Global Biogeochemical Cycles - 2001 - Goldewijk - Estimating Global Land Use Change Over The Past 300 Years The HYDE

GLOBAL BIOGEOCHEMICAL CYCLES, VOL. 15, No.
2, PAGES 417-433, JUNE 2001
Estimating global land use change over the past 300 years:
The HYDE Database
KeesKlein Goldewijk
NationalInstituteof PublicHealth andthe Environment(RIVM), Bilthoven,The Netherlands
Abstract. Testingagainsthistoricaldatais an importantstepfor validatingintegratedmodels

of globalenvironmentalchange.Owing to long time lagsin the climatesystem,thesemodels
shouldaim the simulationof the landusedynamicsfor longperiods,i.e., spanningdecades
up to a century.Developingsuchmodelsrequiresunderstanding of pastandcurrenttrends
andis thereforestronglydatadependent. For thispurpose,a historydatabaseof the global
environment hasbeendeveloped:HYDE. This paperdescribes andanalyzespartsof HYDE
version2.0, presentinghistoricalpopulationandlandusepatternsfor the past300 years.
Resultssuggest,amongotherthings,a globalincreaseof croplandareafrom 265 million ha
in 1700 to 1471 million ha in 1990, while the areaof pasturehasincreasedmore than six fold
from 524 to 3451 million ha. In general,the increaseof man-madeagriculturalland took
placeat the expenseof naturalgrasslands andto a lesserextentof forests.There are
differencesbetweenthe severalregionsin the temporalpaceof theseland useconversions.
The temperate/developed regionsof Canada,United States,USSR, andOceaniaappearto
havehadtheir strongest increaseduringthe 19thcentury,while mostof the
tropical/developing regionswitnessedthe largestlanduseconversions at the endof the last
century.Resultsof this analysiscanbe usedto testintegratedmodelsof globalchangeand
are availableat http://www.rivm.nl/env/int/hyde/.
1. Introduction man-made land cover conversionsover the last 300 years.

HYDE fits very well in initiatives like the aforementioned,
It is beyond doubt that human activitieshave modified the
supportingtheir empiricalbase.
naturalenvironmentconsiderably.Nowadaysit becomesclear
Historical land cover conversions have resulted in
that during the last centuriesthe intensityand scale of these
modifications has increased significantly [Watson et al.,
significantemissionsof greenhousegases,such as carbon
dioxide,to the atmosphere.Severalattemptshave beenmade
1996]. The consequences of thesemodifications to the global
society-biosphere-climate system are often referred to as to quantifythe emissionfluxes,both on regionaland global
global change.The possibleimplicationsof global change scales, resulting in different ways to assessthis issue.
have resultedin variousinternationalefforts to identify the Houghtonet al. [1983] developedan accountingmodel to
underlyingdriving forces.An exampleof suchan effort is the analyzethe responseof terrestrialcarbonstorageto changes
International Human Dimensions of Global Environmental in landuse[seealsoHoughtonandHackler,1995;Houghton,
Change Programme(IHDP), which identifies four science 1999]. The modeltracksthe extentof land area,which is
projects. One of these projects, cosponsoredby the affected by land cover changes such as harvest and
International Geosphere-Biosphere Programme (IGBP), is deforestation,
by applyinghistoricalclearingestimates.[Esser
"Land-Use and Land-Cover Change (LUCC)," whose and Overdieck,1991] developedthe so-calledOsnabruck
objectivesare "to obtain a better understandingof land-use Biosphere Model (OBM). The OBM modelis a budgetmodel
and land-cover changes (e.g., degradation,desertification, that is developedon a high-resolutiongrid. Each grid is
biodiversityloss) and of the physical and human driving characterizedby a specificlandcovertype,climateandsoil.It
forcesbehindtheseprocesses"[ Turner et al., 1990]. Another representschangingland usein a fairly mechanistic way. In
effort is a core project of the IGBP called PAGES (Human the model the demandfor agriculturalland is a functionof
Interactionsin Past EnvironmentalChanges),chargedwith globalpopulation pressure andstatistics
onhistoricalchanges
providing a quantitativeunderstandingof the Earth's past in agriculturalarea, and of productivity.However, it is
assumed that most of the land cover conversions occur in the
climateand environment.Its focus3, the Human Impactson
Terrestrial Ecosystems(HITE) initiative, concentrateson tropics.Richardsand Flint [1994] presentan extensivestudy
of historicalland usechangesin SoutheastAsia for the period
1880-1980.Hall [1995] developeda seriesof algorithmsto
Copyright2001 by the AmericanGeophysicalUnion. represent (1) the tendencyto develop("nibbleawayat") land
next to land that was alreadydeveloped;(2) the jumping over
Papernumber1999GB001232. of relative unfavorable locations; and (3) the patterns and
0886-6236/01/1999GB001232512.00 rates of land-use change caused by population density,
417
19449224, 2001, 2, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/1999GB001232 by Nat Prov Indonesia, Wiley Online Library on [19/04/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
418 KLEIN GOLDEWIJK: 300 YEARS OF GLOBAL HISTORICAL LAND USE CHANGE
economy,andpoliticalfactors.The modelhasbeenappliedto as population,land use, livestock, gross domesticproduct,

