• Ács, F., 1994: A coupled soil–vegetation scheme: Description, parameters, validation, and sensitivity studies. J. Appl. Meteor.,33, 268–284.

  • Ben Mehrez, M., O. Taconet, D. Vidal-madjar, and Y. Sucksdorff, 1992:Calibration of energy flux model over bare soil during HAPEX-MOBILHY experiment. Agric. For. Meteor.,58, 275–283.

  • Braud, I., A. C. Dantas-Antonino, M. Vauclin, J. L. Thony, and P. Ruelle, 1995: A simple soil-plant atmosphere transfer model (SiSPAT) development and field verification. J. Hydrol.,166, 213–250.

  • Budyko, M. I., 1956: Heat Balance of the Earth’s Surface. Gidrometeozidat, Leningrad, 255 pp.

  • Clapp, R. B., and G. M. Hornberger, 1978: Empirical equations for some soil hydraulic properties. Water Resour. Res.,14, 601–604.

  • CNC (Committee for Natural Classification, Chinese Academy of Sciences), 1959: Climate Classification of China. Science Press, 456 pp.

  • Deardorff, J. W., 1977: A parameterization of ground surface moisture content for use in atmospheric prediction model. J. Appl. Meteor.,16, 1182–1185.

  • De Martonne, E., 1925: Traite de Geographie Physique. Paris: Librairie Armand Colin.

  • Dickinson, R. E., 1984: Modeling evapotranspiration for three-dimensional global climate models. Climate Processes and Climate Sensitivity, Geophys. Monogr., No. 29, Amer. Geophys. Union, 58–72.

  • ——, and A. Henderson-Sellers, 1988: Modeling tropical deforestation:A study of GCM land-surface parameterizations. Quart. J. Roy. Meteor. Soc.,114, 439–462.

  • Dorman, J. L., and P. J. Sellers, 1989: A global climatology of albedo, roughness length and stomatal resistance for atmospheric general circulation models as represented by the simple biosphere model (SiB). J. Appl. Meteor.,28, 833–855.

  • GISLP (Group for Integrated Survey of the Loess Plateau, Chinese Academy of Science), 1991: Land Resource of the Loess Plateau. China Science and Technology Press, 355 pp.

  • Grace, J., E. D. Ford, and P. G. Javis, 1981: Plants and Their Atmospheric Environment. Blackwell Scientific, 419 pp.

  • He, M., 1989: Vegetation map of China. Atlas of China Natural Conservation. Science Press, 50–51.

  • Jackson, R. D., D. B. Idso, R. J. Reginato, and P. J. Pinter Jr., 1981: Canopy temperature as a crop water stress indicator. Water Resour. Res.,17, 1133–1138.

  • Jiang, R., 1989: Runoff map of China. Atlas of China Natural Conservation. Science Press, 88.

  • Kustas, W. P., 1990: Estimated values of evapotranspiration with a one- and two-layer model of heat transfer over partial canopy cover. J. Appl. Meteor.,29, 704–715.

  • Li, J., J. Zhang, and C. Huang, 1989: Soil and land use classification map of China. Atlas of China Natural Conservation. Science Press, 36–37.

  • Liu, C., J. Hong, and H. Jin, 1991: Calculation of field evapotranspiration. Field Evaporation—Measurement and Estimation, X. Xie, D. Zuo, and D. Tang, Eds., Meteorology Press, 134–142.

  • Lu, Y., and G. Gao, 1987: Physical Climatology. Meteorology Press, 357–401.

  • Lu, Z., 1992: Simulation and field study of water transport in soil-plant-atmosphere continuum. Part II: The resistance in SPAC system. Research on the Relationship between Crops and Water Moisture, China Science and Technology Press, 304–322.

  • Mahrt, L., and H. Pan, 1984: A two-layer model of soil hydrology. Bound.-Layer Meteor.,29, 1–20.

  • Monin, A. S., and A. M. Obukhov, 1954: Basic laws of turbulent mixing in the ground layer of the atmosphere. Akad. Nauk SSSR Geofiz. Inst. Tr.,151, 163–187.

  • Monteith, J. L., 1981: Evaporation and surface temperature. Quart. J. Roy. Meteor. Soc.,107, 1–26.

  • Moran, M. S., T. R. Clarke, Y. Inoue, and A. Vidal, 1994: Estimating crop water deficit using the relation between surface-air temperature and spectral vegetation index. Remote Sens. Environ.,49, 246–263.

  • Nobre, C. A., P. J. Sellers, and J. Shukla, 1991: Amazonian deforestation and regional climate change. J. Climate,4, 957–988.

  • Noilhan, J., and S. Planton, 1989: A simple parameterization of land-surface processes for meteorological models. Mon. Wea. Rev.,117, 536–549.

  • ——, and P. Lacarrere, 1995: GCM grid-scale evaporation from mesoscale modeling. J. Climate,8, 206–223.

  • Penman, H. L., 1948: Natural evaporation from open water, bare soil, and grass. Proc. Roy. Soc. London,193A, 120–145.

  • Sellers, P. J., Y. Mintz, Y. C. Sud, and A. Dalcher, 1986: A simple biosphere model (SiB) for use within general circulation models. J. Atmos. Sci.,43, 505–531.

  • ——, and Coauthors, 1996: A revised land-surface parameterization (SiB2) for atmospheric GCMs. Part I: Model formulation. J. Climate,9, 676–705.

  • Serafini, J. V., 1987: Estimation of evapotranspiration using surface and satellite data. Int. J. Remote Sens.,8, 1547–1562.

  • Shuttleworth, W. J., and J. S. Wallace, 1985: Evaporation from sparse crops—An energy combination theory. Quart. J. Roy. Meteor. Soc.,111, 839–855.

  • Taconet, O., R. Bernard, and D. Vidal-Madjar, 1986: Evapotranspiration over an agriculture region using a surface flux/temperature model based on NOAA-AVHRR data. J. Climate Appl. Meteor.,25, 284–307.

  • Thom, A. S., 1972: Momentum, mass and heat exchange of vegetation. Quart. J. Roy. Meteor. Soc.,98, 124–134.

  • ——, 1975: Momentum, mass and heat exchange in plant communities. Vegetation and the Atmosphere, J. L. Monteith, Ed., Academic Press, 57–109.

