SCIENCE CHINA Earth Sciences • RESEARCH PAPER • September 2010 Vol.53 No.9: 1358–1364 doi: 10.1007/s11430-010-4028-6 Water consumption in artificial desert oasis based on net primary productivity ZHAO WenZhi1,2*, NIU ZuiRong1, CHANG XueLi3 & LI ShouBo1,2 1 Laboratory of Ecohydrology and Integrated River Basin Science, Cold and Arid Regions Environmental and Engineering Research Institute, Chinese Academy of Sciences, Lanzhou 730000, China; 2 Linze Inland River Basin Research Station, Chinese Ecosystem Research Network, Lanzhou 730000, China; 3 Geography and Resource Management College, Ludong University, Yantai 264025, China Received March 15, 2009; August 4, 2009; published online July 16, 2010 Analysis of the water consumption is the basis for water allocation in oasis. However, the method of estimating oasis water consumption remains a great challenge. Based on net primary productivity (NPP) and the transpiration coefficient, a vegetation water consumption model was developed to estimate the water consumption in desert oasis in ERDAS environment. Our results demonstrated that the ecosystem in the middle reaches of the Heihe oasis consumed water of 18.41×108–21.9×108 m3 for irrigation. Without taking precipitation into account, the water consumption in farmland accounted for 77.1%–77.8% (or about 13.97×108–16.84×108 m3) of the oasis vegetation water consumption and in the farmland protection system accounting for 22%. The growing period precipitation in desert environments is about 7.02×108 m3, and the total annual precipitation is about 8.29×108 m3. The modeled water consumption of desert vegetation, however, is about 4.57×108 m3, equivalent to only 65% of the growing period precipitation or 55% of the total annual precipitation. The modeled value equals to the cumulative precipitation of greater than 5 mm, which is defined as the effective precipitation in arid desert. net primary productivity (NPP), transpiration coefficient, water consumption, oasis Citation: Zhao W Z, Niu Z R, Chang X L, et al. Water consumption in artificial desert oasis based on net primary productivity. Sci China Earth Sci, 2010, 53: 1358–1364, doi: 10.1007/s11430-010-4028-6 Water use in Oasis includes not only the water used for agricultural production and environmental sustainment, but also the water provided for industrial and living. In the arid oasis of China, primary producers account for more than 80% of water consumption. Therefore, analysis of vegetation water consumption is the basis for allocation of water resources in arid oasis [1]. Evapotranspiration is the main process of vegetation water consumption. Many studies on evapotranspiration have been carried out. For example, Brunel et al. [2] used eddy correlation technique and energy balance method to examine the evapotranspiration process *Corresponding author (email: [email protected]) © Science China Press and Springer-Verlag Berlin Heidelberg 2010 in Syrian oasis. Han et al. [3] investigated the annual evaporation law in a typical oasis in the Tarim Basin based on a coupled water-energy balance model with extended Budyko hypothesis. Hu et al. [4] studied the evapotranspiration process in the atmospheric surface layer in the Heihe River Basin with the coupled soil-vegetation-atmosphere water evaporation model. Chen et al. [5] modeled the evapotranspiration process using a distributed hydrological model by using DEM, remote sensing, and hydro-meteorological information. Ji et al. used Penman-Monteith formula and improved Shuttleworth-Wallace model to simulate evaporation and transpiration in desert oasis [6]. Guo [7] used the Priestley-Taylor formula and NOVA AVHRR image to calculate and verify the evapotranspiration in the earth.scichina.com ZHAO WenZhi, et al. Sci China Earth Sci Heihe River Basin. In addition to the above studies and simulations of atmospheric physics mechanisms, there are also methods of water consumption estimation by using theprimary productivity of vegetation and the transpiration coefficient. Transpiration coefficient is defined as the amount of water in weight or volume unit that is required for the production of a unit of plant dry weight. In general, about 300–600 g of water is required for plants to produce one gram of dry matter. Postel et al. [8] used