regions as Costa Rica, Southeast Asia, and sub-Saharan andvalueaddedof industryand/orservicesaswell asspecific
tropical Africa. Recently, Ramankuttyand Foley [1999] data categories concerning energy/economy,
presenteda similarstudyin which they reconstructed global atmosphere/ocean,and the terrestrial environment. These
historical crop cover for the period 1700-1992, based on "driving forces for global environmentalchange" occur at
contemporaryland cover derived from satelliteimages(the severalspatialandtemporalscalesand dimensions,and differ
DISCoverproject,Lovelandand Belward [ 1997]), hindcasted often among regiens. Therefore a geographicallyexplicit
with a combinationof historicalstatisticsand percentage modeling approach is required. Most data have been
croplandin a grid cell. Major differenceswith this studyand organizedat the countrylevel for the period 1890-1990,and
Ramankuttyand Foley [1999] arethe following:(1) this study wherepossiblefor the period 1700-1990.Somedataare also
uses the DISCover IGBP legend classification for available with geographicdetail [Klein Goldewijk, 1997;
contemporary landcover,while RamankuttyandFoley [1999] Klein Goldewijkand Battjes, 1997].
usean aggregationof the OlsonGlobal Ecosystemclasses;(2)
this studyusesan adjustedversionof BIOME for undisturbed
natural land cover [Prentice et al., 1992; Leeroansand van 3. From Statistics to Maps
den Born, 1994], while Ramankuttyand Foley [1999] use a
3.1. Population as a Proxy for Agricultural Activity
BIOME versionof Haxeltine and Prentice [1996]; (3) this
studyexplicitlydealswith pastureland; and(4) the allocation Globalpopulationnumbershavebeenrelativelylow for a
of croplandareasis steeredin this studyby the presenceof longperiodof time. Estimatednumbersfrom 8000 B.C. of 5-
historicalpopulationdensities,rather than by downscaling 10 million peopleincreasedvery slowly with fluctuationsto
contemporary landuseasin RamankuttyandFoley [1999]. -500 million in 1600 A.D. [Grigg, 1974]. Globally, both
This paper describesa databasethat could make a relative and absoluteincreaseswere still very small until the
contribution in particular areas of global environmental 1700s.From that point onwardhumanlifetimesprolonged,
assessments, with emphasison land use patternsduringthe mainly due to better living conditions.Later, death rates
last 300 years. The databasepresentsnot only historical beganto decline, first in Europe and later also in America,
statisticalinformationon a countrybasisbut offersa spatial Australia, and other colonial areas [Durand, 1974]. After
representation on a high-resolution
grid of the resultsas well, World War II, populationgrowthratesaccelerated in the less
whichcanbe usedby integratedmodelsof globalchange. developedregions (Africa, Asia, and Latin America). A
summaryof estimatedglobalpopulationnumbersis presented
in Table 2.
2. General Background of HYDE
Populationdataon a countrybasisfor the period1950-1995
The HYDE databasewas originally designedfor testing were derived from the 1996 Revision of the United Nations
and validation of the IMAGE 2 model. This was the result of
World PopulationProspects[United Nations, 1997]. Mitchell
the 2nd IMAGE AdvisoryBoard Meeting [Solomon,1994], [ 1993, 1998a,b]providescountrypopulationestimates for most
with the reasoning that models used for integrated countriesof the world for a long historicaltime path (1750-
environmentassessments, which make projectionsfor the 1993). Another important source is Maddison [1994] who
next hundredyears, should also be able to simulatethe last presentsestimatesof populationnumbersand GrossDomestic
hundredyears.The objectiveof the IMAGE 2 modelis to fill Product(GDP) for severalcountriesfor the period 1820-1992.
in somemultidisciplinary gapsin globalchangeresearchby In orderto createa consistentdata set, the historicalcountry
providing a disciplinaryand geographicoverview of the data points from Mitchell and Maddisonand were scaledto
society-biosphere-climate system.This steersthe regional matchthe United Nationsdatain 1950. Data gapswere filled in
breakdownof the model (Table 1) and subsequently the with a logistic curve. Country and regionaltotals obtainedin
organization of the HYDE database. IMAGE simulates this way (see Table 3) were checkedagainstother available
greenhouse-gas emissionsstemmingfrom variousindustrial estimates.
activities,energyuse, and land use.Data from 1970-1990 are The startingpoint for the global georeferencedhistorical
used to initiate and calibrate the model runs. Different populationmapsis the 0.5ø x 0.5ø degreelongitude/latitude
assumptionson population,wealth, technology,and policy populationdensitymap of 1994 from the National Center for
developments are evaluatedfor globalchangeconsequences GeographicInformationandAnalysis(NCGIA) [Tobleret al.,
for the period1990-2100.The modelhasbeenused,amongst 1995]. The NCGIA data set was overlayedwith the HYDE
otherthings,as a tool for internationalclimatenegotiations country borders, and grid cells of the NCGIA database
and Intergovernmental Panel on Climate Change (IPCC) belongingto countriesas definedby HYDE were aggregated to
assessments [Alcamoet al., 1995; Watsonet al., 1996]. To countrytotals.The HYDE countrytotalswere adjustedin order
improvethe accuracyandcredibilityof projections madewith to equal the countrytotals of the United Nationspopulation
modelssuchas IMAGE 2, it is crucialthat they are tested database. Finally,thepopulation densities
werescaled to a 0.5ø
againstreliable data sets. Becausea dynamicmodel like x 0.5ø latitude/longitudegrid by usingthe historicalcountry
IMAGE 2 simulates trends, it has a need for time series of totalscombinedwith the samepopulationdistributionas in the
historicaldata. For this purpose,a databasewith historical NCGIA database,under the assumptionthat high population
dataof the globalenvironment (HYDE) was created[Klein densityareasremainin the sameplaceovertime.Of course,this
Goldewo'kand Batties, 1995; 1997]. HYDE can be usedto assumption placesa bias on the distributionof pastpopulated
testglobalchangemodels,coveringbothgeneraltopicssuch areas. Therefore, subnationaldata for large countrieslike
419
KLEIN GOLDEWIJK:300YEARSOF GLOBALHISTORICALLAND USECHANGE
KLEIN GOLDEWIJK:300YEARSOF GLOBALHISTORICALLAND USECHANGE
(.•
•
(.•
,--½
•
(.•
o
(.•
,--½
o
o
.•._•
420
Canada,UnitedStates,Mexico,Brazil,Argentina,India,China, the former Soviet Republics Armenia, Azerbijan, Belarus,