  • Thornthwaite, C. W., and H. G. Wilm, 1944: Report of the committee on transpiration and evaporation, 1943–44. Trans. Amer. Geophys. Union,25, 683–693.

  • Veihmeyer, F. J., 1964: Evapotranspiration. Handbook of Applied Hydrology, V. T. Chow, Ed., McGraw-Hill, 11–25.

  • Wetzel, P. J., and J. Chang, 1987: Concerning the relationship between evapotranspiration and soil moisture. J. Climate Appl. Meteor.,26, 18–27.

  • Wollenweber, G. C., 1995: Influence of fine scale vegetation distribution on surface energy partition. Agric. For. Meteor.,77, 225–240.

  • Xue, Y., 1996: The impact of desertification in the Mongolian and the inner Mongolian grassland on the regional climate. J. Climate,9, 2173–2189.

  • ——, and J. Shukla, 1993: The influence of land-surface properties on Sahel climate. Part I: Desertification. J. Climate,6, 2232–2245.

  • Zhang, R., and Q. Wang, 1989: Comprehensive climatic regionalization of the Loess Plateau. Region, Disaster—Geographical Research, Science Press, 79–101.

  • Zhu, Z., and T. Wang, 1993: Trends in desertification and its rehabilitation in China. Desertific. Control Bull.,22, 27–30.

  • View in gallery

    A theoretical scheme showing different land use types vs WDI and vegetation fraction (veg), and the direction of desertification and deforestation vs these two indices.

  • View in gallery

    Location and climatic classification of the Loess Plateau (Zhang and Wang 1989) and the distribution of 107 meteorological stations used in the present study.

  • View in gallery

    Dominant soil textures (a) and vegetation types (b) derived from the Soil and Vegetation Map of China.

  • View in gallery

    Schematic procedures used to generate surface parameters, evapotranspiration, and WDI.

  • View in gallery

    Distribution of annual-mean E0, Ea, and WDI estimated by the regional water deficit model.

  • View in gallery

    (a) Seasonal changes in E0, Ea, and WDI in the whole study area (mean value of 107 grid cells) and (b) changes in WDI in four climatic subzones (ART, averages of 28 stations; SAT, 28 stations; SAW, 35 stations; and SHW, 16 stations).

  • View in gallery

    Correspondence between frequencies of the grid cells of dominant vegetation types and the estimated WDI.

  • View in gallery

    Sensitivity of (a) Ea and (b) WDI to the weight coefficient ξ in Jan, Apr, Jul, and Oct.

  • View in gallery

    Fraction of desert (%) in total area both at control and desertification cases.

  • View in gallery

    Changes in Ea and WDI after the fraction of desert doubled.

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 183 183 3
PDF Downloads 50 50 0

A Land Surface Water Deficit Model for an Arid and Semiarid Region: Impact of Desertification on the Water Deficit Status in the Loess Plateau, China

View More View Less
  • 1 Laboratory of Geoecology, Graduate School of Environmental Earth Science, Hokkaido University, Sapporo, Japan
© Get Permissions
Full access

Abstract

A land surface water deficit model was developed for a large-scale heterogeneous arid and semiarid area with various soil, vegetation, and land use types, and used to simulate seasonal and spatial variability in potential (E0) and actual (Ea) evapotranspiration and an index of water deficit (WDI). Comparisons with the results of other commonly used models and natural vegetation conditions suggest that this model can give an estimate of the success for large-scale regional studies. By using the model, the authors estimated E0, Ea, and WDI in a grid cell of 0.25° lat × 0.25° long over the Loess Plateau, China. Finally, the sensitivities of the model to both a vegetation parameter and an assumed desertification case were simulated, and several highly sensitive areas were found to be the risk regions to desertification.

* Current affiliation: Eco Frontier Fellow, National Institute for Environmental Studies, Japan.

Corresponding author address: Dr. Qinxue Wang, Institute of Geography, Hokkaido University of Education, Asahikawa, Hokkaido 070-8621, Japan.

Email: qinxue@yahoo.com

Abstract

A land surface water deficit model was developed for a large-scale heterogeneous arid and semiarid area with various soil, vegetation, and land use types, and used to simulate seasonal and spatial variability in potential (E0) and actual (Ea) evapotranspiration and an index of water deficit (WDI). Comparisons with the results of other commonly used models and natural vegetation conditions suggest that this model can give an estimate of the success for large-scale regional studies. By using the model, the authors estimated E0, Ea, and WDI in a grid cell of 0.25° lat × 0.25° long over the Loess Plateau, China. Finally, the sensitivities of the model to both a vegetation parameter and an assumed desertification case were simulated, and several highly sensitive areas were found to be the risk regions to desertification.

* Current affiliation: Eco Frontier Fellow, National Institute for Environmental Studies, Japan.

Corresponding author address: Dr. Qinxue Wang, Institute of Geography, Hokkaido University of Education, Asahikawa, Hokkaido 070-8621, Japan.

Email: qinxue@yahoo.com

1. Introduction

It is widely recognized that land use–cover changes (LUCC), such as desertification in arid and semiarid regions and deforestation in tropical zones, may exert an influence on regional or even global environmental change by changing the hydrological cycle and surface energy balance. The desire to gain a better understanding of the impact of LUCC on the global environment has stimulated many studies of land surface–atmosphere interactions using coupled soil–vegetation–atmosphere transfer (SVAT) models. Over the past decades, there has been significant progress in the development of SVAT parameterizations, which differ in their description of surface processes, amount of input data required, and time and space scales (Braud et al. 1995). They range from simple big-leaf models to multilayer models with higher-order closure formulations, and their spatial scale ranges from local small-scale homogeneity to regional macroscale heterogeneity, and even to a global scale. The simple models usually describe soils covered with one or two vegetation layers, and solve the energy balance equation for each layer (Deardorff 1977; Taconet et al. 1986; Serafni 1987; Noilhan and Planton 1989). Other models, such as BATs of Dickinson (1984) and SiB1 of Sellers et al. (1986), include detailed radiation transfer schemes to estimate the incoming and outgoing short- and longwave radiation components. In the revised SiB2 model of Sellers et al. (1996), satellite data are used to specify the canopy photosynthetically active radiation, leaf area index (LAI), and canopy greenness fraction. All of these models were derived for a realistic description of the interacting processes at the soil–vegetation–atmosphere interface. Use of these models to investigate the impact of Amazonian deforestation and Sahel desertification (e.g., Dickinson and Henderson-Sellers 1988; Nobre et al. 1991; Xue and Shukla 1993; Xue 1996) has shown that land surface changes play an important role in regional climatic anomalies.

