were con ducted. Predicted sediment yieldisvery sensitive totheManning's NewModelDevelopmentsand Improvements nvalue used, reflecting the influence ofshear stress in the sediment New modeling efforts typically included generalizations of detachmentand transport equations. current models or addressed specific catchment submodel com Ormsbee and Khan [1989] conceptually incorporated the ponentweaknesses. Tliomas andBeasley [1986a, 1986b] modified mechanisms of micro and macropore flow into an "Extended ..theANSWERS {Beasley[1977]) model to include interflow so that Kinematic StorageModel" andevaluated themodel on foursteeply it could be used for forested catchments and tested the model on sloping forested watersheds with encouraging results. Geor- seven watersheds (1.3 to 16 ha) in the southeastern US. Thirty gakakos and Kabouris [1989] extended the GIUH watershed metersquare grid elements were used, apparently with no calibra modeling approach to account for subsurface runoff as well as tion.Regressions were used to compareobserved and computed spatialvariationsin landuseand rainfall.Hartley [1987] presented runoffvolumes and peak flows. A significantrelationship between a simplified procedure to estimatehydrographs fromhillslopesfor modeled and observed peak flows was found but runoff volumes use in an erosion model. were underestimated on four ofthe seven watersheds. Zhang and Cimdy [1989] develop a two-dimensional model for Donigian [1986] pointed out that many catchment runoff and overland flow and infiltrationwhich allows for spatial variations waterqualitymodelsdo notconsider the relationshipbetweenwater in roughness, infiltration and microtopography. The model was quality and agricultural practices and must include both field scale tested against characteristic solutions and experimental data. Simulations showthatmicrotopography impartsthe greatestvaria tion in overland flow depth, velocity and direction. Thispaperis not subjectto U.S.copyright.Publishedin 1991by the Woolhiseret al. [1990] described KINEROS, a kinematic wave American Geophysical Union. baseddistributed modelfor unsteadyrunoffand erosioncomputa tions.Newaspectsofthemodelincluded interactiveinfiltration(in

202

/ Goodrichand Woolhiser: CatchmentHydrology 203

which infiltration continues when rainfall ceases if uphill overland lead to significant improvements in model performance as flow supplies water) and a variable width, interactive infiltration measured by the Nash-Sutcliffe efficiency criterion. This finding channel routing algorithm for channel losses. Application to may be due to the model's inability to deal with small scale various USDA-ARS Walnut Gulch Experimental (Arizona) sub- infiltration variability within a model element. watersheds was also presented. Wilcox et al [1989b] analyzed USDA-ARS Reynolds Creek The modeling efforts presented do offer some significantim ExperimentalWatershed(Idaho)dataon fourcatchments(1-83ha). provements but alsopointoutseveralinconsistencies. The models Runoff was a small fraction of the total water budget for all developed by Sivapalan etal. [1987,1990] are specifically formu watersheds and processes associated with frozen soil were sig lated to be scale independent. Yet their development does not nificant. Both factors made it difficult to perform reliable modeling includemicrotopography,theimportanceofwhichwaspointedout with SPUR (Simulation of Prod, and Util. of Rangelands). Sub by Zhang and Cundy [1989] and Hirschi and Barfield [1988a]. sequentapplicationof the SPUR hydrology model [Wilcox et al., Clearly, the scale independent, unifying hydrologic principles 1989a] using prior calibration parameters on monthly runoff discussed by Dooge [1986] have not been found and models must volumes for threesubwatersheds in theReynolds CreekWatershed not be appliedoutsideof the range of scales for whichthey where (26 to 124 ha) were attempted. The Nash-Sutcliffe efficiencies for developed. validationperiods ranging from 4 to 12 years were 0.69, -1.40 and -0.24, indicating that mean monthly runoff was a better predictor Model Comparisons And Assessment than the model for two out of three watersheds. The authors Model comparisons were conducted both with and without concluded that the calibrated SPUR model can adequately predict observed data. In addition, more complex, distributed models, the volume and timing of snowmelt runoff if snow cover and were used in some cases to acquire insights and to improve the snowmelt is relatively uniform over the watershed. They em performanceofless complex, typically lumped catchmentmodels. phasized that calibration is needed and that the model is not Alternatively, simple models were assessed to see how well they expected to be reliable for snowmelt predictions on rangeland approximated more complicated rainfall-runoff models. Model watersheds. assessments were restricted to those cases where simulations were The performance of two derived flood frequency distribution comparedto observed data using promisingassessmentmethods. techniques was evaluated by Moughamian et al. [1987] for three Martinec and Rango [1989] examined a number of model watersheds. Each technique combined rainfall, infiltration and assessment statistics. They suggested that modelers refrain from catchment response models parameterized by measured and as adding unnecessary evaluation criteria which do not provide new sumed rainfall and catchmentcharacteristics. They found that each insight into model performance. With a limited number of criteria method performedpoorly in all ofthe watersheds even though good it is more likely that meaningful, not misleading, interpretations fits for individual events for the parameterized models were are made. A new method ofmodel assessment using the stochastic obtained. They suggested that the probabilistic rainfall and/or integral equation method was developed by Hromadka and Mc- watershed response models may not be generalenough to deal with Cuen [1989a,b]. Confidence intervals can be developed with the the large range ofevents encountered for a 50-year flood frequency proposed methodology with assumed probability distributions. analysis and concluded that fundamental improvements areneeded before derived flood frequency methods can be applied with Rawlsand Brakensiek[1986] compared runoff volume predic confidence. This study reiterates the need for model calibrationand tions made by a one-layer Green-Ampt model and the SCS curve verification over a range ofevent sizes and watershed conditions number technique for 330 runoffevents from 17 small, single land with reliable, well checked, distributed input-output data. use, watersheds. Parameters were estimated without optimization. The following efforts investigated the use ofcomplex models to The Green-Ampt procedure performed betterthan the curve num understand and gain insight into simpler models. Shen et al. [1990] bermethod. Wardetal. [1989] tested seven runoffmodels for their employed a distributed kinematic rainfall-runoff model with ability to predictrunoffpeaks and volumes for four storms for each uniform rainfall using Horton's infiltration relationship on four of26 South African catchments ranging in size from a few hectares hypothetical catchments with 48 assumed combinations ofcatch to 100 km . Land uses included urban, agricultural, forest and ment characteristics. The time ofconcentration (Tc), peak runoff