transpiration coefficient to estimate the evapotranspiration of global terrestrial ecosystems in 1996. Zhao et al. [9] studied the vegetation ecological water requirements in the Ejina Oasis by using a similar method. Compared with the natural oasis, artificial desert oasis has more vegetation types, and as a result, estimation of the water consumption of the artificial desert oasis is more complicated. There are many methods for evapotranspiration estimation, including experimental and theoretical equation [10], statistical model based on experimental data [11], GIS-based model with experimental data and theoretical analysis based on the spectral characteristics, and GIS and experimental data [12, 13]. All of these studies contained two common elements: one is the spatial extent of vegetation types; the other is the transpiration coefficient of vegetation types. The most accurate and advanced way to get the spatial extent of vegetation types is land-use and land-cover interpretation using remote sensing. Various methods have been developed to get the transpiration coefficient, such as the methods using vegetation types [12, 14], soil water balance [11], and phreatic water evaporation [9, 15], and every method has its own merits and pitfalls. Common methods that have been used to analyze and evaluate water consumption in arid plain oasis mainly include the instream-and-outstream methods, the water balance methods, the dissipative hydrological models, and the remote sensing models [1]. Uncertainties, however, exist in the estimation of water consumption by using instreamand-outstream methods and water balance methods because of the circulating utilization of water in desert oasis region. The combined method of NPP and transpiration coefficient can effectively remedy the defect of uncertainty, because it has fixed data sources, the calculation procedure is repeatable, and the results can be used to validate and verify the other physical mechanism models. However, there is still few reports on the calculation of water consumption by using a combined method of NPP and transpiration coefficient in artificial desert oasis. In this study, TM images were used as the sources of information. The NPP was calculated base on the relationship between NDVI and NPP, and then we estimated vegetation water consumption using transpiration coefficient of every vegetation type. The results were interpreted with the same spatial resolution as the TM image (30 m). The heterogeneities of the NPP spatial distribution of the same vegetation types were fully stressed to improve accuracy of the estima- September (2010) Vol.53 No.9 1359 tion. 1 Description of study area The study area is located in the middle reaches of Heihe River with the area of 10.753×103 km2, including Linze, Gaotai, and Ganzhou (Figure 1). Climate in this region is temperate arid with cold and dry winters and warm and wet summers. The average annual precipitation over the study area is 117 mm with high inter-annual variation; about 65% of it falls between July and September. The average annual evaporation is 2390 mm. The average annual temperature is 7.6°C with the highest temperature in summer is 39.1°C and the lowest temperature in winter is −27.3°C. In 2006, the annual precipitation is 98 mm in Linze, 95 mm in Gaotai, and 121 mm in Ganzhou. 2 2.1 Study methods Remote sensing interpretation According to the TM images (Orbital Nos. 133-33 and 133-34, Figure 2(a)) taken on 11 September, 2006, the study area was divided into 9 land-cover types, namely, arbor forest, sparse forest and shrubs, farmland, wetland, desert grassland, water body, gobi desert, bare land, and residential area. Geometric correction of ETM images was performed in the ERDAS IMAGINE environment (using the Albers conical Equal Area Krasovsky projection method). Land use map (Figure 2(c)) was completed by computer visual interpretation in the accuracy level of 1:100000. Figure 1 Location of the study area. 1360 ZHAO WenZhi, et al. Sci China Earth Sci September (2010) Vol.53 No.9 Figure 2 TM image, NDVI and land-cover map of the study area. (a) TM image; (b) NDVI; (c) land-cover map. 2.2 Calculation of NDVI Since the visible light (RED) channel and near-infrared light (NIR) channel of TM scanning