SPECIAL SECTION

CONSERVATION IMPLICATIONS OF CLIMATE CHANGE

Expected climate change impacts on soilerosion rates: A review

MA Nearing, F.F. Pruski, and M.R. O'Neal

ABSTRACT: Global warming is expected to lead to a morevigorous hydrological cycle, includingmore total rainfall and more frequent high intensity rainfall events. Rainfall amounts andintensities increased on average in the United States duringthe 20th century, and according toclimatechange modelsthey are expected to continue to increase during the 21stcentury. Theserainfall changes, alongwith expectedchanges in temperature, solar radiation, and atmosphericCO, concentrations, will have significant impacts on soil erosion rates. The processes involvedin the impact of climate change on soil erosion by water are complex, involving changesin rainfall amounts and intensities, number of days of precipitation, ratio of rain to snow,plant biomass production, plant residue decomposition rates, soil microbial activity, evapo-transpiration rates, and shifts in land use necessary to accommodate a new climatic regime.This paper reviews several recent studiesconducted bythe authors that address the potentialeffects of climate change on soil erosion rates. The results show cause for concern. Rainfall

erosivity levels may be on the rise across much of the United States. Where rainfall amounts

increase,erosion and runoffwill increase at an even greater rate: the ratio of erosion increase toannual rainfall increase is on the order of 1.7. Even in cases where annual rainfall woulddecrease, system feedbacks related to decreased biomass production could lead to greatersusceptibility of the soil to erode. Results also show howfarmers' responseto climate changecanpotentially exacerbate,or ameliorate,the changes inerosionrates expected.

Keywords: Climate change, runoff, sediment, soil erosion, soil loss

The consensus of atmospheric scientistsis that dintate change is occurring, both interms of global air temperature and precipitation patterns. For instance, the year1998 was likely the warmest of the last 1000years in the Northern Hemisphere (IPCC,2001), and the year2001 wassecond warmeston record (NCDC, 2002). Globally, 9 of the10 warmest years since 1860 have occurredsince 1990 (WMO, 2001). Warmer atmospheric temperatures associated with greenhouse warming are expected to lead to amore vigorous hydrological cycle, includingmore extreme rainfall events (IPCC, 1995).Karl and Knight (1998) reported that from1910 to 1996 total precipitation over the contiguous U.S. increased, and that 53% of theincreasecame from the upper 10% ofprecipitation events (the most intense precipitation).The percent of precipitation coming fromdays of precipitation in excess of 50 mm hasalsoincreased significantly.

Soil erosion rates may be expected tochange in response to changes in climate fora variety of reasons,the most direct of whichis the change in the erosive power of rainfall(Nearing, 2001; Pruski and Nearing, 2002a).A second dominant pathway of influence byclimate change on erosion rates is throughchanges in plant biomass. The mechanismsby which climate changesaffectbiomass,andby which biomass changes impact runoff anderosion, are complex (Pruski and Nearing,2002b) For example, anthropogenic increasesin atmospheric carbon dioxide concentrations cause increases in plant production ratesand changes in plant transpiration rates(Rosenzweig and Hillel, 1998), which translate to an increase in soil surface canopy coverand, more importandy, biolo-gical groundcover. On the other hand, increases in soiland air temperature and moisture will likelycause faster rates of residue decompositiondue to an increasein microbial activity. More

precipitation may also lead to an increase inbiomass production. Higher temperaturesmay translate to higher evaporation rates,while more rainfall would tend to lead tohigher soil moisture levels.

Temperature changes also afreet biomassproduction levels and rates in complex ways.Corn biomass production may increase withincreasing temperature, particularly if thegrowing season is extended, but then maydecrease because oftemperature stresses asthetemperature becomes too high (Rosenzweigand Hillel, 1998). Again, biomass changesimpact soil surface cover, which greatlyimpactserosion. Another potential impact ofclimate change isassociated with the changesfrom snowfall to rainfall. If decreased days ofsnowfall translates correspondingly to increases in daysof rainfall, erosion by storm runoffis liable to increase. Even changes in soilsurface conditions, such as surface roughness,sealing, and crusting,may change with shiftsin climate, hence impactingerosionrates.

A more complex, but perhaps dominantfactor in the equation, is the potential forshifts in landusenecessary to accommodate anew climatic regime (Williams et al., 1996).As farmers adapt cropping systems, the susceptibility of the soil to erosive forces willchange. Farmer adaptation may range fromshifts in planting,cultivation, and harvest datesto changes in crop type (Southworth et al.,2000,2002a, b; Pfeiferand Habeck, 2002).

The purpose ofthispaper is to present andinterpret the principal results of four recentstudies conducted by the authors in order toshow the potential impact of climate changeon soil erosionrates, which in turn hassignificant implications for conservation planning.As a whole, these studies provide a goodoverview of both the mechanisms wherebyclimate change is expected to affect soil ero-

Mirk A. Nurfng Is a soil scientist with the U.S.Department of Agriculture-Agricultural ResearchService, Southwest Watershed Research Center inTucson, Arizona. F.F.PrutM is a professor in theDepartment of Agricultural Engineering at the Federal University of Vicosa in Vicosa, Minas Gerais.Brazil. Monte R. O'Neal is a postdoctoral researcher with the U.S. Department of Agriculture-Agricultural Research Service, National Soil ErosionResearch Laboratory in West Lafayette, Indiana.

JIF20M VOLUME 59 NUMBER 1 A3

5-3

wSp

£ ^>

sion rates and the magnitudes of change thatwe might expect from climate change. Wewill discuss first the effects and mechanisms

whereby precipitation change itselfinfluenceserosion response, and then look at some casestudies using output data from GlobalCirculation Models to assess more complicatedinteractions involving precipitation, biomassproduction, and other system factors. Wewill then overview a case study of potentialimpacts of farmers' response to climatechange forthe Midwestern United States.