veld. Models included the rational method, three versions of rate (Qp),and time to peak (Tp) are numerically derived for each American SCS methods, the ILLUDAS time area method, a combination. Simplified, physically based, relationships for Tc, distributed kinematic runoff model and a South African unit Qp, and Tp are then derived using the concept of the controlling hydrograph method. Predictions were made without optimization catchment (for Tc), and an average overland flow plane (for Qp). ofparameters. As might be expected, each method worked best Tc, Qp,and Tpare then computed using the simplifiedrelationships for the particular land use condition it was designed to model. and compared to the numerically derived variables. The simple ^ However theoverallperformance ofallmodels wasdisappointing. \ relationships provided good results on the synthetic data set where Clark's synthetic unit hydrograph method and the Extended input errors are eliminated. They also found that three common Kinematic Storage Model were applied to three steeply sloping synthetic hydrograph methods did not reproduce the results ofthe watersheds in Kentucky and West Virginia by KJmn and Ortnsbee kinematic wave models for Tc, Qp, and Tp nearly as well as the [1988]. Both models were satisfactory when calibrated with exist proposed method. Shen et al. [1990] then derived physically based ing data but only the Extended Kinematic Storage Model survived flood frequency curves which include infiltration considerations. a series ofsimple validation tests. Both Troutman et al [1989] and It was found that soil type and initial soil moisture conditions exert Hetricket al. [1986] found that simulations from a simple model the greatest influence on flood frequency distributions with basin compared reasonably well with more complex model process shape and slope playing secondary roles. representation. Akan and Al-Turbak [1988] used the hydrologic similarity Loague [1990] utilized additional information [Loague and concept to generalize the results ofa kinematic wave model with Gander, 1990] on spatial variation of infiltration to re-evaluate a a Green and Ampt infiltration component to interpret the runoff quasi-physicallybased rainfall-runoff model. The new data did not coefficient ofthe rational formula. A simple rectangular basin with

/ 204 Goodrichand Woolhiser: CatchmentHydrology uniform physical properties with time varying, uniform rainfall At the hillslope and small catchment scale, a high degree of intensity, was used in the approach. The runoff coefficient is spatial variation ofinfiltration propertieswere found by Wilson and viewed as a nondimensional parameter which can be numerically Luxmoore[1988], and Loague and Gander [1990]. Short correla evaluated in terms of basin and rainfall non-dimensional tion scales (10m) were demonstrated in the first two articles and parameters. Using this approach Tc need not be estimated and the Wilson and Luxmoore [1988] concluded that infiltration can be assumption that the peak discharge calculated by the rational treated as a stochastic process regardless ofthe initial soil tension formula is the same return frequency as the rainfall is not required. for modeling hillslope hydrology on the Tennesseewatershed they GarciaandJames [1988] employed the kinematic wave version of studied. HEC-1 to simulate the impact of urbanization on hydrologic For a hillslope with an arbitrary spatial arrangement of in response. Regression relationships were developed to quantify the dividual point infiltration capacities, Hawkins and Cundy [1987] ' effect ofurbanization on the UH. established an envelope of steady state surface runoff responses. GanandBurges [1990a, 1990b] used the quasi-physically based Woolhiser and Goodrich [1988] implemented a simple stochastic , model (S-H) ofSmith andHebbert [1983] to generate "true" time representation of small scale, cross slope, infiltration variability •' series of runoff and evaporation for several climate-soil-basin within a distributed model element. They found significant rainfall geometry combinations on a small hypothetical catchment. This and soil variability interactions. Mean (equivalent) representations generated "true" series was then used to assess the soil moisture ofsaturated hydraulic conductivity were found to be incapable of accounting (SMA) component of the National Weather Service reproducing variable representations for time variable rainfall. Sacramento model for calibration, validation and prediction. For Woolhiserand Goodrich[1988] also investigatedthe importance the scenarios studied, they concluded that the SMA proved unreli of time varying, spatially uniform, rainfall by using observed able and unsuitable to address issues ofscale and heterogeneity. In breakpointrainfallintensities and intensities derived from several particular the SMA model is unreliable in predicting hydrologic disaggregation schemes in a watershed model with a physically response from extreme rainfall (or behavior outside the range of based infiltration componenton a elementary open book watershed. calibration events), is highly sensitive to initial soil moisture, and The derived runoffdistributions from each rainfall disaggregation tends to overestimate outflows during dry to wet transitions. They scheme and the breakpoint data were then compared. They found also found thatparameterestimatesareclimate sequencedependent that disaggregating total rainfall amounts into simple constant and and therefore caution should be taken by investigators such as triangular intensity distributions caused significantdistortion in the Glieck [1987] who used calibrated conceptual models to explore runoff distributions in all but highly damped systems. The disag catchment response to climate change scenarios. gregation scheme of Woolhiser and Osborn [1985] was found to Many of the papers reviewed reiterated the well known con be superior to these simpler forms of disaggregation. Ormsbee clusion that uncertainty in the effective rainfall distribution [1989] also found that the uniform disaggregation grossly under dominated runoff prediction uncertainty. Modeling efforts or estimates peakdischarge frequencies from a continuous hydrologic evaluations based on simulated input-output data may be used to model for four rainfall stations. Ormsbee's [1989] disaggregation gain model insight but are oflittle real world value in light of this scheme did substantially betterthan the uniform case but still under conclusion. Thosemodelevaluationswhich utilized observeddata, estimated observed frequencies. do not paint an encouraging picture of our ability to model The impact of spatial rainfall and soil information on runoff catchment response. Clearly, we cannot sit on our laurels and predictions at the hillslope scale were analyzed by Loague [1988]. proclaim a detailed understanding ofcatchment hydrology. Model Fine scale realizations ofspatial hydraulic conductivityand rainfall ing research, distributed data collection, and objective evaluation fields were aggregated to coarser resolutions and subsequent ofmodels with a large number ofevents covering a wide range of impacts on hillslope rainfall were assessed. At the hillslope scale response conditions must continue. it was found that hydraulic conductivity field knowledge appears Treatment ofCatchment Variability to be more critical than rainfall information. Different