Crusting, Runoff, and Erosion Response to Soil Water Contentand Successive Rainfalls

Yves Le Bissonnais and Michael J. Singer*

ABSTRACTSoil crusts reduce the water infiltration rate and may induce the

erosion process by increasing runoff. We investigated the effects ofinitial soil water content and successive rainfalls on soil crusting, sub-sequent runoff, and erosion. Surface samples of Capay silty clay loam(fine, montmorillonitic, thermic Typic Chromoxerert) and Solano siltloam (fine-loamy, mixed, thermic Typic Natrixeralf) were packed into0.37-m2 plots inclined at 9% slopes and subjected to simulated rainfallapplied at 40 mm h~ l as 3.2-mm-diam. drops falling 2.5 m. Threesuccessive 1-h rainfalls were applied to initially air dried or prewettedsoils. Water splash, runoff, soil splash, and wash materials were col-lected at 5-min intervals. Post-rainfall photographs of crusted soilsurfaces provided additional morphological evidence to support con-clusions drawn from quantitative data. Prewetting reduced crust de-velopment, runoff, and erosion and the differences remained significanteven during the third rainfall. For the initially air dry soils, erosionrates and runoff reached steady state after 25 mm of rain during thefirst rainfall, and were the same for the second and third rainfalls.For prewetted soils, runoff and erosion were lower, did not reachsteady state during the first event, and increased during each subse-quent rainfall. The infiltration rate for initially air dry soils was 20times lower than for prewetted soils after 40 mm of the first rainfall,and remained 10 and 2 times lower after the second and third rain,respectively. The crust was still not completely developed on prewettedsoils after 120 mm of cumulative rainfall.

AGGREGATE BREAKDOWN and soil credibility areoften considered constant properties of soils. They

are usually estimated by standardized measurements

Y. Le Bissonnais, Service d'Etude des Sols et de la Carte P6do-logique de France and Institut National de la Recherche Agron-omique, Centre d'Orleans 45160, Ardon, France; and M.J. Singer,Dep. of Land, Air, and Water Resources, Univ. of California,Davis, CA 95616. Received 17 Oct. 1991. 'Corresponding au-thor.

Published in Soil Sci. Soc. Am. J. 56:1898-1903 (1992).

or deduced from basic physical and chemical char-acteristics of the soils (Kemper, 1965; Bryan, 1968;Henin and Monnier, 1956; Wischmeier et al., 1971;Young, 1984). In both types of measurements, influ-ence of time-dependent parameters such as soil watercontent are not taken into account despite the fact thatthey obviously modify soil physical properties. Sev-eral studies have examined the effect of soil watercontent on aggregate breakdown and erosion (Francisand Cruse, 1983; Kemper and Rosenau, 1984; Farres,1987; Luk, 1985; Le Bissonnais et al., 1989; Trumanand Bradford, 1990), but results sometimes show op-posite effects.

Different results in aggregate breakdown can be ex-plained, in part, by the fact that soil water contentchanges quickly during a rain, and that aggregatebreakdown can occur during wetting (Nijhawam andOlmstead, 1947; Panabokke and Quirk, 1957). Theapparent positive effect of increasing water content onaggregate breakdown could be due to the wettingprocess itself (Cousen and Farres, 1984). Further-more, shear strength of wet aggregates is lower thanthat of dry aggregates, leading to the conclusion thatdry aggregates are more resistant to raindrop impact(Nearing and Bradford, 1985). Such mechanical mea-surements do not take into account slaking, whichoccurs during wetting by compression of entrappedair (Yoder, 1936; Le Bissonnais, 1990). This mech-anism can be much more efficient in breaking downaggregates than raindrop impact.

When no physical soil degradation takes place dur-ing rainfall, for example on soil covered by vegetationor already affected by a well-formed crust, runoff maybe more important on initially wet soils than on drysoils. If degradation occurs during the rainfall, crust-ing can quickly modify the hydraulic properties of thesurface and the crust becomes the infiltration limiting

30 i: A.