In most of the studies discussed in thispaper, the Water Erosion Prediction Project(WEPP) model was utilized (Flanagan andNearing, 1995; Nearing et al., 1989). TheWEPP model accounts for most of the

processes and interactions whereby climatechange impacts runoff and erosion. In theWEPP model, plant biomass production isinfluenced by changes in temperature andsoilmoisture, and the model was modified toincludethe effectsofchanges in atmosphericcarbon dioxide concentrations on biomassproduction. Residue decomposition rates inthe model are sensitive to soil moisture and

temperature. Soil moisture is sensitive torainfall inputs and evapo-transpiration rates,which are sensitive to air temperature.Infiltration rates are impacted by residuecover and soil consolidation. Canopy andground cover are modeled as dynamic variables, and influence infiltration, runoff, anderosion in differentand interactive ways. Themodel iswell suited forstudyingthe complexinteractions involved in assessing climatechange impacts on erosion.

The results of the studieson climate changeanderosion showthat the anticipated effects ofclimate change on both runoffanderosion aresignificant and important. The interactionsinvolved are complex because of the manyinteractions between the processes involved.The trend for the United States is for increased

erosion on average, with significant geographic heterogeneity. This has important implications for conservation programs.

Potential changes In rainfall erosivityRainfall erosivity is, in general terms, theability or powerofrainto cause soilloss. It ismost generally thought of in terms of theR-factor in the Universal Soil Loss Equation(USLE) (Wischmeier and Smith, 1978),hence it wasoriginally derivedbasedon datafrom natural runoff plots located throughoutthe eastern part of the United States. It has

44 JOURNAL OFSOILANDWATER CONSERVATION I|F2004

since been found to be a robust parameter inmany parts of the world (Larionov, 1993;Schwertmann et al., 1990;Yun et al., 2002).Rainfall erosivity is correlated to the productof total rainstorm energy and maximum 30minute rainfall intensity during a storm(Wischmeier and Smith, 1978). The relationship first derived by Wischmeier andSmith is still used today in the RevisedUniversal Soil Loss Equation (Renard et al.,1997), which is the current technologyapplied in the United States for conservationplanning and compliance. Studies using aphysically-based, continuous simulationmodel of erosion have also substantiated the

geographic trends of published R-factors formany parts of the United States, includingwestern states (Baffaut et al.,1996).

Estimations of future climate change provided by Global Circulation Models (GCMs)do not provide the type of detailed storminformation needed to direcdycalculate predicted R-factor changes. Hence, statisticalrelationships between monthly and annualprecipitation and the R-factor must be usedto analyze the GCM output relative to erosivity changes (Nearing,2001).

Renard and Freimund (1994) evaluatederosivity at 155 locations within the continentalUnited States, and developed statisticalrelationships between the R-factor and bothtotal annual precipitation at the locationanda modified Fournier coefficient (Fournier,1960; Arnoldus, 1977), F, calculated frommonthly rainfall amounts as:

12

SpF =

i=i [1]

where:

pi (mm) = average monthly precipitation, andP (mm) = average annual precipitation.

The derived relationships between R-factor and P developed by Renard andFreimund (1994) were:

R-factor = 0.04830P,6I° (r3 = 0.81) [2]and

R-factor = 587.8-1.219P-K).004105PJ

(r> = 0.73) [3]and the relationships between

R-factor and F were

R-factor = 0.7397FM' (r2 = 0.81) [4]and

R-fector = 95.77-6.081F+0.04770F*

(r> = 0.75) [5]

where: the R-factor is in units of (MJ mmha'1 year1). (Note that Equation 13 in thepaperofRenard and Freimund(1994), whichcorresponds to Equation 4 in this paper,containedamisprint.) Equations 2 and 4 provided a better fit on the lower end ofthe data

range, and Equations 3 and 5 fit better on theupper end; therefore Renard and Freimund(1994) recommended using Equation2 whenP was less than 850 mm and Equation 3when P was greater than 850 mm. Likewise,they recommended using Equation 4 whenF was lessthan 55 mm and Equation 5 whenF was greater than 55 mm.

Using theserelationships, precipitation output fromGCMs canbe analyzed for trends inR-factor changes. Nearing (2001) used output from two coupled Atmosphere-OceanGlobal Climate Models,one developed by theU.K. Meteorological Office's Hadley Centre(Gordon et al., 2000; Pope et al., 2000;Woodet al., 1999) and the other from the CanadianCentre for Climate Modeling and Analysis(Boer et al., 2000; Reader and Boer, 1998;McFarlane et al., 1992).

Changes in rainfall erosivity for the twomodels were computed for two time intervals. In the first case, erosivity values from2040 to 2059 were compared to those from2000 to 2019, and in the second caseerosivityvalues from 2080 to 2099 were compared tothose from 2000 to 2019. Erosivity changeswere computed in two ways: i) as a functionof changein average annual precipitation forthe twenty year periods using Equations 2and 3, and ii) as a function of the Fourniercoefficient for the twenty year periods usingEquations4 and 5.

Erosivity results calculated from die HadleyCentre model analyses indicated a generalincrease in rainfall erosivity over large parts ofthe eastern United States, including most ofNew Englandand the mid-Atlantic states as farsouth as Georgia, as well as a general increaseacross the northern states of the United States

and southern Canada. Hadley Centre resultsalso indicated a tendency for erosivity increasesover parts of Arizona and New Mexico.Decreases in erosivity were indicated in otherparts ofthe southwesternUnited States, including parts ofCalifornia, Nevada, Utah,andwestern Arizona. Decreases were also shown over

eastern Texas and a large portion of the southern central plains fromTexas to Nebraska.

ita.©

S.S

*•!<* s

?«Id

^3

^

*5

Tab)* 1. Average magnitudes (absolute- values) of erosivity change calculated for theUnited States during the 21st century (from Nearing, 2001).

Average magnitude of change

Climate model 40 Yr. time Interval 80 Yr. time Interval

Hadley Centre

Canadian Centre

(%)17

29

(%)21

58

Effects of precipitation changes onrunoff and erosionChanges in total rainfall amount at a givenlocation may occur in different ways, primarilyeitherbecause ofan increase in the number of precipitation (wet) days or because ofincreased average precipitation per wet day.Corresponding to a change in the averageamount of rainfall during wet days isgenerally a change in rainfall intensities. Inother words, the distribution of rainfallamounts per day is generally correlated to thedistribution of rainfall intensities. These two

factors ofrainfall changewill influence runoffand erosion in different ways.