andAustraliaderivedfromMitchell[ 1993,1998a,b]wereused, Estonia,Georgia, Latvia, Lithuania, Kazakstan,Kyrgyzstan,
to allow for at least someinternalpopulationchanges.The Moldovia, Russia, Tajikistan, Turkmenistan,Ukraine, and
resultsarepresented
in Plate 1. Uzbekistan were assumed to follow the same trend as the
former USSR. The assumptionwas made that the share of
3.2. Allocation of Agricultural Land croplandof each republic relative to the total area of the
USSR has remained constant over time between 1700 and the
Conversionof land coveris an importantissuein global present. The same applies to the republics of Slovenia,
changestudies.Populationgrowth has resultedin an ever Croatia,Bosnia-Herzegovina, the former YugoslavRepublic
increasingdemand for food, which in its turn resultedin a of Macedonia,and the FederalRepublicof Yugoslaviawhich
worldwide expansionand intensificationof croplandand follow the trend of former Yugoslavia,the Czech Republic
pasture[Turnere! al., 1990]. This expansiontook mostof the and Slovakia which follow Czechoslovakia,and Ethiopia
time place at the cost of natural primary forests and D?R andEritreawhich follow the trendof formerEthiopia.
grasslands. Althoughthe emphasisis laid on tropicalregions 3.2.1. Starting point: Initial land cover. Potentialnatural
when lookingat (recent)deforestation, the sameprocesses vegetationpatternsare simulatedwith a modifiedversionof
occurredin the industrialized countriescenturiesago.Before the BIOME model (original versiondescribedby Prentice e!
the greatoutwardmovementfrom the continent,Europeans al. [1992]; with somemodifications aspresented by Leeroans
had alreadycolonizedtheir continentvigorouslyand had and van denBorn [1994], drivenwith climatefromthe IIASA
clearedmost forestsduring the Middle Ages [Williams, climate databaseof Leemartsand Cramer [1991]). Global
1990]. The expandingpopulationneededmore food and took vegetationpatternscanbe derivedby agglomerations of plant
awaypartsof the forestto createarablelandandpasture.At functional types, each having definable environmental
the sametime therewas an increaseof wooddemanding iron constraints (temperature, moisture, etc.). This dynamic
smelting,productionof charcoal,mining,and shipbuilding approachhas the advantagethat it can anticipateto climate
activities.In EuropeanRussiathe deforestation reallytookoff change.
during the 16th and 17th century, when slash-and-bum The initial map for agricultural land is based on a
agriculturegaveway to a more sedentaryand intensivethree- combinationof a georeferenced datasetandnationalland use
field farmingsystem. statistics.Calibrationcriteria are that "(i) geographicpatterns
Startingpoint in this studyfor the land useestimatesis the of agriculturalland are consistentwith the nationalstatistics
statisticaldatabaseof the FoodandAgriculturalOrganization from the FAO [1996], (ii) it supportsthe computationof the
of the United Nations (FAO) [FAO, 1996], which was transientvegetationresponseto climate change,and (iii)
consideredmost authorative.The categories"Arable Land computedgeographicpatterns of individual crops and
and PermanentCrops" were used for cultivated land and potentialnationalvegetation resemble the patternssimulated
"?ermanentPasture"for pasture.The FAO presentsfor each in the TerrestrialVegetationModel of IMAGE 2" [Alcamoe!
countrydata for the period 1961-1994. Additionally,for the al., 1998].Initial agricultural
land coveris derivedfromthe
United Statesstate level information on croplandarea was DISCover land use data set, classified accordingto the
used,as describedby Ramankuttyand Foley [1998] for the InternationalGeosphereBiosphere?rogrammeGlobalLand
period 1850-1990 (N. Ramankuttypersonalcommunication, CoverLegend[Lovelandand Belward,1997]. The scheme
2000). Data for the period 1700-1850 were estimatedby consistsof 17 classes,of which "Croplands"and "Urban and
taking into accountthe colonizationof the different states built-up"were given the sameweight, and then usedfor
over time, which resultedin different curvesper state. For allocation of the amount of "arable Land and ?ermanent
India, Sri Lanka, Bangladesh,Myanmar, Laos, Thailand, Crops"aspresented by FAO. If theneedfor arableareacould
Vietnam, Brunei, Indonesia and the Philippines,estimated not be satisfiedwith thesecategories,the DISCover classes
croplandareasfrom Richardsand Flint [ 1994] were usedfor "Savannas"(9), "Grasslands"(10) and "Cropland/Natural
1880, 1920, 1950, and 1980; the years in between were VegetationMosaic"(14) were also giventhe sameweight
interpolated.The 1700 estimatefor countriesbelongingto the (but lower then the "Cropland"and "Urbanand Built-up")
regionsof Latin America, SouthAsia, FormerUSSR, China, and used to allocate the rest. This approachis somewhat
Pacific Developed,and SoutheastAsia were scaledto match differentasgivenby Ramankutty
andFoley[ 1999],wherethe
the totals as estimated by Richards [1990] and Klein DISCover data set classified under the Olson Global
Goldewijk and Battjes [1997]. The analysis of Richards Ecosystems
(OGE) frameworkwasusedin combination
with
[1990] is drawn from the originalwork of Houghtone! al. the use of BIOME-3 model of Haxeltine and Prentice [ 1996]
[ 1983], andKlein Goldewijkand Battjes[ 1997] addedpasture to fill in regionsdominatedby landuse.
estimatesto the data of Richards [1990]. The rest of the 3.2.2. The allocation of cropland was done by applying
countries in the world were scaled to match the global the followingrules. (1) The total amountof croplandon a
estimateof Richards[ 1990]. sub-national(when available)or countryscalewas allocated
Estimatesfrom Richards[ 1990] for the world, Houghton accordingto historicalpopulationdensitymaps(as described
e! al. [1991] for Latin America, and Klein Goldewijk and in section3.1). The grid cells with the highestpopulation
Battjes [1997] for other regionshave been usedto estimate densitieswere first assignedto cropland,thenthosewith the
the changein pasture areas.Again, countriesin one region secondhighestdensity,etc.,until thetotalamountof cropland
were assumedto follow the same regional trend if no was allocatedin that unit. (2) The allocationof croplandis
additionalcountryspecificdata were available.Furthermore, restrictedto the agriculturalareaas determinedby the initial
KLEIN GOLDEWIJK: 300 YEARS OF GLOBAL HISTORICAL LAND USE CHANGE 423
landcovermap,whichis considered as beingrepresentative estimateof Matthews[1983] of 1760Mha croplandin 1980is