LE BISSONNAIS & SINGER: CRUSTING, RUNOFF, AND EROSION RESPONSE

15 -i

1899

10 20 30 40Cumulative rainfall (mm)

50

0 10 20 30 40Cumulative rainfall (mm)

Fig. 1. Runoff rate vs. cumulative rainfall for the (a) air-driedand (b) prewetted Solano soil for three successive rainfalls.

factor (Mclntyre, 1958; Epstein and Grant, 1973).Aggregate breakdown, crusting, and erosion are dy-namic processes (Chen et al., 1980; Moore and Singer,1990; Truman et al., 1990). To better understand theprocesses, we must consider their intermediate andnon-steady-state stages, which are often the most im-portant in field conditions. Thus it is necessary toevaluate the effect of the time-dependent parameters.

The objectives of this investigation were to studyand explain the effect of initial soil water content oncrust formation, runoff, and erosion, and to accountfor duration of this effect during three successive rain-falls.

MATERIALS AND METHODSSurface samples (0-15 cm) of Capay silty clay loam and

Solano silt loam were sampled from cultivated fields inYolo County, California (Table 1). Each soil was air driedand crushed to pass through a 15-mm sieve before packinginto 0.37-m2 square plot boxes. Soil was packed to a depthof 10 cm over a 10-cm layer of sand at a nominal bulkdensity of 1.2 Mg m~3. Two initial soil water conditionswere studied: air dried and prewetted. Prewetted sampleswere saturated with distilled water by capillary action dur-ing a 24-h period, then allowed to drain for 2 h to approx-imate a field-moist condition. Soil water content was notmeasured prior to the rainfalls. Three replicate rainfalls weremade on freshly packed samples of both soils at each initialsoil water condition. Crusts were formed on a surface in-clined at a 9% slope to facilitate runoff. Soils on each plotwere subjected to a simulated rainfall at an intensity of 40mm h^1 for 1 h using a multiple-drop rainfall simulator

10 -

I

10 20 30 40Cumulative rainfall (mm)

50

0 10 20Cumulative rainfall (mm)

Fig. 2. Splash rate vs. cumulative rainfall for the (a) air-driedand (b) prewetted Solano soil for three successive rainfalls.

(Munn and Huntington, 1976). Rainfall characteristics in-cluded: 3.2-mm drop diam.; 2.5 m height of fall; 5.9 ms-1 drop velocity. Distilled water was used in all simula-tions. The second rainfall was applied at the same intensityand duration 24 h later. The soil surface on each plot wascovered between rainfalls to limit soil water loss. The thirdrainfall, with the same characteristics, was applied 7 d afterthe second rainfall on each plot where the soil surface hadbeen allowed to air dry. Soils in the plot boxes were nottilled or manipulated between rainfalls.

Samples of runoff, water splash, and soil material weretaken every 5 min throughout each 1-h rainfall. Runoffsamples were collected from a plot end spout. Splash sam-ples were collected from splash boards and troughs thatsurrounded the plots. After measuring splash and runoffvolumes, water was evaporated from the samples, and theremaining soil material was weighed. Data from splash and

Table 1. Selected physical and chemical properties of the Solanoand Capay soil materials.

Canon-Organic exchange Exchangeable

C capacity NaSoU Sand SOt Clay

SolanoCapay

_ ,, _ i

297121

b

405462

~b

298417

10.59.0

18.326.5

•••»b0.890.14

1900 SOIL SCI. SOC. AM. J., VOL. 56, NOVEMBER-DECEMBER 1992

50-i A.

0

50^

40-

10 20 30 40Cumulative rainfall (mm)

£ 10

0

B.

0 10 20 30 40 50Cumulative rainfall (mm)

Fig. 3. Infiltration rate vs. cumulative rainfall for the (a) air-dried and (b) presetted Solano soil for three successiverainfalls.

T10 20 30 40

Cumulative rainfall (mm)50

10 20 30 40 50Cumulative rainfall (mm)

Fig. 4. Soil splash rate vs. cumulative rainfall for the (a) air-dried and (b) prewetted Solano soil for three successiverainfalls.

wash samples from each of the three replicate rainfalls weresummarized in terms of sediment production rates. Datawere analyzed by one-way analysis of variance usingSuperAnova software for the Macintosh (Abacus Concepts,Berkeley, CA).

