Printer Friendly

Runoff index values for frozen soil areas of the Pacific Northwest.

The hydrology of the Palouse and Nez Perce Prairies, part of the United States' Northwest Wheat and Range Region (NWRR)(Austin 1981) located in eastern Washington, north-eastern Oregon, and northern Idaho, is highly impacted by winter processes. More than 60% of the region's precipitation falls between November and April (Molnau and Mesu 1983; Zuzel et al. 1993), which makes this period critical in determining crop soil-water. In addition, the interactions among rain, snowmelt, and freezing and thawing soils lead to high erosion hazard.

In a 1-year study in north-central Oregon, Zuzel et al. (1982) found that snowmelt, frost melt, and rainfall on thawing soils contributed to approximately 86% of the soil erosion in winter wheat following a summer fallow. Data from the current 13-year study indicates the proportion of total soil loss from thawing soil events to range from 30 to 60% of the average annual erosion depending on crop treatment.There has been a major effort in recent years to develop computerized agricultural models for predicting everything from crop yields to soil loss. CREAMS (Knisel 1980), Erosion Productivity Impact Calculator (EPIC) (Williams et al. 1984), Water Erosion and Prediction Project (WEPP)(Lane and Nearing 1989) and Agricultural Non-Point Source model (AGNPS)(Young et al. 1987) are all examples. In general, the hydrologic components of many models perform very poorly in the Palouse (McConkey 1992) because of their dependence on the NRCS Curve Number (formerly the SCS Curve Number) technique (Soil Conservation Service 1964; 1972; 1985; 1991), which was developed primarily for large events during non-winter conditions. The Curve Number technique performs poorly when applied to the typically low rainfall intensities and snowmelt conditions of the Palouse winters.

The Curve Number technique is dependent on the CN parameter, which is determined pseudo-qualitatively from the hydrologic characteristics and conditions of an area. Despite many well-documented problems with the Curve Number method and the CN parameter (Freebairn et al. 1989; Gray and deMonsabert 1983; Hailey and McGill 1983; Hjelmfelt 1980; Osborn and Renard 1983; and Ritter and Gardner 1991), it is a widely used and accepted technique for estimating runoff from precipitation. Its simplicity and low input requirements make the Curve Number method attractive to applied modelers.

This study focused on examining winter runoff index (RI) values calculated using Curve Number relationships from measured plot runoff at the Palouse Conservation Field Station near Pullman, Washington, and exploring simple methods of predicting appropriate RI values for the conditions observed. One goal of this study was to refrain from undue complexity, i.e., maintain the Curve Number method's simplicity of parameter estimation, while attempting to improve the method's accuracy.


The NRCS Runoff Curve Number method is a simple but widely used technique for determining event runoff from storm precipitation data. The procedure appeared in the 1964 SCS National Engineering Handbook (Soil Conservation Service 1964) with revisions published in 1972, 1985, and 1991 (Soil Conservation Service 1972; 1985; 1991). It assumes that runoff is related to storm precipitation through the following relationship:

Q = [(P - [I.sub.a]).sup.2]/(P - [I.sub.a]) + S for p [greater than or equal to] [I.sub.a] (1)


Q = Runoff in inches

P = Rainfall in inches

[I.sub.a] = Initial abstraction in inches

S = Potential maximum retention plus initial abstraction in inches.

The initial abstraction term [I.sub.a] accounts for interception, infiltration, and surface storage prior to runoff. It is frequently assumed, based on an empirical relationship presented by NRCS, that the initial abstraction [I.sub.a] is a fixed portion of S:

[I.sub.a] = 0.2S (2)

Data to which this relationship was fitted show large scatter but errors appear random in nature. Thus, substituting the expression for [I.sub.a] from equation (2) into equation (1),

Q = [(P - 0.2S).sup.2]/P + 0.8S for P [greater than or equal to] 0.2S (3)

The curve number CN is related to S by the following relationship:

CN = 1000/S + 10 (4)

The Curve Number procedure was developed primarily to estimate runoff for small watersheds rather than individual fields. However, it is used in a number of models that deal with areas ranging from field size to watersheds of a few square miles.

For purposes of predicting runoff under a given set of conditions, soils were classified by their infiltration and transmission characteristics into the following four hydrologic soil groups:

A) (Low runoff potential) High infiltration when wetted and moderate transmission rates when saturated.

B) Moderate infiltration when wetted and moderate transmission rates when saturated.

C) Slow infiltration rates when wetted and slow transmission rates when saturated. Usually will have an impeding layer.

D) (High runoff potential) Very slow infiltration rates when wetted and very slow transmission rates. May have a nearly impervious layer.