for presentagriculturalactivity.This areais regardedas a somewhathigher. Compared to the original undisturbed
maximumboundaryfor agricultural activities.Croplandwas global forest area, some 7% was cleared for agricultural
notallocatedoutsidethatarea(seeFigure1). purposesin 1700.
3.2.3. The allocationof pasturewas done by applying The global amount of pasture has also increasedat a
the following rules. (1) The total amountof pasturein a stunningrate duringthe last 300 years,from 524 Mha in 1700
countryor statewas also allocatedaccordingto population to 3451 Mha in 1990 (a 6.6-fold increase,Table 5). Until
density,while excludingthosegrid cellsthat were already 1700, some6% of the naturalundisturbedgrasslands, steppe
allocatedto croplandon the basisof the above-described
rules and savannas were already estimated to have been
(2) The allocationof croplandis restrictedto the agricultural transformedinto agriculturalland; this number increasedto
areaas determined by the initial land covermap.This map 16% in 1850 and resulted in a total 49% loss in 1990. The
servesas a maximumboundarywere agriculturalactivitiesare studyof Ramankuttyand Foley [ 1999] estimatedlossesof 3,
assumedto be feasible(see Figure 1). A summaryof the 6, and 20%, respectively.Differencescan be attributedto the
methodsusedto estimatehistoricalland useis presented
in factthattheydid not includeconversion to pasture.Especially
Figure 2. in the temperateregions, large areas of natural grasslands
wereploughedand convertedto cropland(e.g., the prairiesin
4. Results and Discussion the United Statesand steppeareasin the formerUSSR). Most
of the tropicalnaturalgrasslands were convertedinto pasture
Resultsof the presentedanalys•sshouldbe considered for raising livestock (e.g., the camposin Brazil and the
carefullyas the studyknowsmanyuncertainties. It is very pampasin Argentina),or wild grazinganimalswere simply
hard to comparebetweenthe severaldatasourcesbecauseof deposedby domesticherds(Subsaharan Africa). In someparts
the use of different classifications and methodologies.of the world pristine grasslandswith no natural history of
However,according to thisstudy,the globalareaof cropland large herbivores(New Zealand and South America) were
apparentlyhasincreased5.5-fold in lessthanthreecenturies, invadedby settlersand ranchers,encouragedby a growing
from 265 Mha in 1700to 1471Mha in 1990(whichis equal marketfor beef andwool [Richards,1986].
to the FAO estimate,seeTable 4). This is a lower estimate The increasein croplandhas not been distributedevenly
thanRamankutty andFoley[1999] who estimated 400 Mha in over the different regions. For example, Canada and the
1700 and 2000 Mha in 1990, but similar to the estimate of United Stateshave experienceda relatively slow increaseof
[ Williams,1990] of 1400 Mha and the 1500 Mha estimateof their agriculturalareasfor a longtime. This was merelydueto
[Richards,1990], althoughboth were for the year 1980. The the different agriculturalpracticesof the indigenouspeople
Notusedforagricultural
activities Based on:
1 Usedforagncultural
activities - DISCoverwithIGBP legend;Lovelandand Belward[1997].
- Statisticsfrom FAO [1996].
Figure 1. Agriculturalareasasdefinedby the IMAGE 2 model,basedon theDISCoverdatabase

with theIGBP legendclassificationandcalibratedwith FAO [ 1996]statistics.
424 KLEIN GOLDEWIJK:300 YEARSOF GLOBALHISTORICALLAND USE CHANGE
Step 1' populationdensity
Population
density
map
of
1994
derivedfromTobleret al. (1995)
1994
World PopulationProspects,The
1996 Revision, United Nations
PopulationDivision,yearly nationaltotals
Ruleforhistorical
population
density
map:
1950
Yearly(sub-)nationaltotalsfrom:
- Downscaling
densities
until
country/state International Historical Statistics
totalsmatchthehistoricaltotalestimate (Mitchell, 1993-1998)
&
Country/stateCensusdata
(varioussources)
&
Logisticcurvesfor 'no data' and
'fillinggaps'
Populationdensitymaps for 1700
the period 1700- 1994
Step 2: land use
Natural undisturbed land cover Currentagriculturalarea based on

computedby the BIOME model DISCover(IGBP legend)& FAO (1996)
,
Current land use map
Set of rules for allocationcroplandand pasture:

Amount of cropland and pasture
per countrybasedon FAO (1996),
- Allocate croplandfirstaccordingto population and per state based on historical
densttymaps until total amount of cropland statistics
per country/stateis distributed
- No allocationoutsideinitialagriculturalarea
- Allocatepastureuntiltotalamountof pasture
per country/state/sdtstributed,grid cellsalready 1961
allocatedto croplandare excluded
_ • Historical
-•'
land
use
estimates,
per
country
orstate(several
sources)
Historical land use maps

