Simulating the vegetation-producing process in small watersheds in the Loess Plateau of China

Jourmal ofJournal of Arid LandArid L and2012, 4(3): 300 -309"Pdoi: 10.3724/SP.J.1227.2012.00300Science Pressjal.xjegi.com; www.chinasciencejournal.comSimulating the vegetation-producing process in smallwatersheds in the Loess Plateau of ChinaKaiBo WANG', ZhouPing SHANGGUAN2*'State Key Laboratory of Loess and Quatemary Geology, Institute of Earth Environment, Chinese Academy of Sciences, Xian710075, China;“ Northwest Agriculture and Forestry University, Yangling 712100, ChinaAbstract: Small watersheds are the basic composition unit of the Loess Plateau in China. An accurate estimationof vegetation net primary productivity (NPP) is of great significance for eco-beneft evaluation in small watershedmanagement in this region. Here we describe the development and testing of a vegetation producing processmodel (VPP) of a small watershed in the Loess Plateau. The model couples three modules: radiation adjustment;soil hydrological processes; and vegetation carbon assimilation. Model validation indicates that the VPP model canbe used to estimate the NPP of small watersheds in the region. With the VPP model, we estimated the spatial NPPdistributions in the Yangou watershed for 2007. The results show that in the Yangou watershed the NPP is relativelylow, averaging 168 9 C/(m*.a). Trees and shrubs have a higher NPP than crops and grasses. The NPP is larger onthe partly shaded and shaded slopes than on the partly sunny and sunny slopes. The NPP on the slopes increasesgradually on 0- -20° slopes and decreases slightly on slopes steeper than 20°. Our simulation indicates that thevegetation type is the most important factor in determining the NPP distribution in small watersheds in the LoessPlateau.Keywords: small watershed; net primary productivit; vegetation-producing process model; solar radiation; soil hydrologicalprocess; vegetation carbon assimilationEcosystem degradation due to soil erosion is one oftors, and reflects the yielding and adaptive capacity ofthe most serious environmental problems facing hu-plant communities. NPP is closely linked to the carbonman beings (Pimentel, 2006; Zuazo and Pleguezuelo,cycle, climate and land use changes and ecosystem2008). Soil erosion is also the key issue that constrainsrecovery assessment, and characterizes the regionalecological restoration and sustainable agriculture iproducing and carrying capacity (Nemani et al., 2003;the Loess Plateau region of China (Liu, 1999). Estab-Xia and Shao, 2008; Xu et al, 2009; Armeth et al,lishment of vegetation has been shown to be the most2010). It follows that research projects on NPP arepositive and effective measure in controlling soil ero-important foundations for ecological restoration and asion (Shangguan et al, 2004; Zheng, 2006; Zuazo andrange of other environmental issues.Pleguezuelo, 2008). Therefore, afforestation and in-NPP is the balance between carbon gained by grosscrease in vegetation coverage have become the mostprimary productivity (GPP) and carbon lost from theimportant considerations in soil conservation andrespiration of all plant parts. It is determined not onlyecosystem restoration in the Loess Plateau region ofby the biological characteristics but also the surround-China.ing environmental conditions of plants, including solarThe net primary productivity (NPP) of vegetation isenergy, tenl中国煤化工d moisture. In aa measure of the interaction between vegetational bio-Received 2011-10YHCNMHGlogical characteristics and external environmental fac-"Corresponding author: ZhouPing SHANGGUAN (E-mail: shangguan@ms iswc.ac.cn)No.3KaiBo WANG et al: Simulating the vegetation-producing process in small watersheds in the Loess Plateau of China301small area of a few hectares, NPP can be determined1 Model developmentby field measurements. Direct NPP measurement1.1 Model structureover large areas is difficult, however, and modelsimulation is suggcstcd as an effective method (Pot-The VPP model comprises three modules: topographicter et al, 1993; Cramer et al, 1999; Li et al, 2004;correction of solar radiation, simulation of soil hydro-Piao et al, 2008). Since the implementation of thelogical process, and vegetation carbon assimilationInternational Biological Program (IBP) in the 1960s,(Fig.1).a large number of NPP models have been developedfor estimating regional and global NPP, such aCASA (Carnegie-Ames-Stanford Approach) (Potterζ slopeyet al, 1993), TEM (Terrestrial Ecosystem Model)(Melillo