Vous êtes sur la page 1sur 6

See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/298179011

at: https://www.researchgate.net/publication/298179011 Applicability of SCS curve number method for a California

Article in Journal of Soil and Water Conservation · January 2000

CITATIONS

18

3 authors:

33 PUBLICATIONS

363 CITATIONS

READS

112

California, Davis 33 PUBLICATIONS 363 CITATIONS READS 112 Michael J Singer University of California, Davis 111

111 PUBLICATIONS

4,503 CITATIONS

SEE PROFILE SEE PROFILE
SEE PROFILE
SEE PROFILE
133 PUBLICATIONS 2,308 CITATIONS SEE PROFILE
133 PUBLICATIONS
2,308 CITATIONS
SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Rangeland Water Quality Project View project View project

Soils of the United States just published by Springer View project View project

All content following this page was uploaded by Kenneth W Tate on 24 December 2016.

The user has requested enhancement of the downloaded file.

Copyright © of 2000

Conservation Society. All www.swcs.org55(2):226-230Journal rights reserved.

Soil and

Soil Water

and Water

Conservation

Applicability of SCS Curve

Number Method for a California Oak Woodlands

D. Lewis, M.J. Singer, K.W. Tate

(AGNPS) (Young et al., 1985). CN defines runoff volume (Q as a function of total precipitation (I?) and the potential maximum soil water retention after runoff begins (S) through Equation (1) (Haan et al. 1994; SCS 1985; Ralli- son and Miller 1981).

(P - 0.2s)2

Q=

P + 0.8s

P>O.2S

ABSTMCE The curve number (CN) method devebped by the Soil ConservationService (now

Watershed has not been extensively tested in western

regtons of th~United States. We used a 17 year rainfill and runof record?om a California Oak woodlurid watershed to compare the accuracy of CN as prescribed in the SCS National Enginewing Handbook (NEH-4) with two alternative method. Each method predicted mean annual peak runoff that was not signijkanth dtffweentf;om observed runofl and correlation between estimated and observed runof’om each of the three method was statistically signif;- cant. However, the highest correlation coeftcient showed that only 50% of the variability in the dntd were ni~ountedforby any of the methods. The NEH-4 method underpredicted maximum flowsfor tbe &best flow years. The more conservative Hjelmfilt method morefrequently over- predicted peak flow. Overpredictionprovides a measure of safety when using the CN technique.

NRCS) for predicting peak runofffiom

Keyword: Fore-rt hydrology,oak woodland, watershed management, wildland hydrology

The value for S is calculated using the curve number (CN) from equation (2).

The standard procedure (called the NEH-4 procedure after the publication that describes it) is to select a CN for a given soil series, land use, and land treat- ment from tables developed through analysis of watershed data (SCS 1985). Each soil series was assigned to one of

elating storm runoff volume to rain- Rall volume is of direct importance to

drainage and flood structure design. However, there are few watersheds in which rainfall and runoff are routinely measured. This is particularly true for the millions of hectares of upland Oak grass- land watersheds in California. These wa- tersheds are important for cattle grazing, second home development, wildlife habi- cat, and water. Even in watersheds with rainfall and runoff records, the relation- ship between annual peak runoff and rainfall is not simple. For example, for the Schubert experimental watershed, only 39% of the variability in annual peak runoff was explained by rainfall with a regression slope of 0.63 (Figure 1). This result suggests that there is considerable risk in over or under designing flood con-

trol structures using this simple relation- ship. A more accurate method is required that can better account for conditions found in Mediterranean Oak woodland watersheds, such as Schubert. The curve number approach (CN) was developed by the Soil Conservation Service to predict peak annual runoff on ungaged watersheds. Its most common

- -.

David Lewij is a Research Assistant and Michael Singer i3 n P~ofisssorof SoilScience in the Depart- ment of Land, Air, and Water Resources Depart- ment of the Uniuersirp.of California. Kenneth Tate 1) 17 Rungeland Watershed Hydrologist in the Department of Agronomy and Range Science at the I!niversrry orCaL@rnia.

n

E

E

v

E

C

3

a

6o50 t

0

10

20

Figure 1

30

40

50

60

Rainfall (mm)

70

80

Figure 1. Schubert watershed annual peak runoff as a function of correspond- ing annual rainfallfrom 1981 to 1997. Correlation coefficient 1.2= 0.39.