1700
for 1700 - present
Figure2. Scheme
forcalculations
ofhistorical
landuse.
beforethecolonization by European butoncestarted, of agricultural

settlers, area.In absolute terms,thetopthreeregions
theexpansion of agricultural
landreallytookoff after1850. with the largest increasein cropland areaare the former
The areaof cropland in bothcountries morethan USSR(196Mha),followed
increased by theUnitedStates (184Mha)
70-fold in 300 years.Other regionswith a considerable andSouthAsia(162Mha).However,it should be notedthat
increase in croplandareaareOceania(25-fold,mainlydueto fortheperiod1950-1990 thesituation isquitedifferent.USA,
Australia),CentralAmerica(15-fold), and Southeast Asia OECD Europe, eastern
Europe, eastern Asia, and Japan show
(l 1-fold). Someregions(Canada,United States,former decreasing cropland areas,albeitsmallamounts. Duringthis
USSR,Oceania)showa "peak"in theirexpansion duringthe period,SouthAmericahas knownthe largestabsolute
19th century(see Table 6), while mostof the developing increase(50 Mha), followedby SouthEastAsia (37 Mha),
regionsexperienced theirlargestgrowthduringthe second Oceania(25 Mha), andWesternAfrica(25 Mha).
half of the 20th century. See Plate 2 for a spatial Accordingto HYDE, the clearingratesremainedrather
representationof theresultsfor 1700,1800,1900,and1990. constant for a longperiodof timeresulting in a lossof global
Theregions withthelowestincrease areOECDEuropeand forestarea of 15% in 1850, but owingto continuousgrowth
eastern Europe,buttheystill doubledtheirareaof cropland. of the worldpopulationand colonization of the continents,
Increaseof production in suchregionshasbeenapparently this number has increasedto 29% in 1990 (Table 7a and 7b).
realizedby improvedmanagement, bettertechnology,and Ramankutty and Foley [1999] estimatedsomewhatlower
intensiveuse of artificial fertilizersratherthan by expansion forestlossrates,but the trendis similar(5% in 1700, 10% in
425
USE CHANGE
LAND
HISTORICAL
OF GLOBAL
300 YEARS
GOLDEWIJK:
KLEIN
II
KLEIN GOLDEWIJK: 300 YEARS OF GLOBAL HISTORICAL LAND USE CHANGE
428
429
C
KLEIN GOLDEWIJK: 300 YEARS OF GLOBAL HISTORICAL LAND USE CHANGE
r
Table 7a. Comparison

of LandUseChanges
to OtherEstimates
in AbsoluteTerms
Forest/ Steppe/ Shrubland Tundra/ Cropland Pasture Total

Reference Year Woodland Savanna/ Hot Desert/
Grassland Ice Desert
Matthews[1983]
Pre-agricultural 61.5 33.9 13.0 23.1 0.9 132.4
1980 52.4 27.4 12.1 22.9 17.6 132.4
Richards
[1990],aftertheoriginal
workofHoughton
etal.[1983]b
1700 62.2 68.6 - - 2.7 133.4
1850 59.7 68.4 - - 5.4 133.4
1980 50.5 67.9 - - 15.0 133.4
Williams[! 990]
186O .... 5.7
1978 .... 14.2
KleinGoldewijk
andBattjes[1997],afterRichards
[1990]
1700 62.2 63.3 - - 2.7 5.3 133.4
1850 59.7 60.6 - - 5.4 7.8 133.4
1980 50.5 34.5 - - 15.0 33.4 133.4
Ramankutty
andFoley[!999]
Undisturbed 55.3 33.4 17.9 23.6 0.0 - 130.1
1700 52.8 32.3 17.4 23.5 4.0 - 130.1
1850 49.9 31.4 17.1 23.5 8.2 - 130.1
1992 43.9 26.7 15.9 23.3 20.3 - 130.1
ThisStudy
Undisturbed 58.6 34.3 9.8 31.4 0.0 0.0 134.1
1700 54.4 32.1 8.7 31.1 2.7 5.2 134.1
1850 50.0 28.7 6.8 30.4 5.4 12.8 134.1
1990 41.5 17.5 2.5 26.9 14.7 31.0 134.1
aEstimates
aregiven
inmillion
km2.Note
thattheamount
ofpasture
after
1970
does
notcorrespond
completely
withthestatistical
input.Thisisduetotheallocation
scheme.
Because theinitiallandcovermapwascalibrated
fortheIMAGEtoward1970theamount of
pastureafter1970exceedsforsomecountries
theareainwhichit canbeallocated
resulting
in someareas
a fraction
ofpasture
whichcan
notbeallocated
(10%ofworldtotalin 1990mostly
intheUnited
States,
Eastern
Africa,MiddleEastandSoutheast
Asia).
bPasture
ishere
included
intheSavanna/Grassland
categorie.
Table 7a. Comparisonof Land Use ChangeResultsto OtherEstimatesin AbsoluteTerms.
1850, and 21% in 1990). Natural forestswere clearedat a (45-fold increasefrom 0.014 Mha in 1700 to 0.647 Mha in
massscalein returnfor woodand arableland (e.g., deltaic 1990), Canada(32-fold from 0.9 to 28 Mha) andOceania(15-
rain forests of Burma, the mixed deciduous-coniferous
of fold from 29 Mha to 430 Mha in 1990). Again, the increaseof
European USSR, the "mallee scrub" forests of South pasturearea in absolutetermsresultsin a differentranking.
Australia)or replacedby plantations(Atlanticcoastalforests The top three regionsare East Asia with a total increaseof
in Brazil).Also,numerous wetlandshavebeendrained(the 442 Mha over the period 1700 - 1990, followedby Oceania
Midwest of the United States, USSR, Finland, Southeast (401 Mha), andSouthAmerica(398 Mha).
Asia), and (semi)arid areas have been brought under A numberof uncertaintiescanbe identifiedto the approach
ambitious
irrigation
schemes
andbecame
suitable
forgrowing chosen here to estimate historical land cover. First it is
crops,for exampleMidwest United States,India, Mexico, questionablewhether population density is an acceptable
SouthAfrica,andAustralia[Richards,1986]. proxy for the allocationof croplandand pasture.It seems
The expansion of pasturein the USA regionhasbeenas reasonableto assumethat where people are living, there
vigorousasthe increasein cropland.From2.5 Mha in 1700 it alwayshasbeen somesortof agriculturalactivity,especially
has increased almost 100-fold to 239 Mha in 1990. Other in the past. With the current availability of historical
regionswhich experienced
a substantial
increaseare Japan populationnumbers,on the national scale as well as on the
KLEINGOLDEWIJK:
300YEARSOFGLOBAL
HISTORICAL
LANDUSECHANGE 431
Table7b.Comparison
ofLandUseChanges
toOtherEstimates
inPercentages
Forest/ Steppe/ Shrubland,