Pruski and Nearing (2002a) conducted amodelingstudyspecifically to assess the relativeinfluence of these two mechanisms of rainfall

changeon runoffand erosion. They used theWEPP model to simulate runoff and erosion

using a combination of three soil classes (siltloam,sandyloam,and clay), three slopegradients, four cropping systems, andthree locations(Corvallis, Oregon; Temple, Texas; and WestLafayette, Indiana). The model was calibratedto provide reasonable cropping, soil moisture,erosion, and runoff responses for the 108 different (3x3x4x3) scenarios.

Precipitation inputs for the scenarios wereadjusted in the following way. Total annualprecipitation was modifiedby 0%,± 10%, and± 20% relative to the historical values for

each of the three locations studied. The

relative amount of rainfall per month wasmaintained at historical(baseline) levels. Themodifications were made in three ways: 1)The number of days of precipitation wasincreasedor decreased by the desiredamount(0%, ± 10%, and ± 20%), with the averageamount of rainfall per day remaining con

stant. This modification was done by adjusting the transitional probabilities of wetfollowing wet (P:W/W) and wet followingdry (P:W/D) days for input to the climategenerator, CLIGEN (Nicks et al., 1995). 2)The average amount of precipitation perwetday was modified by the desired amount,maintainingthe number ofwet days as aconstant. In this case, usingCLIGEN, changes inthe average amount of precipitation per dayalso change the precipitation intensities in astatistically representative manner based onrelationships between thesevariables forgeneral geographic areas (Nicks and Gander,1994). Thus rainfall intensities were adjustedappropriately with rainfall amount. 3) Halfthe change was made by changingthe number of wet days, and half the change by theamount of rain

per day. For example, for the precipitationscenario of +10% from the baseline, +5%change was a result of increased number ofprecipitation days, and+5%wasaresult oftheamount ofprecipitation per day.

The overall average responses to precipitation change are reported in Table 2 as sensitivity values, i.e., the average percent changein runoff or soil loss for each one percentchange in precipitation. In every case studied, both runoff and erosion increased withincreasing precipitation (and vice versa), andin all but one case the change was greaterthan linear. Overall, runoff was more sensitive to precipitation change than waserosion.In the case of soil loss for the precipitationscenario of changing wet days only, erosionincreased with increasing precipitation (andvice versa), but with a sensitivity less thanone. The reason for this has to do with

biomass feedbacks. With other factors equal,increased precipitation caused an increase inbiomass production, which increases theresistance of the system to erosion. This actsas a partial compensating factor in regards tothe precipitation influence on erosion.Runoff is influenced by biomass, but lessstrongly than is erosion.

Erosivity results calculated from theCanadian Centre for Climate Modeling andAnalysis model also showed an increase inerosivity across the northern United States,including New England, and southernCanada. The Canadian Centre model results

also indicated a reduction in erosivity acrossmuchofthe southern plains, again fromTexasto Nebraska. The Canadian Centre model

did not show consistent results for the south

eastern United States. Results ofthe computations using annual precipitation indicatedchanges in parts of the southeast UnitedStates tending toward lower erosivity, whileresults ofthe erosivitycomputationsusing theFournier coefficient indicated the possibilityof little change or increases over part of thatregion. This suggests a change in the distribution of monthly rainfall patterns throughthe year.

Erosivity results calculated from theCanadian Centre for Climate Modeling andAnalysis show majordifferences compared tothe results from the Hadley results in thesouthwestern United States, includingCalifornia, Arizona, Nevada, and Utah.Whereas the Hadley Centre model resultssuggest a definite trend towards lower erosivity in this area, the Canadian Centre forClimateModelingandAnalysis model resultssuggest a definite, strong trendtoward greatererosivity throughthe 21stcentury in this area.

The overall results indicated that changesmay be significant, though varying widelyfrom region to region. From the most conservative of the four methods used, Nearing(2001) estimated that the average magnitudeof change (as either increase or decrease)over the country as a whole would be 17%(Table 1) different from current erosivity atthe location. At the other extreme, one ofthe four methods used predicted an averagemagnitude of change of 58% from currentconditions. Regardless of which method isused, the results suggest that changes in erosivity will be geographically variable andquite large at certainlocations.

Table 2. Sensitivities of changes In runoff and erosion to changes In average annualprecipitation: the ratio of % _ runoff or erosion to % _ change In precipitation (from Pruskiand Nearing, 2002a).

Average normalized sensitivity to changesIn average annual precipitation

Runoff

Erosion

Number of wet days

1.28

0.85

Amount of rain per day

JIF2004

2.50

2.38

Combined

1.97

1.66

VOLUME 59 NUMBER 1 45

The sensitivity analyses showed that bothsoil loss andrunoffweremuch moreimpactedby annual precipitation change when thechange in precipitation came in the form ofamount and intensity ofrainfall per day, ratherthan the number of days of rainfall. Pruskiand Nearing (2002b) assessed the results relative to the empirical equations derived byRenard and Freimund (1994) for the R-factor, discussed above, and determined thatusing the combination ofboth changing thenumber of days of rain and the rainfallamount (and implicitly intensity, as well) gavea result that followed the empirical data mostclosely. Thus we conclude that,other factorsnot considered, we can expect about a 2%and 1.7% change in runoff and erosion,respectively, for each 1%change in total precipitation under climate change. Table 2reports average sensitivity values for all thesystem scenarios studied. Individual scenarios varied a bit from the mean. For runoff,where the average sensitivity for the combined case was 2.0, the values for the scenarios ranged from 1.6 to 2.2. For erosion thecorresponding values ranged from 1.5 to 2.0.