runoff Pursuit of the treatment and understanding of spatial and tem evaluation variables (peak rate, time to peak and runoff volume) poral variability of catchment characteristics, states, and rainfall were also found to require different information levels. inputs is one ofthe most fruitful directions to further ourknowledge A promising, although simulation based, methodology to ad ofhydrologic catchment behavior. Parameters and processes con dress the myriad variability at the catchment scale is the concept trolling hydrologic response operate at many different time and ofa Representative Elementary Area (REA) put forth by Woodet length scales. Consistenttreatment ofthese processes requires that al. [1988]. Thisconceptis analogous to theRepresentativeElemen observed and model (numerical) scales be commensurate. taryVolume conceptemployed foryearsingroundwater investiga- H The variation in storm size, relative to catchment size, was tions. For scales smaller than the REA, spatial variability of explored by Eagleson and Qinliang [1987] using a kinematic wave relevant variables must be treated in a more detailed manner. For * based runoff model with a conceptual rainstorm model. They scales greaterthantheREA,detailed variability neednotbetreated A concludedthatboththefirstandsecond momentofpeakstreamflow explicitly but average or macroscale relationships may be used. A decrease rapidly with increasing values ofthe catchment to storm modified version ofBeven andKirkby's [1979] TOPMODEL was scale ratio. Milly and Eagleson [1988] demonstrated the critical used to test the existence of an REA for Coweeta watershed need to incorporate some representation ofareal storm variability topography using simulated fields ofrainfall and soils parameters. in large area hydrologic models. Milly and Eagleson [1987] also They found that an REA (approximately 1 knr) does exist in the studied the effects of spatial variability of soils and vegetation on context of catchment hydrologic response and is more strongly spatially and temporally averaged hydrologic fluxes. An equivalent influenced by basin topography than rainfall length scales. homogeneous soil that would reproduce the average behavior ofa In a somewhat analogous manner at the hillslope scale Julien heterogeneousarea was found only in the case wherea small degree andMoglen [1990] assessed the impacts ofspatially varied slope, ofinitial variability existed. Theyconcluded that becauseoftypical roughness, width and excess rainfall intensity on runoffbehavior. nonlinear catchment response behavior, spatial integration to the They derived a length scale based on surface characteristic and catchment scale will be difficult. excess rainfall duration assuming uniform kinematic overland Goodrichand Woolhiser: CatchmentHydrology 205 flow. For overland flow with slope lengths much less than the resolutionofcatchment descriptors (topography, soils, vegetation), lengthscale, spatialvariabilityhas littleinfluenceon runoffas the and in the speed of obtaining them will not result in improved systemwill be predominately in a runoff equilibrium condition. hydrologic prediction unless appropriate mod«ls are used. For lengths greater than the computed lengthscale,variability of surfaceand input parameters will have a much greater effect on Experimentalcatchmenthydrology runoff(nonequilibriumconditions). This length scale can serve as The relativelylarge number of reports dealing with experimental a guideline for model grid size selection. catchment research is encouraging. A primary conclusion of the Several innovative techniques for the treatment of catchment U.S. National Research Council [1991] is that hydrology is a data variability are evident in the papers reviewed. In several cases poor science. Continued efforts to collect data, coupled with hybrid schemes, incorporating both stochasticand deterministic specific hypotheses and modeling efforts is essential for a more elements, have shown promise. They represent an attempt for thorough understanding ofcatchment behavior. A catalog of data balanced model treatment ofdata and hydrologic processes known sourcesand case studies ofwatershedprojectsin the westernUnited with various degree of uncertainty. These efforts are encouraged. States and Canadian Provinces was compiled by Callaham[1990]. Areport on longterm interdisciplinary research and datacollection Automation at the Coweeta watershed was presented by Swankand Crossley Increases in model complexity typically yielded concomitant [1988]. A nationwide effort to understand and model erosion increasesin inputdatarequirements,catchmentdescriptors,model processes (WEPP: Water Erosion Prediction Project [Lane and parameterization and output interpretation. For relatively simple Nearing, 1989]) has also spurred detailed, interdisciplinary data models, the overall modeling effort was aided by spreadsheet collection and modeling investigations. This and other water software [Walker et al., 1989; Dymond and McDonnell, 1988]. quality efforts [Burkart et al., 1990] underscored the importance For more complicated, distributed catchment modeling, extraction of knowledge of hydrologic behavior before transport ofsoils and of basin network, drainage area, and topographic characteristics chemicals can be adequately predicted [Lawrence et al., 1988]. wasundertakenbyMar* [1988], Band[19S9],Mooreetal. [1988], The majority ofthe experimental work reviewed can be broadly and JensonandDomingue [1988], Catchment descriptors, such as grouped into categories of hydrologic effects ofwatershed treat percent impervious area were automatically estimated from ment, chemical hydrograph separation, and macropore studies. remotely sensed video data by Draper and Rao [1986]. Their Catchment response to logging, range management practices, methodology provided estimates of impervious area as accurately reforestation and conversion oftree and brush areas to grasslands as manual methods with significant cost savings. Automation of were addressed by Blackburn et al. [1990], Higgins et al. [1989], topographic and land use information for HEC-1 and subsequent Heede[1987], Riekerd[1989], Trunbleetal. [1987], Davis [1987], graphical output was described by Cline et al. [1989]. For new Wright et al. [1990], and Keppeler andZiemer [1990]. Greater model users, decision support systems provided an overall aid in response differences between control and treatment catchments for the selection ofmodel input values by accessing large data bases, small runoff events (or dry years) and smaller or negligible interfacing with expert systems, suggesting default values and differences for larger events were noted by Trimble et al. [1987], providing graphics [Arnoldand Summons, 1988]. Wright et al [1990], and Keppeler andZiemer [1990]. Wrightet Geographic Information Systems (GIS) were employed to es al [1990] pointed out problems in previous paired watershed timate SCS curve numbers and subsequent runoff peaks and studies in which no large runoff events occurred in either the volumes by White [1988], and BerryandSailor [1987]. However, pretreatment or post treatment periods, thereby influencing con Berry and Sailor [1987] noted that if only a one-time calculation clusions regarding logging effects on large flows. ofbasin parameters is required, the automation and added precision A large body ofwork on chemical hydrograph separation was of using a