et al, 1993), BEPS (Boreal Ecosystem Pro-PPFDductivity Simulator) (Liu et al, 1997). These modelshave produced great achievements in simulating regional and global NPP and clarifying the distributionLAIand magnitude of regional and global carbon sourcesand sinks, and in evaluating the response of terres-日「ETTGPPtrial ecosystems to global changes. However, these|器f(向)models cannot accurately simulate NPP in relativelyNPP Rsmall watersheds since they do not consider in detailthe site features of different ecosystems and thus ig-Fig. 1 Structure of the VPP model. The rectangles representprimary process variables; the ovals represent key variablesnore environmental spatial heterogeneity. As a result,influencing or linking the hydrological process and carbon as-they either overestimate or underestimate the NPP in .impacting the slope solar radiation. The solid lines representsmall watersheds, particularly those with complexenergy, water and carbon flow directions and the dashed linesterrains. However, small watersheds are the basicdenote the effects of environmental variables. Qeat solar radiationcomposition unit of the Loess Plateau and also theon flat terrains; Qslope, solar radiation on slope land; Rm。net solarradiation; PPFD, photosynthetically active radiation photon flux; P,unit for ecological management and soil conservation.precipitation; 1, canopy interception; ET, evapotranspiration; Rof,It is therefore of great theoretical and practical sig-surface runoff; 0, mean soil moisture content at depths of 0- -2 m;LAI, leaf area index; f(B), soil- water stress cofficient; Ao, leafnificance for the evaluation of vegetation restorationphotosynthetic rate; Ao canopy photosynthesis; GPP, gross pri-and watershed ecosystem management in the Loessmary productity; NPP, net primary productivity, Rg, total respira-tion.Plateau.An accurate NPP estimation is significant forWe constructed a solar radiation distribution modelcorrectly summarizing the outcome and lessons itbased on the digital elevation model (DEM) of thewatershed management and providing reference andstudy area in line with the complex terrain fcatures ofhilly loess regions in China. Using flatland solar radia-guidance for subsequent watershed managementtion as a base, we integrated the effect of slope andsteps in the Loess Plateau. In addition, it will pro-aspect to simulate solar radiation for the hilly areas.vide a useful tool for predicting NPP responses inFor the hydrological and carbon assimilation processthe Loess Plateau under future climate and land usemodules, leaf area index (LAI) and soil-water stressconditions. The major objective of this study is tocoefficient (f(0)) are the two key variables that connectdevclop and tcst a vcgetation-producing processsoil-water balance and vegetation primary production.model (VPP) that includes the effects of terrain andThe hydrological processes in the Loess Plateauis appropriate for small watersheds in the Loessmainly c中国煤化工rception, evapo-transpiratiol:YHCNMHGariations in soil302JOURNAL OF ARID LANDVol. 4water storage. Changes of variables in the water bal-thesis and drives the ecosystem carbon cycle. In flatance equation can induce variation in the soil waterterrains, the solar radiation measured at one pointcontent, thereby affecting the physiological and eco-can be extrapolated over a large area virtually un-logical processes in plant and altering the primarychanged. In complex terrain, however, the spatialproduction of vegetaion. Carbon assimilation at leaf variability in elevation, slope and aspect can createlevel in the VPP model is simulated using the hyper-very strong local gradients in solar radiation thatbolic equation developed by Prioul and Chartiercan induce large-scale variation in vegetation pro-(1977), scaled-up to the canopy level by the big-leafductivity. In the VPP model, we considered the ef-model (Sellers et al, 1992) for the relatively less andfect of slope and aspect on solar radiation but notcasily obtained variables. Finally, the NPP is com-the effect of altitude, since the study area possessesputed as the difference between total plant respirationcomplex terrains but relatively small differences inand gross primary production.elevation.1.2 Model descriptionSolar radiation in complex terrains is termed *slopesolar radiation' (Qslope) as opposed to solar radiation inThe model is driven by meteorological, soil, biologi-flat terrains (Q}a) derived from meteorological data.cal and spatial raster data. The detailed input data areQslope