% Tundra/ Cropland,
% Pasture,
% Total, %
Period Woodland,% Savanna/ Hot Desert/
Grassland,% Ice Desert, %
Matthews[! 983]
Preb- 1980 - 14.9 33.9 13.0 23.1 0.9
Richards
[1990], aftertheoriginalworkofHoughton
et al. [1983]
1700 - 1850 -4.0 -0.3 - - 102.6
1850 - 1980 -15.3 -0.7 - - 179.5
1700 - 1980 -18.7 -1.0 - - 466.4
Williams[i 990]
1860 - 1978 - - 148.9
KleinGoldewijkandBattjes[1990], afterRichards[1990]
1700 - 1850 -4.0 -4.4 - - 102.6 48.7
1850 - 1990 -15.3 -43.0 - - 179.5 326.5
1700 - 1990 -18.7 -45.5 - - 466.4 534.0
Ramankuttyand Foley [1999]

Und.c- 1700 -4.5 -3.1 -2.5 -0.2
Und. - 1850 -9.7 -6.0 -4.2 -0.4
Und.- 1992 -20.5 -20.1 -10.9 -1.1
1700- 1850 -5.4 -3.0 - 1.7 -0.2 103.0
1850- 1992 -12.0 -15.0 -7.0 -0.7 147.0
1700 - 1992 - 16.8 - 17.5 -8.6 -0.9 401.2
This Study
Und.- 1700 -7.2 -6.4 - 11.3 - 1.1
Und.- 1850 -14.8 -16.1 -30.8 -3.1
Und.- 1990 -29.2 -48.8 -74.4 -14.4
1700- 1850 -8.2 -10.3 -22.0 -2.0 100.7 145.3
1850 - 1990 -16.9 -39.0 -63.1 -11.6 171.6 141.9
1700 - 1990 -23.7 -45.3 -71.2 -13.4 444.5 493.5
Estimatesare givenin %.
Pre- Pre-agricultural.
Und. = UndisturbedPotentialVegetation.
Table7b. Comparison
of LandUseChangeResults
to OtherEstimates
in Percentages.
subnational
scale,this approachhasthe advantage
that some independency
of countries)
area majorobstacle
to provide
migrationpatternscanbe reproduced
(e.g.,with statelevel consistent
timeseries(seee.g.,Ramankutty andFoley[1999]
present for theUnitedStates).Illustrative
informationUnitedStates),in contrastto downscaling to thatis the factthateven
agriculturalareas[e.g.,Ramankutty andFoley,1999].Second todaymanynational census estimatesdifferquiteoftenwith,
the allocationof pastureto land covertypessuchas savanna for example,theFAOestimates. Solving theproblem of lack
andgrassland/steppe canalsoresultin unforeseen of nationaldataby downscaling
difficulties. a groupof countrytotalsin a
For examplesomegrass-/shrubland areasin southernAfrica regionto matcha regionalnumberseemshardlyacceptable
areknownto be rampantby plantsthatarenot digestiblefor but is oftenthe only way of gettinga first-orderestimate.
cattleand thereforevery unlikelyto be usedas pasture(R. Fourthusinginitial landuse/cover mapsfromthe DISCover
Leemans,personalcommunication, 2000). Third the scarcity data set, calibrated with the FAO statistics,can be
of data for historicalland use estimates,in particularfor the questionableaswell.Satellite-based datacanbeverydifficult
19thand 18th century,is a seriousproblem.In addition,the to use for some regions.For example,accordingto the
different definitions and classifications of land use classes analysisof Veldkamp et al. [1998],the DISCoverdataset
andthe changingcountryboundaries (e.g.,asresultof wars, classifies
largepartsof Europeas"cropland/natural mosaic".
This makesit very difficultto usesuchpromisingdatasetsfor Haub, C., How manypeoplehave ever lived on Earth?,Population
studies like this one. Today,February,5, 1995.
Haxeltine,A., andC.I. Prentice,An equilibiriumterrestrialbiosphere
model based on ecophysiological constraints, resource
5. Conclusions availability,andcompetition amongplantfunctional types,Global
Biogeochem.Cycles,! 0(4), 693-709, 1996.
A goodknowledgeof historicland usetrendsis important
Houghton,R.A., The annualnet flux of carbonto the atmosphere
to understand current and future developments. Such from changesin land use 1850-1990, Tellus,Ser.51B,298-313,
knowledge can only emerge when historic databaseswith 1999.
adequatecoverageand resolutionare developedand being Houghton,R.A., and J.L. Hackler, Continentalscaleestimatesof the
distributedfreelyto the scientificcommunity.This attemptto biotic carbon flux from land cover change: 1850- 1980, Num.
create a global georeferenceddatabasehas its obvious Datapackage-050,CarbonDioxideInfor. Anal. Cent.,Oak Ridge
limitations and inconsistencies, but the structure of the Natl. Lab., Oak Ridge,Tenn., 1995.
databaseis clear, and the data are easily accessible,while Houghton,R.A., J.E. Hobbie,J.M. Melillo, B. Moore, B.J. Peterson,
providing an independentestimatefor historical land use G.R. Shaver,and G.M. Woodwell, Changesin the carboncontent
of terrestrial biota and soils between 1860 and 1980: A net release
conversions.The scientific community involved in global
of CO2 to the atmosphere.Ecol. Monogr.,53(3), 235-262, 1983.
environmentalchangemodelingcouldclearlybenefitfrom the
translation of more historical statistical information into a
Houghton,R.A., Lefkowitz, D.S. and Skole, D.L., Changesin the
landscapeof Latin America between 1850 and 1985. I.
digital (spatial) format. Therefore a better access of Progressive lossof forests,ForestEcologyand Management,38,
researchersto national and sub-nationalstatistics,especially 143-172, 1991.
on land use/cover,would be of greatimportance.Inclusionof Klein Goldewijk,C.G.M., and J.J. Batties,The IMAGE 2 hundred
more subnational land use data, such as location and size of year (1890-1990) data baseof the global environment(HYDE).
cities,couldstill makefurtherimprovements to this database. Rep. 482523001, National Institute of Public Health and the
Environment (RIVM), Bilthoven,The Netherlands,1995.
Acknowledgements.The author wishes to thank Lex Klein Goldewijk, C.G.M., and J.J. Battjes,A hundredyear (1890 -
Bouwman, Eric Kreileman, Rik Leemans, Bert de Vries, and 1990)databasefor integratedenvironmental assessments
(HYDE,
version1.1). Rep.422514002,NationalInstituteof PublicHealth
Jaap van Woerden of RIVM for helping out with the
andtheEnvironment(RIVM), Bilthoven,The Netherlands,1997.
computations and usefulcommentsaboutthe manuscript,and
Klein Goldewijk,K., The role of historicalGIS datain integrated
Navin Ramankutty from the Climate, People, and models of global change,in Proceedingsof the Third Joint
Environment Program (CPEP), Institute of Environmental European Conference and Exhibition on Geographical
Studies,University of Wisconsin,Madison, for generously Information,pp. 317-328, IOS Press,Vienna,1997.
providing historical state level information for the USA. Leemans,R., and W. Cramer, The IIASA databasefor meanmonthly
valuesof temperature,precipitationand cloudinesson a global
References terrestrialgrid, Rep.RR-91-18,InternationalInstitutefor Applied