use is to determine a design discharge (SCS 1985). The CN approach is simple, predictable, and stable (Ponce and Hawkins 1996). Because of these charac- teristics, it has been used in hydrologic models such as Chemicals, Runoff, and Erosion From Agricultural Management

Systems (CREAMS) (Kiesel 1980),

Groundwater Loading Effects of Agricul- tural Management Systems (GLEAMS) (Leonard et al., 1986) and Agricultural Non-Point Source Pollution Model

four hydrologic soil groups, according to the soil’s minimum infiltration rate. Group A soils have low runoff and high infiltration rates; group B soils have mod- erate infiltration rates; group C soils have low infiltration rates; and group D soils have high runoff potential (SCS 1986). Curve number values reported in NEH-4 (SCS 1985) are for antecedent soil moisture condition I1 (AMC-11). This is one of three antecedent soil mois- ture conditions established to account for

Copyright © of 2000

Conservation Society. All www.swcs.org55(2):226-230Journal rights reserved.

Soil and

Soil Water

and Water

Conservation

the runoff conditions that exist at the time of each storm event. AMC-I1 is defined by either average conditions, median CN, or antecedent rainfall tables (SCS 1985). Median CN is the CN that divides runoff and rainfall data into equal numbers. Quantitative and qualitative description of average conditions is not clear, but it appears to correspond to con- ditions when annual flooding occurs. The definition of AMC-I1 is critical to the successful use of CN because CN for AMC-I and AMC-I11 are calculated from CN for AMC-I1 by these equations:

CN (I) =

CN (1111) =

4.2CN(II)

10- 0.058CN(II)

23CN(II)

10 + O.l3CN(II)

(3)

(4)

climate has also been questioned (Ponce and Hawkins 1996). The objective of this study was to test the accuracy and appro- priateness of the NEH-4 CN technique for the study site and the environment it represents and to compare this method with two other published CN techniques.

Materials and methods

Site description. The Schubert water- shed is located approximately 96 km (60 mi) northeast of Sacramento, California in Yuba County, on the University of California Sierra Foothill Research and Extension Center (SFREC). The 103 ha (254 ac) watershed ranges in elevation from 152 m (500 ft) to 427 m (1,400 fi) with slopes that average 18% and range

from 2% to 50%. The annual rainfall distribution reflects California’s Mediter- ranean climate with average monthly rainfall greatest in the wet, cool winters

In general, condition I (ARIC-I) is for

and

absent in the dry, hot summers.

storm events occurring with little to no preceding rainfall. Condition I11 (AMC-

Average annual rainfall and stream flow during the 17 year study were 708 mm

111) is for storm events that follow consid-

(27.9 in) and 344 mm (13.5 in). The

erable rainfall when infiltration rates are at their lowest. Antecedent rainfall tables provide a

averages include extreme variability among years, with minimum and maxi- mum annud rainfall of 367 mm (14.5 in)

predetermined convention for AMC-I, 11,

and

1,212 mm (47.8 in) and minimum

and I11 designation based on five day

and

maximum annual stream flow of 88

antecedent rainfall. The comparison of these definitions by Hjelmfelt (1991)

(3.5 in) and 848 mm (33.4 in). Dominated by blues Oaks (QUPYCUS

mm

identified a consistency in runoff predic-

douglarii) and intermixed with interior

tion between the average conditions and

live

Oaks (Q. wislizenii) and foothills

median curve number if annual peak runoff events are used. The antecedent rainfall table definition was not dis- missed, but was not advised by Hjelmfelt (1991) because of problems identified by SCS in its application. Results from field calibration of CN, using measured rainfall and runoff, have identified considerable variability in CN selection and runoff volume calculations (Hawkins 1993; Hjelmfelt 1991;

pine (Pinus sabiniana), the vegetation of the site typifies foothill Oak wood- lands (Griffin 1988). Distributed uneven- ly, these trees create a mosaic of open grasslands, savanna, and woodlands (Epifanio 1989). Annual grasses and legumes dominate the understory, pro- viding > 90% ground coverage. Species diversity and production differs under the Oak canopy and in the open grasslands (Jackson et al. 1990; and Jansen 1987).