calculation was based on the hypothesis that thegiven as follows: (1) Meteorological data (dailyratio of solar radiation to extraterrestrial radiation inmaximum, minimum and average temperatures, pre-flat terrains is equal to the ratio of the slope solar ra-cipitation, relative humidity, solar radiation, sunshinediation to the slope extraterrestrial radiation over thehours, and average wind speed); (2) Spatial rasterslope (Zeng et al, 2005):data (maps of vegetation type, slope, and aspect); (3)RftatVegetation physiological and ecological parametersLslope = Qflat x(1)(average plant height, community coverage, plantphenology, plant photosynthetic characteristic pa-Where, Rrat and Rslope are the horizontal extraterrestrialrameters, and initial biomass of each plant compo-radiation and slope extraterrestrial radiation, respec-nent); (4) Soil property parameters (soil bulk density,tively.soil field capacity, wilting water content, and initialThe values of Rgat and Rslope were calculated by Eqs.2 and 3 (Allen et al, 1998; Fu, 1998), and 031 and 0s2soil moisture content).The model runs at a daily time step with the gridin Eq.3 were calculated after Fu (1998):24x 60size of 20 mx20 m. Simulation results can be ex-G;ed,ported for time steps of per day, per month and perx(a, sin φsin 8 + cos φcosδ cos os ),year. Output includes slope solar radiation (Qslope),net solar radiation (Rn), photosynthetically activeRsopeGx d,usinδx(@%2 - o1)radiation (Qpar), mean soil moisture content at thedepth of 0- 2 m, evapotranspiration (ET), runoff (Rof),+ vcosδx(sin 0).2 -sin@1)(3)canopy interception (I) and leaf area index (LA), and- wcosδx(cos 0,2 - COs@,1 ).the GPP, NPP, maintenance and growth respirationWhere, Gsc is the solar constant (MJ(m: min)); d, is the(Rm, Rg) of the vegetation.inverse relative Earth-Sun distance (dimensionless); φThe detailed calculation method and process of theis the latitude of the study area (rad); δ is the solarmodel are described below.decimation (rad); 0s1 and CO32 are the sunrise hour an-1.2.1 Solar radiationgle (rad) and the sunset hour angle (rad) of slope land,Solar radiation is the most important variable irrespectively; and u, v and w are the intermediate vari-NPP simulation as it controls the energy balance ofables which were calowlatard gftar R4(1998).the watershed, thus deeply influencing the soil waterAfter cal中国煤化工ation (R,) on thebalance. It is the energy source of plant photosyn-slope surfacYHCNMHGs.4 6, and theNo.3KaiBo WANG et al: Simulating the vegetation-producing proccss in small watersheds in the Loess Plateau of China303photosynthetically active radiation (Qpar) and photo-(Li, 2001). As the amount of precipitation is far ftomsynthetically active radiation photon flux (PPFD)meeting the demand of evapotranspiration in the re-were computed by Eqs. 7 and 8 (Zhou et al, 1984):gion, we have introduced a soil-water stress coffi-R,=Rs-Rnl,cient (0)) into the calculation of the evapotranspira-Rs =Qaiopex(1-C,),(5)tion for the watershed. f(0) is shown as Eq. 15 (Xiaand Shao, 2008).Tmax + TminRm=σx(0.34- 0.14vea )1=里Ps,(10)(6)1.35一- 0.35S, = Imx[1-exp( -0.5LA1)],(11)(P-0.2S)x(P-0.2S)Lpar = Qsope (0.384 + 0.053logea),(7)P+0.8SP20.2S，(12)Rof=0P<0.2SPPFD =4.57x106 Qpe(8)3600xns= 25400_ 254,(13)Where, Rns and Rnl are the net shortwave radiationCN(MJ/(m-day)) and net long wave radiationET=K.xf(0)x ET,(14)(MJ/(m2.day)); C, is the Albedo (dimensionless); Tmaxand Tmin are the maximum absolute temperature (K)θ≥θ°and minimum absolute temperature (K), respectively; .σ is the Stefan- Boltzmann constant (MJ/K*/(m'.day));f(8)={θ-0,日,≤θ≤θ*,(15)8,-0,ea is the actual vapor pressure (kPa); n is the actualθ≤θ,daylight hours; and N is the theoretical daylight hours.1.2.2 Soil-water balance，_900In the arid and semi-arid Loess Plateau region of0.4080(R, -G)+r-+27342(e.-ea). (16)China, precipitation is the main source of soil moisture.O+r(1+ 0.34u2 )Also, in most areas the water table is deep and the un-Where, S, is the water storage capacity of vegetationsaturated soil layer may be up to tens of meters thick,canopy (mm); Imax is the maximum interception ofso that almost no deep leakage occurs (Yang and Shao,vegetation (mm); S is the potential maximum retention2000). Therefore， according to the principle ofafter the beginning of runoff (mm); CN is the curvesoil-water balance, the variation in soil water contentnumber (dimensionless); K。is the crop coefficientwas calculated as follows:(dimensionless); f and 0, are respectively the fieldsθ=P-I-ET-Rr.