SystemsAnalysis(IIASA), Laxenburg,Austria,1991.
Alcamo, J., G.J.J.Kr½ilcman,and M. Krol, An overviewof IMAGE Leemans, R., and G.J. van den Born, Determining the potential
2.1. Brochure of the National Institute of Public Health and the global distributionof naturalvegetation,cropsand agricultural
Environment(RIVM), Bilthovcn, The Netherlands,1995. productivity,l,Vat.Air Soil Pollut., 76(1-2), 133-162, 1994.
Alcamo,J., R. Lecmans,and E. Kr½ileman,(Eds.), Global Change Loveland,T.R., and A.S. Belward, The IGBP-DIS global lkm land
Scenariosof the 21th Century,Pergamon,Tarrytown,N.Y., 1998. coverdataset,DISCover:first results.Int. J. of RemoteSens.,18,
Bogue,D.J., The Populationof the UnitedStates,Historical Trends 3289-3295, 1997.
and Future Projections,The Free Press,New York, 1985. Matthews,E., Global vegetationand land use:New high-resolution
Dcmeny, P., Population,in The Earth as Transformedby Human databasesfor climatestudies,d. Clim., Appl. Meteorol.,22, 474-
Action, edited by B.L. Turner II ½t al., pp. 41-54, Cambridge 487, 1983.
Univ. Press,New York, 1990. Maddison,A., Monitoring the l,Vorld Economy,Organisationfor
Durand, J.D., Historical Estimates of Worm Population.' An EconomicCo-operation andDevelopment (OECD), Paris.1994.
Evaluation, Analytical and Tech. Rcp. No. 10, Table 2, Univ. of McEvedy,C. andR. Jones,Atlas of WormPopulationHistory,pp.
Penn.,PopulationCenter,1974. 368, Penguin,London,1978.
Durand, J.D., Historical estimatesof world population.Pop. and Merrick, T.W., World populationin transition,PopulationBulletin,
Der. Rev., 3(3), 253-296, 1977. 41(2), 53, 1986.
Esser,G., and D. Overdi½ck(Eds.), Osnabrt•ckBiosphereModel.' Mitchell, B.R., International Historical Statistics, The Americas:
Structure, Construction,Results. Modern Ecology, Basic and 1750 - 1988,pp. 817, MacMillan, Indianapolis, Ind., 1993.
AppliedAspectsSer., ElsevierSci., New York, 1991. Mitchell, B.R., International historical statistics Europe.' 1750 -
Food and AgriculturalOrganizationof the United Nations (FAO), 1993,pp. 959, FourthEd., MacMillan, Indianapolis, Ind., 1998a.
FAOSTAT PC software, Rome, 1996. Mitchell, B.R., International Historical Statistics,Africa, Asia &
Grigg, D.B., The Agricultural Systems of The WorM.' An Oceania.' 1750 - 1993, pp. 1113, Third Ed., MacMillan,
Evolutionary Approach, Cambridge Univ. Press, New York, Indianapolis,Ind., 1998b.
1974. OxfordUniversity,OxfordEconomicAtlasof the WorM,New York.,
Grigg,D., The industrialrevolutionandland transformation,
in Land 1972.
Tranformation in Agriculture, edited by M.G. Wolman and Prentice, I.C., W. Cramer, S.P. Harrison, R. Leemans, R.A.
F.G.A. Fournicr,pp. 79-109, JohnWiley, New York, 1987. Monserud,A.M. and Solomon,A global biome modelbasedon
Hall, C.A.S., Modelingspatialandtemporalpatternsof tropicalland plantphysiologyand dominance,
soil propertiesandclimate,J.
usechange.•. Biogeogr.22, 2547-2551, 1995. Biogeogr.,19, 117-134, 1992.
KLE1N GOLDEWIJK: 300 YEARS OF GLOBAL HISTORICAL LAND USE CHANGE 433
Ramankutty,
N. and J.A. Foley,Estimating
historicalchanges
in Tuve, G.L., Energy, environment,population,and food, pp. 264,
global land cover: Croplandsfrom 1700 to 1992. Global Wiley-Interscience,New York, 1976.
Biogeochem.Cycles,13(4), 997-1027, 1999. United Nations, Long-rangeworld populationprojections:Two
Richards, J.F., World environmentalhistory and economic centuriesof populationgrowth,1950-2150,rep.New York. 1992.
development, in Sustainable
Development
of The Biosphere, UnitedNations,World PopulationProspects,1950- 2050, The 1994
editedby W.C. Clark and R.E. Munn,pp. 53-73, Cambridge Revision,Rep. ST/ESA/SER.A/145,UN PopulationDivision, New
Univ. Press, New York, 1986. York. 1995.
Richards,
J.F.,Landtransformation,
in TheEarthasTransformed
by UnitedNations,World PopulationProspects,1950- 2050, The 1996
HumanAction, editedby B.L. TurnerII et al., pp. 163-178, Revision.rep. UN PopulationDivision,New York. 1997.
CambridgeUniv. Press,New York, 1990. Veldkamp, J.G., C. Heunks, V.F. van Katwijk, M. Barkhof, C.A.
Richards,J.F., andE. Flint, historicland useand carbonestimatesfor Muecher,and B.J. ten Brink, Towardsan improvedmethodology
South and Southeast Asia 1880-1980, Data set NDP-046 for pan-Europeanland cover classification,Report NRSP-2 no.
CDIAC/ORNL,CarbonDioxideInfor. Anal. Cent.,Oak Ridge 98-06, National Institute of Public Health and the Environment
Natl. Lab., Oak Ridge,Tenn., 1994. (RIVM), Bilthoven,The Netherlands,1998.
Solomon,A. (Ed.), Reportfrom the IMAGE 2 advisoryboard Watson, R., M.C. Zinyowera, and R.H. Moss, (Eds.), Climate
meetingin Amsterdam,
20-22June1994.NRPReportno. 00-13, Change 1995. Impacts, adaptationsand mitigation of climate
Nat. Res. Prog. on Global Air Pollut. and Clim. Change, change:scientific-technical
analysis,pp. 878, CambridgeUniv.
Bilthoven,The Netherlands,1994. Press,New York, 1996.
Stern,D.I., andR.K. Kaufmann,Estimatesof globalanthropogenic Williams, M., Forests.in The Earth as Transformedby Human
sulphate emissions
1860-1993,Workingpaper series9602, Cent. Action, editedby B.L. Turner II et al., pp. 179-202, Cambridge
for EnergyandEnv. Stud.,BostonUniv., Boston,Mass., 1996. Univ. Press,New York, 1990.
Winkler Prins, De grote Winklet Prins Wereldatlas.Elsevier Sci.,
Thomlinson,J.R., M.I. Serrano,T.D. Lopez, T.M. Aide, and J.K.
New York, 1987.
Zimmerman,Land-usedynamicsin a post-agricultural Puerto
Woytinsky, W.S., and E.S. Woytinsky, World Population and
Ricanlandscape (1936-1988).Biotropica,28(4 PartA), 525-536,
1996.
Production, The Lord Baltimore Press,New York, 1953.
World ResourcesInstitute (WRI), World Resources,1986-1987,
Tobler,W., U. Deichmann,J. Gottsegen,andK. Maloy, The global
Basic Books, New York, 1987.
demographyproject. Tech. Rep. TR-95-6, National Center for
GeographicInformationand Analysis(NCGIA), SantaBarbara,
K. Klein Goldewijk, National Instituteof Public Health and the
California, 1995.
Environment (RIVM), P.O. Box 1, 3720 BA, Bilthoven, The
Turner, B.L.II., W.C. Clark, R.W. Kates, J.F. Richards, J.T. Netherlands.
(kees.klein.
goldewijk•rivm.nl).
Mathews,and W.B. Meyer (Eds), The Earth as Transformed
by
Human Action, pp. 713, CambridgeUniv. Press,New York, (ReceivedSeptember20, 1999; revisedFebruary9, 2000;
1990. acceptedMay 26, 2000)