GIS might not be offset by the required digital data conducted during the review period [Caine, 1989; McDonnell et collection efforts. Addressing this point, Johnson [1989] presented al, 1990; Swistock et al, 1989; Burns, 1989; Nolan and Hill, an integrated catchment modeling approach based explicitly on 1990; Pionke et al., 1988; Lawrence et al., 1988; and McDonnell, geographic information system (GIS) functions. Interactive 1990]. Hydrograph separation of "new" runoff water from "old" graphics were used to quickly develop catchment digital databases groundwater is directly related to the acid precipitation neutraliza for soils, topography, land use and both point and radar derived tion hypothesis. This hypothesis states that the water chemistry is rainfall distributions. A variety ofhydrologic models ranging from determinedprimarily by its residence time within the soil. This line unit hydrograph methods to variable source area models with of research has spurred analyses of flow generation mechanisms various infiltration and evapotranspiration components were where "old" groundwater or soil water has been observed to be a parameterized and used in his modeling system. large portion oftotal storm runoff. An automated model development and building methodology Nolan and Hill [1990] found that flood wave effects offer an was described by Arnold and Williams [1988]. They utilized a explanation to the apparentincompatibilityofobserved "old water" fourth generation simulation language based on queuing theory to domination ofearly storm runoffwater for a test catchment. Flood modelthe runofffrom a small homogeneous watershed. Analogous waves, composed primarily of old channel water, tend to reach queuing theory operators and constructs were used to model basin outlets before "new" storm runoffderived from impervious processes of precipitation, surface runoff, lateral flowand percola catchment areas arrives. This can occur when new water rapidly tion. The feasibility of the modeling approach was demonstrated enters the channel from a localized upstream source and generates for a runoff producing storm on a small Ohio catchment. a flood wave which travels down the channel at a greater velocity Progress in automation ofcatchment hydrologic modeling has than the mean water velocity. They were able to confirm the been good and continued research in this area is encouraged. presence of flood waves in a California catchment by using data Advances will further technology transfer to the practitioner and from a gaging station 600m upstream from the basin outlet. They free the researcher from many tedious data and model layout tasks. concluded that the rapid influx of old water did not originate from However, practitioners must remember that improvements in the thedisplacementofsoilmoisture as was proposed inearlier studies. 206 Goodrichand Woolhiser: Catchment Hydrology

For small events (< 2mm/hrpeak flow overthe catchmentarea) tance of first order channels as a major source ofsurface runoff. in a New Zealand catchment, McDonnell [1990] concluded that The general shape of frequency histograms of first order channel near-stream, valley bottom (old) groundwater could be discharged distances from the basin outlet were quite similar to hydrographs in sufficient quantitiesto account for storm period streamflow. For from general storms for basins in Pennsylvania. They concluded larger storms ( 2mm/hr peak flow) old water production again thatthe surface runoffhydrograph shape was closely controlled by dominated flow as it was derived from hillslope-hollow drainage the distribution of first order channel distances. into steeply sloping firstorderchannels. Fastdrainageofoldwater The study ofcatchment geomorphology has been used as a key into the first order channels was facilitated by crack infiltration, to acquire insight into scale issues and basin response behavior. slope water tabledevelopment and lateralpipe flow ofstoredwater. Other investigations, more strictly related to catchment geomor Pionke et al [1988]also provided a detailed analysis ofdata from phology, could not be reviewed in detail due to space limitations the USDA-ARS Mahantango Creek watershed in Pennsylvania to but are listed for reader convenience. They include Deutsch and explain storm response runoffgenerating mechanisms. They con Ramos [1986], Hjelmfelt [1988], Gupta and Mesa [1988], Gupta cluded that the highly dynamic near stream hydrologic environ and Waymire [1989], Tarboton et al. [1988, 1989], and ment, although areally a small portion ofthe catchment, exerted a Montgomery and Dietrich [1989]. major control over streamflow hydrology and chemistry. In a series of papers Wood and Hebson [1986], and Sivapalan Macropore characterization studies were conducted by Wilson et al [1987,1990] discussed basin response similarity and derived andLuxmoore [1988], Edwards etal [1988], and Ursicand Esher a methodology utilizing TOPMODEL. The model is expressed in [1988].Thelatter investigation demonstrated that increases in small dimensionless form with consideration of water table and initial mammal burrowing activity can significantly increase detention- local storage deficit conditions prior to a storm. Five dimensionless retention storage of rainfall in the upland Coastal Plains of the catchment similarity parameters and three conditions are used in southern United States. the model equations. Under the assumptions oftheir analysis, two Jarrett [1990] described an ongoing research investigation by catchments are considered hydrologically similar ifthey are iden the USGS to improve the understanding of hydrologic and tical in the five parameters, regardless ofscale. hydraulic processes in mountainous regions. The goal of this Theoretically derived flood frequency distributions were also multidisciplinary investigation is improving paleohydrologic tech described in terms of scaled dimensionless parameters with con niques with thorough data analysis as well as identification ofthe siderationofrelative catchment to storm scales by Sivapalan et al sources and magnitude ofdata errors. [1990]. They concluded that for infiltration excess dominated Experimental studies are essential if we are to attain a truly catchments, with assumed uniform Hortonian overland flow scientific understanding of hydrologic phenomena. The simul generation, the flood frequency distribution is completely defined taneous measurement of solute concentrations as well as water by a rainfall-soil scale parameter and the scaled catchment area fluxes and stores promises to provide data that will aid our emphasizing the need for accurate rainfall distribution information. understanding ofrunoff generation mechanisms. However, inter Theseanalyses provided atheoretical framework to compare storm pretations ofintensive but short-term studies must be made care response for catchments with different characteristics at different fully because they sample a very limited set ofcatchment states. scales for both infiltration and saturation excess runoffgeneration. Ideally such short term intensive studies should be carried out on Much ofthis work was also summarized in a review by Wood et an experimental catchment that has been monitored for a long al [1990]. period oftime. The length scale derived byJulien andMoglen [1990] (also see above), based on concepts ofkinematic equilibrium, can also serve Basinscale and similarity