(9capacity (mm) and wilting water content (mm); θ isThe variables for the soil-water balance equationthe critical water content when plant suffered soil wa-were calculated from Eqs. 10-16 in which the precipi-ter stress (mm); G is the soil heat flux densitytation (P) was obtained from meteorological data, the(MJ/(m^ *day)); T is the daily average air temperaturecanopy interception (I) was calculated according to Xu(°C); u2 is the wind speed at 2-m height (m/s); es is theet al. (2005), the modified SCS-CN model was used tosaturation vapor pressure (kPa); s is the slope vapourcalculate the surface runoff (Rof) (Liu et al, 2005),pressure curve (kPa/°C); and》is the psychrometricand the FaAo Penman-Monteith equation was used toconstant (kPa/°C).caleulate the evapotranspiration (ET) (Allen et al,1.2.3 Carbon cycling1998).(1) Leaf area index (LAI)In most areas of the Loess Plateau the potentialLAI is another kev variable in the VPP model in that itevaporation is about 800- -1,000 mm per year, whilehas a signif中国煤化工ydrological proc-the average yearly precipitation is about 400- -600 mmess and carlYHCNMHGd.Itchangesthe304 .JOURNAL OF ARID LANDVol. 4precipitation distribution by canopy interception, im-byperbolic equation; Rd is the dark respiration of thepacts the quantum and allocation ratios of evaporationplant (umol/(m*s)); f(D) is the temperature stressand transpiration, and influences the light interception,function (dimensionless); f(0) was computed by Eq.thus deeply affecting the simulated carbon assimila-15.tion and primary productivity of vegetation.(3) Canopy photosynthesisWe assumed that LAI was a function of plant grow-The VPP model is capable of simulating daily NPPth stages in the VPP model. It was computed as thevariations; but since our main aim was to simulate theproduct of the Q functions (Qt)) and the feature leafannual average NPP, we adopted the big-leaf model ofarea index (maximum LAI, LAlmax) for each vegetationcanopy photosynthesis simulation in order to simplifytypc; the definition and detailed calculation process ofhe overall model (Sellers et al, 1992). The big-leafQ functions were found in Gao (2003). The LA/maxmodel assumes that canopy carbon fluxes have thewas calculated using an empirical regression equationsame relative responses to the environment as an indi-based on the measured data in the hilly Loess Plateauvidual leaf, and that scaling from leaf to canopy isregion (Huang, 2003):therefore linear. The canopy photosynthesis (Ac) wasL.AI = LAIm xQr(u).(17)calculated by:(2) Leaf photosynthesis. A.= [u A(L)dL=[ Axe-KIdL =1-e-KXLAIK--A. (20)We used the hyperbolic equation of light responsecurve developed by Prioul and Chartier (1977) toThe daily gross primary productivity (GPP) wascompute the instantaneous leaf photosynthesis (Eqcalculated by Eq. (21):18). The hyperbolic equation has some physicalGPP= A. xnx12.01x10-*.(21)meaning and it is much simpler than Faquhar 's modelWhere, n is the sunshine hours; 12.01 is the molar(Farquhar et al, 1980). The photosynthetic charac-mass of carbon atoms (g/mol), and 10 ”is the unitteristic parameter used in the hypcrbolic cquation canconversion factor from Ac (umol/(m-.s)) to GPP (g C/be easily obtained by measuring the photosynthetic(m2.day)). .light response curve. The only variable needed to runVegetation uses the GPP partly to maintain itsthe model is the photosynthetically active radiationgrowth needs, the remainder being consumed byphoton flux (PPFD). Leaf photosynthetic rate is af-growth and maintenance respiration. The NPP of ve-fected not only by air temperature and soil moisture,getation was estimated by subtracting the total respira-but also by the plant growth stage. As the photosyn-tion (R。= Rm+Rg) from the GPP (Eq. 22):thetic characteristic parameter in Eq. 18 was meas-NPP=GPP-Rm-Rg,(22)ured in July (the period of most vigorous plantRm=2 RS2oIKT-2)xM，(i=1,s,r)， (23)growth in the Loess Plateau), in addition to the tem-perature and soil moisture stress function, we intro-Rg =rxGPP.(24)duced a scasonal correction factor for photosynthesis,Where, Rm is maintenance respiration (g C/(m' -day));ka, defined as the ratio of the mean leaf photosyn-Rg is growth respiration (g C/(m' .day)); R”is thethetic rate to the maximum leaf photosynthetic ratemaintenance respiration coefficient; Q1o is the tem-during the relevant period, to co-determine the actualperature sensitivity factor; Te is canopy temperatureleaf photosynthetic rate (Eq. 19):(°C); M; is the initial biomass of the plant (g/m2 ); and rA=(中. PPFD+ Amax)/2k-Rg-is the growth respiration coefficient (dimensionless).N(φ PPFD+ Amx) -4φPPFD.k.Amax /2k, (18)3 Model validation and applicationA=k,xAgx f(T)x f(O).