Land use–cover changes have seriously occurred in arid and semiarid regions of northern China mainly due to intensive and continuous human-induced disturbances such as excessive reclamation, overgrazing, and denudation. According to aerial photographs, TM imagery analysis and field investigation (Zhu and Wang 1993), sandy desertified land in the arid and semiarid regions of northern China has increased by 25 200 km2 from 1975 to 1987, an annual average increase of 2100 km2. The desertified area in some regions has almost doubled in size over the last few decades. The main areas of desertified land on the Loess Plateau are extending to 47 counties on the Mu Us sandy land and the area along the Great Wall. According to GISLP’s (1991) investigation, the total area of desertified land is 118 000 km2, of which 35 000 km2 is severely desertified, 29 500 km2 is moderately desertified, and 52 800 km2 is slightly desertified.

Impacts of desertification are most clearly manifested by the land surface water status. The aim of this study was to establish a water deficit index (WDI) by using a regional water deficit model and to evaluate the impact of desertification on the WDI. The regional model is a simplified but interdisciplinary one, which combined meteorological measurements with soil, vegetation, and land use data derived from remote-sensing measurements.

WDI, recently used by Moran et al. (1994) to define a soil-canopy water deficit status, was derived from the original concept of Crop Water Stress index (CWSI) introduced by Jackson et al. (1981). CWSI is directly determined by the minimal, maximal, and mean values of the foliage and air temperature difference. However, it is only applicable to conditions of full vegetation coverage. Moran et al. (1994) developed a graphic method that allowed the index to be estimated for a partial canopy incorporating, in addition, fractional vegetation coverage. Based on the simulation results, however, Moran et al. (1994) pointed out that further refinement of WDI should take into account coupled flux exchanges between the soil–vegetation–atmosphere continuum (SVAC). The present study presents a new approach that combines major land surface properties for estimating the index in a large heterogeneous arid and semiarid area with various soil, vegetation, and land use types. The concept of the model is described in section 2. Parameterization of the model is shown in section 3. Study area, parameters, and data processing are given in section 4. The feasibility of the model is discussed in section 5, and the sensitivity of the model’s outputs to desertification is analyzed in section 6.

2. Basic model

There have been many studies of land-surface aridity using various indices such as those of De Martonne (1925), Budyko (1956), Penman (1948), and CNC (1959) (Table 1). However, these indices were developed from empirical or semiempirical relationships between measured evapotranspiration and climatic factors such as net radiation, air temperature, and precipitation, and do not capture physical processes in the SVAC.

In the present study, a physically based index representing the water deficit status of a large-scale heterogeneous area was derived based on the results of recent SVAT studies. The water deficit index of Moran et al. (1994) is defined as
i1520-0442-12-1-244-e5
where E0 and Ea are the potential and actual evapotranspiration. WDI varies from 0 to 1. WDI = 0 means that the land surface is extremely humid and covered by well-watered forest or water-saturated soil, and WDI = 1 means that the surface is in an extremely arid condition or completely covered by desert. Figure 1 illustrates a theoretical scheme showing the relationship between WDI and the vegetation fraction (veg), which can be used to show different land use types versus different WDI and veg. The directions (arrows in Fig. 1) of land use changes such as deforestation and desertification can also be shown by this scheme.

To estimate WDI, we developed a regional land surface parameterization, which is described in the following section.

3. Parameterization

The parameterization includes three main parts: land surface resistances, radiation transfer, and energy balance in the evapotranspiration processes.

a. Estimation of surface resistances

1) Aerodynamic resistances

From the point of view of diffusion, the aerodynamic resistances of momentum (ram), heat (rah), and water vapor (rav) have the following relationships (Grace et al. 1981; Thom 1972):
i1520-0442-12-1-244-e6
where the friction velocity, u* (symbols are listed in appendix), can be calculated from the reference level wind speed using similarity theory (Monin and Obukhov 1954). Thom (1975) gave a simplified function:
i1520-0442-12-1-244-e7
Here κ is the von Kármán constant (=0.41); z0 is the roughness length, which depends on the vegetation types (Table 2); and d is surface displacement height, which can be estimated by (Wollenweber 1995)
dhCd1/4
The aerodynamic resistance of momentum (ram) was determined by using Monteith’s (1981) Eq. (5), which has been proven by Lu (1992), by a comparison of its results with seven other models with consideration to stratification stability, to be appropriate for regional study:
i1520-0442-12-1-244-e9

2) Canopy resistances

Among land surface resistances, canopy resistances—that is, stomatal and leaf boundary resistances—play a special role in surface interactions. A physically based stomatal resistance, which depends on incoming solar radiation (Q), maximum irradiance (Qmax), soil moisture (ws), wilting point (wwilt), and leaf area index (LAI), was selected with c1 = 0.03 (Taconet et al. 1986) as follows:
i1520-0442-12-1-244-e10
where the minimum stomatal resistance rsto depends on vegetation types (Table 2).
The leaf boundary resistance (rlb) was calculated according to Kustas’ (1990) formula for estimating latent heat fluxes over a partial canopy cover:
rlbAlu1/2
where l is a characteristic length scale for an average leaf width (0.05 m) and A is a constant (90 s1/2 m−1).