as an indicator of hillslope similarity. They found that the time to equilibrium is essentially the same for spatially varied conditions Understanding catchment response over a wide range ofbasin or spatiallyaveraged surface parameters. In addition, runoff sen scales and identifying easily derived measures that quantify dif sitivities were typically the same for uncorrelated and correlated ferences and similarities ofcatchment behavior remain important surface parameters. and active area ofresearch. Progress in these fields is paramount ifour collective knowledgeofcatchments is to aid in understanding The future regional and global hydrologic phenomena [see Wood this issue]. An improved understanding of many conjunctive use, water Gupta et al [1986] is referenced here for completeness as it quality, and climate change issues requires parallel advancement provides relevant information on the topic. in catchment hydrology. Unfortunately in the present atmosphere In the quest to scale up point based process knowledge Morel- ofprioritized research,this fact is often overlooked. Clearly ifwe Seytoux [1988] argued that nature embodies both the elements of cannotaccurately predict catchment runoffresponse, prediction of chance and the descriptive laws of physics. Therefore, excessive sediment or chemical transport by water are even more suspect. process description at one scale is lost through the processes of El-Kadi [1989] reviewed numerous currently available water integration in time and space, and via expectation. This justifies shed models to assess howthey might be used to describe surface- model simplification as long as the essential behavior is retained. groundwater interactions for conjunctive water use management. He illustrated how simplifications can be made so that straight The review concluded that it is not yet feasible to extend the models forward scaling integration can be accomplished in a physical- examined to handle conjunctive use for both surface and ground stochastic modeling framework. water resources. A need therefore exists to improve the surface In exploring runoffresponse at the catchment scale, Rogersand water-groundwater linkage in new and existing watershed models Singh [1986] studied the number of first order channels and the while ensuring that model components treat complexity and scale distribution of first order channel lengths from the outlet. They in an integrated manner. found that the number of first order channels was inversely Simulations by Davis and Goldstein [1988] demonstrated that proportional to basin soil infiltration capacity and cite the impor the effectiveness ofmitigation strategies for lake acidification are Goodrichand Woolhiser: Catchment Hydrology 207 critically dependent uponknowledge of the dominant hydrologic tion) with high quality, distributed catchment data. Simple, graphi flowpaths within a catchment. Siegel[1988] argued thatunlessa cal comparisons of observed and simulated sequences for a small more detailed understanding of wetland hydrology, water number ofevents, with few correlation statistics, provide insuffi chemistry, and biota is acquired it will be extremely difficult to cientknowledgeoflikely operational model performance. Several assessthe cumulative impactsofchangingcatchmentconditions. objective evaluation criteria, as well as graphical comparisons [Martinec andRango, 1989; Willmott et al, 1985]; whiletesting Hakonson et al [1986] pointed out the shortcomingsofcurrent overa wide range of input-output conditions [Sorooshian et al, water balance model feedback mechanisms to account for long 1983; Gan and Burges, 1990a, 1990b], should be considered a term, time dependent, changes in hydrologic, biotic and soils regimes. Their particular studyconcerned the longtermviability minimum for model testing. The problem of parameter iden- of low level radioactive waste trench caps. However, models tifiability and proper model structure should also be addressed capable of dealing with evolvingregimeswill also be critical for [Sorooshian and Gupta, 1983, 1985; Hendrickson etal, 1988]. Verification using stormflowmeasures will be difficult if runoff evaluation ofclimate change scenarios. per unit area is of the same magnitude as rainfall measurement Understanding catchmentprocessesand responseasa function errors (low signal to noise ratio). This is especially true when oftime and space scales in a variety of hydroclimatic regimes is uncertainty in catchment rainfall can be large for the typical crucialifwe are to deal with the impacts ofglobal climate change assumption ofuniform rainfall from a nearby gage. and feedbacksofthe Earth's large scale dynamic climate. To attack For distributed model testing, internal (subwatershed) response this problem in a cost effective manner, hydrologic models must behavior should also be verified. Klemes [1986] described a be reformulatedto directlyutilize availableandanticipated remote ly sensed data. However, for predictive capability, these refor systematic methodology for the caUbration and verification of mulations must remain process based and be verified with hydrologic simulation models. His work is cited by many as an example of how modeltesting should be donebut apparently is experimental data. heeded by few. Without thorough testing to establish model A new age ofinterdisciplinaryexperimental andmodeling study confidence, subsequent conclusions regarding catchment behavior is uponus. Futurelargescale fieldexperiments, suchas STORM, are often speculative. GEWEX (Global Energy and Water Balance Experiment) and To perform defensible modelcalibration and verification, well NASA EarthObserving System efforts, offer exciting possibilities checked datasets from a variety ofhydroclimatic regimes must be to couple hydrologic models over a broad range of scales and made easily available in electronic format. El-Kadi [1989] directly incorporate remotely sensed data. reiterated that most watershed models have been tested with some Futureresearch inthearea ofcatchmentmodel formulation must field data but a need exists for a set ofstandardized test problems address the determination of proper model component repre with respective data sets. Rainfall-runoff data and a watershed sentation (system complexity), the scales over which the com topographic map alone are insufficient for distributed catchment ponents are valid, and integration of the components. Pulling model testing. Soils, land use information, soilwater content, and sub-model components from varying sources and patching them internal(subwatershed) response measurements are also required. together without considering the issues of commensurate com The importance and integration of field data into our research plexity and scale treatment will only invite trouble. Nor can efforts cannot be overemphasized. integrated model formulation ignore the sources of uncertainty Acknowledgements: Special thanksareextendedtoJean-Marc Faures for inducedby uncertaininput and parameterdistributions. conducting a computerliterature search, to Laura Yohnka for expedient Objective model assessment is an area we must improve upon. computer typesetting andtoTim Keefer, KenRenard,DaveDawdy,Steve Far too many published works have not followed a rigorous Burges and two anonymous reviewers for thorough reviews and helpful protocolofsplit sample model calibrationand verification(valida suggestions.