Boughton 1989). This variability is expected (SCS 1985) and points to the importance of conducting site-specific and regional calibration studies with long term watershed research data, Consider- able long term research of this nature has been conducted throughout the eastern

Oak stocking varies from 90 to 200 trees ha with a minimum canopy coverage of 70% (Dahlgren et al., 1997). Soils at Schubert are classified as Auburn-Sobrante-Las Posas Association (Herbert and Begg 1969). These are shallow to moderately deep, medium

and midwestern regions of the United

textured, gravely and rocky soils (classi-

States (Van Cleve and Martin 1991;

fied

as fine, mixed, thermic Typic Hap-

Haan et al. 1994; Hawkins 1993; Hjelm-

loxeralfs). Formed in basic metavolcanic

felt 1991), but little long term research has been reported for the west. More

(greenstone) bedrock of the Smartville complex (Beiersdorfer 1979), they are

importantly, the available data raises the

rich

in Fe-oxides. Field mapping found

question of the accuracy and applicability

that

58% of the site is comprised of the

of CN for western climatic conditions (Hawkins 1993).The ability to adapt the CN approach to regional geology and

Sobrante-Las Posas rocky loams, 39% Sobrante-Auburn very rocky loams, and 1% each Sobrante-Las Posas very rocky

hams and Auburn-Las Posas-Argonaut gravely loams (Epifanio 1989). Soil pit observations within the Schubert water- shed consistently identified very stony clay subsoils with 40% average clay con-

tent (Huang 1997). This clay content, along with the bedrock, at the site limit deep percolation of water (Dahlgren and Singer 1994). The site is primarily managed as range- land with light to moderate beef cattle grazing during January to March and August to October. Average AUM/ha was

0.22 (0.54 AUM ac). Additional manage-

ment included conversion of 14% of the watershed area from Oak cover to open grasslands between 1984 to 1986 (Epifanio 1989). Instrumentation. In 1978 the stream was instrumented with a three foot Par- shall flume for stage height measurements during high flow events and a tandem one foot 90 degree V notch weir for stage height measurements during low flow periods. Stage height was continually measured by floats inside stilling wells beside the flume and weir; these data were recorded on charts. From these data, the annual, seasonal, monthly, and daily discharge was calculated. Streamflow volume, converted to depth by dividing discharge by the watershed area, is referred to as runoff. It was assumed that runoff was the sum of overland and sub- surface flow (Haan et al. 1994). Rainfall quantity and intensity were measured with tipping bucket rain gauges at three different elevations within the study site. Although measurements were first recorded in 1978, consistent data collec- tion began in 1981, resulting in the 17 year record used in this study. Data analysis. We compared the NEH-4 CN method to two alternate methods for determining CN. The first was that presented by Hjelmfelt (1991). In his review of 14 watershed data sets, annual peak runoff and corresponding rainfall were used to calculate the proba- bility distribution of S using Equation (1). With a lognormal distribution of S, the mean or 50% probability of S used with Equation (2) determined the CN for AMC-11. Frequency values at the 10% and 90% probabilities for S were then used to determine CN for AMC-I and AMC-111 in the same manner. The second method was an asymptotic determination of CN, recommended by Hawkins (1993). Frequency matching was employed by equating rainfall (P) and runoff (Q) event return periods. Rainfall and runoff depths were sorted

SECOND

QUARTER

zoo0

227

Copyright © of 2000

Conservation Society. All www.swcs.org55(2):226-230Journal rights reserved.

Soil and

Soil Water

and Water

Conservation

separately and reassigned as rank-ordered pairs of equal return periods. As first interpreted by Hjelmfelt (1980), this method of rank-ordering does not require that storm runoff events be matched with respective rainfall events, only that equal rainfail and runoff return frequencies be matched. For each rank-ordered pair, S was calculated using the following qua- dratic equation developed by Hawkins

(1993):

S = 5(Pt 24- 4Q2 + 5PQ)’

(Q,P,Smm,in) (5)

These S values were then substituted into Equation (2) to calculate CN for each QP pair. The relationship between the calculated CN and rainfall volume was then plotted and used to determine an AMC-I1 CN value. Using the CNs determined by each method, runoff was calculated for the 17 peak annual runoff events. Analysis of variance and multiple comparison tests were used to determine differences in means among the estimated and observed runoff (Hesel and Hirsch 1995). In addi- tion, sum-of-least-squares and regression analysis of observed runoff and estimated runoff were used to assess the effective- ness of the methods.