radiometer are very sensitive to plant leaves and their growing period, the normalized difference vegetation index (NDVI) derived from RED and NIR is the best choice in reflecting and defining the relative abundances and activities of vegetation [16]. Because the relationship between NDVI and NPP, as well as the relationship between NDVI and LAI, was widely used to estimate regional vegetation productivities [16, 17], it was used in this study to calculate the regional vegetation productivity of the study area based on the relationship between NDVI and NPP. The calculation formula is NDVI=(λNIR− λRED)/(λNIR+λRED). According to this formula, the NDVI values can be calculated with the corrected composite images through ERDAS software (Figure 2(b)). 2.3 Measurement of the vegetation spectrum In the study area, the prominent vegetation types are farmland, protection forest, desert vegetation, and wetland. The farmland is dominated by corn and wheat, with a few other crops. The protection forest is composed of Populus gansuensis, Tamarix spp., and Haloxylon ammodendron, which are distributed around the farmland or between the desert and oasis. The desert vegetation is distributed on the fringe of oasis, while the wetland is distributed along the Heihe River in the central of the oasis. We used the ASD FieldSpec Portable Handheld Spectroradiometer to measure the spectrum of corn, wheat, desert vegetation, and wetland in July 2008. To be convenient in the measurement of vegetation spectrum, the TM images used in this study were taken in September when the biomass is at its highest level for most vegetation types. Significant errors, however, could exist in the spectrum estimation because the harvest season of wheat had been closed in this time period. Fortunately due to the relatively far smaller coverage of wheat field, and the fact that the reaped fields were usually not reseeded in the study area, wheat field can be easily differentiated from the other vegetation types in the NDVI images (as the particularly low amongst the high value regions). Areas in farmland with NDVI less than 0.637 were defined as wheat fields and were assigned the average spectrum. The interpreted result showed that the area of wheat was about 175.38 km2 in 2008, which was only about 8.81% of farmland. The data statistics suggest that the NPP was proportional linearly to the NDVI, and the significance was more than 0.05 (Figure 3, Table 1). Spectrum of vegetation types were measured during the period of July 20, 2008 to July 28, 2008, and almost all the measurements were performed in Zhangye, Linze, and Gaotai, except the spectral measurement of wheat in Minle where wheat was still at the end of filling stage due to the altitude effect (wheat in other regions has been reaped in July). Measuring method: 5 times on every vegetation type plot between 10 am to 4 pm, and took the average spectrum as the result. Aboveground and underground biomasses have been also determined by dry-weight method. Based on the measurement of the vegetation type spectrum, aboveground and underground biomass, regression relationships have been established between the spectrum and NPPs. Table 1 Linear regression equations between NDVI and Bim Correlation F P coefficient Corn 0.855 48.869 <0.01 Bim=5113.701NDVI−2645.286 Wheat Bim=1788.966NDVI+559.681 0.809 43.579 <0.01 0.826 75.248 <0.01 Wetland Bim=965.269NDVI−136.754 Desert Bim=1730.176NDVI−200.126 0.626 6.442 <0.05 Vegetation 2.4 Linear regression equation Determination of transpiration coefficient There are two important parts in the water consumption calculation by using 3S technology. One is spatial extent of vegetation types, and another is definition of the transpiration coefficient and the variance of water consumption in different regions. The first one has been worked out in ZHAO WenZhi, et al. Figure 3 Sci China Earth Sci 1361 September (2010) Vol.53 No.9 Relationship between NDVI and aboveground biomass (Bim). many other results, whereas the variance of water consumption for the same type of vegetation is difficult to determine, especially in arid desert oasis. In this study, water consumption is calculated in every grid with the regression equations in Table 1 and the equation below: desert vegetation 300 (dominated