Runoff and erosion at selected

locationsThe complex interactions of climate changeimpacts on runoff and erosion rates becomeevidentwhen looking at the modeling resultsusing specific examples of climate changescenarios. Pruskiand Nearing (2002b) investigated the changes expected in runoff anderosion as a function of the climate changesestimated for the 21st century using outputfrom the Hadley Centre model (Gordon etal., 2000; Pope et al., 2000; andWood et al.,1999) under corn and wheat managementsystems at eight locations in the UnitedStates. Climate data from eight locationswere studied: Atlanta, Georgia; Cookeville,Tennessee; Corvallis, Oregon; Pierre, SouthDakota; Syracuse, Nebraska; Temple, Texas;West Lafayette, Indiana; andWichita, Kansas.

Data spanning eleven decades, from 1990to 2099, were obtained from the GlobalCirculation Models developed by the U.K.Meteorological Office's Hadley Centre(Gordon et al., 2000; Pope et al.,2000; andWood et al.,1999). The data obtained fromthe Hadley modelwere the monthly values oftotal precipitation, mean temperature, andtotal, downward, surface, short-wave, solarradiation flux, from which monthly meanvalues of precipitation, mean temperatureand

% ', JOURNAL OFSOfl AND WATER CONSERVATION J|F 2004

solar radiation werecomputed. Perturbationsfrom the historical data for precipitation,temperature, solar radiation, and atmosphericcarbon dioxide levels were then made, asdescribed in more detail by Pruski andNearing (2002b). Soils were chosen from themost common at each location, and the simulatedcrops were chisel plow corn andno-tillwinter wheat.

The pathways whereby modeled runoffand erosionrates were affectedby the climatechange inputs were quite complicated. Thebasic results from the study of Pruski andNearing (2002b) are shown in Table 3. Interms ofannual values, there are eight (2x2x2)possible combinations (Groups) of trends inchanges for precipitation, runoff,and erosion;for example, increasing precipitation accompanied by increasing runoff and increasingerosion is one possibility (see bottom ofTable 3). Six of the eight possible combinations (Groups) were observed in the modeling results.

Only three of the eight locations studiedinvolved increasing precipitation, and for oneof those (West Lafayette, Indiana), the changewassmall (2.6%). Forthe two siteswhere theprecipitation trend was positive and wherethe significance level of the change wasgreater than 50%, the modeling data fell intoresults Group 1: increasing precipitationaccompanied by both increasing runoff andsoil loss (Table 3). For the West Lafayettesite, several of the soil/cropping scenariosmodeled fell into results Group 5 (Table 3):increasing precipitation, decreasing runoff,and increasing erosion. The explanation forthese results is related to the seasonal distribu

tions of the changes in precipitation, runoff,and erosion through the year. The predictedchanges in precipitation for West Lafayette,Indiana, as with several ofthe other locations,were not similar on a month to month basis,and the changes in monthly runoff anderosion were dissimilar as well. As an exam

ple, for the case of corn growing on theDrummer soil modeled precipitationdecreased during the growing season ofJunethrough September,which causedreductionsin runoff and erosion, while increases inprecipitation in April and May causedincreases in both runoffand erosion. The net

result was a decrease in annual runoffwith an

increase in annual erosion.

For the sites where modeled precipitationdecreased, the situation was a bit more complex. Some of the results fell into results

Group 2, where decreases in precipitation leddirectlyto decreases in runoffand erosion. Inthese cases the decreased hydrologic drivingforces of erosion dominated the processes ofchange. In other cases biomass reductionsdue to moisture stresses from less rainfall led

to increased modeled runoff and/or erosion.

The biomass reductionhada more significanteffect on the erosion than on runoff, whichproduced Group 4 results whereby erosionincreased even while runoff decreased. Both

runoff and erosion are sensitive to biomass

changes, but erosion is affected more than isrunoff. Erosion is affected by plant canopy,which reduces the impact energy of rainfall;by crop residues, which protect the soil fromraindrop impact and drastically reduce rilldetachment rates and sediment transportcapacities; and by sub-surface roots anddecaying residue, which mechanically holdthe soil in place and provide a medium formicro-organisms to thrive. The decrease ofbiomass production with decreased rainfallthus counteracted the decreased erosivity ofthe rain and runoff for Group 4 results.

The results of this study (Pruski andNearing, 2002b) suggested that in locationswhere precipitation increases are significant,we can expect runoff and erosion rates toincrease at an even greater rate than theprecipitation. The results also point out thaterosional response to climate change may bevery complex. Where rainfall decreases werepredicted, predictederosion rates were just aslikely to increase as to decrease. Given theseresults, along with the likelihood of overallincreases in heavy storms during the nextcentury (Karl et al.,1996),the overall story isone of increased erosion rates under climate

change for the coming century.

Influence of fanners' response toclimate changeAs climate and economic conditions change,farmers will respond with changes in theircrops and cropping practices. Such changesmay have a dramatic influence on erosionrates, and predicting what those changesmight be is a complicated process.

Southworth et al. (2000, 2002a, b) performed a series of studies to examine future

yield changes and optimal planting datesunder climate change in five states of theMidwest United States. Pfeifer and Habeck

(2002) expanded on these results to determine the most economically viable croprotations for farmers under climate change.

a 5"

^3

21 ft

5-3a a

.*§36o

5£?

si-

3hi

Q.

Ei0)

e>

e2at

co

H"8

«£

5

88

?£

8!i

tHtH

tHtH

tH

©<

ua>

a>o

o

COCO

q^

fjtH

OJ

CO

CO

^tH

O^

<»

CO

CM

CM

CN

oqr-_<

*q

qcm

O0>

WP

Jrl

h-

NrlN

rlrld

EE#

_«*

*"?.2.

nH

io

in

mo

)O

IN

HrllO

O

iricm

<rico

>*

*»tH

CM

tH

CM

CO

CM

CO

OU

>

coo>

odr^

coid*

ID

NO

rin

iO

CM

CM

CM

CM

CM

tH

_w

C_

_o

|S

o|S

0).2

*?0)

.5*?

SlF

OiF (0

HH

HrlrlH

O)

00

H0

0O

CO

t-500

CO*?