Results and discussion

Hydrologic group designations for soils in the watershed were Argonaut D; Auburn D; Sobrante; C. Group designa- tion for Las Posas was not available. Although Sobrante associations with Argonaut, Auburn, and Las Posas soils make up the largest portion of the water- shed, hydrologic group D was selected for the watershed to ensure the most conserv- ative estimate of runoff, following the rec- ommendations in NEH-4 (SCS 1985) for hydrologic group selection. In addition to the hydrologic group designation, the site was characterized as pasture and range in good hydrologic condition, according to the definition by soil group, cover type, and cover density (SCS 1985). Curve numbers were lowest using NEH-4, inter- mediate for the asymptotic method and highest for the S probability method (Table 1). The 10, 50, and 90% S proba- bility values were taken from Figure 2. As- ymptotic analysis was based on Figure 3. The relationship in Figure 3 of the derived CN values approaching a con- stant value as rainfall increases is “stan- dard,” corresponding to high lateral flow conditions (Hawkins 1993). This is in agreement with conditions of lateral flow proposed for this watershed (Huang

Table 1. Curve numbers for the Schubert watershed determined from NEH-4, S-frequency return period, and asymptotic methods.

Method

NEH-4

S-probability

Asymptotic

Note: AMC-antecedent

AMC-I

AMC-II

63

80

86

91

70

85

moisture condition as defined by SCS (1985).

AMC-Ill

91

99

94

1997; Dahlgren and Singer 1994), as well as preferential flow paths observed in other forested watersheds (Mosley 1979; Freeze 1974; Dunne and Black 1970). During the 17 years, peak annual events occurred under CN-I conditions once, under CN-I1 conditions three times, and under CN-I11 conditions 13 times, according to NEH-4 (SCS 1985) (Table 2). Analysis of variance indicated a significant difference between means cal- culated by the S probability and NEH-4 methods (p = 0.009) by the Kruskal- Wallis test (Hesel and Hirsch 1995). However, all three CN applications pre- dicted runoff values with no significant differences to the observed values (Figure 4). Predicted mean peak-annual runoff was highest for the S-probability method and lowest for NEH-4 (Table 2). Com- parison of estimated peak runoff to observed runoff indicates that the S prob- ability procedure overestimated runoff 12 years and underestimated it five years (Table 2). In comparison, runoff estima- tions from the NEH-4 and asymptotic methods were below observed values 12 and 10 years and above observed values five and seven years, respectively. This high variability between actual and ob- served runoff for all three CN methods is expected (SCS 1985). The errors are ran- dom and the differences in estimated and observed values are neither all minus or all plus. Accuracy determinations de- scribed in NEH-4 (SCS 1985) emphasize that estimated runoff results reflect the same return period frequency as observed runoff rather than exact agreement of runoff values.

Figure 2

Figure 3

100

5

5

90

80

70

60

50

5

40

30

1

It

A

20

‘1=50

I00

150

Rainfall (mm)

10

30

50

70

90 Figure 3. Schubert watershed asymptotic determination of curve number. Data-derived curve numbers are plotted

Exceedence Probability (%)

Figure 2. Schubert watershed maximum soil water reten- tion (S) return period frequency for 17 year peak annual runoff events.

as a function of corresponding rainfall volume.

Note: The solid line represents the threshold of runoff initiation (P > 0.2s).

228

1OLf:N4L

OF

SOIL

AEU’D WATER

CONSERVATION

Copyright © of 2000

Conservation Society. All www.swcs.org55(2):226-230Journal rights reserved.

Soil and

Soil Water

and Water

Conservation

Table 2: Observed and estimated runoff for 17 peak annual runoff events in the Schubert watershed.

 

Cumulative

Observed

NEH-4

S-frequency

AMC

Rainfall

Runoff

Runoff

Runoff

Year

Condition

(mm)

(mm)

(mm)

1981 Ill

20

9

6

1982 Ill

64

38

41

1983 Ill

49

28

28

1984 II

53

30

16

1985 Ill

54

36

32

1986 111

70

62

47

1987 Ill

8

4

0

1988 II

26

8

2

1989 Ill

66

35

43

1990 Ill

33

12

14

1991 Ill

37

19

18

1992 Ill

62

35

40

1993 Ill

26

21

9

1994 II

36

5

6

1995 Ill

64

44

41

1996 I

59

27

5

1997 Ill

41

57

21

Mean

45

28

22

St. Dev.