Land use and treatment effects are added to the hydrologic soil groups to produce hydrologic soil-cover complexes. A table of CN values was developed for various land use and treatment classes for each of the four hydrologic soil groups.

Antecedent soil moisture as a result of rains just prior to a storm are considered as affecting the potential maximum retention plus initial abstraction parameter S. Due to ambiguity in defining antecedent moisture conditions (AMC) based on rainfall in the recent revisions of the National Engineering Handbook (Soil Conservation Service 1991), AMC effects have been ignored in this study.

Experimental procedure

The experiment, located at the Palouse Conservation Field Station (PCFS), about 3km (2mi) northwest of Pullman, Washington, ran for 13 winters from the fall of 1978 to the spring of 1991. Average annual precipitation at the site is about 540mm (21in), approximately 250mm (10in) (46%) of which falls during the primary erosion season, December through March. The soil at the site is Palouse silt loam (fine silty, mixed Mesic-Pachic, Ultic, Haploxeroll), a hydrologic class B soil.

Rectangular runoff plots were placed on a predominantly south-facing hillside with slope steepnesses of 15% to 26%. Plots were bordered by galvanized metal strips, 200mm (8in) high, forced 100mm (4in) into the soil, leaving about 100mm above the surface. At the downhill end of each plot the borders angled to a V-shape, directing runoff to a 50mm (2in) tube at the apex. Runoff ran down the tube into a collection tank from which samples were obtained. The plot widths were 3.66m (12ft) and lengths were between 12.0m and 45.9m (39ft and 151ft) [each treatment included at least one 22.2m (73ft) plot]. In the last four seasons all the plots were 22.2m long.

Runoff volume was determined from the depth of water collected in each tank. In general, runoff measurements and sediment samples were taken daily, at which time the tanks were emptied. This routine was usually maintained even if an event lasted for more than one day, i.e., total runoff for some events consisted of the sum of several days' measurements.

Each plot was assigned one of the following four crop management treatments: continuous bare fallow (CBF), winter-wheat following a summer fallow (WW/SF-T), winter-wheat following a small grain of winter wheat or spring grain (WW/SG-T), and winter wheat following winter wheat no-till seeded (WW/WW-NT). The WW/SF-T and WW/SG-T treatments are consistent with the cropping sequences found in the medium to high precipitation regions of the Palouse and the greater Northwestern Wheat and Range Region (NWRR); the CBF treatment, which was used as a control and WW/WW-NT treatment represent extreme cropping managements. Each treatment was applied to two or three plots. The WW/SF-T and WW/SG-T plots were tilled to duplicate field conditions, but because tillage speeds were low, residue cover and surface roughness tended to be slightly higher than found in the field.


The procedure of analysis was as follows:

1) Calculate an RI parameter for each event by solving equations (3) and (4) for CN and replacing CN with RI:

RI = 100/[1 + 0.5 (P + 2Q - [(4[Q.sup.2] + 5PQ).sup.0.5])] (5)

2) Separate winter events from non-winter events. Winter was somewhat arbitrarily chosen to be the period from November 1 to March 31. This time period encompassed all the frozen soil events except one.

3) Separate winter runoff events into frozen-soil events and nonfrozen-soil events. Events were classified based on frost tube data taken from representative plots. An event was classified as frozen if frost was present during any time within the period of a runoff event. Many frozen-soil events included thawing soils. If an event was not classified as frozen it was considered a nonfrozen-soil event.

4) For both nonfrozen-soil and frozen-soil events average RIs were determined for each treatment. All events with rainfalls below the initial abstraction [defined in equation (2)] were removed from determination of these averages. Due to the RI's dependence on the initial abstraction, averages had to be calculated iteratively.

5) Runoff was calculated for all the winter events using the RIs determined by the method described above, and the results were compared with the measured runoff values.

6) Runoff was calculated for the summer events for the CBF plots only, using the same method described in (4). Summer events were not broken down by treatments because only the CBF plots were left in place during the summer months.

The events analyzed to this point include only about 50% of the total runoff measured. No events involving snowmelt, about 40% of the total runoff, were included in the analysis. For about 10% of the total runoff, it was difficult to clearly define the associated storms. Future analysis will include the snowmelt events.

Results and discussion

Results are presented in Tables 1 through 3. Because the numbers of runoff plots and events vary among treatments, the runoff totals in Table 1 and the means in Table 2 cannot be meaningfully compared, as absolute values, among treatments. Therefore, discussion of the results in these two tables is limited primarily to differences in measured and predicted runoff values among treatments and between frozen and nonfrozen conditions. With respect to the means presented in Table 3, these are the means of all the observations within a treatment for a given condition. While the number of observations varies among treatments due to differing numbers of runoff plots in different treatments, differing numbers of runoff plots in different years for each treatment, and differing number of events for each treatment each year, the mean values in Table 3 are the best source for comparison of absolute runoff values among treatments. The events reported here include only about 50% of the total measured runoff. No snowmelt events are included.