CMCM*

OO

)O

)*J

'St-T

t

O4

)0>

OO

t)

CM

OH

CO

O)

CO

^ri

rim

sd

CM

*t

CO

CM

CO

CO

EE<

*

(O"

•H

oo

co

st

oo

co

in

co

co

tH

co

r-

r-

iriodd

cococo

rlrlCN

(O

00

CO

O(O

CO

co-*

co3

coiri

tH

00

CO

00

**

OCO

tH

CM

CM

-H

tH

0)

(0

mx

So

isQo

JCjg

COJC

<«4

tf«

3£

(0

35

mx

2

Copyright©

2004Soiland

Water

Conservation

Society.Allrights

reserved.Journalo

fSoilandW

aterC

onservation59(I):43-50w

ww

.swcs.org

CM

(D

CM

sr

^^>

cocao

qoqq

(p<r

cp"iri

cot-i

^7

^tHtHCM

,co£.5,£_.a

•—•

qq

qcocoq

9'

iriiri

r-'CO

t-i•

>•

_j

Eh°°.

s

io

in

th

co

cm

to

tH

CO

CO

O^

O)

CM

CO

CM

tH

tH

O

q^tinoooqq

tH

irit-i

odcoCM

ocd

in

ooo

co

rlrlH

dH

d

CC

S»

So

QO

.<

OQ

.<

00

00

00

CM

CM

CM

9O

«fl

OH

rTfff00

00CO

00Q>

Q00

<*

h-^f

in

cd

in

<»

<*

"^

CM

CM

CM

3.S

L&

SS

.2.^f

in"^

fr^q

PTcm

cor^

coco

oi-<

*C

OST

tHtH

'

^t

*•C

M0

00

00

0

coco

co

a>a>

_

iq'cM'cJ^r

cm"^

irico

iriiri

cot-i

'co

mt

10(0

CO

•Jr.C

t.iS

."O

T3

coin

cocT

r-n"

CMCM

CM*<*

tHO

HC

M•

•7*A

AA

«S

*«

«s^r^^

tictio

jen

^a

oo

o

ECM°J

COd

^~

tHtH0

0C

Osf

tH«sf

OO

OIO

rlH

O)

sf

•*<

*CM

CMt-i

x-i

tHtH

coq

ino

cd^

to

orid

do

iCO

^CO

CMO

-HH

Hrld

iH

H

tffllfX

OJ

XO

J

inco

i».

ino

oco

iricm

iri^r"

co^t

''

'CM

tH

CM

Eco

<°.«

co

O(0

CM

tH

CM

00

d^

co

t-i^rcm

CM

CM

H

q^

ootor-in

cd

t-iTt

odt*rJ

^00CMtH3"CO

CM

-HCM

CM

tH

H

CO

tH

t*

I-

OtH

CM

CO*(^COCO1^

''

*CM

CM

tH

Eco'I

£6

*7

<9

^t

CO

sf

^t

CO

tH

oo

iDiq

qo

ia;

tHO

CO

CO

tHC

OC

OC

OC

MO

O

H'to

oso

oi

cid

oid

oN

CO

tHo

oin

CM

^*

CO

CO

tHC

MC

MH

zs

1^

2o

•«o

o«

o2

|o

2|

xoo

|2x

co|2

^^-

I"

CM

CM

CM

2,2

,O

COX

3CO

oo

^q

in6

qP

ooo>

odo

^d

CMCM

CM^

7^

SS

6S

SH

SO

ffflCN

CMco

t-i^f

«*r«^

co

Eoo^

S

CMC

OQ

*t

tHC

Oin

coq

cocm

^,

CO»*"

COCM

COCM

CM

CM

CM

qt-j

ftco

oosf

t^

dd

co

ri

00

(Or»

intH

CM

tHtH

tHtH

tH*H

in

tH

tH

in

in

min

tn

3S

*.0

)'a

'a,o

'o(5

to^H

ofincM'cM

^S

Sri

(O(O

iri(D

IflN

ID

SS

HH

HH

£0X

3(0

CD

T3

.O-O

"O

SH

NffiW

HC

Oq

cM

co

t-idtd

t^co

iri'

^tH

•tH

tH

EEae

_oo

«°.S

CM

CO

CM

OO

(O

CO

OO

cq

tor-co

tmo

oo

olriC

0(0

(DC

MC

MC

MC

MtH

tHtH

tH

qcn

tHco

cM

OC

om

irid

co

rH

rH

dco

co

(o

o^tn

co

(o

co

cM

CM

CM

CM

CM

CM

tHC

MC

M

in

oS

$£

a<5

5>.2

olbl^

in

idso

To determine the most profitable futurerotations they used the Purdue UniversityCrop/Livestock Linear Programming model(PC/LP) (Dobbins et al., 1994) to model sixcrop rotations with various combinations ofvarieties. Fromthese studies they predictedalarge increase in the planted area of soybeansand a decrease in the area planted to wheatacross much of the study area.

The Southworth et al. (2000, 2002a, b)and Pfeifer and Habeck (2002) studies investigated how these predicted changes in cropping might impact erosion rates. Theyapplied simulation methods similar to thosedescribed above used by Pruski and Nearing(2002b), including the application ofthe CO2sensitiveWEPP model and Hadley Centreclimate scenarios for the 21st century.

The study results indicated that bothrunoff and soil loss increased in future

scenarios compared to current conditions for10 of the 11 Midwestern regions studied(Table 4). The WEPP model predictedincreases of+18% to +274% in soil loss,withassociated increases in runoff. Soil loss and

runoff patterns frequendy followed those ofannual precipitation.

The studies of Southworth et al. (2000,2002a, b) predicted that corn yield woulddecrease under climatechange, which causeddecreased predicted soil residue cover andhence led to increased predicted erosion. Inalmost every case, however, soybean yieldswere predicted to increase while the cornyields were decreasing. Changes in 2040 to59 yields atoptimal planting dates were-31%to +18% for corn and +9% to +101% forsoybeans relative to the baseline. The dropincorn yield appeared to lead to increasederosion even when precipitation decreased.In eastern Wisconsin and southwestern

Wisconsin, where annual precipitationdecreased but runoff and soil loss increased

(Pruski and Nearing's Group 8),Juh/ precipitation (important to corn's silking period)decreased, and predicted corn yield decreased(Table 4). Therefore, the predicted loss ofcrop cover caused the predicted increase ofrunoff and soil loss.