18

17

16

 

r2

0.49

P=

(0.002)

Asymptotic

Runoff

(mm)

(mm)

20

9

64

48

49

34

40

22

51

39

70

54

8

1

15

5

63

50

33

19

37

23

62

46

26

13

24

10

61

48

19

9

41

26

38

27

20

18

0.50

0.50

(0.001)

(0.001)

Regression analysis of estimated runoff on observed runoff was highly significant for all three methods (Table 2). These re- sults indicated that all three CN methods improved the prediction of annual peak runoff compared to rainfall alone (Figure 1). In addition, they indicated that a lower S value and higher CN-11, generat- ing a near one-to-one relationship for rainfall and runoff in AMC-111 condi- tions are needed. If predicted peak runoff is statistically the same as measured peak runoff for all three methods, there is a reason to select one method over another. If flood predic- tion is a major reason for using the CN procedure, the S probability method is

the most conservative and has the largest built-in safety factor for the experimental watershed because it most frequently overpredicted peak runoff compared to the other two methods. For example,

1986 had the highest observed

water year

runoff (Table 2). The S probability method estimated 70 mm (2.8 in) of runoff compared to 47 mm (1.8 in) and 54 mm (2.1 in) for the NEH-4 and as- ymptotic methods, respectively. Flood control structures built based on the S probability predicted runoff will better contain the peak runoff than structures based on either the NEH-4 or asymptotic methods. Engineering based on overpre- diction of peak flows may have the largest built-in safety factor, but may also have

n

E

E

v

5

2

=I

U

70-

60 -

50 -

40 -

30

-

20-

10 -

0-

iI

cObserved NEH-4 S-probability Asymptotic

Curve Number Method

Figure 4. Analysis of variance among observed and estimated peak annual runoff. A line within the box shows means.

Note: 25th and 75th percentiles are shown as horizontal ends of the boxes and the whisker lines furthest from the boxes show 10th and 90th percentiles.

the highest dollar cost, which is not con- sidered in this analysis. These results do not explain, however, why all three methods for determining CN were at best only able to explain 51% of the runoff variability for the 17 annual peak runoff events. Other studies showed

that curve number explained more than 75% of the variability (Hawkins 1993; Hjelmfelt 1991). One possible answer to this question is identification of stream- flow sources in forested watersheds. Water infiltration in such settings usually exceeds rainfall intensity (Dunne and

SECOND

QUARTER 2000

229

Copyright © of 2000

Conservation Society. All www.swcs.org55(2):226-230Journal rights reserved.

Soil and

Soil Water

and Water

Conservation

Black 1970) because macro pore water velocities are two and three orders of magnitude higher than saturated soil hydraulic conductivity (Mosley 1979). The resulting streamflow is generated pri- marily from subsurface flow with only minimal contributions from overland flow (Mosley 1979; Freeze 1974). Curve number accounts for retention with re- gard to antecedent conditions through S, corresponding soil properties, and site conditions as they influence overland flow. However, it was not developed to account for site-specific contributions of macropore and subsurface flow to stream flow. This point is important because subsurface flow has been suggested as a contributor to stream flow at the study site (Dahlgren and Singer 1994) and could account for the lower r’ between observed and estimated runoff than has been found for other studies. It also implies that regardless of how CN is de- rived, it must be used with caution for California and other Mediterranean Oak woodlands.

Conclusion

Estimated mean annual peak runoff was not significantly different from ob- served runoff regardless of the method used for determining CN, confirming the ability of the curve number technique to relate rainfafl to runoff for peak events. The S probability method (Hjelmfelt 1991) most frequently overestimated runoff and is, therefore, a more conserva- tive predictor than the NEH-4 method. The best method for CN calculation ex- plained only 51% of the variability in peak annual runoff with a regression slope of 0.73. In this Oak and grassland watershed, it is likely that macro pore and lateral flows contributed to runoff and were not accounted for with any of the CN techniques. This should serve as cau- tion to researchers and resource managers utilizing CN for hydrologic modeling and flood management in California’s Oak Woodlands and other watersheds in which macropore flow is prevalent.

Acknowledgments Partial support-fir this work came?om the cllnivet-sity of California International Agricalturd Development Graduate Group and the Justro-Shields fund. Thanks to Mike Conner, Dave Labadie, and the Sierra Foothills Research and Extension Center stafjw continued szlpport in sam-

pling and instrumentation. Their work, combined with that of Charlette Epifanio and Xiaohong Huang, as researchers and custodians of the Schubert Watershed record, have generated the long term data used in this study.