3) Soil surface resistance

Soil surface resistance was estimated by the empirical formula of SiB2 (Sellers et al. 1996):
rsoilexp(8.206 − 4.255ws)

b. Effective parameters for vegetation and soil

To compute regionally averaged fluxes over a large area, averaging operators were chosen to calculate effective parameters. It is assumed that one grid cell can be classified into several land use classes for which the fractions are δi (i = f, gs, c, d, etc., which represent forest, grass, cultivated land, desert, etc., respectively) of each category is known, together with its surface properties αi, rstoi, LAIi, and z0i (Table 2). Then, the effective parameters of surface properties α, rsto, LAI, and z0 can be estimated as the mean weighted by the fractional coverage of the different vegetation types (Noilhan and Lacarrere 1995):
i1520-0442-12-1-244-e13

Seasonal changes in the parameters were considered calculated using the monthly time series of total and green-leaf area indices, vegetation coverage, albedo, and surface roughness length for the major SiB vegetation types (Dorman and Sellers 1989).

c. Radiation transfer

A semiempirical model is used to describe the net radiation transfer with respect to vegetation and bare soil surface (Shuttleworth and Wallace 1985):
i1520-0442-12-1-244-e17
where Rn, which is total net radiation, can be written as
i1520-0442-12-1-244-e18
Shortwave radiation is divided into visible (Rvi) and near-infrared (Rni) components, which are estimated by (Ferenc 1994)
i1520-0442-12-1-244-e19
where Q and q are potential direct solar radiation and indirect shortwave radiation in clear sky (W m−2), and a and b are the empirical coefficients (Liu et al. 1991) shown in Table 3.

d. Estimation of evapotranspiration

A modified version of the Penman–Monteith equation was derived for vegetation and bare soil surface:
i1520-0442-12-1-244-e20
The total actual evapotranspiration can then be estimated by
EaEavEas
However, if the surface is saturated by water vapor—that is, rst = rlb = rsoil = 0 and rav = rah—then Eqs. (20) and (21) become
i1520-0442-12-1-244-e23
Then, the potential evapotranspiration can be calculated by
E0E0υE0s
One important parameter in Eqs. (20)–(24), δ, which is called the shielding factor (Ben Mehrez et al. 1992), is calculated by
δe(−ξ×LAI)
This factor changes from 0 to 1 depending on the leaf area index (LAI) and ξ, a weight coefficient of LAI that depends on soil, vegetation, and land use types. Here ξ = 0 means vegetation has no leaves or bare soil, and ξ = 1 means full-leafed vegetation ground. In section 5, the dependence of Ea and WDI on this parameter will be discussed. Soil heat flux (G) can be estimated simply as a fraction of Rn dependent only upon the range of annual air temperature.

e. Treatment of soil water content

Many studies (e.g., Mahrt and Pan 1984; Wetzel and Chang 1987) have shown that the relationship between soil moisture and evapotranspiration depends strongly on soil, atmospheric, and vegetation conditions. In the present study, soil moisture is estimated by the so-called force restore method proposed by Deadroff (1977) and modified by Noilhan and Planton (1989):
i1520-0442-12-1-244-e27
where P should be treated as the difference between precipitation and runoff. According to the China Map of Runoff (Jiang 1989), the annual runoff is less than 50 mm in most of study area and is around 100 mm for only a small area in the subhumid zone. Because this model is developed mainly for an arid and semiarid region, we neglected the runoff in this study. The neglect might cause a small increase in Ea and decrease in WDI. The two dimensionless coefficients C1 and C2 were estimated for different soil textures by
i1520-0442-12-1-244-e28
and values and parameters used in Eqs. (27)–(28) versus soil textures were employed from Noilhan and Planton (1989). The initial data of ws (1990–91) was obtained from 24 agricultural meteorological stations in the Loess Plateau of China.

4. Study area, parameters, and data

The Loess Plateau of China was selected as the study area because of its heterogeneity of land surface and its sensitive ecological and environmental conditions. The plateau is located in northern China (34°–41°N, 100°–115°E) with a total area of 632 520 km2, accounting for 6.3% of the entire land area of China (Fig. 2). Most of the plateau lies on the transitional border between the monsoon climatic zone and the continental arid climate zone. The annual precipitation is 200–600 mm, and the mean temperature range is 5°–12.5°C from northwest to southeast. According to climatic classification (Zhang and Wang 1989), the plateau can be divided into four subzones (Fig. 2): arid temperate zone (ART), semiarid temperate zone (SAT), semiarid warm temperate zone (SAW), and subhumid warm temperate zone (SHW). From ART to SHW—that is, from northwest to southeast—global solar radiation decreases gradually, whereas air and soil temperatures, precipitation, and humidity increase. There are no apparent differences in the mean wind speed over the whole plateau. The coverage of forest and shrubs in SAW is greater than that in other zones due to more mountainous regions. The fraction of typical-grass land and cultivated land increases from ART to SHW, whereas that of short-grass land decreases gradually.

Meteorological data from 107 stations, including monthly mean (1951–91) air and ground temperatures, precipitation, sunshine duration, wind speed, and air vapor pressure, were used in this simulation. It should be recognized that the data from meteorological stations cannot logically be applied to the various vegetation types being studied. Using this dataset is just due to difficulty to obtain ground-based observations over such a large heterogeneous area for its very pilot study. Theoretically the application of remote-sensing data might greatly complement this shortage, which leaves us many problems to be solved in the next step. Soil and vegetation parameters were mainly derived from the dominant soil textures and vegetation types in each 0.25° lat × 0.25° long grid cell. The map of soil texture was derived from the Soil Map of China (Li et al. 1989) and investigated data (Table 4) of GISLP (1991). The classification is based on the dominant soil texture and distinguishes between five types (Fig. 3a): sand, sandy loam, light loam, medium loam, and heavy loam. The soil moisture parameters, wsat, wfc, and wwilt, are prescribed from this map with reference to Clapp and Hornberger’s (1978) experiment. Land cover classification (Fig.3b) was based on the Vegetation Map of China (He 1989) and was then used to construct maps of parameters. The values of parameters assigned to each type of dominant vegetation were determined based on those of BATs shown in Table 2. The fraction coverage of each land-cover category δi, which was derived from remote-sensing measurements (GISLP 1991), was used to compute effective parameters with Eqs (13)–(16).

5. Implementation and discussion

a. Implementation of the model

1) Flowchart of simulation

Fundamentally, the model equations are based on micrometeorology, and are used on short timescale (time step = 1 day). However, some meteorological factors such as S and P are monthly total values, which need to be interpolated into diurnal values. In the present study, we simply used monthly total values divided by the number of days in each month and obtained diurnal values. On the other hand, the meteorological measurements of 107 stations were interpolated into each 0.25° lat × 0.25° long grid cell with the distance weight least squares methods, and the land surface parameters in each grid cell were derived from various maps of China mentioned above. Figure 4 shows the flowchart of data processing and simulation.