Increases in future runoff and soil loss

would likely have been even larger if theeffect of changing from corn-soybeans tocontinuous soybeans had been more accuratelymodeled. Looking at the same climatewith two different rotations showed that the

change from corn-soybeans to continuoussoybeans could either increase or decrease

48 JOURNAL OFSOILANDWATER CONSERVATION J|F 2004

predicted runoff and soil loss, from -23% to+23% for the future scenarios. However,erosion research literature has shown that

continuous soybeans will increase soil lossrelative to rotational soybeans (Laflen andMoldenhauer, 1979; Laflen and Colvin,1981). Comparing results for continuoussoybeans and the corn-soybean rotation fortwo sample regions for each time periodshowed that while continuous soybeans hadless canopycoverthan corn-soybeans, continuous soybeans had greater ground cover,when averaged over the 100 years. Thus,WEPP predicted a decrease in erosionbecause of increased soybean ground cover.Given our current understanding of the system, such a decrease would not be expected(Laflen and Moldenhauer, 1979; Laflen andColvin, 1981). Therefore, the soil loss estimates inTable 4 could actually underestimatesoil loss, since the WEPP model was apparently over-accounting for the erosionalimpact of increase in soybean cover. Southcentral Michigan and northern Indiana werethe only regions to have soybeans withoutcorn, and modeled future soybean yieldsthere increased markedly (over 90% perperiod). This was also the only region toconsistendy show a decrease in predictedfuture soil loss. As aconsequence, evenin theone region with a predicted decrease in soilloss, the result is in doubt because the soybeans should probably have caused greatererosion. The reason that WEPP did notaccurately model the soybean cover effectonerosion is because the functions in the modeldo not adequately differentiate the effectivenessof residue type on erosion impact.

The loss of wheat in rotations also probably contributedto the increase in soilerosion.Despite the relatively small area of wheat, itmay be expectedthat the loss ofwheat fromrotations hadan impacton soil loss comparable to thatofthe adoption ofcontinuoussoybeans (Laflen and Colvin, 1981;Edwards andOwens, 1991). In a separate set of erosionsimulations that were performedby applyingthe baseline conditions, calibrations, croprotations,and plantingdates to future climateconditions, soil loss under continuous wheatwas 1/6 to 1/3 that of continuous soybeans(data not shown).

The results ofsensitivity testing with planting date showed that soil loss increased substantially with later planting dates for cornand soybeans, but not for wheat. For southern Illinois, where runoff and soil loss

increased (Table 4), examination of monthlysoil loss showed a clear May peak whichincreased significantly from the baseline to2040-59. Corn was being planted 2 weekslater (May 14) and soybeans 1 week later(May 24) than 1990-99, so the delayed planting date caused a longer time for soil toremain uncovered during April and Mayrains, which could have intensified soil loss.InWisconsin, however, where soybean planting occurred 4 to 6 weeks earlier in 2040-59than the baseline, the extended periodofcropcover could have been reducing the soil lossfrom an even greater increase than might haveoccurred otherwise.

The erosion simulations from this studyhad more widely varying results than otherstudies not taking into account changes inmanagement. This makes sense, consideringthat changes in croptypes and cropping practices such as planting dates would have amajor influence on erosion rates.

Results and DiscussionThe data generated from Global CirculationModels, which is used for input for many ofthe studiesdescribed in this paper, aredefinitive neither in terms of precise magnitude ofclimate change expected nor in terms of thegeographical distribution of the expectedchanges. This is evident, for example, whenone compares the results from the study ofNearing (2001) that uses both the HadleyCentre and the Canadian Centre for Climate

Modeling andAnalysis results. The Canadianmodel suggests greater magnitudes ofchange(Table 1), and results for California for thetwo models are opposite in sign. The important aspects, therefore, of the results of thestudies described here relate to recognizingthat change in climate will result in changein erosion rates and in understandingthe processes and factors that dominate theerosional changes.

All of the erosion/climate change studiesto date suggest that increased rainfall amountsand intensities will lead to greater rates oferosion,and there appears to be little doubtthat both average rainfall amounts and intensities are on the rise nationally. Thus, thereappears to be little doubt that erosionwill alsobe on the increase nationally, unlessamelioration measures are taken.

In terms of processes, rainfall amounts andintensities are certainly the most direct andimportant factors controlling erosionalchangesunder climate change. The results of

*1I*co2s.S

?*

ii§•3

^3

Table 4. Precipitation, runoff, anderosion estimated for 1990-99,and changes estimated for 2040-69 and 2080-99 with changes In cropmanagement.

Change CttsngB Chanfa Change Changs ChangeSon bi In hsoO In (n biwll

Crop Pwdpttatton Runoff leu* Crap proctpKsUcn runoff' lost* pradpttftJen runoff lost*(mm) (mm) (Ma*) rotation* {%) (%) (%) Group' (%) (*) {%) Croup-

199049 1990-99 199049 199049 2040-59 204059 204059 204059 204049 2080-99 208049 208049 208049and 208049