by Tamarix spp.), protection forest (dominated by Populus gansuensis) 513, wetland (dominated by Phragmites communis and Searia viridis, which has a similar transpiration coefficient of 300 as Searia viridis, Sorghum sudanemse and corn). n VEWR = Ki × ∑ Bim × Wue × A, i =1 where VEWR is vegetation water consumption (g), Ki is the ratio of NPP and aboveground biomass, which is about 1.6933:1 for herbs in the study area; Wue is transpiration coefficient; A is the area of every grid, which is 34 m×34 m in this study; and n is the quantity of grids. Bim was calculated using the equations in Table 1. Wue was derived by referring to the earlier research results in arid region. Usually, about 300–600 g of water is required for plants to produce one gram of dry matter. However, different plants species have different water requirement amounts, for example, the transpiration coefficient of Searia viridis is 285, Sorghum sudanemse 304, corn 349, wheat 557, Brassica campestris 714 and Medicago sativa 844 [18]. Huang gave out the transpiration coefficients of 138–344 for some arid desert plants, such as Haloxylon ammodendron, Caragana Korshinskii, Artemisia salsoloides, Caligonum mongolicum, Hedysarum scoparium and Tamarix spp.; the corresponding value for Elaeagnus angustifolia is 383 and Populus gansuensis 513 [19]. Of course, the amount of water requirement is also related to other ecological factors, such as solar radiation intensity, temperature, atmospheric, soil regimes, plant development stage, etc. By referring to the published results and comprehensively considering the influencing factors, the transpiration coefficients in this study were defined as: corn 349, wheat 557, 2.5 Build the water consumption model The water consumption model was built in Model Maker in ERDAS 9.1. This program module has its own algorithmic language and program library, information of water consumption (or NPP) in per unit area can be easily expressed as gray values by using user-defined syntax and conditional statement. The calculating procedure is shown in Figure 4. 2.6 The water consumption at regional scale The water consumption at regional scale was calculated by using the following equations: n VSWV = (1 / ki ) × ∑ (VEWR − ERi ), i =1 RVEWR = VSWV + ERi , where VSWV and RVEWR are irrigation water use and consumption at regional scale, ERi is effective rainfall, n is number of vegetation type, and ki is water use efficiency of irrigation. The value of k depends on the irrigation times in the area. The water use efficiency of farmland irrigation is about 0.42–0.52 in growing periods according to the monitoring data from the Linze Station. The water use efficiency of protection forest irrigation was determined by experts marking methods, and the value is 0.42–0.60 (Table 2), 1362 Figure 4 Table 2 ZHAO WenZhi, et al. Sci China Earth Sci Calculating procedure of the water consumption model. Water use efficiency of irrigation Sparse forest Vegetation Farmland Arbor forest and shrubs type Water efficiency 1.00 0.42–0.52 0.50–0.60 of irrigation Wetland 1.00 suggesting that 40%–58% of irrigation water is lost in the form of soil water evaporation and deep percolation. 3 Results As can be seen in Table 3, farmland takes the greatest part of the water consumption (about 51.4%) in the study area, followed by desert vegetation (28.7%) and arbor forest (12.4%). Across all the vegetation types, desert grassland and Gobi desert depend on natural rainfall to sustain, whereas the others mainly depend on river or irrigation. In 2006, the water consumption we calculated is 4.57×108 m3 in desert vegetation, while the total precipitation is 8.29×108 and 7.02×108 m3 in growing period (Tables 4, 5). These results suggest that the water consumption is less than the precipitation in desert vegetation. The total precipitation of the other vegetation types is 2.50×108 m3 and in growing period is 2.18×108 m3 (Figure 5). According to the Tables 2, 3, 5 and those equations, the water consumption we calculated is about 18.41×108– Figure 5 September (2010) Vol.53 No.9 Precipitation of the study area. 21.92×108 m3 (Table 6), which have taken the loss of irrigation and ineffective rainfall into account. In the water consumption, farmland is taking a part of 77.1%–77.8%, followed by arbor forest (16.1%–16.4%). 