2) Distribution of estimated values

Monthly E0, Ea, and WDI in each grid cell of 0.25° lat × 0.25° long were estimated based on our dataset and the annual-mean distributions are shown in Fig. 5. As seen in Fig. 5, all of the estimated values have apparent regional differences. Here E0 was large (about 1000 mm) in the southeast (SHW), then decreased to 700–900 mm in the middle (SAW), and finally increased to a maximum of about 1400 mm in the northwest (ART); Ea ranged from a maximum of about 500 mm in the SHW to a minimum of about 100 mm in the ART;and WDI ranged from a minimum of about 0.5 to a maximum of about 0.9. The reason for the “peak–valley–peak” pattern of E0 was due to a higher temperature in the SHW zone and larger solar radiation in the ART zone. However, Ea was dependent to a large extent on soil moisture conditions and the vegetation fraction. Under natural conditions, Ea decreases gradually from southeast to northwest (from SHW to ART) due to the monsoon climate. As a result, WDI in the four regions is clearly in the order of ART > SAT > SAW > SHW throughout the year. Although the area near Yinchuan is in the ART, the estimated values of WDI are relatively smaller than other parts of the ART. This exception is due to intensive irrigation in this area, a major grain growing area in China.

3) Seasonal changes in estimated values

Seasonal variations in E0, Ea, and WDI were very clear in the study area. The averages of 107 grid cells over the whole plateau are shown in Fig. 6a, in which E0 and Ea had the lowest values in January and the highest values in June, whereas WDI had the lowest value in July and highest value in December.

To analyze regional differences in seasonal changes, the averages of E0, Ea, and WDI in four subzones (ART, 28 stations; SAT, 28 stations; SAW, 35 stations; and SHW, 16 stations) were computed. The results showed that the characteristics of seasonal variation were similar in the four subzones but their amplitudes were different according to the zone. One interesting characteristic is that regional differences are small in winter, but large in summer (Fig. 6b).

b. Discussion on the feasibility of the model

The validity of the model was usually demonstrated by comparing the estimated values with ground-based measurements. Since the outputs of the model, such as Ea and WDI, are difficult to observe for various vegetation types in a large-scale area, the feasibility of our model can be only partially validated by 1) comparing the estimated value of Ea with those of other commonly used equations listed in Table 5 and 2) analyzing the correspondence between estimated WDI and natural vegetation condition.

1) Comparison with the results of other evapotranspiration equations

In past decades, the lack of basic data and the difficulties in measurement required in field methods have accounted for the great efforts made to develop evapotranspiration equations that can relate the evapotranspiration with some readily available climatic data. There are many methods of estimating potential and actual evapotranspiration, including physically based theoretical approaches, water, and heat balance analytical approaches and empirical approaches based on the relation between measured evapotranspiration and climatic conditions. However, theoretical approaches are difficult for large time and spatial scales. A number of equations have been suggested for different purposes and scales (Veihmeyer 1964). Some typical ones commonly used are listed in Table 5. Many studies conducted in China on these equations have been reviewed by Lu and Gao (1987), who suggested that the mean errors of these methods are within about ± 15%, in which Budyko’s and Scheiber’s methods are better with the mean errors of −0.2% to 5.5% and −5.2% to 13.3%, respectively. The results of these methods were used to compare with those of the present model, which was shown in Table 6. It can be clearly seen that the average values of the present model are very close to these of Budyko’s and Scheiber’s equations in different climatic zones, which suggests that our model can output reasonable values of Ea, comparable to those estimated by other commonly used models. Here it should be stressed that the present model has its advantage of being able to output regional evapotranspiration values because it has been coupled with land surface properties.

2) Correspondence with natural vegetation condition

According to the definition, WDI indicates the water deficit status of land surface, which is most easily manifested on natural vegetation conditions. Different water deficit conditions are correspondence with different vegetation types, thus the relationship between the estimated WDI and the distribution of vegetation types will convince us whether this model can output reasonable values or not. For this reason, we counted the frequencies of grid cells for dominant natural vegetation types illustrated in Fig. 7. It can be found that different vegetation types have different distribution patterns of WDI; for example, the large fraction of forest has small WDI (<0.70), whereas that of semidesert and desert land have large WDI (>0.75). Shrubs have a large range (0.60–0.86), whereas the major part of grassland is in the range between forest and desert. This result is consistent with the actual situation, thus supporting WDI as an indicator of the surface water deficit on the regional scale.

Finally, note that high variability of land surface and difficulty of measuring Ea in such a large-scale area makes the complete validation very complex. Attempts made above provide only a partial validation of the model for large-scale regional studies.

6. Sensitivity analysis: Impact of desertification

The results of this study confirmed that the land surface water deficit is closely related to both climatic conditions and surface properties. The land surface is undergoing great changes due to land use and can influence the exchange of momentum, energy, and water fluxes with the atmosphere. We concentrate in this study on ξ, the weighted coefficient of LAI, and the fractional area of desert land (δd). The sensitivity to ξ will give us a general concept of the transition from completely bare ground to full-leafed vegetation ground. The sensitivity to δd is of particular interest because the effect of desertification can be taken into account through variation of this parameter. Theoretically, removing vegetation would be expected to be accompanied by an increase in surface temperature, which should increase E0. However, there is another negative feedback—that is, removing vegetation would lead to an increase in albedo, then a decrease in net radiation, and hence a reduction in E0. As a result, the sensitivity of E0 to both ξ and δd was found to be very small. For this reason, sensitivity analysis was carried out only for Ea and WDI. Here note that meteorological factors and soil parameters are used as average values in the sensitivity study.

a. Sensitivity to ξ

As mentioned in section 3d, ξ changes from 0 to 1—that is, from off-leafed vegetation or bare ground to full-leafed vegetation ground. Changes in ξ would at first cause changes in the shielding factor of δ and would then influence each flux. Figure 8 shows the change in averaged Ea and WDI of 107 stations corresponding to the change in ξ. It can be clearly seen that Ea increases while WDI decreases exponentially when ξ varies from 0 to 1. This change was especially notable in summer. Therefore, it can be stated that the larger the vegetation leaf area is, the bigger is the actual evapotranspiration, and the smaller is the water deficit index.