(19)Where, Ao is leaf photosynthetic rate; φ is the appar-The VPP model was. developed and validated in theent quantum yield; Amax is the leaf saturated photo-Yangou wal中国煤化Iral Loess Plateausynthetic rate (umol/(m'-s)); k is the curvature of theof China (1| YHC N M H G36°32N; Fig. 2);o.3KaiBo WANG et al: Simulating the vegetation-producing process in small watersheds in the Loess Plateau of China305the region is 8.6 km in length, and covers an area ofmm and an average temperature of 9.8°C. The soil isabout 47 km*. The landform is typical hilly loess, at ancalcareous silt loam (calcic Cambisol) belonging toelevation of 980- 1 ,400 m. It has a semi-arid continen-the Entisol order (USDA texture classification system),tal climate, with an annual mean precipitation of 560derived from loess to depths generally beyond 50 m.100°E105°E10°E115°E40°NYellow RiverA30°N110°E01,200 2,400 mFig.2 Location of Yangou watershed in the Loess Plateau of ChinaThe natural biome is the northern extended area of watershed using Arcgis 9.3 (ESRI, Inc, Redlands,deciduous broadleaf forest and the zonal climax vege-USA). As more than 90% of the watershed is coveredtation is oak forest dominated by Quercus liaotungen-by loess, we assumed homogeneous soil properties:sis. Robinnia pseudoscacia and Populus spp. are thebulk density is 1.25 g/m', field capacity is 24.3%, andmain plantation tree species that account for 18.6% ofwilting moisture is 6.3% (Yang and Shao, 2000). Soilthe total watershed area (Wang et al, 2008). Sophoramoisture content was measurcd at the 0- 2 m dcpth fordavidi, Caragana and Spiraea schneideriana are thethe different vegetation types in late May, July andmain shrub species. Artemisia sacrorum and ArtemisiaSeptember of 2007. Five widely distributed speciesgiraldi are the main herb species. In addition, cropswere chosen to represent different vegetation types insuch as maize and millet are widely planted on thethe watershed: Robinnia pseudoscacia (tree), Cara-flatlands and terraces; apple trees are mainly distrib-gana microphylla (shrub), Artemisia sacrorum (grass),uted over the hilltops and upper slopes.Malus pumila (orchard) and Zea mays (crop). RelatedThe values of the parameters used in the VPP modelphysiological and ecological parameters of the fivewere obtained either by field measurements or fromvegetation types were measured to drive and validatepublished sources. The climate data were gatheredthe model. The leaf area index was measured with afrom Yan' an meteorological station, about 3 km fromLAI-2000 plant canopy analyzer (LI-COR, Inc, Lin-the study area. Since, as stated above, the area of thecoln, USA) and the above ground biomass was har-watershed is only 47 km*, we were able to ignore thevested to validate the NPP. The initial soil and vegeta-spatial variation in meteorological elements such astion parameters for the study area used in the modelprecipitation and temperaturc. However, the effects ofwcre obtaincd from both published reports and fieldtopography on solar radiation were taken into account,observations (Liu, 2003; Li et al, 2004; He et al.and we established a distributed solar radiation model2005; Piao et al, 2005; Wu et al., 2005; Xu et al,based on the DEM of the study area. The vegeta-2005).tion-type map was digitized based on the land use mapAt prese中国煤化工models posesof the Yangou watershed produced in 2007. The slopemajor diffic:YHCNMHGobalscale,anditand aspect maps were created from the DEM of theis also difficult at the watershed scale. In this study,306JOURNAL OF ARID LANDVol. 4we tested the VPP model through comparing fieldmeasured data indicated that the VPP model was ablemeasurements and the results obtained by a selectionto estimate the NPP in the Loess Plateau.of other relevant models. Figure 3 shows the above-Table 1 NPP comparison between the VPP model and theground NPP for different communities in 2007, to-other modelsgether with previously published data (Liu, 2003; Liet al, 2004; Wu et al, 2005; Xu et al, 2005). Be-Land cover typeVPP VSIM CASA AVIM CEVSA(g C(m'a))cause the NPP measurements of the five land-coverWoodland220221245530560types were carried out on different stand ages and siteShrubland16640120conditions, a large difference is evident between theGrassland10955144275270measured maximum and minimum NPP values andCropland1477570606the simulated