Even though the NRCS CN is not applicable to frozen soil conditions, and probably winter-time conditions as a whole, runoff was predicted using the published CN approach to emphasize the problem faced by engineers attempting to use the method under these "inappropriate" conditions. Table 1 clearly shows that runoff predicted with unadjusted CNs was typically inaccurate by an order of magnitude. The WW/WW-NT results suggest the degree of sensitivity of the predicted runoff to the runoff parameters, CN and RI; a 2% to 4% difference in CN/RI resulted in RI or CN values increasing the error in predicted runoff by more than 50%. Bondelid et al. (1982) suggested that errors in predicting CNs will fluctuate by an average of about 3, [TABULAR DATA FOR TABLE 1 OMITTED] [TABULAR DATA FOR TABLE 2 OMITTED] [TABULAR DATA FOR TABLE 3 OMITTED] which is about a 4% error with respect to the RIs in Table 1 for WW/WW-NT. An interesting note is that the difference between frozen and nonfrozen mean RIs is less than 1.3% for all the treatments except WW/SG-T. The significance of this observation is not particularly clear within the scope of this study. The expected trend towards a higher CN for frozen conditions compared to nonfrozen is consistent in all treatments except WW/WW-NT. The reason for this aberration is not readily apparent.

While the statistics presented in Table 3 indicate the high variability between individual predicted and measured event runoffs (R.D.[greater than]60% for all treatments and greater than 100% for all but two treatments), the differences in mean measured and predicted event runoffs are only about 22% on average. The CBF treatments showed the best agreement between measured and predicted event runoffs for both frozen and unfrozen conditions. The worst correlations between measured and predicted event runoff were found in WW/SG-T and WW/WW-NT, with the absolute mean event runoff of typically less than 1mm (0.04in), the exception being WW/SG-T for frozen conditions which had a mean of 3.1 mm (0.1 in).

The statistics for annual runoff totals, as expected, indicated better agreement between measurements and prediction than those for their corresponding individual runoff events. A paired students-t test reveals no statistical difference in the average annual runoff amounts for both frozen and non-frozen events at the 95% confidence interval. The difference between measured and predicted mean annual runoff totals for each treatment is about the same as for mean event runoff amounts - 20%. The R.D.s indicate that a lot of variation exists between individual measured and predicted annual runoff totals.

The RI value for the CBF treatment was much larger than for other treatments, as was expected. Based on mean total runoff (not given in this paper), it was expected the Runoff Index values for the WW/SF-T, WW/SG-T, and WW/WW-NT would fall into order below the CBF. The similarity in mean RI values for the frozen condition for WW/SF-T and WW/SG-T was not expected, and the similarity in mean RI values for the nonfrozen condition for WW/SG-T and WW/WW-NT was also not expected.

The calculated Runoff Index values for all treatments were much larger than assigned Curve Numbers for comparable treatments in areas of the U.S. subjected to more intense and larger summer events. This is consistent with results of other research in the Pacific Northwest (Molnau and Mesu 1983). Use of Curve Number CN values similar to the Runoff Index RI calculated in this study will improve runoff prediction in the frozen soil area of the Pacific Northwest. Further analysis is needed to deal with snowmelt conditions.


Runoff plot data from the Palouse Conservation Field Station near Pullman, Washington, were analyzed using runoff Curve Number relationships to determine Runoff Index values for non-snowmelt winter and summer events. The RI values were highly variable. The RI values for the winter events were much higher than Curve Number values for similar treatments in areas dominated by summer storms. These Runoff Index values can be used to improve the performance of models using Curve Number based hydrologic components.


Austin, M.E. 1981. Land resource regions and major land resource areas of the United States. USDA Agriculture Handbook No. 296. 156 p.

Bondelid, T.R., R.H. McCuen, and T.J. Jackson. 1982. Sensitivity of SCS Soil Conservation Service Models to Curve Number Variation, Hydraulic Design, Peak Discharge, Runoff. Water Resources bulletin. Minneapolis, Minnesota. American Water Resources Association. Feb. 1982. 18(1):111-116.

Churchill, E.A. Jr. 1981. Hydrologic Relationships on a Watershed in the Eastern Palouse. M.S. thesis. Graduate School, University of Idaho, Moscow.

Freebairn, D.M., S.C. Gupta, C.A. Onstad, and W.J. Rawls. 1989. Antecedent rainfall and tillage effects upon infiltration. Soil Science Society of America Journal. 53(4): 1183-1189.