b. Sensitivity to δd

To analyze the sensitivity of the model to desertification, the fraction of desert in each grid cell was assumed to expand gradually from a control case (δd)—that is, current conditions, to a desertification case (δd + Δδd = 2δd), that is, doubling of the desert fraction. Here it is assumed that if δd + Δδd ≥ 100, then δd + Δδd = 100. With an increase in the fraction of desert land, on the contrary, it can be assumed that the grassland fraction is δg − Δδd, which physically means that desertification occurs only at desert edges and is preceded by progressive degradation of grassland. Of course, if there is no desert in a grid cell, then, δd + Δδd = 0. Thus, the case where desertification occurs initially in a grid cell was not taken into consideration. Figure 9 shows fraction of desert in total area both at control and desertification cases.

Differences [dEa (%) and dWDI] between desertification case and control case were simulated under the assumption mentioned above. Figure 10 shows the results of dEa (%) and dWDI in summer over the whole plateau. It can be found that there are highly sensitive areas distributed in the north central part of the plateau near Yulin and the northwestern boundary area of the Loess Plateau. In these regions, vegetation destruction might cause an obvious decrease in evapotranspiration (−5%–−20%) and an increase of WDI (0.01–0.1)—that is, an aggravation of land surface water stress. Therefore, these regions should be carefully treated or protected. Unfortunately, due to the continuous increase in population and exploration of natural resources, these regions are being subject to intensive human activities, which will continue in the future.

7. Conclusions

A regional water deficit model has been developed for a large arid and semiarid region with heterogeneous land surface properties. This model can be used to estimate the regional evapotranspiration (E0, Ea) and water deficit index (WDI) with a grid cell of 0.25° lat × 0.25° long by combining meteorological measurements, soil, vegetation, and land use data derived from remote-sensing observations.

The feasibility of the model has been partially verified by comparisons with the results of other commonly used models and natural vegetation condition, which suggests that the model can give a reasonable estimate for large-scale regional studies.

Sensitivity analysis showed that changes in Ea and WDI caused by desertification are larger in arid and semiarid subzones than in subhumid zones, and also larger in summer than in other seasons. Several highly sensitive geomorphic units, such as the area near Yulin and the northwestern boundary area of the Loess Plateau, were investigated. These regions can be regarded as risk regions that are easily affected by vegetation destruction.

Finally, we would like to point out that this is a pilot study on SVAT in the Loess Plateau of China, an area for which there are insufficient measurements for SVAT parameterizations. Needless to say, the comparisons made above only partially show the validity of the model, but we can say that on a regional scale, this model is an interesting first attempt to bridge the gap between LUCC and climate changes in regions where there has been no intensive investigation such as that carried out in the Amazonian and Sahel regions. In the future, it is expected that this model can be improved in terms of both resolution and precision along with the progress of remote sensing and ground-based measurements.

Acknowledgments

As a joint researcher of the Group for Integrated Survey of the Loess Plateau, Chinese Academy of Sciences (CISLP, CAS), the first author participated in investigation and data collection for a project, Rational Utilization of Agricultural Climate Resources in the Loess Plateau, during 1986–91, which enabled us to conduct this research. Thanks are also due to the Toka Foundation of Education and Cultural Exchange, Japan, which provided financial aid for this project.

REFERENCES

  • Ács, F., 1994: A coupled soil–vegetation scheme: Description, parameters, validation, and sensitivity studies. J. Appl. Meteor.,33, 268–284.

  • Ben Mehrez, M., O. Taconet, D. Vidal-madjar, and Y. Sucksdorff, 1992:Calibration of energy flux model over bare soil during HAPEX-MOBILHY experiment. Agric. For. Meteor.,58, 275–283.

  • Braud, I., A. C. Dantas-Antonino, M. Vauclin, J. L. Thony, and P. Ruelle, 1995: A simple soil-plant atmosphere transfer model (SiSPAT) development and field verification. J. Hydrol.,166, 213–250.

  • Budyko, M. I., 1956: Heat Balance of the Earth’s Surface. Gidrometeozidat, Leningrad, 255 pp.

  • Clapp, R. B., and G. M. Hornberger, 1978: Empirical equations for some soil hydraulic properties. Water Resour. Res.,14, 601–604.

  • CNC (Committee for Natural Classification, Chinese Academy of Sciences), 1959: Climate Classification of China. Science Press, 456 pp.

  • Deardorff, J. W., 1977: A parameterization of ground surface moisture content for use in atmospheric prediction model. J. Appl. Meteor.,16, 1182–1185.

  • De Martonne, E., 1925: Traite de Geographie Physique. Paris: Librairie Armand Colin.

  • Dickinson, R. E., 1984: Modeling evapotranspiration for three-dimensional global climate models. Climate Processes and Climate Sensitivity, Geophys. Monogr., No. 29, Amer. Geophys. Union, 58–72.

  • ——, and A. Henderson-Sellers, 1988: Modeling tropical deforestation:A study of GCM land-surface parameterizations. Quart. J. Roy. Meteor. Soc.,114, 439–462.

  • Dorman, J. L., and P. J. Sellers, 1989: A global climatology of albedo, roughness length and stomatal resistance for atmospheric general circulation models as represented by the simple biosphere model (SiB). J. Appl. Meteor.,28, 833–855.

  • GISLP (Group for Integrated Survey of the Loess Plateau, Chinese Academy of Science), 1991: Land Resource of the Loess Plateau. China Science and Technology Press, 355 pp.

  • Grace, J., E. D. Ford, and P. G. Javis, 1981: Plants and Their Atmospheric Environment. Blackwell Scientific, 419 pp.

  • He, M., 1989: Vegetation map of China. Atlas of China Natural Conservation. Science Press, 50–51.

  • Jackson, R. D., D. B. Idso, R. J. Reginato, and P. J. Pinter Jr., 1981: Canopy temperature as a crop water stress indicator. Water Resour. Res.,17, 1133–1138.