mean NPP values. However, the simu-Orchard1990lated and observed mean NPP values were highlyconsistent (Fig. 3), with a relative error of less than4 Results and discussion10%.In Yangou watershed, the NPP was between 0 and 300400■Observed-min.g C/(m^.a), averaging 168 g C/(m^-a) (Fig. 4b). The■Observed-meanNPP was significantly different between the different00 t口Simulated-mean0Observed-max.vegetation types. Woodland had the highest NPP av-200-eraged 220 g C/(m^-a), which was twice that of thegrassland; orchard had the second highest NPP which荟100was about 90% that of the woodland; shrub and crop-land had medium NPP which were respectively aboutWoodland Shrubland Grassland Cropland Orchard76% and 68% that of the woodland (Table 1). Overall,the trees and shrubs had higher NPP values than cropsFig. 3 Observed and simulated NPP for the different land coverand grasses. The NPP distribution map of the Yangoutypes in the study area. The observed NPPs of the woodland,watershed was highly consistent with its land use mapshrubland and orchard land were mainly obtained from Xu et al.(Fig. 4), indicating that vegetation type was the deci-(2005); and the NPPs of the grassland and cropland were ob-sive factor affecting NPP distribution in a small wa-tained based on our field measurement.tershed.Table 1 shows the average annual NPP estimatedAt different aspects and on different slopes, thby the different models. The NPP simulated by ourNPP did not differ greatly in the Yangou watershedmodel was close to that by VSIM (Vegetation- Soil(Figs. 5, 6). The average NPP at different aspectsIntegrated Model; Xu et al, 2005) but less than thoseranged from 154 to 174 g C(m*a). The NPP was lar-of CASA (Carnegie-Ames Stanford Approach; Piaoger on the partly shaded and shaded slopes than on theet al, 2005)， AVIM (Atmosphere-Vegetationpartly sunny and sunny slopes. The NPP on the partlyInteraction Model; He et al, 2005) and CEVSAsunny slopes was almost equal to the average value of(Carbon Exchangc bctwcen Vegetation, Soil, and thehe watershed. The average NPP on the slopes variedAtmosphere; Li et al, 2004) which gave the averagebetween 153 and 176 g C/(m*-a), increasing graduallyNPP for different vegetation types of China, probablyon the 0° 20° slopes, peaking on the 10°- 20° slopes,because these models were originally developed forand decreasing slightly on slopes steeper than 20°.large-scale simulation. The VSIM model was devel-Within a given range, NPP increased with the in-oped for small watersheds and was validated withcrease in solar radiation. However, soil water deficitappropriate parameters for the hilly loess region (Ta-can severely constrain plant growth, thus lowering theble 1); it is therefore likely that its estimated NPPNPP, espe中国煤化工rid regions. Thevalue was more accurate for the Loess Plateau. Thehigher lev|Y片CNMHGsunnyandpartlyvalidation and model comparison on the basis of thesunny slopes resulted in higher evapotranspiration andNo.3KaiBo WANG et al: Simulating the vegetation- producing process in small watersheds in the Loess Plateau of China307NPP (g C/(m*a))Land cover type| 100- .150Grassland.150- 200Shrubland200 -250g WoodlandI 250- -300Orchard.300- 350I FarmlandSettlement and water01,200 2.400m1,200 2,400 m(a) Land use map(b) Net Primary Productivity NPP) dstribution map of 2007Fig. 4 Land use map and NPP map of Yangou watershed in 2007. The land use map was based on interpretation of a TM remotesensing image obtained in the same year. The NPP map was calculated by the VPP model.lower soil water content, causing lowered NPP com-80「pared to shady and semi-shady slopes.合170Although the NPP was greatly influenced by varia-tions in solar radiation, soil moisture and other envi-60 tronmental elements on the scale of the site, the spatialdistribution pattern of vegetation has an important in-气150fluence on NPP distribution in the Yangou watershed.40L-山llThe NPPs of the woodland and shrubland were gener-IMIally much larger than for the cropland and grassland.SlopeThe forests and shrubs, as the two main vegetationFig. 