References

Akan, A. O., and A. A. Al-Turbak, The Berry, J. K., and J. K. Sailor, Use of a 71 (29):980-981,988,1990. rational formula interpreted using a geographic information system for storm Burns, D. A., Speciation and equilibrium physically-based mathematical model. runoff prediction formsmallurbanwater relations of soluble aluminum in a head Notxlic Hydrol, 19:41-52,1988. sheds. Environ. Manage., ll(l):21-27, water stream at base flow and during rain Arnold, J. G., and N. B. Sammons, 1987. events. Water Resour. Res., 25(7):1653- DecisionSupport system for selecting in Beven, K. J., Runoff productionand flood 1665,1989. puts to a basin scale model. Water frequency in catchments of order n: An Caine, N., Hydrograph separation in a small Resour. 5«//.,24(4):749-759, 1988. alternative approach, in Scale Problems alpine basin based on inorganic solute con Arnold, J. G., and J. R. Williams, in Hydrology, Ed. by V. K. Gupta, I. centrations. J. Hydrol, 112:89-101,1989. Hydrologic modeling using a simulation Rodriquez-Iturbe, and E. F. Wood, D. Callaham, R. Z., Case studies and catalog of languagebased on queuing theory. Water Reidel, Hingham, Mass., 1986. watershed projects in western provinces Resour. 5wtf.,24(4):861 -868,1988. Beven, K. J., and M. J. Kirkby, A physi and states. Wildland Resources Center, Band, L. E., A terrain-based watershed in cally-based variable contributing area Div. of Agric. and Nat. Resour., Univ. of formation system. Hydrol. Proc, 3:151- model of basin hydrology, Hydrol. Sci. Calif., 145 Mulford Hall. Berkely, Calif., 162,1989. Bull, 24(l):43-69,1979. 94720, Report 22, June, 1990. Beasley, D. B., ANSWERS: A mathemati Blackburn, W. H., R. W. Knight, J. C. Cline, T. J., A. Molinas, and P. Y. Julten, cal model for simulating the effects of Wood, and H. A. Pearson, Stormflow An AUTO-CAD-based watershed informa land use and managemnet on water and sediment loss from intensively tion system for the hydrologic model HEC- quality.Ph.D. Thesis, Purdue Univ., W. managed forest watersheds in east Texas. 1. Water Resour. Bull., 25(3):641-652, Lafayette, Indiana,266 p., 1977. Water Resour. Bull., 26(3):465-477, 1989. Beck, M. B., Water quality modeling: A 1990. Davis, E. A., Chaparral conversion and review of the analysis of uncertainty. Burkart, M. R., C. A. Onstad, and G. D. streamflow: Nitrate increase is balanced Water Resour. Res., 23(8): 1391-1492, Bubenzer, Research on agrichemicals in mainly by a decrease in bicarbonate. Water 1987. water resources. Eos: Trans. AGU, Resour. /tes.,23(l):215-224,1987. 208 Goodrichand Woolhiser: CAtchmentHydrology

Davis, J. E., and R. A. Goldstein, Simulated Hakonson, T. E., G. R. Foster, L. J. Lane, steeply sloping forested watersheds. J. response of an acidic Adirondack lake and J. W. Nyhan, The application of an Hydrol, 109:325-349,1989. watershed to various liming mitigation agricultural water balance and erosion Klemes, V., Operationaltesting ofhydrologi- strategies. Water Resour. Res., 24(4):525- model in environmental science. In: cal simulation models. Hydrol. Sci. J., 532,1988. Agricultural Nonpoint Source Pollution: 31(1):13-24,1986. Deutsch, S. J., and J. A. Ramos, Space-time Model Selectionand Application;Develop Lane, L. J., and M. A. Nearing, USDA modeling of vector hydrologic sequences. ments in Environmental Modeling. El Water Erosion and Prediction Project: Water Resour. flu//., 22(6):967-981,1986. sevier, New York, p. 173-190,1986. Hillslope profile version. ARS-NSERL Dooge, J. C. I., Looking for hydrologic Hammer, D. E., and R. H. Kadlec, A model Report #2, Nat. Soil Eros. Res. Lab., W. laws. Water Resour. Res., 22(9):46S-58S, for wetland surface water dynamics. Water Lafayette, In., 1989. 1988. Resour. /?«.,22(13):1951-1958,1986. Lawrence, G. B., C T. Driscoll, and R. D. Donigian, A. S., Integration of runoff and Hartley, D. M. Simplified process model for Fuller, Hydrologic control of aluminum receiving water models for comprehensive water sediment yield from single storms. chemistry in an acidic headwater stream. watershed simulation and analysis of Part I-Model formulation. Trans. ASAE Water Resour. Res., 24:659-669,1988. agricultural management alternatives. In: 30(3):710-717,1987. Loague, K., and G. A. Gander, R-5 Agricultural Nonpoint Source Pollution: Hawkins, R. H., T. W. Cundy, Steady-state Revisited, 1: Spatial variability of infiltra Model Selection and Application; Develop analysis of infiltration and overland flow tion on a small rangeland catchment. Water ments in Environmental Modeling. El for spatially-varied hillslopes. Water Resour. /?«.,26(5):957-972,1990a. sevier, New York, p. 265-275,1986. Resour. 2to//.,23(2):251-256,1987. Loague, K., R-5 Revisited, 2: Reevaluation Draper, S. E., and S. G. Rao, Runoffpredic Heede, B. H., Overland flow and sediment of a quasi-physically based rainfall-runoff tion using remote sensing imagery. Water delivery five years after timber harvest in a model with supplemental information. Resour. *«//.,22(6):941-949,1986. mixed conifer forest, Arizona, U.S.A. /. Water Resour. Res., 26(5):973-987,1990. Dymond, R. L., and A. J. McDonnell, Hydrol, 91:205-216,1987. Loague, K. M., Impact of rainfall and soil Determination of ungaged tributary flows Hendrickson J. D., S. Sorooshian, and L. E. hydraulic property information on runoff and intersegmental reservoir flows using Brazil, Comparison of Newton-type and predictions at the hillslope scale. Water spreadsheet software. Water Resour. Bull., direct search algorithms for calibration of Resour. Res.,24(9):1501-1510,1988. 24(5):1065-1072,1988. conceptual rainfall-runoff models. Water Mark, D. M., Network models in geomor Eagleson, P. S., and W. Qinliang, The role Resour. /to.,24(5):691-700,1988. phology. In: Modelling Geomorphological of uncertain catchment storm size in the Hetrick, D. M., C. C Travis, P. S. Shirley, Systems, M. G. Anderson (Ed.), John moments of peak streamflow. J. Hydrol., and E. L. Etnier, Model predictions of Wiley and Sons, Ltd., p. 73-97,1988. 