Central (MS)Wisconsin

792.4 54.9 3.3 (MS.S) 0.5 53.8 150.0 1 0.2 57.7 122.0 1

EastCentral (MS.S.SW)Indiana/Wast Central

Ohio

889.2 85.6 3.8 (MS) 10.2 9.9 34.4 1 8.4 5.0 18.7 1

Eastern (MS.MWS.SW)Illinois

867.S 117.8 6.3 (MS) 8.7 16.4 32.6 1 2.1 16.6 68.9 1

Eastern (MS.W)Wisconsin

800.3 59.5 3.8 (MS, S) -1.1 125.4 129.3 8 -6.6 112.7 119.3 8

Michigan (MS.W)Thumb

730.8 38.6 1.9 (MS,S) 14.2 49.2 105.0 1 9.7 41.6 67.9 1

North (MS.S.W)WesternOhio/South Eastern

Michigan

826.4 56.2 1.9 (MS,S) 13.8 309.5 273.7 1 7.9 284.9 225.5 1

SouthCentral (MS.W)Michigan/Northern Indiana

885.6 49.6 3.5 (S) 6.8 •26.1 -13.0 7 0.9 •37.9 •38.0 7

Southern (M,MS,MWS,SW) 1106.6Illinois

205.2 11.5 (MS,MWS) 9.6 18.6 37.5 1 9.6 18.8 33.6 1

South (M.MS.MWS.SW) 1106.2 143.1 8.4 (MS) 8.4 6.3 18.2 1 11.8 10.6 30.8 1Western

Indiana

SouthWestern

Wisconsin

(MS.SW) 802.9 97.8 5.5 (MS.S) -2.1 120.8 147.2 8 -6.2 118.1 134.7 8

Western

Illinois

(MS.SW) 933.7 119.1 8.2 (MS) 7.9 7.7 18.9 1 1.2 3.1 24.5 1

' Runoff and soil loss are averaged over all crop rotations, averaged over all three soil types for each region. All changes are relative tobaseline conditions (199049).

* M a com, 8 a soybeans, W = wheat; single letter = continuous crop, multiple letters = rotation.

* After Pruski and Nearing (2002a).

simulation studies suggest that erosion willincrease approximately 1.7% for each 1%change in annual rainfall. This result is inbasic agreement with our understanding ofthe relationship between rainfall erosivityandrainfall amounts that we know from measured

erosion data. The results also indicate that the

dominant factor related to the change in erosion rate is the amount and intensity of rainthat falls in the storm, rather than the numberofdays of precipitation in a year.

The second dominant process related toerosion and climate change is biomass production. Biomass levels will change underclimate change due to changes in temperature,moisture, and atmospheric carbon dioxide levels, and biomass ranks right next torainfall in terms ofits impact on erosion rates(Nearing et al. 1990). The change in erosion rates as a function of biomass is perhapsthe most complex to understand because of

the many interactions and counter-effects,and to the greater effect of biomass onerosioncompared to runoff.

The third major process of erosion ratechanges under climate change, and the wildcard, is landuse. Detailed landuse changes asa function of future climates (both weatherrelated and economic climates) are nearlyimpossibleto predictwith anydegree ofaccuracy. Nonetheless, the study conducted byO'Neal et al. (in press) is suggestive that landuse changes will occur and that they willimpact erosion rates. The trend from thoseresults suggests that erosion will increase as afunction of future land use changes in theMidwest United States, largely because of ageneral shift away fiom wheat and corntowardssoybean production. Other scenariosarepossible that would leadto different results.

References CitedArnoldus,H.MX. 1977. Methodology used to determine the

maximum potential average annual soil loss due to sheetand riH erosion in Morocco. FAO Soils Bulletin 34:39-51.

Baffaut, C, M.A. Nearing,andA.D.Nicks. 1996. Impactofclimate parameters on soilerosion usingCLIGEN andWEPP. Transactions of the American Society ofAgricultural Engineers 39:447-457.

Boer, G.J.,G. Flato,M.C. Reader, and D. Ranuden, 2000. Atransientclimate changesimulationwith greenhouse gasand aerosol forcing: experimental design and comparison with the instrumental record for the 20th century.Climate Dynamics 16(6):405-425.

Dobbins, C.L..Y Han, P. Preckel, and D.H. Doster. 1994.

Purdue Crop/Livestock LinearProgram: User'sManual.E-EC-6. Purdue University,West lafette, Indiana.

Edwards.W.M. and L.B. Owens. 1991. Largestorm effectson total soil erosion. Journal of Soil and WaterConservation 46(l):75-78.

Flanagan, D.C and M.A. Nearing. 1995. U.S. Departmentof Agriculture-Water Erosion Prediction Project:Hillslope profile and watershed model documentation.National Soil Erosion Research Laboratory Report No.10. U.S. Department of Agriculture-AgriculturalResearch Service National Soil Erosion Research

Laboratory,West Lafayette,Indiana.

JIF2004 VOLUME 59 NUMBER 1 49

3 5"

etft

dI

^3

Fournier, F. I960. Climat et Erosion. Univenitaries de

France, Paris.

Gordon C, C. Cooper. CA. Senior, H. Banks, J.M.Gregory,T.C. Johns. J.F.B. Mitchell, and R.A. Wood. 2000. Thesimulation ofSST.scaice extents andoceanheat transports

in aversion of the Hadley Centrecoupledmodel withoutflux adjustments. ClimateDynamics 16:147-168.

Intergovernmental Panel on Climate Change (IPCC)WorkingGroupI. 2001. Summaryforpolicymakers: Areport of Working Group I of the IntergovernmentalPanelon Climate Change.

Intergovernmental Panel on Climate Change (IPCC).1995. Second Assessment Synthesis of Scientific-Technical Informationrelevant to interpretingArticle 2of the UN Framework Convention on Climate

Change. Intergovernmental Panel on ClimateChange.Geneva. Switzerland. 64 pp.

Karl T.R.. R.W. Knight. D.R. Easterling, and R.G. Quayle.1996. Indices of climatechange for the United States.Bulletin of the American Meteorological Society77(2):279-292.

Karl T.R. and R.W. Knight. 1998.Seculartrend ofprecipitation amount, frequency, and intensity in the UnitedStates.Bulletin of the American Meteorological Society79:231-242.

Laflen. J.M. andW.C. Moldenhauer. 1979. Soil and waterlosses from corn-soybean rotations. Soil Science SocietyofAmericaJournal43(6):1213-1215.

Laflen.J.M.andT.SCorvin. 1981. Effect of cropresidue on soilloo from continuous row cropping. Transactions of theAmerican SocietyofAgricultural Engineers 24:605-609.

Larionov, G.A. 1993. Erosion and deflation of soils.