Photographs of the plot surfaces were taken after eachrainfall to document surface conditions. Selected photosfrom one soil are presented here. No other measurementsof crust characteristics were made.

RESULTSResults were similar for both soils, thus we present

curves for only the Solano soil (Fig. 1-5), and reportcorresponding Capay values in Tables 2 and 3.

Water Runoff and SplashFor the initially dry Solano soil, runoff started after

the soil received between 7 and 10 mm of rain duringthe first rainfall (Fig. la). The runoff rate increasedprogressively to a steady state of 25.5 mm h"1 after23 mm of rain (Table 2). About the same final valuewas reached after 7 mm of rain during the secondrainfall on a wet surface, and after 27 mm of rainduring the third rainfall on the dry crusted surface(Fig. la).

For the prewetted plots, maximum runoff rate dur-ing the first rainfall was 2 mm h~ J at the end of therainfall. The runoff rate increased during the secondrainfall to 19.3 mm h-1, and eventually reached the

same final value as the initially dry plots at the endof the third rainfall (Fig. Ib).

Rate of water splashed from the initially dry soilclosely followed the runoff behavior, starting at thesame time and reaching a final rate of =12.5 mm h-1

at the end of each rainfall (Fig. 2a). Splash was lowerfor prewetted soil than for dry soil, but started beforerunoff began. Final splash rates were 2.9, 6.3, and8.2 mm h-1, respectively for the successive rainfalls(Fig. 2b). Results were similar for the Capay soil,except runoff and splash started earlier.

Infiltration rates were calculated as the differencebetween rainfall applied and runoff plus splash leavingthe plots (Table 2). Final infiltration rate was =2 mmh"1 for the air-dry Solano soil for the three rainfalls(Fig. 3a). For the prewetted Solano soil, final infil-tration rates were 35, 14, and 5 mm h"1, respectively,for successive events (22, 13, and 4 mm h"1 for theCapay soil) (Fig. 3b).

Detachment and ErosionAmounts of splashed and washed soil material were

related to splash and runoff rates (Fig. 4 and 5). Rel-ative differences between air-dried and prewetted soilswere greater for splashed and washed soil materialthan for water loss. Even when runoff at the end ofthe third rainfall was the same, the final erosion ratewas lower for prewetted soil than for air-dried soil,indicating a lower sediment concentration in runoffwater (Table 3) These data should only be considered

LE BISSONNAIS & SINGER: CRUSTING, RUNOFF, AND EROSION RESPONSE 1901

10 20 30 40Cumulative rainfall (mm)

50

B.

0 10 20 30 40Cumulative rainfall (mm)

Fig. 5. Soil .erosion rate vs. cumulative rainfall for the (a) air-dried and (b) prewetted Solano soil for three successiverainfalls.

as relative values and have little absolute significancebecause they are related to the experimental deviceand to the technique of measurement.

DISCUSSION AND CONCLUSIONCrusting Dynamics

If crusting (and sealing) is a process of soil surfacedegradation due to rainfall and leads to a decrease inwater intake, then the dynamics of this process can bewell characterized by the relationship between infil-tration rate and cumulative rainfall. Because the ex-perimental device represents only a finite portion ofthe soil surface, we have to take into account the splashrate to calculate the infiltration rate, which is givenby:

infiltration = rainfall — (runoff + splash)

The decrease in infiltration rate during a rainstormcould result from both crusting and a decrease in thehydraulic gradient. Although we did not measure thehydraulic gradient, the study of three successive rain-falls with different water contents allows us to distin-guish between these two effects because the hydraulicgradient differed between the dry and prewetted soilonly for the first rainfall. We assumed that, after thefirst rainfall, the water content of the soil from thetwo treatments would be similar. Supporting the as-

Table 2. Splash, runoff, and infiltration and rainfall to steadystate, for rainfall simulations on Solano and Capay soils atdifferent initial soil water.