  • Jiang, R., 1989: Runoff map of China. Atlas of China Natural Conservation. Science Press, 88.

  • Kustas, W. P., 1990: Estimated values of evapotranspiration with a one- and two-layer model of heat transfer over partial canopy cover. J. Appl. Meteor.,29, 704–715.

  • Li, J., J. Zhang, and C. Huang, 1989: Soil and land use classification map of China. Atlas of China Natural Conservation. Science Press, 36–37.

  • Liu, C., J. Hong, and H. Jin, 1991: Calculation of field evapotranspiration. Field Evaporation—Measurement and Estimation, X. Xie, D. Zuo, and D. Tang, Eds., Meteorology Press, 134–142.

  • Lu, Y., and G. Gao, 1987: Physical Climatology. Meteorology Press, 357–401.

  • Lu, Z., 1992: Simulation and field study of water transport in soil-plant-atmosphere continuum. Part II: The resistance in SPAC system. Research on the Relationship between Crops and Water Moisture, China Science and Technology Press, 304–322.

  • Mahrt, L., and H. Pan, 1984: A two-layer model of soil hydrology. Bound.-Layer Meteor.,29, 1–20.

  • Monin, A. S., and A. M. Obukhov, 1954: Basic laws of turbulent mixing in the ground layer of the atmosphere. Akad. Nauk SSSR Geofiz. Inst. Tr.,151, 163–187.

  • Monteith, J. L., 1981: Evaporation and surface temperature. Quart. J. Roy. Meteor. Soc.,107, 1–26.

  • Moran, M. S., T. R. Clarke, Y. Inoue, and A. Vidal, 1994: Estimating crop water deficit using the relation between surface-air temperature and spectral vegetation index. Remote Sens. Environ.,49, 246–263.

  • Nobre, C. A., P. J. Sellers, and J. Shukla, 1991: Amazonian deforestation and regional climate change. J. Climate,4, 957–988.

  • Noilhan, J., and S. Planton, 1989: A simple parameterization of land-surface processes for meteorological models. Mon. Wea. Rev.,117, 536–549.

  • ——, and P. Lacarrere, 1995: GCM grid-scale evaporation from mesoscale modeling. J. Climate,8, 206–223.

  • Penman, H. L., 1948: Natural evaporation from open water, bare soil, and grass. Proc. Roy. Soc. London,193A, 120–145.

  • Sellers, P. J., Y. Mintz, Y. C. Sud, and A. Dalcher, 1986: A simple biosphere model (SiB) for use within general circulation models. J. Atmos. Sci.,43, 505–531.

  • ——, and Coauthors, 1996: A revised land-surface parameterization (SiB2) for atmospheric GCMs. Part I: Model formulation. J. Climate,9, 676–705.

  • Serafini, J. V., 1987: Estimation of evapotranspiration using surface and satellite data. Int. J. Remote Sens.,8, 1547–1562.

  • Shuttleworth, W. J., and J. S. Wallace, 1985: Evaporation from sparse crops—An energy combination theory. Quart. J. Roy. Meteor. Soc.,111, 839–855.

  • Taconet, O., R. Bernard, and D. Vidal-Madjar, 1986: Evapotranspiration over an agriculture region using a surface flux/temperature model based on NOAA-AVHRR data. J. Climate Appl. Meteor.,25, 284–307.

  • Thom, A. S., 1972: Momentum, mass and heat exchange of vegetation. Quart. J. Roy. Meteor. Soc.,98, 124–134.

  • ——, 1975: Momentum, mass and heat exchange in plant communities. Vegetation and the Atmosphere, J. L. Monteith, Ed., Academic Press, 57–109.

  • Thornthwaite, C. W., and H. G. Wilm, 1944: Report of the committee on transpiration and evaporation, 1943–44. Trans. Amer. Geophys. Union,25, 683–693.

  • Veihmeyer, F. J., 1964: Evapotranspiration. Handbook of Applied Hydrology, V. T. Chow, Ed., McGraw-Hill, 11–25.

  • Wetzel, P. J., and J. Chang, 1987: Concerning the relationship between evapotranspiration and soil moisture. J. Climate Appl. Meteor.,26, 18–27.

  • Wollenweber, G. C., 1995: Influence of fine scale vegetation distribution on surface energy partition. Agric. For. Meteor.,77, 225–240.

  • Xue, Y., 1996: The impact of desertification in the Mongolian and the inner Mongolian grassland on the regional climate. J. Climate,9, 2173–2189.

  • ——, and J. Shukla, 1993: The influence of land-surface properties on Sahel climate. Part I: Desertification. J. Climate,6, 2232–2245.

  • Zhang, R., and Q. Wang, 1989: Comprehensive climatic regionalization of the Loess Plateau. Region, Disaster—Geographical Research, Science Press, 79–101.

  • Zhu, Z., and T. Wang, 1993: Trends in desertification and its rehabilitation in China. Desertific. Control Bull.,22, 27–30.