96:329-344,1987. watershed hydrologic components: Com Martinec, J., and A. Rango, Merits ofStatis Edwards, W. M., L. D. Norton, and C E. parison and verification. Water Resour. tical Criteria for the performance of Redmond, Characterizing macropores that fi«//.,22(5):803-810,1986. hydrological models. Water Resour. Bull., affect infiltration into nontilled soil. Soil Higgins, D. A., A. R. Tiedemann, T. M. 25(2):421-432,1989. Set. Soc. Am. /.,52(2):483-487,1988. Quigley, and D. B. Marx, Streamflow McDonnell, J. J., A rationale for old water El-Kadi, A. I., Watershed Models and their characteristics of small watersheds in the discharge through macropores in a steep, applicability to conjunctive use manage blue mountains of Oregon. Water Resour. humid catchment. Water Resour. Res., (in ment. Water Resour. Bull, 25(1):125-137, Bull, 25(6): 1131-1149,1989. press), 1990. 1989. Hirschi, M. C, and B. J. Barfield. KYER- McDonnell, J. J., M. Bonell, M. K. Stewart, Gan, T. Y., and S. J. Burges, An assessment MO-A physically based erosion model, and A. J. Pearce, Deuterium variations in of a conceptual rainfall-runoff models's Part I. Model development. Trans. ASAE, storm rainfall: Implication for stream ability to represent the dynamics of small 31(3):804-813,1988a. hydrograph separation. Water Resour. hypothetical catchments 1. Models, model Hirschi, M. C, and B. J. Barfield. KYER- Jto.,26(3):455-458,1990. properties, and experimental design. Water MO-A physically based erosion model, - Milly, P. C D., and P. S. Eagleson, Effect Resour. Res.,26(7):1595-1604,1990a. Part II. Model sensitivity analysis and test of spatial variability on annual average Gan, T. Y., and S. J. Burges, An assessment ing. Trans. ASAE, 31(3): 814-820, 1988b. water balance. Water Resour. Res., of a conceptual rainfall-runoff models's Hjelmfelt, A. T., Jr., Fractals and the river- 23(11):2135-2143,1987. ability to represent the dynamics of small length catchment-area ratio. Water Resour. Milly, P. C. D., and P. S. Eagleson, Effect hypothetical catchments 2. Hydrologic fl«//.,24(2):455-459,1988. of storm scale on surface runoff volume. responses for normal and extreme rainfall. Hromadka, T. V., and R. H. McCuen, Water Resour. ifcw.,24(4):620-624,1988. Water Resour. Res., 26(7):1605-1619, Evaluation of rainfall-runoff performance Montgomery, D. R., and W. E. Dietrich, 1990b. using the stochastic integral equation Source areas, drainage density and channel Garcia, A., Jr., and W. P. James, Urban method. Stoch. Hydrol. and Hydraul, initiation. Water Resour. Res., 25(8):1907- runoff simulation model. J. Water Resour. 3:217-226,1989a. 1918,1989. Planning Manag., ASCE, 114(4):399-413, Hromadka, T. V., JJ, and R. H. McCuen, Moore, I. D., E. M. O'Loughlin, and G. J. 1988. An approximate analysis of surface runoff Burch, A contour based topographic model Georgakakos, K. P., and J. C. Kabouris, A model uncertainty. J. Hydrol, 111:321- for hydrological and ecological applica streamflow model using physically-based 3601989b. tions. Earth Sur. Proc. Landforms, instantaneous unit hydrographs. J. Jarrett, R. D., Hydrologic and hydraulic re 13(4):305-310,1988. Hydrol, 111:107-131,1989. search in mountain rivers. Water Resour. MoreUSeytoux, M. J., Soil-aquifer-stream Glieck, P. H., The development and testing Bw//.,26(3):419-429,1990. interactions - A reductionist attempt toward of a water balance model for climate im Jenson, S. K., and J. O. Domingue, Extract physical-stochastic integration. /. Hydrol, pact assessment: Modeling the Sacramento ing topographic structure from digital 102:355-379,1988. basin. Water Resour. Res., 23(6):1049- elevation data for geographic information Moughamian, M. S., D. B. McLaughlin, and 1061,1987. systems analysis. Photogram. Eng. and R. L. Bras, Estimation of flood frequency: Gupta, V. K., and E. Waymire, Statistical Remote Sens., 54(11):1593-1600,1988. An evaluation of two derived distribution self-similarity in river networks para Johnson, L. E., MAPHYD - A digital map- procedures. Water Resour. Res., meterized by elevation. Water Resour. based hydrologic modeling system. 23(7):1309-1319,1987. /&w.,25(3):463-476,1989. Photogram. Eng. and Remote Sens., 55(6): Nolan, K. M., and B. R. Hill, Storm-runoff Gupta, V. K., and O. J. Mesa, Runoff 911-917,1989. generation in the permanent creek generation and hydrologic response via Julien, P. Y., and G. E. Moglen, Similarity drainage, West Central California - an ex channel network geomorphology - recent and length scale for spatially varied over ample of flood-wave effects on runoffcom progress and open problems. J. Hydrol, land flow. Water Resour. Res., position. J. Hydrol, 113:343-367,1990. 102:3-28,1988. 26(8):1819-1832,1990. Ormsbee, L. E., and A. Q. Khan, A Gupta, V. K., I. Rodriquez-lturbe, and E. F. Keppeler, E. T., and R. R. Ziemer, Logging parametric model for steeply sloping Wood (Ed.), Scale Problems in Hydrology, effects on streamflow: Water yield and forested watersheds. Water Resour. Res., D. Reidel, Hingham, Mass., 246 pp., summer low flows at Casper Creek in 25(9):2053-2065,1989. 1986. northwestern California. Water Resour. Ormsbee, L. E., Rainfall disaggregation Haan, CT. Parametric uncertainty in Res., 26(7): 1669-1680,1990. model for continuous hydrologic modeling. hydrologic modeling. Trans. ASAE 31(1): Khan, A. Q., and L. E. Ormsbee, A com J Hydraul. Eng., ASCE, 115(4):507-525, 137-146,1989. parison of two hydrologic models for 1989. Goodrich and Woolhiser: Catchment Hydrology 209