Splash Runoff Infiltration

Soil

Rainfall Rainfall RainfallInitial Rainfall to to tosoil event Final steady Final steady Final steady

water no. rate state rate state rate state

Solano Air dry

Prewetted

Capay Air dry

Prewetted

123123123123

mm/h12.6df12.8d12.4d2.9a6.3b8.2c

12.8y12.4y11.3y6.2x7.2xz7.9yz

mm301430-t3330371733—2730

mm/h

25cd27d25c2a

19b26cd26z26z25zllx20y28z

mm231027—3327231327—3030

mm/h1.9a0.2a2.5a

35.1d14.3cS.lb0.7x1.9xy3.4y

22.1w12.6z4.3y

mm301430—3330371733—3030

t Values within a column followed by the same letter are not significantlydifferent at P = 0.05 using Fisher's protected least significant difference.Soils were not compared; n = 3 for each rainfall.

t Steady state not achieved.

sumption, differences in hydraulic gradient do not ex-plain the results because soil with the lower gradient(prewetted soil) had the higher infiltration rate, notlower, as might be expected if gradient alone deter-mined the infiltration rate. We concluded that an in-crease in initial soil water content from air dry to nearfield capacity significantly reduced sealing and runoff,and this effect remained noticeable after three 40-mm-rainfalls. For example, after 25 mm of rain, the runoffrate was 25 mm h-1 for dry soil vs. no runoff forprewetted soil (Fig. 1).

Additional evidence for the importance of sealingand crusting is provided by photographs of the surfacemorphology of the plots (Fig. 6). The surface of theinitially dry soil became very smooth by the end of

Table 3. Splashed material, soil erosion, runoff sedimentconcentration, and rainfall to steady state, for rainfallsimulations on Solano and Capay soils at different initialsoil water.

Splash Runoff Infiltration

SoU

Rainfall Rainfall RainfallInitial to to tosoil Rainfall Final steady Final steady Final steady

water no. rate state rate state rate state

Solano Air dry

Prewetted

Capay Air dry

Prewetted

123123123123

g/5 min4.7ct4.1c2.8b0.2a0.5al.la

5y4y7z

1.5x1.2x1.8x

mm-t2024—1024—1334341418

g/5 min28d24d17cO.Sa3abSb

20y16x3li3v5v

lOw

mm301034—3024342430—3020

g/L35d28c21b7a4a6a

25y20x40z

8w9w

llw

nun281034—3424342030—2428

t Values within a column followed by the same letter are not significantlydifferent at P = 0.05 using Fisher's protected least significant difference.Soils were not compared; n = 3 for each rainfall.

t Steady state not achieved.

1902 SOIL SCI. SOC. AM. J., VOL. 56, NOVEMBER-DECEMBER 1992

Fig. 6. Solano soil surface conditions (a) air dried after first rainfall, (b) prewetted after first rainfall, (c) air dried after secondrainfall, and (d) prewetted after second rainfall.

the first rainfall, the surface of the prewetted soil re-mained rough with many unbroken aggregates afterthe second rainfall and even after the third rainfallCracking of the crust on initially dry soil due to dryingbetween the second and third rainfalls delayed the startof runoff during the third rainfall, compared with thesecond but did not increase the final infiltration ratebecause the cracks closed quickly during the rain (Fig.la). This has been reported by others (Hardy et al.,1983; Levy et al., 1986). Cracking did not affectprewetted soil because there was no continuous crustat the soil surface.

Infiltration rate for initially dry soil was 20 timeslower than for prewetted soil after 40 mm of the firstrainfall. It remained 10 and 2 times lower after thesecond and third rainfalls, respectively. This is evi-dence that a seal or crust was still not completelydeveloped on prewetted soil after 120 mm of cumu-lative rainfall. Infiltrated volume, calculated by inte-gration of the curves in Fig. 3, shows that <25% ofthe total rain applied during the three rainfalls infil-trated the initially dry soil, compared with > 68% forthe prewetted soil (30 and 63% for Capay, respec-tively). Initial soil water content affected both erosionand water storage. The differences observed in crust-ing dynamics between the two initial water contentsis explained by a difference in aggregate breakdownand crusting processes (Le Bissonnais et al., 1989;Arshad and Mermut, 1989).