4 Discussion and conclusions The relationship between NPP and evapotranspiration is useful to calculate the water consumption of ecosystems. Postel et al. [8] have used this method to estimate the evapotranspiration of global terrestrial ecosystems in 1996. Zhao et al. [9] used the similar method to study the vegetation ecological water requirement in Ejina oasis. The core of this method is to estimate the transpiration coefficient and the distribution of vegetation as veracious as possible. The transpiration coefficient of global terrestrial ecosystems is estimated as 500 by Postel. Based on the plant physiology research in arid regions, the transpiration coefficient is estimated to be 300–500 in our study, depending on the vegetation types. Besides, the NDVI and NPP are calculated at a grid scale in order to improve the precision of the results. Because of the circulating utilization of water in desert oasis region, the water consumption is uncertain using water balance method. Our results can verify the results of the water balance method. Chen et al. [5] have modeled the hydrological processes of the middle reaches in the Heihe River using a distributed hydrology model based on observation data. The result shows that the precipitation is 193 mm, while the evapotranspiration is 294.5 mm. Multiplying the area of their study area, the rainfall water is about 34.5×108 m3, while the evapotranspiration is 52×108 m3. Therefore, the required irrigation water is about 18.13×108 m3. It is slightly less than our estimation of 18.41×108– 21.92×108 m3. This difference may be caused by circulating utilization of water. According to the statistics from the Zhangye Water Authority, the actual irrigation water was ZHAO WenZhi, et al. Sci China Earth Sci 1363 September (2010) Vol.53 No.9 Table 3 Water consumption based on NPP Vegetation type Farmland Arbor forest Sparse forest and shrubs Wetland Desert Total Water consumption (×108 m3) 8.18 1.97 0.74 0.45 4.57 15.91 Percent (%) 51.41 12.38 4.66 2.83 28.72 100.00 Table 4 The area of every land-cover type in 2006 (km2) Land-cover type Farmland Arbor forest Sparse forest and shrubs Desert grassland Wetland Water body Residential area Gobi desert Bare land Total Table 5 Ganzhou 1061.23 28.69 Linze 457.50 4.44 Gaotai 472.01 20.84 Total 1990.74 53.97 39.93 25.49 21.70 87.12 675.74 9.35 8.98 147.25 1597.84 96.40 3665.40 189.19 43.55 15.77 36.99 1893.38 12.95 2679.25 525.61 112.51 52.48 46.11 3124.86 32.63 4408.74 1390.53 165.41 77.23 230.35 6616.08 141.98 10753.39 Precipitation of every vegetation type (×108 m3) Period Ganzhou Growing 1.17 period Farmland Whole year 1.29 Growing 0.03 period Arbor forest Whole year 0.031 Growing 0.04 Sparse forest & period shrubs Whole year 0.05 Growing 0.01 period Wetland Whole year 0.01 Vegetation type Linze Gaotai Total 0.39 0.36 1.91 0.45 0.45 2.19 0.004 0.02 0.05 0.004 0.02 0.06 0.02 0.02 0.08 0.03 0.02 0.10 0.04 0.09 0.13 0.04 0.11 0.16 Table 6 The water consumption which have taken the loss of irrigation and ineffective rainfall into account (×108 m3) Vegetation type Effectively irrigation Arbor forest Maximum 16.84 3.89 Minimum 13.97 3.25 1.91 0.05 Effectively precipitation Water consumption Farmland Maximum 18.75 3.94 Minimum 15.78 3.30 about 22.35×108 m3 in 2006, suggesting that 0.43×108– 3.94×108 m3 of the irrigated water was probably discharged into the river or groundwater [20]. The result from Chen et al. [5] shows that almost all the precipitation in the desert around oasis is consumed by evapotranspiration. However, in our research, the water consumed by desert vegetation is only 55% of the precipitation. Earlier research suggests that most rainfall in this area is less than 5 mm, about 46.2% of the total precipitation [21]. Generally, small rainfall less than 5 mm was considered ineffective because it can easily be intercepted and evaporated, and thus unproductive for vegetation. Base on that, the effective rainfall (53.8% of total precipitation) is almost consumed by desert vegetation. Our results also show that, without taking precipitation into account, the water consumption for farmlands is about 77.1%–77.8% of the total water consumed in desert oasis, which is about 13.97×108–16.84×108 