Gray, D.D., and S.M. deMonsabert. 1983. Antecedent moisture condition probabilities: Closure. [Author's Reply], Journal of Irrigation and Drainage Division. ASCE 109 (2):305-307.

Hailey, J.L., and H.N. McGill. 1983. Runoff curve number based on soil-cover complex and climatic factors. Microfiche no. 8302957, ASAE, St. Joseph, Michigan, ASAE.

Hjelmfelt, A.T. Jr. 1980. Empirical investigation of the curve number technique. Journal of Hydraulics Division. ASCE 106(9):1471-1476.

Knisel, W.G., (ed.). 1980. CREAMS: A Field-scale model for chemicals, runoff, and erosion from agricultural management systems. U.S. Department of Agriculture, Conservation Research Report No. 26, 640 pp.

Lane, L.J., and M.A. Nearing (eds.). 1989. USDA-water erosion prediction project: Hillslope profile model documentation. NSERL Report No. 2, USDA-ARS National Soil Erosion Research Laboratory. West Lafayette, Indiana.

McConkey, B.G. 1992. Modeling Cold Region Agricultural Hillslope Hydrology. Ph.D. dissertation. Department of Crop and Soil Sciences. Washington State University.

Molnau, M., and F.P. Mesu. 1983. Hydrographs for a winter runoff regime watershed. PNR 83-209. For presentation at the PNR Annual Meeting, ASAE, Victoria, British Columbia, Canada, Oct. 12-14, 1983.

Osborn, H.B., and K.G. Renard. 1983. Antecedent moisture condition probabilities: Discussion. Journal of Irrigation and Drainage Division. ASCE 109(2):300.

Ritter, J.B., and T.W. Gardner. 1991. Runoff curve numbers for reclaimed surface mines in Pennsylvania. Journal of Irrigation and Drainage Division. ASCE 117(5):656-666.

Soil Conservation Service. 1964. National Engineering Handbook. Section 4. Hydrology, U.S. Department of Agriculture. U.S. Government Printing Office, Washington, D.C.

Soil Conservation Service. 1972. National Engineering Handbook. Section 4. Hydrology, U.S. Department of Agriculture. U.S. Government Printing Office, Washington, D.C.

Soil Conservation Service. 1985. National Engineering Handbook. Section 4. Hydrology, U.S. Department of Agriculture. U.S. Government Printing Office, Washington, D.C.

Soil Conservation Service. 1991. National Engineering Handbook. Section 4. Hydrology, U.S. Department of Agriculture. U.S. Government Printing Office, Washington, D.C.

Williams, J.R., C.A. Jones, and P.T. Dyke. 1984. A modeling approach to determining the relationship between erosion and soil productivity. Trans. ASAE 27(1):129-144.

Wischmeier, W.H., and D.D. Smith. 1978. Predicting rainfall erosion losses: A guide to conservation planning. USDA, Agriculture Handbook No. 537. U.S. Government Printing Office, Washington, D.C.

Young, R.A., C.A. Onstad, D.D. Bosch, and W.P. Anderson. 1987. AGNPS, Agricultural Non-point-source pollution model. A Watershed Analysis Tool. U.S. Department of Agriculture, Conservation Research Report No. 35, 80 pp.

Zuzel, J.F., R.R. Allmaras, and R.N. Greenwalt. 1982. Runoff and soil erosion on frozen soil in northeastern Oregon. Journal of Soil and Water Conservation 37(6):351-354.

Zuzel, J.F., R.R. Allmaras, and R.N. Greenwalt. 1993. Temporal distribution of runoff and soil erosion at a site in northeastern Oregon. Journal of Soil and Water Conservation 48(4):373-378.

D.K. McCool is an agricultural engineer with the USDA-Agricultural Research Service, stationed at the Biological Engineering Department, Washington State University, M.T. Walter is a research assistant, and L.G. King is an agricultural engineer with the Biological Systems Engineering Department, Washington State University, Pullman 99164-6120. This paper is a contribution from USDA-ARS, Land Management and Water Conservation Research Unit, Pullman, Washington, in cooperation with the College of Agriculture and Home Economics Research Center, Washington State University, Pullman.
COPYRIGHT 1995 Soil & Water Conservation Society
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1995 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Title Annotation:Special Issue: Water Research and Management in Semiarid Environments
Author:McCool, D.K.; Walter, M.T.; King, L.G.
Publication:Journal of Soil and Water Conservation
Date:Sep 1, 1995
Previous Article:Utilizing fractal principles for predicting soil hydraulic properties.
Next Article:The surface infiltration model.

Terms of use | Privacy policy | Copyright © 2018 Farlex, Inc. | Feedback | For webmasters