Initial Water Content and Crusting Processes

From the water and sediment output results and sur-face morphology, we concluded that the primary ef-fect of higher initial soil water content was to decreaseaggregate breakdown and seal or crust formation. Assuggested by other published work, increased aggre-gate stability of prewetted soil, compared with air-drysoil, is certainly due to a reduction in slaking (LeBissonnais et al., 1989; Truman et al., 1990). Re-duced slaking decreases the production of small easilyremovable particles, reduces splash, maintains infil-tration rate, and slows seal and crust formation. Wepropose that this pattern, initialized during the wettingof the first millimeter of the surface, produces a pos-itive feedback, resulting in less soil erosion. For theprewetted soil, the amount of splashed material re-mained small during the three rainfalls, while runoffincreased 10-fold. We interpret this to indicate thatthe process is detachment limited rather than transportlimited.

For the initially air-dried soil, the final amount ofsplashed material decreased throughout the three rain-falls, while final runoff rate was the same. This sug-gests that the crust formation during the three rainfallsdecreased soil detachability under steady-state runoff.Particles are more difficult to detach from the crustthan from the initial aggregates even if runoff in-creases the water depth at the surface (Hardy et al.,

LE BISSONNAIS & SINGER: CRUSTING, RUNOFF, AND EROSION RESPONSE 1903

1983). This effect of initial soil water content on ag-gregate breakdown was previously observed in aggre-gate stability studies (Panabokke and Quirk, 1957;Henin and Monnier, 1956; Young, 1984; Truman etal., 1990; Le Bissonnais, 1990) and these results alsoshow the close relationship between aggregate break-down mechanisms, crusting, and erosion.

Erosion ConsequencesAs a consequence of reduced runoff and detach-

ment, prewetting reduced erosion. For the initially air-dried soil, erosion started as soon as runoff started andreached a maximum value at the end of the first rain-fall. This high erosion was due to high detachment(splash amount) and high transport capacity (runoff).One of these two processes without the other is in-sufficient to induce erosion. During the second rain-fall, on a wet sealed surface, erosion and runoff startedearly. During the third rainfall, on a dry crusted sur-face, the amount of eroded material was significantlylower for the prewetted soil than for the initially air-dry soil because of a lower amount of detachment(Fig. 3 and 4). Only a small amount of particles canbe detached from the crust and the previously de-tached particles have already been transported fromthe plots. These results correspond with those of Levyet al. (1986), who found that crust stability did notdecrease after drying.

For prewetted soil, no wash-sediment loss occurredduring the first rainfall because there was almost norunoff. Erosion remained low during the second andthird rainfalls (8 and 3.5 times lower respectively,than for the air-dried soil) despite increased transportcapacity. This can be explained in part by lower splashdetachment rates. Although not measured in this ex-periment, Le Bissonnais (1990) has shown that thegreater size of the detached particles is also partlyresponsible for lower wash losses from prewetted soil.

SUMMARYThese results show that initial surface soil water

content influences aggregate breakdown mechanisms,crust formation and seal processes, and subsequentrunoff and erosion. The effect was durable in timethroughout three 40-mm-rainfalls, but there existedcomplex interactions between aggregate breakdown,crusting, runoff, and erosion. Because of these inter-actions, one of these properties cannot be directly de-duced from another, unless the other factors influencingthe processes and the stage of crust development aredefined. Particularly, erosion is not a simple conse-quence of splash. Interrill erosion (as simulated here)can be higher on an initially dry surface than on a wetone during the first stages of seal or crust formationbecause of the high rate of splash and runoff. Theninterrill erosion should decrease and may be lowerthan for the uncrusted surface, because of the greaterresistance to detachment of the crust formed on theinitially dry surface. Rill erosion should increase be-cause of the increasing runoff rate over the crustedsurface.

ACKNOWLEDGMENTSThis work was funded in part by a grant from the Kearney

Foundation of Soil Science, K.K. Tanji, director.