6 NPP distribution on dfferent slopes in Yangou watershedtypes, were distributed more on the partially shadedin 2007.1, 0°- -5°; II, 5*-10*; II, 10*-15*; Iv, 15° -20°; V,18020°- 25*; vI, >25*.slopes than on the sunny slopes, resulting in lower目NPP on the sunny slopes than on the partially shaded可160-slopes (Table 2). The distribution of land cover types150on different slopes showed that vegetation typeschanged with slope variation, which probably caused140山slight differences in NPP for the different slopes. OfSNSSNSSHHAspectthe five land cover types, the cropland had the highestdistribution percentage on the <10° slopes and wood-Fig. 5 NPP distribution at dfferent aspects in Yangou watershedin 2007. The dash-dot line represents the average NPP of theland had thehizhar* dintib.norcentage on thewatershed. SN, sunny slope; SSN, semi-sunny slope; SSH,slopes of中国煤化工d wit the NPPsemi-shady slope; SH, shady slope.distributionYHCNMH G308JOURNAL OF ARID LANDVol. 4Table 2 Distribution of the land cover types at diferent aspects in Yangou watershedLand cover typeAspectWoodlandShrublandGrasslandCroplandOrchardArea (hm2) Percentage (%) Area (hm?) Percentage (%) Area (hm2) Percentage (%) Area (hm2) Percentage (%) Area (hm2) Percentage (%)SN164.43.190.9.3127.92.9135.83.1SSN286.864323.87.3228.1.1244.05.5182.0 .4.0SH400.69.487.911.0191.4303.86.8168.33.8SSH205.84.267.66.0133..0167.093.9ww1 ,038.023.31,243.828.0743.316.7842.719.0579.913.0Note: ww, whole watershed.Table 3 Distribution of the land cover types on diferent slopes in Yangou watershedSlope(9)_0-5Area (hm) Perentage (%) Arca (hm) Percnge (%) Area (hm) Percentage (%) Area (hm) Percentage (%) Area (hm) Prcentage (%)29.50.741.90.930.249.91.22.90.5S-1063.41.477.647.881.5.1.847.610-151000.2.397.457.8105.82.4.5315- 20140.53.2120.52.788.020- -25171.0178.0107.62.13393.099.4.2>25533.312.0719.416.2421.4351.17.9254.0,.71,038.023.4| ,243.82716.18.95 Conclusionssoil water. Simulation results of the VPP modeiagreed with the observed mean values and simu-Taking a small watershed as a unit for soil andlation results of the VISM mode! developed forwater conservation is an important experience ofthe Loess Plateau. The simulation and validationthe ecological environment construction in theresults indicate that the VPP model can be used toLocss Platcau. NPP is a critical parameter inestimate the NPP of small watersbeds in the re-evaluating the restoration levcl of degraded eco-gion. Furthermore, the prediction showed the highsystems and indicating the response of terrestrialconsistency of NPP distribution with the land useecosystem to global change. Therefore, it is ofmap of the Yangou watcrshed, which indicatesgreat significance in assessing the effectiveness ofthat vegetation type is the most important factorregional eco-environmental governance and guid-in determining the NPP distribution in small wa-ing future ecological environment construction.tersheds in the Loess Plateau. However, moreCompared to the soil erosion model at a smallmeasured NPP data and a further validation shallwatershed scale in the Loess Plateau, research ine aquired in other small watersheds of the LoessNPP model in this area is less and should be fur-Plateau.ther strengthened.In this study, we built an NPP model at a smallAcknowledgementswatershed scale in the Loess Plateau. In our model,particular attention was paid to the impact factorsThis research was financed by the Strategic Priority Researchon the NPP in this region, including solar radia-Program (XDA05050403) and the Key Research Program oftion and soil water that affect the process of plantChinese Academy of Sciences (KZZD-EW-04). We thank theanonymous reviewers for their constructive and valuabletranspiration and photosynthesis, and terrain thatcomments and the editors for their responsible and patientaffects the redistribution of solar radiation andwork on the perfection of this paper.References中国煤化工Allen R G Pereira LS, Raes D, et al. 1998. Crop Evapotranspira-Food andAFled Nations.tion- Guidelines for Computing Crop Water Requirerments, Rome:Arneth A, HarMYHC N M H Cesia bgochehNo.3KaiBo WANG et al: Simulating the vegetation-producing process in small watersheds in the Loess Plateau of China309cal feedbacks in the climate system. Nature Geoscience, 3(8);:1999. Science, 300: 1560- -1563.525- -532.Piao Ss L, Fang J Y, Zhou L M, et al.2005. Changes in vegetation netprimary productivity from 1982 to 1999 in China. Global Biogeo-models of terrestrial net primary productivity (NPP): overview andchemical Cycles, 19(2): GB2027.key resuts. Global Changc Biology, 5(Suppl. 