Global Biogeochemical Cycles - 2001 - Goldewijk - Estimating Global Land Use Change Over The Past 300 Years The HYDE

Uploaded by

Copyright:

Available Formats

You might also like

Global Biogeochemical Cycles - 2001 - Goldewijk - Estimating Global Land Use Change Over The Past 300 Years The HYDE

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Global Biogeochemical Cycles - 2001 - Goldewijk - Estimating Global Land Use Change Over The Past 300 Years The HYDE

Uploaded by

Copyright:

Available Formats

GLOBAL BIOGEOCHEMICAL CYCLES, VOL. 15, No.

2, PAGES 417-433, JUNE 2001

Abstract. Testingagainsthistoricaldatais an importantstepfor validatingintegratedmodels

1. Introduction man-made land cover conversionsover the last 300 years.

economy,andpoliticalfactors.The modelhasbeenappliedto as population,land use, livestock, gross domesticproduct,

Canada,UnitedStates,Mexico,Brazil,Argentina,India,China, the former Soviet Republics Armenia, Azerbijan, Belarus,

landcovermap,whichis considered as beingrepresentative estimateof Matthews[1983] of 1760Mha croplandin 1980is

Figure 1. Agriculturalareasasdefinedby the IMAGE 2 model,basedon theDISCoverdatabase

Step 1' populationdensity

Step 2: land use

Natural undisturbed land cover Currentagriculturalarea based on

Current land use map

Set of rules for allocationcroplandand pasture:

Historical land use maps

beforethecolonization by European butoncestarted, of agricultural

Table 7a. Comparison

Forest/ Steppe/ Shrubland Tundra/ Cropland Pasture Total

Table 7a. Comparisonof Land Use ChangeResultsto OtherEstimatesin AbsoluteTerms.

Forest/ Steppe/ Shrubland,

1850 - 1990 -15.3 -43.0 - - 179.5 326.5

1700 - 1990 -18.7 -45.5 - - 466.4 534.0

Ramankuttyand Foley [1999]

1850- 1992 -12.0 -15.0 -7.0 -0.7 147.0

1700 - 1992 - 16.8 - 17.5 -8.6 -0.9 401.2

1700 - 1990 -23.7 -45.3 -71.2 -13.4 444.5 493.5

You might also like