m3; the other 22% water is used by the farmland protection system. In conclusion, the results of water consumption are in line with the report from Zhangye Water Authority, and are slightly larger than the result of water balance method. The Sparse forest and shrubs Wetland 0.74 0.45 0.08 0.13 0.91 0.48 Total 21.92 18.41 2.17 24.08 20.47 vegetation water consumption model based on NPP, which is estimated using the relationship established by vegetation spectrum with NDVI, is feasible in artificial desert oasis. This work was supported by National Basic Research Program of China (Grant No. 2009CB421302), Key Project of Knowledge Innovation Program of Chinese Academy of Sciences (Grant No. KZCX2-XB2-04-01), and National Natural Science Foundation of China (Grant No. 40930634). 1 2 3 4 5 6 Lei Z D, Yang H B, Ni G H, et al. Analysis on water consumption of oasis in arid area (in Chinese). Water Resour Hydropower Eng, 2006, 37: 15–20 Brunel J P, Ihab J, Droubi A M, et al. Energy budget and actual evapotranspiration of an arid oasis ecosystem: Palmyra (Syria). Agri Wat Manage, 2006, 84: 213–220 Han S J, Hu H P, Tian F Q. Annual evapotranspiration in oases of the Tarim basin based on a coupled water-energy balance (in Chinese). J Tsinghua Univ Sci Technol, 2008, 48: 2070–2082 Hu Y J, Gao Y X, Wang J M, et al. Some Achievements in scientific research during HEIFE (in Chinese). Plateau Meteorol, 1994, 13: 225–236 Chen R S, Kang E S, Yang J P, et al. Simulation of water resources transformation in the midstream area of the Heihe River Basin (in Chinese). J Glaciol Geocryol, 2003, 25: 566–573 Ji X B, Kang E S, Zhao W Z, et al. Analysis on supply and demand of water resources and evaluation of the security of water resources in 1364 7 8 9 10 11 12 13 14 ZHAO WenZhi, et al. Sci China Earth Sci irrigation region of the middle reaches of Heihe River, northwest China (in Chinese). Sci Agr Sin, 2005, 38: 974–982 Guo X Y. Distribution of evapotranspiration in Heihe River basin using remote sensing (in Chinese). Prog Nat Sci, 2005, 15: 1266–1270 Postel S L, Daily G C, Ehrlich P R. Human appropriation of renewable fresh water. Science, 1996, 271: 785–788 Zhao W Z, Chang X L, He Z B. Study on vegetation ecological water requirement in Ejina Oasis. Sci China Ser D-Earth Sci, 2007, 50: 121–129 Wang R H, Song Y D, Fan Z L, et al. Estimation on ecological water demand amount in four sources and one main stream area of Tarim basin (in Chinese). Soil Water Bull, 2001, 15: 19–22 He Z B, Zhao W Z, Fang J. Ecological water requirements of vegetation in the middle reaches of Heihe River (in Chinese). Acta Ecol Sin, 2005, 25: 705–710 Wang F, Wang H, Chen J M, et al. A study of ecological water requirements in Northwest China Part II: Application of remote sensing and GIS (in Chinese). J Nat Resour, 2002, 17: 129–137 Chen X N, Xu Z X. Plausible impact of global climate change on water resources in the Tarim River Basin. Sci China Ser D-Earth Sci, 2005, 48: 65–73 Jia B Q, Ci L J. The primary estimation of water demand by the eco en- 15 16 17 18 19 20 21 September (2010) Vol.53 No.9 vironment in Xinjiang (in Chinese). Acta Ecol Sin, 2000, 20: 243–250 Wang H X, Liu C M. Advances in crop water use efficiency research (in Chinese). Adv Water Sci, 2000, 11: 99–104 Jin L F. Estimation of Grassland Yield in Inner Mongolia using Landsat TM. In: Inner Mongolia Grassland Ecosystem Research Station, Chinese Academy of Sciences, eds. Grassland Environmental Protection (in Chinese). Beijing: Science Press, 1986. 58–627 Zhang N, Yu G R, Zhao S D, et al. Ecosystem productivity process model for landscape based on remote sensing and surface data (in Chinese). Chin J Appl Ecol, 2003, 14: 643–652 Sun R Y, Li B, Zhuge Y, et al. General Ecology (in Chinese). Beijing: Higher Education Press, 1993. 41–48 Huang Z C, Shen W S. Plant-Water Relation and Drought Tolerance in Arid Zone (in Chinese). Beijing: China Environmental Science Press, 2000. 31–39 Wang Q C, Hu G L, Chen H N. Analysis on benefit of recent integrated control projects for Heihe River Basin in Zhangye City (in Chinese). J Des Res, 2008, 28: 498–506 Zhang L J, Zhao W Z. Daily precipitation pattern and its temporal variability in Heihe River basin (in Chinese). J Des Res, 2008, 28: 741–747
© Copyright 2026 Paperzz