1): 1-15.PiaoS L, Ciais P, Friedlingstein P, et al. 2008. Net carbon dioxideFarquhar G D, Von Caemmerer s, Berry J A. 1980. A biochemicallosses of northem ecosystems in response to autumn warming. Na-mode! of photosynthetic CO2 assimilation in leaves of Cs species.ture, 4S1: 49-52. .Planta, 149(1): 7890.Pimentel D.2006. Soil erosion: a food arnvironmental threat. Envi-FuB P. 1998. The dfflerences and variations in components of radiationronment, Development and Sustainability, 8(1): 119-137.budget on underlying surfaces of different topographies. ScientiaPotter C S, Randerson J T, Field C B, et al.1993. Terrestrial ecosystemAtmospherica Sinica2: 178-190.production: a process model based on global satellite and surfaceGao Q, Reynolds J F. 2003. Historical shrub-grass transitions in thedata. Global Biogeochemical Cycles, 7(4): 811 -841.northern Chihuahuan desert: modeling the effects of shifting rainfallPrioul J L, Chartier P. 1977. Partitioning of transfer and carboxylationscasonality and event size over a landscape gradient. Global Changecomponcnts of intracllular resistance to photosynthetic CO2 fixa-Biology, 9(10): 1475- 1493.tion: a critical analysis of the methods used. Annals of Botany,He Y, Dong W J, JiJ1, et al. 2005. The net primary production simula41(4): 789- -800.tion of trrestrial ccosystems in China by AVIM. Advances in EarthShangguan ZP, Shao M A. Li Y s, et al. 2004. Impacts of forest vege-Science, 20(3): 345- -349.tation on the soil water cycle in Loess Plateau. Journal of ChineseHuang Y M.2003. Study on ecological aspects of water balance proc-Soil and Water Conservation, 35: 177- 185.ess at watershed level in hilly regions of the Loess Plateau, China: aWang L, Wang Q, Wei s, et al.2008. Soil desiccation for loess sois oncase study in Zhifanggou watershed. Ph.D. Disseration, Bejing:natural and regrown areas. Forest Ecology and Management, 255(7):Beijing Normal University.2467- -2477.LiK R, Wang s Q, Cao M K. 2004. Vegetation and soil cartbon storageWuF Q, Zhou Z L, Liu H B.2005. Productivity of crop-fruit ecologicalin China. Science in China: Serics D, 47(1): 49 -57.agriculture in middle-south Loess Plateau. Chinese Journal of Ap-Li Y s. 2001. Efects of forest on water circle on the Loess Platcau.plied Ecology, 16(2): 262 -266.Journal of Natural Resources, 16(5): 427-432.Xia Y Q, Shao M A. 2008. Soil water carrying capacity for vegetation:Liz W, Cai Q G, Zeng G M.2004. Regional simulation on land pro-a hydrologic and biogeochemical process model solution. Ecologi-ductivity in the Loess Plateau based on GIS and soil erosion. Re-cal Modelling.214(2- 4):112- -124.Xu H M, Jia H K, Huang Y M.2005. A simulation model of net pri-Liu G B. 1999. Soil conservation and sustainable agniculture on themary production at watershed scale in billy area of Loess Plateau,Loess Plateau: challenges and prospects. Ambio, 28(8): 663- 668.China. Acta Ecologica Sinica, 25(5): 1064-1074.Liu J, Chen J M, Cihlar J, et al.1997. A process-based boreal ecosys-XuX, GaoQ, Liu Y H, et al.2009. Coupling a land use model and anem productivity simulator using remote sensing inputs. Remoteecosystem model for a crop pasture zone. Ecological Modelling,Sensing of Environment, 62(2): 158 175.220(19): 2503- -2511.Liu. J H.2003. Spatial and temporal variation of soil mosture content andYang W z, Shao M A. 2000. Study on Soil Moisture in the Loess Pia-vegetation productivity at a small watershed of the Loess Plateau.teau. Bejing: Science Press.MSc. Dissertation, Yangling: Northwest A & F University, China.Zeng Y, Qiu X, Liu C, et al. 2005. Distributed modeling of direct solarLiuXZ, KangSz, Liu D L, et al. 2005. SCS model based on geo-radiation on rugged trrain of the Yellow River basin. Actagraphic information and its application to simulate raifall-unoffGeographica Sinica, 60(4): 680 -688.relationship at typical small watershed level in Loess Plateau.ZhangN, YuG R, Yu Z L, et al. 2003. Simulation of temporal andTransactions of the Chinese Society of Agricultural Engineering,spatial distribution of natural vegetation light utilization efficiency21(): 93-97.based on 3S. Acta Phytoecologica Sinica, 27(3): 325- 336.Luo T X.1996. Patterns of net primary productivity for Chinese majorZhengF L. 2006. Effect of vegetation changes on soil erosion on theforest types and their mathematical models. Ph.D. Dissertation, Bei-Loess Plateau. Pedosphere, 16(4): 420 427.jing: Chinese Academy of Sciences.Zhou Y H, Xiang Y Q, Shan F Z.1984. A climatological study on theMelillo J M, McGuire A D, Kicklighter D w, et al. 1993. Global cli-photo-synthetically active radiation. Acta Meteorologica Sinica, 42:mate change and trrestril net primary production. Nature, 363387-397.234 240.Zuazo V H D,中国煤化工osion and runofI pre-Nemani R R, Keeling C D, Hashimoto H, e1 al.2003. Climate-drivenvention by0HCNMHG'forSusanableDeincrcases in global terrestrial net primary production from 1982 to