Moscow University Press. Moscow,Russia 200pp.McFarlane. N.A.. G.J. Boer,J.-P. Blanchet,and M. Lazare,

1992. The Canadian Climate Centre second-generation general circulation model and its equilibriumclimate. JournalofClimate 5:1013-1044.

NationalClimatic Dau Center (NCDQ, 2002. Climate of2001—Annual Review. National Oceanic and

AtmosphericAdministration,Ashevillc, North Carolina.Nejring. M.A. 2001. Potential changes in rainfall erosivity

in the U.S. with climate change during the 21stcentury. Journal of Soil and Water Conservation56(3):229-232.

Nearing. M.A.. L.D. Ascough, and J.M. Laflen. 1990.Sensitivity analysis of theWEPP hiBslope profileerosionmodel. Transactions of the American Society ofAgricultural Engineers 33:839-849.

Nearing, M.A., G.R. Foster, L.J. Lane, and S.C. Finkner.19X9. A process-based soil erosion model for U.S.Department of Agriculture-Water Erosion PredictionProject technology. Transactions of the AmericanSocietyofAgricultural Engineers32(5): 1587-1593.

Nicks A.D.andCA. Gander. 1994.CLIGEN: A weathergenerator forclimate inputsto waterresource andother models. Pp. 903-909. In: American Society of AgriculturalEngineers, International Conference on Computers inAgriculture. 5th American Society of AgriculturalEngineers, St,Joseph, Michigan.

NicksA.D..L.J. Lane,andG.A.Candcr. 1995. Weather generator. Pp. 2.1-2.22. In: U.S. Department of Agriculture-Water Erosion Prediction Project: Hillslope Profile andWatershedModel Documentation (National Soil ErosionResearch Laboratory Report No. 10),D.C. Flanagan andM.A. Nearing (eds.) U.S. Department of Agriculture-Agricultural Research Service National Soil ErosionResearch Laboratory,West Lafayette, Indiana.

[_J0 | JOURNAL OF SOIL AND WATER CONSERVATION J|F 2004

Pfeifer, R.A. and M. Habeck. 2002. Farm level economic

impacts of climate change. Pp. 159-178. in: O.C.Doering, J.C. Randolph, J. Southworth, R.A. Pfeifer,(eds.) Effects of Climate Change and Variability onAgricultural Production Systems. Kluwer AcademicPublishers, Boston, Massachusetts.

Pope,V.D., M.L. Gallani. PR. Rowntree. R.A. Stratton.2000. The impactof new physical parameterizations inthe Hadley Centre Climate Model - HadAM3.Climate Dynamics 16:123-146.

Pruski, F.F. and M.A. Nearing. 2002a. Climate-inducedchanges in erosion during the 21st century for eightUS. locations. Water Resources Research 38(12):article no. 1298.

Pruski. F.F. and M.A. Nearing. 2002b. Runoff and soil lossresponses to changes in precipitation: a computer simulation study.Journal of Soil and Water Conservation57(1):7-16.

Reader, M.C. and G.J. Boer. 1998. The modification ofgreenhouse gas warming by the direct effectof sulphateaerosols. Climate Dynamics 14:593-607.

Renard, K.G. and J.R. Freimund. 1994. Using monthlyprecipitation dau to estimatethe R-factor in the revisedUSLE. Journalof Hydrology 157:287-306.

Renard, K.G., G.R. Foster. G.A.Weesies, D.K. McCool, and

D.C. Yoder. 1997. Predictingsoil erosionby water- aguide to conservation planning with the reviseduniversal soil loss equation (RUSLE). AgriculturalHandbook No. 703, U.S. Government PrintingOffice,Washington, D.C.

Rosenzweig,C. andD.Hillel,199K. Climatechangeandtheglobal harvest. Potential impacts of the greenhouseeffect on agriculture. Oxford University Press, Inc.New York.

Schwertmann, U, W. Vogl, and M. Kainz. 1990.Bodenerosion (lurch Wasser. Eugen Unger and Co.Publ. Stuttgart,Germany. 64 pp.

Southworth. J., M. Habeck, R.A. Pfeifer, J.C. Randolph, O.Doering, J. Johnston, and D. Gangadhar Rao. 2002a.Changes in soybean yields in the Midwestern UnitedStates as a result of future changes in climate,climatevariability, and COi fertilization. Climatic Change53:447-475.

Southworth, J., R.A. Pfeifer,and M. Habeck. 2002b. Cropmodeling results under climate change for the upperMidwestern United States. Pp. 127-158. In: O.C.Doering, J.C. Randolph, J. Southworth, R.A. Pfeifer(eds.). Effects of Climate Change and Variability onAgricultural Production Systems. Kluwer AcademicPublishers, Boston, Massachusetts.

Southworth, J.,J.C. Randolph, M. Habeck,O.C. Doering,R.A. Pfeifer, D.Gangadhar Rao,andJ.J.Johnston. 2000.Consequences of future climate change and changingclimate variability on maize yields in the MidwesternUnited States. Agriculture, Ecosystems, and Environment 82:139-158.

Williams, J. M.A. Nearing, A. Nicks, E. Skidmore. C.Valentine, K. King, and R. Savbi. 19%. Using soil erosion models for global change studies.Journal of Soiland Water Conservation 51(5):381-385.

Wischmeier,W.H. and D.D. Smith. 1978. Predicting rainfall

erosion losses—A guide to conservation planning.Agricultural Handbook No. 537, U.S. GovernmentPrinting Office,Washington, D.C.

World Meteorological Organization (WMO), 2001. WMOstatement on the status of the global climate in 2001.World Meteorological Organization, Geneva,Switzerland.

Wood, R.A., A.B. Keen. J.F.B. Mitchell, and J.M Gregory.1999. Changing spatial structure of the thermohalinecirculation in response to atmospheric CO? forcing in aclimate model. Nature 399:572-575.

Yun, X., Liu, B.Y., and M.A. Nearing. 2002. A practicalthreshold for separating erosiveand non-erosive storms.Transactions of the American Society of AgriculturalEngineers 45(6):1843-1847.

1?

^3tvtj.ft 3-

1