Pionke, H. B., J. R. Hoover, R. R. river networks. Water Resour. Res., Similarity and scale in catchment storm Schnabel, W. J. Gburek, J. B. Urban, and 24(8):1317-1322,1988. response. Rev. Geophys., AGU, 28(1):1- A. S. Rogowski, Chemical-hydrologic in Tarboton, D. G., R. L. Bras, and I. 18,1990. teractions in the near-stream zone. Water Rodriguez-Iturbe, Scaling and elevation in Wood, E. F., M. Sivapalan, K. Beven, and Resour. Res.,24(7):l 101-1110,1988. river networks. Water Resour. Res., L. Band, Effects of spatial variability and Rawls, W.J., and D.L. Brakensiek. Com 25(9):2037-2051,1989. scale with implications to hydrologic parison between Green-Ampt and curve Thomas, D. L., and D. B. Beasley. A physi modeling./. Hydrol, 102:29-47,1988. number runoff predictions. Trans. ASAE. cally-based forest hydrology model I: Woolhiser, D. A., and D. C. Goodrich, Ef 29(4): 1597-1599,1986. development and sensitivity of com fect of storm rainfall intensity patterns on Riekerd, H., Influence of Silvicultural prac ponents. Trans. ASAE. 29(4):962-972, surface runoff. /. Hydrol, 102:335-354, tices on the hydrology ofpine flatwoods in 1986a. 1988. Florida. Water Resour. Res., 25(4):713- Thomas, D. L., and D. B. Beasley. A physi Woolhiser, D. A., and H. B. Osborn, A 719,1989. cally-based forest hydrology model II: stochastic model of dimensionless Rogers, W. F., and V. P. Singh, Some Evaluation under natural conditions. Trans. thunderstorm rainfall. Water Resour. geomorphic relationships and hydrograph ASAE,29(4):973-9%1,1986b. Res.,2l(4):5l1-522,1985. analysis. Water Resour. Bull, 22(5):777- Trimble, S. W., F. H. Weirich, and B. L. Woolhiser, D. A., R. E. Smith, and D. C. 784,1986. Hoag, Reforestation and the reduction of Goodrich, KINEROS, A kinematic runoff Shen, H. W., G. J. Koch, and J. T. B. water yield on the southern piedmont since and erosion model: Documentation and Obeysekera, Physically based flood fea circa 1940. Water Resour. Jtes.,23(3):425- user manual. U.S. Dept. of Agricul., tures and frequencies. J. Hydraul. Eng., 437,1987. ARS, ARS, 130 pp., 1990 (with diskette). ASCE, 116(4):494-514,1990. Troutman, B. M., M. R. Karlinger, and D. U.S. National Research Council, Oppor Siegel, D. I., Evaluating cumulative effects P. Guertin, Basin-scale relations via con tunities in the Hydrologic Sciences. Na of disturbance on the hydrologic function ditioning. Stoch. Hydrol. and Hydraul, tional Academy Press, 340 pp., 1991. of bogs, fens, and mires. Environ. 3:111-133,1989. Wright, K. A., K. H. Sendek, R. M. Rice, Manage., 12(5):621-626,1988. Ursic, S. J., and R. J. Esher, Influence of and R. B. Thomas, Logging effects on Sivapalan, M., K. Beven, and E. F. Wood, small mammals on stormflow responses of streamflow: Storm runoff at Casper Creek On hydrologic similarity, 2. A scaled pine-covered catchments. Water Resour. in northwestern California. Water Resour. model for storm runoff production. Water Bull, 24(0:133-139,1988. Res., 26(7): 1657-1668,1990. Resour. Res.,23(12): 2266-2278,1987. Walker, J. F., S. A. Pickard, and W. C Zhang, W., and T. W. Cundy, Modeling of Sivapalan, M., E. F. Wood, and K. J. Sonzogni, Spreadsheet watershed modeling two-dimensional overland flow. Water Beven, On hydrologic similarity, 3. A for nonpoint-source pollution management Resour. rt«..25(9):2109-2035,1989. dimensionless flood frequency model using in a Wisconsin Basin. Water Resour. Bull, Additional References a generalized geomorphologic unit 25(1):139-147,1989. American Society of Agricultural Engineer hydrograph and partial area runoffgenera Ward, A. D., B. J. Middleton, C T. Haan, ing, Proceedings 1988 International Sym tion. Water Resour. Res., 26(l):43-58, and G. V. Campbell. An evaluation of posium on Modeling Agricultural, Forest, 1990. small catchment flood estimation techni and Rangeland Hydrology, ASAE Pub. 07- Smith, R. E., and R. H. B. Hebbert, Mathe ques. Trans. ASAE3l(l): 611-619, 1989. 88. ASAE, St. Joseph, Mich., 1988. matical simulation of interdependent sur White, D., Grid based application of runoff Anderson, M. G. (Ed.), Modelling Geomor- face and subsurface hydrologic processes, curve numbers. J. Water Resour. Planning phological Systems, John Wiley and Sons, Water Resour. Res., 19(4):987-1001, and Manage., ASCE, 114(6):601 -612, Ltd., 458 p., 1988. 1983. 1988. Hooper, R. P., and C. A. Shoemaker, A Sorooshian, S., and Gupta, V. K., Automatic Wilcox, B. P., K. R. Cooley, and C. L. comparison of chemical and isotopic calibration of conceptual rainfall-runoff Hanson. Predicting snowmelt runoff on hydrograph separation. Water Resour. models: The question of parameter obser sagebrush rangeland using a calibrated /fcw.,22(10):1444-1454,1986. vability and uniqueness. Water Resour. SPUR hydrology model. Trans. ASAE Kadlec, R. H., Overland flow in wetlands: Res., 19(l):260-268,1983. 31(3): 1351-1357,1989a. Vegetation Resistance. ASCE, J. Hydraul. Sorooshian, S., and Gupta, V. K., The Wilcox, B. P., C. L. Hanson, J. R. Wight, £hg.,116(5):691-706. analysis ofstructural identifiability: Theory and W. H. Blackburn, Sagebrush ran Kawas, M. L. (Ed.), New directionsfor sur and application to conceptual rainfall- geland hydrology and evaluation of the face water modeling. Proceedings of the runoff models. Water Resour. Res., SPUR hydrology model. Water Resour. Third Scientific Assembly of the Interna 21(4):487-495,1985. *«//., 25(3):653-666,1989b. tional Association of Hydrological Scien Sorooshian, S., Gupta, V. K., and Fulton, J. Willmott, C. J., S. G. Ackleson, R. E. ces, IAHS Pub. 181, IAH Press, Institute L., Evaluation of maximum likelihood Davis, J. J. Feddema, K. M. Klink, D. R. ofHydrology, Wallingford, UK, 1989. parameter estimation techniques for con Legates, J. O'Donnell, and C. M. Rowe, Sinniger, R. O. and M. Monbaron (Ed.), ceptual rainfall-runoff models: Influence of Statistics for the evaluation and comparison Hydrology in Mountainous Regions II: Ar calibration data variability and length on ofmodels. /. Geophys. Res., 90(C5):8995- tificial Reservoirs Water and Slopes. (Proc. model credibility. Water Resour. Res., 9005,1985. oftwo Lausanne Sym., Aug., 1990) IAHS 19(0:251-259,1983. Wilson, G. V., and R. J. Luxmoore, In Pub. NO. 194,1990. Swank W. T., and D. A. Crossley, Jr. filtration, macroporosity, and mesoporosity distributions on two forested watersheds. D.C. Goodrich and D.A. Woolhiser, USDA-ARS, (editors), Forest Hydrology and Ecology at AWMRU, 2000 E. Allen Rd., Tucson, AZ 85719. Coweeta, Springer-Verlag New York Inc., Soil Set. Soc. Am. J., 52(2):329-335, Secaucus, NJ, 469 p., 1988. 1988. (Received Sept. 29,1990; Wood, E. F., and C. S. Hebson, On revised Jan. 4, 1991; Swistock, B. R., D. R. Dewalle, and W. E. acceptedJan. 10,1991.) Sharpe, Sources ofacidic storm flow in an hydrologic similarity, 1. Derivation of the Appalachian headwater stream. Water dimensionless flood frequency curve. Resour. Jta.,25(10):2139-2147,1989. Water Resour. Res., 22(11): 1549-1554, Tarboton, D. G., R. L. Bras, and I. 1986. Rodriguez-Iturbe, The fractal nature of Wood, E. F., M. Sivapalan, and K. Beven,