REFERENCES CITED Beiersdorfer, R.E. 1979. Metamorphic petrology of the Smarrville Complex, Northern Sierra Neva- da foothills. Master’s thesis, Universiry of California. Boughton, W.C. 1989. A review of the USDA SCS curve number method. Australian Journal of Soil Research 27:511-523. Dahlgren, R.A., and M.J. Singer. 1994. Nutrient cycling in managed and non-managed oak woodland-grass ecosystems. Final Report to Integrated Hardwood Range Management Program, University of California. Dahlgren, R.A., M.J. Singer, and X. Huang. 1997. Oak tree and grazing impacts on soil properties and nutrients in a California oak woodland. Biogeochemistry 39:45-64. Dunne, T., and R.D. Black. 1970. An experimental investigation of runoff production in permeable soils. Water Resources Research 6:478-490. Epifanio, C.R. 1989. Hydrologic impacts of blue oak harvesting and evaluation of the modified USLE in the northern Sierra Nevada. Master’s thesis, University of California. Freeze, R.A. 1974. Streamflow generation. Reviews of Geophysics and Space Physics 12:627-647. Griffin, J. 1988. Oak woodland. Pp 383-415. In: M. Barbour and J. Major (eds). Terrestrial vegetation of California. New York: John Wiley & Sons. Haan, C.T., B.J. Barfield, and J.C. Hayes. 1994. Design hydrology and sedimentology for small catchments. San Diego: Academic Press. Hawkins, R.H. 1993. Asymptotic determinations of runoff curve numbers from data. Journal of Irri- gation and Drainage Engineering 119:334-345. Herbert, F.W., and E.L. Begg. 1969. Soils of the Yuba area California. Soil Survey Report. Deparrment of Soils and Plant Nutrition, University of California. Hesel, D.R., and R.M. Hirsch. 1995. Statistical methods in water resources. Studies in Environ- mental Science 49. Elsevier,Amsterdam. Hjelmfelt, A.T., Jr. 1991. Investigation of curve number procedure. Journal of Hydraulic Engi- neering 117:725-737. Hjelmfelt, A.T., Jr. 1980. Empirical Investigation of curve number technique. Journal of Hydraulics Division 106:1471-1476. Huang, X. 1997. Watershed hydrology, soil, and biogeochemistry in an oak woodland annual grassland ecosystem in the Sierra Foothills, Cali- fornia. Ph.D. diss., University of California. Jackson, L.E., R.B. Straws, M.K. Firestone, and J.W. Bartolome. 1990. Influence of tree canopies on grassland productivity and nitrogen dynamics in deciduous oak savanna. Agricul- ture, Ecosystems and Environment 32:89-105. Jansen, H. 1987. The effect of blue oak removal on herbaceous production on a foothill site in the northern Sierra Nevada. T.R Plumb and N.H. Pillsbury (Technical Coordinators). Multiple- use management of California’s hardwood resources. In the proceedings of the Symposium of the Pacific Southwest Forest Range Experi- ment Station, Berkeley, CA.

230

JOURNAL OF

SOIL

AND

WATER

CONSERVATION

United States Department of Agriculture (USDA). 1980. Knisel, W.G. (ed). CREAMS: A field- scale model for chemicals, runoff and erosion from agricultural management systems. Conser-

vation Research Report No. 26, Southeast Area. Washington, D.C. Leonard, R.A., W.G. Knisel, and D.A. Still. 1986. GLEAMS: Groundwater loading effects of agri- cultural management systems. American Society

of Agricultural Engineers. Paper No. 86-2511.

Chicago, IL. Mosley, M.P. 1979. Streamflow generation in a forested watershed. New Zealand. Water Re- sources Research 15:795-806. Ponce, V.M., and R.H. Hawkins. 1996. Runoff curve number: Has it reached maturity?Journal

of Hydrologic Engineering 1:11-19. Rallison, R.E. and N. Miller. 1981. Past, present, and future SCS runoff procedure. In: V.P. Singh (ed). Rainfall-runoff relationship. Water Resources Publication. SCS. 1985. Hydrology, Section 4. In: Soil Conser-

vation Service National Engineering Handbook.

U.S. Department of Agriculture (USDA), Washington, D.C. SCS. 1986. Urban hydrology for small watersheds.

Technical Release No. 55. Soil Conservation Service, U.S. Department of Agriculture, Wash- ington, D.C. Van Cleve, K., and S. Martin (eds). 1991. Long

term ecological research in the United States:

A network of research sites, 1991. Long-term

Ecological Research Office, University of Wash-

ington. Young, R.A., C.A. Onstad, D.D. Bosch, and W.P. Anderson. 1985. Agricultural nonpoint surface pollution models (AGNPS) I and I1 model documentation. Pollution Control Agency, St. Paul, and U.S. Department of Agriculture, Agricultural Research Service, Washington,

D.C.