APPENDIX

List of Symbols

  1. A   Constant for leaf boundary resistance (=90 s1/2 m−1)
  2. a1, a2   Empirical coefficient for global radiation
  3. b   Slope of the retention curve for soil types (Clapp and Hornberger 1978)
  4. c   Empirical coefficient for Thornthwaite equation (Thornthwaite 1944)
  5. c1   Coefficient for stomatal resistance (=0.03)
  6. C1, C2,C1sat, C2ref   Parameters in soil moisture equations (Noilhan and Planton 1989)
  7. Cd   Mean drag coefficient for individual leaves (= 0, 2)
  8. Cp   Specific heat at constant pressure (J kg−1 K−1)
  9. d   Surface displacement height (m)
  10. d1   Depth of top soil layer (10 cm)
  11. d2   Depth of subsoil layer (50 cm)
  12. E0, E0υ,E0s   Potential evapotranspiration from land surface, vegetation, and bare soil (mm)
  13. Ea, Eav, Eas   Actual evapotranspiration from land surface, vegetation, and bare soil (mm)
  14. ea   Air vapor pressure at level za (hPa)
  15. es(Ta)   Saturated vapor pressure at temperature Ta (Pa)
  16. G   Surface soil heat flux (W m−2)
  17. h   Mean height of vegetation (m)
  18. I   Thornthwaite’s temperature-efficiency index
  19. IB   Budyko’s radiative aridity index
  20. IM   De Martonne’s aridity index
  21. IP   Penman’s aridity index
  22. IZ   CNC’s aridity index
  23. L   Latent heat of vaporization (J kg−1)
  24. l  A characteristic length scale for an average leaf width (=0.5 m)
  25. LAI   Leaf area index
  26. P   Precipitation reaching the soil surface (mm)
  27. Q   Potential direct solar radiation in clear sky (W m−2)
  28. q   Potential indirect shortwave radiation in clear sky (W m−2)
  29. Qmax   Maximum irradiance (W m−2)
  30. ram, rah, rav   Aerodynamic resistance for momentum, heat, and vapor, respectively (s m−1)
  31. rlb   Leaf boundary resistance (s m−1)
  32. Rn, Rns, Rnv   Total net radiation, net radiation for the bare soil, and vegetation, respectively (W m−2)
  33. rsoil   Soil surface resistance (s m−1)
  34. rst   Stomatal resistance (s m−1)
  35. rsto   Minimum stomatal resistance (s m−1)
  36. RviRni   Visible and near-infrared shortwave radiation, respectively (W m−2)
  37. S   Possible sunshine duration (h)
  38. s   Sunshine duration (h)
  39. T>10   Accumulated temperature of ≥10°C
  40. TsTIa   Soil surface temperature and air temperature, respectively (K)
  41. u   Wind speed at height z
  42. u*   Friction velocity (m s−1)
  43. veg   Vegetation fraction
  44. WDI   Water deficit index
  45. ws   Volumetric water content (cm3 cm−3)
  46. wsat   Saturated volumetric water content (cm3 cm−3)
  47. wwilt   Wilting point (cm3 cm−3)
  48. z   Height of the atmosphere reference level (m)
  49. z0   Roughness length for momentum
  50. τ   Time step (1 day)
  51. κ   Von Kármán’s constant (=0.41)
  52. ξ   Weight coefficient of leaf area index
  53. αvb, αnt   Albedo to visible and near-infrared radiation, respectively
  54. γ   The psychrometer constant (hPa K−1)
  55. Δ   Rate of change in saturated vapor pressure with temperature Ta (hPa K−1)
  56. δ   Shielding factor of vegetation
  57. δd   Fraction area of desert land
  58. δi   Fractional area of each land use category in a grid cell
  59. ε   Emissivity of surface
  60. ρ   Density of dry air (kg m−3)
  61. ρw   Density of soil water (kg m−3)
  62. σ   Stefan–Boltzmann constant (5.67 × 10−8 W m−2 K−4)

Fig. 1.
Fig. 1.

A theoretical scheme showing different land use types vs WDI and vegetation fraction (veg), and the direction of desertification and deforestation vs these two indices.

Citation: Journal of Climate 12, 1; 10.1175/1520-0442(1999)012<0244:ALSWDM>2.0.CO;2

Fig. 2.
Fig. 2.

Location and climatic classification of the Loess Plateau (Zhang and Wang 1989) and the distribution of 107 meteorological stations used in the present study.

Citation: Journal of Climate 12, 1; 10.1175/1520-0442(1999)012<0244:ALSWDM>2.0.CO;2

Fig. 3.
Fig. 3.

Dominant soil textures (a) and vegetation types (b) derived from the Soil and Vegetation Map of China.

Citation: Journal of Climate 12, 1; 10.1175/1520-0442(1999)012<0244:ALSWDM>2.0.CO;2

Fig. 4.
Fig. 4.

Schematic procedures used to generate surface parameters, evapotranspiration, and WDI.

Citation: Journal of Climate 12, 1; 10.1175/1520-0442(1999)012<0244:ALSWDM>2.0.CO;2

Fig. 5.
Fig. 5.

Distribution of annual-mean E0, Ea, and WDI estimated by the regional water deficit model.

Citation: Journal of Climate 12, 1; 10.1175/1520-0442(1999)012<0244:ALSWDM>2.0.CO;2

Fig. 6.
Fig. 6.

(a) Seasonal changes in E0, Ea, and WDI in the whole study area (mean value of 107 grid cells) and (b) changes in WDI in four climatic subzones (ART, averages of 28 stations; SAT, 28 stations; SAW, 35 stations; and SHW, 16 stations).

Citation: Journal of Climate 12, 1; 10.1175/1520-0442(1999)012<0244:ALSWDM>2.0.CO;2

Fig. 7.
Fig. 7.

Correspondence between frequencies of the grid cells of dominant vegetation types and the estimated WDI.

Citation: Journal of Climate 12, 1; 10.1175/1520-0442(1999)012<0244:ALSWDM>2.0.CO;2

Fig. 8.
Fig. 8.

Sensitivity of (a) Ea and (b) WDI to the weight coefficient ξ in Jan, Apr, Jul, and Oct.

Citation: Journal of Climate 12, 1; 10.1175/1520-0442(1999)012<0244:ALSWDM>2.0.CO;2

Fig. 9.
Fig. 9.

Fraction of desert (%) in total area both at control and desertification cases.

Citation: Journal of Climate 12, 1; 10.1175/1520-0442(1999)012<0244:ALSWDM>2.0.CO;2

Fig. 10.
Fig. 10.

Changes in Ea and WDI after the fraction of desert doubled.

Citation: Journal of Climate 12, 1; 10.1175/1520-0442(1999)012<0244:ALSWDM>2.0.CO;2

Table 1.

Several commonly used indices of aridity.

Table 1.
Table 2.

Land cover types and their characteristic parameters. Values of these parameters were determined according to BATs (Dickinson and Henderson-Sellers 1988).

Table 2.
Table 3.

Monthly coefficients a and b for calculation of global radiation over northern China.

Table 3.
Table 4.

Main soil types and their depth, textures, and clay content in the Loess Plateau of China (GISLP 1991).

Table 4.
Table 5.

Some commonly used evapotranspiration equations.

Table 5.
Table 6.

A comparison of the estimated Ea in different climatic zones by various models (ART, averages of 28 stations; SAT, 28 stations; SAW, 35 stations; SHW, 16 stations, and total area, 107 stations).

Table 6.
Save